what is hypothesis of extended cognition

Extended cognition

2019   The Routledge Encyclopedia of Philosophy Online (edited by Tim Crane), Routledge: London

Last updated 7 April 2020

• 1   Article Summary
• 1   Extended cognition
• 2   Arguments for extended cognition
• 3   Objections to extended cognition
• 4   Extended cognition’s kindred views
• 4   Related Articles
• 4   Articles that might consider linking to this article

Article Summary

Extended cognition takes the idea that your mind is ‘on’ your smartphone literally. It says that human cognitive states and processes sometimes spill outside our heads and into objects in our environment. Alleged examples include not just smartphones, but also use of simpler technology (pencil and paper to perform a calculation), our own body (ticking off our fingers when we count), and other people (our spouse who remembers appointments so we don’t have to).

There are three main arguments for extended cognition.

Functionalist arguments rely on similarities in functional structure between extended processes and (actual or possible) internal cognitive processes. Cognition extends because the physical mechanisms that support it ‘work the same way’ in both cases. Inference to the best explanation arguments canvas the benefits that extended cognition would bring to psychology. We should believe that cognition extends because it would make our psychological theories more unified, elegant, and fruitful. Second-wave arguments emphasise the ways in which our brains integrate with our environment. Cognition extends because brains, bodies, and environment are so tightly intertwined that, when we solve certain cognitive tasks, they count as a single system.

Extended cognition is attacked on many fronts. It has been claimed that it generates absurdly high levels of extension (‘cognitive bloat’); that it is inferior to the more conservative hypothesis of embedded cognition ; that its arguments confuse causal coupling with constitution ; and that its alleged cases fail to satisfy some proposed mark of the cognitive .

Extended cognition concerns only the cognitive, information-processing aspects of mental life. It has, however, inspired similar claims about extension for other aspects of the mind, including conscious experience, emotions, moods, intentional agency, knowledge, and the self.

Extended cognition is part of a wider ‘4E cognition’ research programme. The four Es stand for extended, embedded, embodied, and enactive cognition. Each E offers a closely related, albeit distinct perspective on the role of the environment in cognition. Other forms of externalism about the mind – content externalism , direct realism about perception , collective intentionality , and group cognition – are less closely related to extended cognition.

1 Extended cognition

Humans love their devices. Smartphones, tablets, computers, paper notebooks, calendars, and to-do lists play a pervasive role in our lives. These pieces of technology cause us to think and to behave in certain ways, and we in turn modify them to influence our future thought and behaviour. Often we don’t appreciate how much we rely on them until they’re gone. Empirical work in psychology suggests that this dependence runs very deep. Even during relatively undemanding tasks – e.g. copying a simple coloured pattern made of puzzle pieces – we off-load information processing onto the environment to reduce work for our brains (Ballard et al. 1997) . Once one recognises this, one sees it everywhere: a bartender lines up cocktail glasses of different shapes to remember a complex order; a mathematician uses pencil and paper to guide their steps in a calculation; a child uses their fingers to count off the days until their next birthday. These observations reveal that intelligent, adaptive human thought and behaviour need not always be produced by the brain alone. It often involves a two-way interaction between the brain, body, and world (Dennett 1996; Hutchins 1995; Simon 1969) .

The hypothesis of extended cognition (HEC) goes beyond this relatively uncontroversial observation in a controversial way. Environmental processes don’t merely interact with our brain to produce thought and behaviour. Those environmental processes have as much claim to be mental or cognitive as their brain-based collaborators. Human cognition literally extends into smartphones, tablets, notebooks, to-do lists, and cocktail glasses.

HEC is frequently misunderstood, so it is worth taking care here.

HEC applies the slightly puzzling concept of ‘extension’ to cognitive processes. It is important to realise that HEC is not a claim about cognition considered in some abstract non-physical sense, or about cognition considered from a first-person phenomenological point of view. It is a claim about cognition’s physical, mechanistic basis. That physical basis has a location and an extent. Traditional wisdom has it that the physical basis of human cognition (its ‘realiser’ in functionalist terms) is located exclusively inside the human brain. HEC says that this is false: a mixture of brain and environmental states/processes underlie (realise) cognition.

What counts as the ‘environment’ may not always be precisely defined. For our purposes, the environment is anything outside the brain or central nervous system. Thus, your cognitive processes might extend into external technology (smartphones, notebooks), your non-neural body (fingers, limbs, tendons), naturally occurring objects around you (useful sticks and stones), tokens of public language (appropriate sound waves and ink marks), or the brains of other people (your spouse).

There are several claims that are often mistakenly associated with HEC.

HEC does not say that the environment is as important as the brain for cognition. The brain may remain the principal player behind our thought and behaviour. What HEC says is that the brain is not the only player: it is not the sole locus of human cognition (see Clark 2007a on ‘organism-centered’ cognition) . HEC does not say how much of cognition extends. It is compatible with some, even most, of our cognitive life being wholly internal and brain-based. Only some instances of human cognition need extend for HEC to be true. HEC is also silent about whether extended cognition is a good thing for the agent in question. Sometimes it is (e.g. when a smartphone increases your memory capacity or improves your calculating ability), but sometimes it isn’t (Borsboom et al. 2018; e.g. when a smartphone sustains unhealthy cognitive biases or pathological worries about your self image, see Sprevak 2011) . HEC similarly says nothing about whether extended cognition is the outcome of a voluntary choice by the agent (or anyone else). We may be drawn into entanglements with the environment without our awareness or consent. Finally, HEC does not say anything about whether external objects have an independent mental life of their own. As far as HEC is concerned, a smartphone sitting by itself need have no more mental life than an isolated neuron sitting inside a Petri dish. It is only when a smartphone (or a neuron) stands in the right relationship to the rest of our brain that it becomes part of our cognitive life and endowed with mental/cognitive properties.

2 Arguments for extended cognition

There are three main arguments for the hypothesis of extended cognition (HEC). All of them appear in Clark and Chalmers’ (1998) classic paper and each can be illustrated using their thought experiment:

Inga has normal human memory. One day, she hears of an interesting exhibition at the Museum of Modern Art (MoMA). Inga thinks, recalls that MoMA is on 53rd Street, and sets off. Otto suffers from a mild form of Alzheimer’s disease and he always writes down useful information in his notebook. On the same day, Otto also hears of the exhibition at MoMA, retrieves the address from his notebook, and sets off.

Clark and Chalmers say that Otto has a belief that MoMA is on 53rd Street. This belief – and Otto’s cognitive processes associated with storing and recalling it – extends outside Otto’s brain and into his notebook. Before proceeding to the arguments, it is worth noting that this is not meant to be a realistic description of Alzheimer’s disease or any actual real-life case. Clark and Chalmers are describing an idealised set of interactions with the environment that they say would give rise to extended cognition. They claim that more complex versions of these interactions occur in common real-life cases (e.g. smartphone use) and similar arguments could be given for these more complex cases to those given for Otto and Inga below.

Functionalist arguments. The claim here is that Otto and Inga are identical in a way that matters to functionalism. Functionalism is a philosophical theory that says that the functional role of a physical state/process determines whether that state/process is mental/cognitive. Functionalism about beliefs says that what makes a brain state a belief is that it functions in a way appropriate to a belief. Clark and Chalmers observe that Otto’s notebook functions for him in roughly the same way as Inga’s biological memory functions for her. The state of Otto’s notebook interacts with his desires and other beliefs in a manner similar to that in which the state of Inga’s biological memory interacts with her desires and other beliefs. Exposure to new information causes Otto to modify the state of his notebook. Exposure to new information causes Inga to modify the state of her biological memory. Otto’s notebook causes him to stop at 53rd Street. Inga’s biological memory causes her to stop at 53rd Street. The ‘coarse-grained’ functional role of the stored information appears to be the same in both cases. Clark and Chalmers conclude that just as Inga has a belief that MoMA is on 53rd Street, so Otto has a belief (partly realised in his notebook) with the same content.

This argument relies on accepting functionalism about beliefs, and also on the ‘parity principle’:

If, as we confront some task, a part of the world functions as a process which, were it done in the head , we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process. (Clark & Chalmers 1998 p. 8, emphasis in original)

The parity principle says that we should not be prejudiced against treating Otto’s notebook as a realiser of one of his cognitive states merely because it is outside his head. Recent versions of the functionalist argument for HEC use the parity principle to block an objection to the original functionalist argument: Otto’s notebook and Inga’s biological memory are not exactly alike in their functional roles. This flaw was pressed by early critics of HEC (Adams & Aizawa 2001; Rupert 2004) and cited as a reason to think that Otto’s notebook and Inga’s biological memory don’t have the same claim to mental status. However, although Otto and Inga are not exactly functionally alike, Otto is functionally identical to a hypothetical intelligent being who uses a notebook inside its head to store information (e.g. an extraterrestrial who uses internal ink-marks to encode some of its memories). If it is possible for an intelligent being to have a fully cognitive internal memory system like Otto’s notebook, and we accept the parity principle and functionalism, then we should say that Otto’s actual notebook counts as a case of extended cognition. (For a discussion of this argument, see Sprevak 2009.)

Inference to the best explanation. The idea here is that HEC affords some explanatory advantage to psychology. The argument takes the form of an inference to the best explanation (IBE) : we should believe that HEC is true because of the explanatory benefits its acceptance would bring to psychology. Clark and Chalmers say that HEC would unify otherwise unrelated patterns of human thought and behaviour. Otto’s and Inga’s cases illustrate this point. Inga arrived at 53rd Street because she wanted to go to MoMA and believed that it was there. Assuming HEC is true, Otto arrived at 53rd Street because he wanted to go to MoMA and believed that it was there. HEC allows us to explain both agents as falling under the same belief–desire psychology, irrespective of whether they use internal or external resources for their information storage. Always explaining Otto’s behaviour differently from that of Inga, and in terms of Otto’s internal beliefs about his notebook, seems to introduce needless complexity. The notebook is a constant in Otto’s life, similar to Inga’s biological memory. At least for certain explanatory purposes, it seems redundant to always point to the notebook when explaining Otto’s behaviour. Accepting HEC arguably makes the notion of belief in psychology ‘deeper and more unified, and … more useful in explanation’ (Clark & Chalmers 1998 p. 14) . (For a discussion of this argument, see Sprevak 2010.)

Second-wave arguments. These arguments tend to downplay the considerations behind the functionalist and IBE arguments and instead emphasise ‘complementarity’. What is meant by this is that external states and processes need not be similar – either functionally or in terms of their role in psychological explanation – to any internal ones. What justifies HEC is a tight, reciprocal integration between brains, bodies, and the environment. This approach has affinities with dynamical systems theory, which models the brain, body, and environment as a single system whose elements cannot be prised apart when solving a cognitive task (Thelen & Smith 1994) . In their original paper, Clark and Chalmers call this position ‘active externalism’: all parts (internal and external) are ‘active’ in the sense that they work together in a non-redundant fashion to solve a cognitive task. The resulting interaction need not be similar to any (actual or possible) internal cognitive process, and there is no reason why it should be treated as such in psychological explanation. Otto’s notebook isn’t cognitive because it is functionally like Inga’s memory, or because it is explanatorily beneficial for psychologists to group them together. Otto’s notebook is cognitive because it is systematically, reciprocally, and inextricably integrated with his brain during certain cognitive tasks. This integration is intended, at least in certain cases, to have a normative dimension: it arises from, and it is maintained by, social practices, norms, and institutions. (For a reconstruction of second-wave arguments, see Menary 2010a; Sutton 2010.) (Note that talk of a ‘second wave’ may suggest that the ‘first wave’ was a failure, but it is by no means clear that either the functionalist or IBE argument cannot be made to work, or that all three arguments for HEC cannot be combined into a unified package.)

3 Objections to extended cognition

The hypothesis of extended cognition (HEC) has, unsurprisingly, encountered considerable resistance. Here are four common objections.

Cognitive bloat . Fans of HEC say that human cognition extends in certain cases and to a certain degree, but they usually want to avoid claiming too much cognitive extension. Too much extension might make our existing mental concepts pointless, absurd, or otherwise unfit for purpose. We might end up saying you believe everything on the Internet or that you and I believe the same things just because we happen to live in the same environment. This is called the problem of ‘cognitive bloat’. Avoiding cognitive bloat requires drawing some principled line between those relationships with the environment (functional/explanatory/integrational) that entail cognitive extension and those that do not. Drawing such a line has not proved easy. Consider that not only your smartphone, but also your office computer, local library, friends, co-workers, and in many cases the whole Internet contain information on which you routinely draw to guide your behaviour in a way not dissimilar to that in which Otto relies on his notebook. Consider too that when you and I live in the same environment, we often draw on the same set of environmental resources to guide our respective behaviours. An argument for HEC needs to guard against becoming a reductio ad absurdum of the view. Introducing restrictions to avoid cognitive bloat, however, may cut against the assumptions required to motivate any form of HEC at all – for example, they may violate the parity principle (see Sprevak 2009) .

Embedded cognition . The hypothesis of embedded cognition (HEMC) is an alternative to HEC that claims to offer all of the latter’s explanatory benefits but without the ‘cost’ of saying that notebooks have mental/cognitive properties. HEMC and HEC agree on almost all points: extended processes are functionally similar to internal cognitive processes; grouping extended and internal processes together is sometimes useful in psychological explanation; brains are integrated with the environment when we solve certain cognitive problems. HEMC and HEC differ in that HEC says that extended cognitive processes are 100% mental/cognitive whereas HEMC says that only the internal, brain-based parts of those processes are mental/cognitive. According to HEMC, extended processes are composed of separate mental and non-mental parts – and the mental parts reside exclusively inside the head. According to HEC, extended processes have a uniformly mental/cognitive nature inside and out. Critics of HEC argue that we should abandon HEC for HEMC because the latter provides the explanatory benefits of the former but it is more ontologically parsimonious (it attributes fewer mental/cognitive properties to the world) and it is more conservative (it fits better with traditional internalist psychology) (see Rupert 2004; Sprevak 2010) .

The coupling/constitution fallacy . Arguments for extended cognition often describe a two-way causal coupling between our brains and the environment. One might be tempted to say that cognition extends because of this coupling relationship. However, just because X and Y are causally coupled does not mean that X is part of Y. The expansion of a bimetallic strip in a thermostat is coupled to the motion of atoms in the surrounding air in the room, but that does not mean that the strip’s expansion is a process that extends into the atoms of the air (Adams & Aizawa 2007 p. 91) . Similarly, your brain and the environment may be causally coupled but that does not mean that your brain processes extend into the environment. This objection appears to most directly threaten second-wave arguments for extended cognition, since those arguments emphasise two-way causal coupling between the brain, body, and environment. An advocate of HEC needs to find something beyond mere coupling that justifies cognitive extension. A natural resort would be to appeal to either the explanatory value of treating the coupled system as cognitive or the functional similarity between the coupled system and an internal (actual or possible) case of cognition. However, this would collapse second-wave arguments into their first-wave versions (see Adams & Aizawa 2001, 2007) . Alternatively, a second-wave defender of HEC might stress the normative nature of the coupling relationship and explore how this might render the relationship constitutive rather than merely causal.

The mark of the cognitive . A ‘mark of the cognitive’ is a set of necessary or sufficient conditions for a physical state/process to be cognitive. A mark of the cognitive is often sought as a way of overcoming disagreements between philosophers about HEC. If one could agree on a mark of the cognitive, one could use it to decide whether disputed cases (like Otto’s notebook) really are cases of extended cognition. Proposed marks of the cognitive tend, however, to be at least as controversial as the cases they are supposed to help decide. Various marks of the cognitive have been proposed, including: (i) a cognitive process must involve non-derived representational content and be functionally similar to actual cases of internal, brain-based human cognition (Adams & Aizawa 2007) ; (ii) a cognitive process must be part of an integrated, persisting system, and it must causally contribute to a wide range of cognitive phenomena (Rupert 2009) ; (iii) a process is cognitive so long as it ‘belongs’ to a subject in the sense of causally contributing to the intentional content of that subject’s personal-level states (Rowlands 2010) . Critics of HEC tend to favour conservative marks of the cognitive like (i) or (ii), which rule out many cases of extended cognition. Advocates of HEC tend to favour liberal marks of the cognitive like (iii), which allow in many cases of extended cognition. There is currently no consensus about which is the correct mark of the cognitive, and one may worry that no such set of necessary or sufficient conditions exists. Just as there seems to be no single set of necessary or sufficient conditions that determines which physical processes are healthy , living , or cancerous , so there might be no set of necessary or sufficient conditions that determines which physical processes are cognitive or mental (Allen 2017) .

4 Extended cognition’s kindred views

Philosophical work on the hypothesis of extended cognition (HEC) has focused not only on arguments for, and objections against, the view, but also on tracing HEC’s connections to other claims about the mind. Two types of kindred view stand out: claims about other aspects of the mind extending and other forms of philosophical externalism about cognition .

Other aspects of the mind extending. Clark and Chalmers (1998) restricted their claim about extension to the cognitive, information-processing aspects of the mind (e.g. storage and retrieval of dispositional beliefs). They flagged up conscious experience as an aspect of the mind that is likely to escape their arguments; conscious experience could be entirely brain-based. Noë (2004) disagrees. Using variants of the IBE and second-wave arguments of §2, Noë argues that the physical mechanisms of consciousness extend outside the brain and into the body and environment. Clark (2009) replies that while such extensions are theoretically possible, they are unlikely to happen because our brain’s causal interface with the world (our eyes and our hands) has a narrow bandwidth which (information-rich) conscious experience cannot cross.

Consciousness is not the only source of novel claims about mental extension. Carter et al. (2016) say that emotions extend when the physical processes involved in their cognitive-appraisal component extend. Colombetti & Roberts (2015) argue that all kinds of affective mental states (emotions, moods, sentiments, and temperaments) extend when external objects are used to regulate them. Vierkant (2014) argues that intentional agency and willpower extend. Pritchard (2010) and Carter & Kallestrup (2016) explore extension claims for epistemic states like knowledge . Clark (2007b) argues that the self extends. Essays in Anderson et al. (2018) explore extension claims about other aspects of the mind, including processes that differ significantly from the task-based, problem-solving examples favoured by Clark and Chalmers (e.g. friendship and imaginative engagement with a text).

Other forms of externalism about cognition. HEC is often described as part of a wider research programme called ‘4E cognition’. The other Es are embedded , embodied , and enactive cognition. We have already met embedded cognition with the hypothesis of embedded cognition (see §3). Embodied cognition says that cognition depends in some way on the physical nature of our bodies. Different versions of embodied cognition cash this claim out differently. In some versions, embodied cognition simply expresses a body-based version of HEC: cognition depends on our body when our physical bodies partly realise our cognitive processes (e.g. when a child counts on her fingers, her mathematical cognising extends into the physical movement of her fingers). However, embodied cognition may also be used to express other forms of externalism about the mind. Lakoff & Johnson (1999) say that the specific shape of the human body influences the semantic content of our concepts, including abstract logical and mathematical concepts. This is something about which HEC is largely silent. Enactive cognition says that cognition consists in a looping interaction between perception and action that involves brain, body, and world. This has obvious affinity with second-wave arguments for HEC (see §2), but specific versions of enactivism might differ in various ways from HEC about how they see this integration working. Some enactivists reject HEC’s framing in terms of representations and information processing: ‘sensorimotor’ enactivists say that cognition consists in implicit, non-representational bodily skills (Noë 2004) ; ‘autopoietic’ enactivists say that cognition consists in biologically inspired relationships, such as autopoiesis, adaptivity, and sense-making (Di Paolo 2005) .

HEC has a more distant relationship to other kinds of philosophical externalism, such as content externalism, direct realism about perception, collective intentionality, and group cognition. Content externalism says that the representational content of our cognitive states does not supervene on the internal physical state of our brains. HEC has almost nothing to say about this. HEC’s claim is about the location of the physical vehicles of cognition, not about the factors that determine their representational content (Hurley 2010) . Direct realism about perception says that perception involves a perceiver standing in a special ‘perceiving’ relation to a real external object beyond her brain. HEC assumes almost nothing about whether this view is correct. Indeed, HEC is frequently stated in intentionalist (as opposed to direct-realist) terms. Collective intentionality is the claim that groups of individuals sometimes have shared mental states, such as belief, knowledge, and intention. Closely related to this is the notion of group cognition : a cognitive state or process that is properly attributed to a group of individuals rather than to the individuals that compose that group. While both of these last views agree with HEC that cognition is not found exclusively inside individual heads, HEC is a claim about individual human cognition extending, not about shared or group cognition.

HEC may be combined with a variety of other externalist claims about the mind, but it makes a separate, unique claim about how the human mind spreads out into world.

Related Articles

• Collective intentionality
• Consciousness
• Content: Wide and narrow
• Embodied cognition
• Emotions, nature of
• Functionalism
• Inference to the best explanation
• The realization of knowledge

Articles that might consider linking to this article

• Metaphysics of knowledge

Acknowledgements

I would like to thank Ken Aizawa, Andy Clark, Mike Wheeler, and Rob Rupert for helpful comments on a draft of this entry.

Bibliography

• Subject List
• Take a Tour
• For Authors
• Subscriber Services
• Publications
• African American Studies
• African Studies
• American Literature
• Anthropology
• Architecture Planning and Preservation
• Art History
• Atlantic History
• Biblical Studies
• British and Irish Literature
• Childhood Studies
• Chinese Studies
• Cinema and Media Studies
• Communication
• Criminology
• Environmental Science
• Evolutionary Biology
• International Law
• International Relations
• Islamic Studies
• Jewish Studies
• Latin American Studies
• Latino Studies
• Linguistics
• Literary and Critical Theory
• Medieval Studies
• Military History
• Political Science
• Public Health
• Renaissance and Reformation
• Social Work
• Urban Studies
• Victorian Literature
• Browse All Subjects

How to Subscribe

• Free Trials

In This Article Expand or collapse the "in this article" section The Extended Mind Thesis

Introduction, general overviews.

• Anthologies
• Intellectual Precursors and Inspiration
• Extended Functionalism
• “Second Wave” Approaches
• Perception and Action
• Language and Thought
• Consciousness, Self, and Agency
• Extended Mind and Epistemology
• Material Culture and Cognitive Archaeology
• Group and Social Cognition

Related Articles Expand or collapse the "related articles" section about

About related articles close popup.

Lorem Ipsum Sit Dolor Amet

Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia Curae; Aliquam ligula odio, euismod ut aliquam et, vestibulum nec risus. Nulla viverra, arcu et iaculis consequat, justo diam ornare tellus, semper ultrices tellus nunc eu tellus.

• Collective Intentionality
• Epistemology
• Externalism and Internalism in the Philosophy of Mind
• Metaphysics of Mind
• Perception, Cognition, Action
• Philosophy of Cognitive Science
• Philosophy of Neuroscience
• Philosophy of Psychology
• Philosophy of Technology

Other Subject Areas

Forthcoming articles expand or collapse the "forthcoming articles" section.

• Alfred North Whitehead
• Feminist Aesthetics
• Find more forthcoming articles...
• Export Citations
• Share This Facebook LinkedIn Twitter

The Extended Mind Thesis by Julian Kiverstein , Mirko Farina , Andy Clark LAST REVIEWED: 30 May 2019 LAST MODIFIED: 27 November 2013 DOI: 10.1093/obo/9780195396577-0099

The extended mind thesis (EMT) claims that the cognitive processes that make up the human mind can reach beyond the boundaries of individual to include as proper parts aspects of the individual’s physical and sociocultural environment. Proponents of the extended mind story thus hold that even quite familiar human mental states (such as states of believing that so-and-so) can be realized, in part, by structures and processes located outside the human head. Such claims go far beyond the important, but less challenging, assertion that human cognizing leans heavily on various forms of external scaffolding and support. Instead, they paint mind itself (or better, the physical machinery that realizes some of our cognitive processes and mental states) as, under humanly attainable conditions, extending beyond the bounds of skin and skull. Extended cognition in its most general form occurs when internal and external resources become fluently tuned and deeply integrated in such a way as to enable a cognitive agent to solve problems and accomplish their projects, goals, and interests. Consider, for instance, how technological resources such as pens, paper, and personal computers are now so deeply integrated into our everyday lives that we couldn’t accomplish many of our cognitive goals and purposes without them. The extended mind thesis claims that technological resources have become so thoroughly enmeshed with our internal cognitive machinery that they now count as part of the machinery of thought itself. Underlying (but distinct from) the extended mind thesis is a commonplace observation thatintelligent problem solving of the kind we find in adult humans isn’t something the naked brain can achieve all on its own, but is instead the outcome of the brain and body operating together in an environmental and often technologically loaded setting. As humans, we possess the kinds of high-level cognitive skills and abilities we do in no small part because of the many tools we use for thinking. The extended mind thesis goes further, however, by claiming that it is mere prejudice to suppose that all cognition must take place within the confines of the organism’s skin and skull. Cognitive science, it is then claimed, shouldn’t only concern itself with the more or less enduring processes taking place inside the heads of cognitive agents. Cognitive scientists should also investigate, on a kind of equal footing, temporary, soft-assembled wholes that mesh the problem-solving contributions of the brain and nervous system with the body and physical and sociocultural environment.

The section below on Extended Mind and Epistemology was prepared with the assistance of Spyridon Orestis Palermos. Many thanks to Orestis, and to Duncan Pritchard and John Sutton, for their help and advice.

There are a number of monographs and essays that provide useful introductions to the extended mind thesis, and to the work in cognitive science that (partially) motivates it. Clark and Chalmers 1998 provides the seminal statement of the extended mind thesis, and much of the current debate is in part based on the arguments of this paper. Clark 1997 draws on a wide body of empirical research in robotics, artificial life, connectionism, developmental psychology, and economics to make a case for extended cognition. Haugeland 1998 contains an early statement of the view that minds can profitably be thought of as complex systems that emerge from the dynamic couplings of brain, body, and world. It also described (contested) criteria for determining whether the mechanisms that support cognition can be “partitioned-off” from body and world. Rowlands 1999 develops an argument that evolution favored essentially hybrid cognitive processes. Drawing on research from (more or less) mainstream psychology, Rowlands constructs accounts of visual perception, memory, thought, and language as extended cognitive processes. Wheeler 2005 shows how the extended mind thesis has its philosophical foundations in Heidegger’s phenomenology, and explains how Heidegger’s philosophy, when combined with the extended mind thesis, contains the seeds of a solution to the frame problem, which has so far proved an insuperable problem for orthodox cognitive science. Wilson and Clark 2009 offers a useful taxonomy of extended cognition along two dimensions: the nature of the external resources recruited by cognizers, and the durability of extended cognitive systems. Adams and Aizawa 2008 assembles and further develops a number of important challenges to the extended mind thesis, and also advances an alternative “intracranialist” view of cognitive systems. The introduction to the Menary 2010 contains a state-of-the-art survey of the debate surrounding the extended mind.

Adams, Frederick, and Kenneth Aizawa. The Bounds of Cognition . Oxford and Malden, MA: Blackwell, 2008.

A comprehensive critique of the arguments for the extended mind.

Clark, Andy. Being There: Putting Brain, Body and World Together Again . Cambridge, MA: MIT Press, 1997.

A book-length exploration of cognition as embodied and as environmentally scaffolded. This wide body of empirical research helps to motivate and support the extended mind thesis. For a similar survey of recent research (mostly in robotics), see the opening chapter of Clark 2008 (cited under Textbooks ).

Clark, Andy, and David J. Chalmers. “The Extended Mind.” Analysis 58.1 (1998): 7–19.

DOI: 10.1093/analys/58.1.7

The original statement of the extended mind thesis (including the “parity arguments” described later in this entry). Reprinted In Philosophy of Mind: Classical and Contemporary Readings , edited by David Chalmers (Oxford: Oxford University Press, 2002).

Haugeland, John. “Mind Embodied and Embedded.” In Having Thought: Essays in the Metaphysics of Mind . By John Haugeland, 207–240. Cambridge, MA: Harvard University Press, 1998.

An influential early statement of the view that minds should be thought of as product of the dynamic coupling of brain, body, and world. Contains a useful (but also controversial) criterion for fixing the boundaries of cognitive systems.

Menary, Richard. “Introduction: The Extended Mind in Focus.” In The Extended Mind . Edited by Richard Menary, 1–26. Cambridge, MA: MIT Press, 2010.

A useful survey of the current debate surrounding the extended mind. Forms the introduction to an important collection of essays.

Rowlands, M. The Body in Mind: Understanding Cognitive Processes . Camdridge, UK: Cambridge University Press, 1999.

DOI: 10.1017/CBO9780511583261

Chapter 6 (pp. 119–147) of this book points to the work of George Miller to show the many limitations that internal working memory has to face when we remember things. Rowlands argues that at least some memory processes must be understood as the result of a series of interactions between a remembering organism and its milieu, and on this basis claims that working memory is essentially “hybrid” in character.

Wheeler, Michael. Reconstructing the Cognitive World . Cambridge, MA: MIT Press, 2005.

Suggests that the extended mind thesis is part of a new movement in cognitive science that has its philosophical foundations in Martin Heidegger’s existential phenomenology. Uses the extended mind thesis to propose a solution to the frame problem, a notoriously stubborn problem for orthodox cognitive science.

Wilson, Robert A., and Andy Clark. “How to Situate Cognition: Letting Nature Take Its Course.” In The Cambridge Handbook of Situated Cognition . Edited by Philip Robbins and Murat Aydede, 55–76. Cambridge, UK: Cambridge University Press, 2009.

Provides a useful taxonomy of extended cognition and argues that the resulting account of cognitive extension can be used to respond to a number of prominent objections to extended cognition.

back to top

Users without a subscription are not able to see the full content on this page. Please subscribe or login .

Oxford Bibliographies Online is available by subscription and perpetual access to institutions. For more information or to contact an Oxford Sales Representative click here .

• About Philosophy »
• Meet the Editorial Board »
• A Priori Knowledge
• Abduction and Explanatory Reasoning
• Abstract Objects
• Addams, Jane
• Adorno, Theodor
• Aesthetic Hedonism
• Aesthetics, Analytic Approaches to
• Aesthetics, Continental
• Aesthetics, Environmental
• Aesthetics, History of
• African Philosophy, Contemporary
• Alexander, Samuel
• Analytic/Synthetic Distinction
• Anarchism, Philosophical
• Animal Rights
• Anscombe, G. E. M.
• Anthropic Principle, The
• Anti-Natalism
• Applied Ethics
• Aquinas, Thomas
• Argument Mapping
• Art and Emotion
• Art and Knowledge
• Art and Morality
• Astell, Mary
• Aurelius, Marcus
• Austin, J. L.
• Bacon, Francis
• Bayesianism
• Bergson, Henri
• Berkeley, George
• Biology, Philosophy of
• Bolzano, Bernard
• Boredom, Philosophy of
• British Idealism
• Buber, Martin
• Buddhist Philosophy
• Burge, Tyler
• Business Ethics
• Camus, Albert
• Canterbury, Anselm of
• Carnap, Rudolf
• Cavendish, Margaret
• Chemistry, Philosophy of
• Childhood, Philosophy of
• Chinese Philosophy
• Cognitive Ability
• Cognitive Phenomenology
• Cognitive Science, Philosophy of
• Coherentism
• Communitarianism
• Computational Science
• Computer Science, Philosophy of
• Computer Simulations
• Comte, Auguste
• Conceptual Role Semantics
• Conditionals
• Confirmation
• Connectionism
• Consciousness
• Constructive Empiricism
• Contemporary Hylomorphism
• Contextualism
• Contrastivism
• Cook Wilson, John
• Cosmology, Philosophy of
• Critical Theory
• Culture and Cognition
• Daoism and Philosophy
• Davidson, Donald
• de Beauvoir, Simone
• de Montaigne, Michel
• Decision Theory
• Deleuze, Gilles
• Derrida, Jacques
• Descartes, René
• Descartes, René: Sensory Representations
• Descriptions
• Dewey, John
• Dialetheism
• Disagreement, Epistemology of
• Disjunctivism
• Dispositions
• Divine Command Theory
• Doing and Allowing
• du Châtelet, Emilie
• Dummett, Michael
• Dutch Book Arguments
• Early Modern Philosophy, 1600-1750
• Eastern Orthodox Philosophical Thought
• Education, Philosophy of
• Engineering, Philosophy and Ethics of
• Environmental Philosophy
• Epistemic Basing Relation
• Epistemic Defeat
• Epistemic Injustice
• Epistemic Justification
• Epistemic Philosophy of Logic
• Epistemology and Active Externalism
• Epistemology, Bayesian
• Epistemology, Feminist
• Epistemology, Internalism and Externalism in
• Epistemology, Moral
• Epistemology of Education
• Ethical Consequentialism
• Ethical Deontology
• Ethical Intuitionism
• Eugenics and Philosophy
• Events, The Philosophy of
• Evidence-Based Medicine, Philosophy of
• Evidential Support Relation In Epistemology, The
• Evolutionary Debunking Arguments in Ethics
• Evolutionary Epistemology
• Experimental Philosophy
• Explanations of Religion
• Extended Mind Thesis, The
• Faith, Conceptions of
• Feminist Philosophy
• Feyerabend, Paul
• Fichte, Johann Gottlieb
• Fictionalism
• Fictionalism in the Philosophy of Mathematics
• Film, Philosophy of
• Foot, Philippa
• Foreknowledge
• Forgiveness
• Formal Epistemology
• Foucault, Michel
• Frege, Gottlob
• Gadamer, Hans-Georg
• Geometry, Epistemology of
• God and Possible Worlds
• God, Arguments for the Existence of
• God, The Existence and Attributes of
• Grice, Paul
• Habermas, Jürgen
• Hart, H. L. A.
• Heaven and Hell
• Hegel, Georg Wilhelm Friedrich: Aesthetics
• Hegel, Georg Wilhelm Friedrich: Metaphysics
• Hegel, Georg Wilhelm Friedrich: Philosophy of History
• Hegel, Georg Wilhelm Friedrich: Philosophy of Politics
• Heidegger, Martin: Early Works
• Hermeneutics
• Higher Education, Philosophy of
• History, Philosophy of
• Hobbes, Thomas
• Horkheimer, Max
• Human Rights
• Hume, David: Aesthetics
• Hume, David: Moral and Political Philosophy
• Husserl, Edmund
• Idealizations in Science
• Identity in Physics
• Imagination
• Imagination and Belief
• Immanuel Kant: Political and Legal Philosophy
• Impossible Worlds
• Incommensurability in Science
• Indian Philosophy
• Indispensability of Mathematics
• Inductive Reasoning
• Instruments in Science
• Intellectual Humility
• Intentionality, Collective
• James, William
• Japanese Philosophy
• Kant and the Laws of Nature
• Kant, Immanuel: Aesthetics and Teleology
• Kant, Immanuel: Ethics
• Kant, Immanuel: Theoretical Philosophy
• Kierkegaard, Søren
• Knowledge-first Epistemology
• Knowledge-How
• Kristeva, Julia
• Kuhn, Thomas S.
• Lacan, Jacques
• Lakatos, Imre
• Langer, Susanne
• Language of Thought
• Language, Philosophy of
• Latin American Philosophy
• Laws of Nature
• Legal Epistemology
• Legal Philosophy
• Legal Positivism
• Leibniz, Gottfried Wilhelm
• Levinas, Emmanuel
• Lewis, C. I.
• Literature, Philosophy of
• Locke, John
• Locke, John: Identity, Persons, and Personal Identity
• Lottery and Preface Paradoxes, The
• Machiavelli, Niccolò
• Martin Heidegger: Later Works
• Martin Heidegger: Middle Works
• Material Constitution
• Mathematical Explanation
• Mathematical Pluralism
• Mathematical Structuralism
• Mathematics, Ontology of
• Mathematics, Philosophy of
• Mathematics, Visual Thinking in
• McDowell, John
• McTaggart, John
• Meaning of Life, The
• Mechanisms in Science
• Medically Assisted Dying
• Medicine, Contemporary Philosophy of
• Medieval Logic
• Medieval Philosophy
• Mental Causation
• Merleau-Ponty, Maurice
• Meta-epistemological Skepticism
• Metaepistemology
• Metametaphysics
• Metaphilosophy
• Metaphysical Grounding
• Metaphysics, Contemporary
• Metaphysics, Feminist
• Midgley, Mary
• Mill, John Stuart
• Mind, Metaphysics of
• Modal Epistemology
• Models and Theories in Science
• Montesquieu
• Moore, G. E.
• Moral Contractualism
• Moral Naturalism and Nonnaturalism
• Moral Responsibility
• Multiculturalism
• Murdoch, Iris
• Music, Analytic Philosophy of
• Nationalism
• Natural Kinds
• Naturalism in the Philosophy of Mathematics
• Naïve Realism
• Neo-Confucianism
• Neuroscience, Philosophy of
• Nietzsche, Friedrich
• Nonexistent Objects
• Normative Ethics
• Normative Foundations, Philosophy of Law:
• Normativity and Social Explanation
• Objectivity
• Occasionalism
• Ontological Dependence
• Ontology of Art
• Ordinary Objects
• Other Minds
• Panpsychism
• Particularism in Ethics
• Pascal, Blaise
• Paternalism
• Peirce, Charles Sanders
• Perception, The Problem of
• Perfectionism
• Persistence
• Personal Identity
• Phenomenal Concepts
• Phenomenal Conservatism
• Phenomenology
• Philosophy for Children
• Photography, Analytic Philosophy of
• Physicalism
• Physicalism and Metaphysical Naturalism
• Physics, Experiments in
• Political Epistemology
• Political Obligation
• Political Philosophy
• Popper, Karl
• Pornography and Objectification, Analytic Approaches to
• Practical Knowledge
• Practical Moral Skepticism
• Practical Reason
• Probabilistic Representations of Belief
• Probability, Interpretations of
• Problem of Divine Hiddenness, The
• Problem of Evil, The
• Propositions
• Psychology, Philosophy of
• Quine, W. V. O.
• Racist Jokes
• Rationalism
• Rationality
• Rawls, John: Moral and Political Philosophy
• Realism and Anti-Realism
• Realization
• Reasons in Epistemology
• Reductionism in Biology
• Reference, Theory of
• Reid, Thomas
• Reliabilism
• Religion, Philosophy of
• Religious Belief, Epistemology of
• Religious Experience
• Religious Pluralism
• Ricoeur, Paul
• Risk, Philosophy of
• Rorty, Richard
• Rousseau, Jean-Jacques
• Rule-Following
• Russell, Bertrand
• Ryle, Gilbert
• Sartre, Jean-Paul
• Schopenhauer, Arthur
• Science and Religion
• Science, Theoretical Virtues in
• Scientific Explanation
• Scientific Progress
• Scientific Realism
• Scientific Representation
• Scientific Revolutions
• Scotus, Duns
• Self-Knowledge
• Sellars, Wilfrid
• Semantic Externalism
• Semantic Minimalism
• Senses, The
• Sensitivity Principle in Epistemology
• Shepherd, Mary
• Singular Thought
• Situated Cognition
• Situationism and Virtue Theory
• Skepticism, Contemporary
• Skepticism, History of
• Slurs, Pejoratives, and Hate Speech
• Smith, Adam: Moral and Political Philosophy
• Social Aspects of Scientific Knowledge
• Social Epistemology
• Social Identity
• Sounds and Auditory Perception
• Space and Time
• Speech Acts
• Spinoza, Baruch
• Stebbing, Susan
• Strawson, P. F.
• Structural Realism
• Supererogation
• Supervenience
• Tarski, Alfred
• Technology, Philosophy of
• Testimony, Epistemology of
• Theoretical Terms in Science
• Thomas Aquinas' Philosophy of Religion
• Thought Experiments
• Time and Tense
• Time Travel
• Transcendental Arguments
• Truth and the Aim of Belief
• Truthmaking
• Turing Test
• Two-Dimensional Semantics
• Understanding
• Uniqueness and Permissiveness in Epistemology
• Utilitarianism
• Value of Knowledge
• Vienna Circle
• Virtue Epistemology
• Virtue Ethics
• Virtues, Epistemic
• Virtues, Intellectual
• Voluntarism, Doxastic
• Weakness of Will
• Weil, Simone
• William of Ockham
• Williams, Bernard
• Wittgenstein, Ludwig: Early Works
• Wittgenstein, Ludwig: Later Works
• Wittgenstein, Ludwig: Middle Works
• Wollstonecraft, Mary
• Privacy Policy
• Cookie Policy
• Legal Notice
• Accessibility

Powered by:

• [66.249.64.20|185.80.150.64]
• 185.80.150.64

Cognitive ability and the extended cognition thesis

• Published: 12 March 2010
• Volume 175 , pages 133–151, ( 2010 )

Cite this article

• Duncan Pritchard 1  

1140 Accesses

112 Citations

3 Altmetric

Explore all metrics

This paper explores the ramifications of the extended cognition thesis in the philosophy of mind for contemporary epistemology. In particular, it argues that all theories of knowledge need to accommodate the ability intuition that knowledge involves cognitive ability, but that once this requirement is understood correctly there is no reason why one could not have a conception of cognitive ability that was consistent with the extended cognition thesis. There is thus, surprisingly, a straightforward way of developing our current thinking about knowledge such that it incorporates the extended cognition thesis.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price includes VAT (Russian Federation)

Instant access to the full article PDF.

Rent this article via DeepDyve

Institutional subscriptions

Similar content being viewed by others

The Mark of Understanding: In Defense of an Ability Account

Sven Delarivière & Bart Van Kerkhove

Knowledge how, ability, and the type-token distinction

Garry Young

Kant on cognition and knowledge

Marcus Willaschek & Eric Watkins

Adams F., Aizawa K. (2001) The bounds of cognition. Philosophical Psychology 14: 43–64

Article   Google Scholar  

Adams F., Aizawa K. (2008) The bounds of cognition. Blackwell, Oxford

Google Scholar  

Chisholm R. (1977) Theory of knowledge. Prentice-Hall, Englewood Cliffs

Clark A. (1998) Being there: Putting brain, body and world together again. MIT Press, Cambridge

Clark A. (2007) Curing cognitive hiccups: A defense of the extended mind. Journal of Philosophy 104: 163–192

Clark A. (2008) Memento’s revenge: The extended mind, extended. In: Menary R. (eds) The extended mind. Ashgate Press, Aldershot

Clark A., Chalmers D. (1998) The extended mind. Analysis 58: 7–19

Cowart, M. (2006). Embodied cognition. In B. Dowden & J. Fieser (Eds.), Internet Encyclopaedia of Philosophy . Martin, TN: University of Tennessee. http://www.iep.utm.edu/e/embodcog.htm .

Gertler B. (2007) Overextending the mind?. In: Gertler B., Shapiro L. (eds) Arguing about mind. Routledge, London, pp 192–206

Gettier E. (1963) Is justified true belief knowledge?. Analysis 23: 121–123

Goldman A. (1976) Discrimination and perceptual knowledge. The Journal of Philosophy 73: 771–791

Goldman A. (1986) Epistemology and cognition. Harvard University Press, Cambridge

Greco J. (1999) Agent reliabilism. Philosophical Perspectives 13: 273–296

Greco J. (2000) Putting skeptics in their place: The nature of skeptical arguments and their role in philosophical inquiry. Cambridge University Press, Cambridge

Book   Google Scholar  

Greco J. (2002) Knowledge as credit for true belief. In: DePaul M., Zagzebski L. (eds) Intellectual virtue: Perspectives from ethics and epistemology. Oxford University Press, Oxford

Greco J. (2007) The nature of ability and the purpose of knowledge. Philosophical Issues 17: 57–69

Greco J. (2008) What’s wrong with contextualism?. The Philosophical Quarterly 58: 416–436

Greco J. (2009) The value problem. In: Haddock A., Millar A., Pritchard D.H. (eds) Epistemic value. Oxford University Press, Oxford, pp 313–321

Haddock A., Millar A., Pritchard D.H. (2010) The nature and value of knowledge: Three investigations. Oxford University Press, Oxford

Lackey J. (2007) Why we don’t deserve credit for everything we know. Synthese 158: 345–361

Lehrer K., Cohen S. (1983) Justification, truth, and coherence. Synthese 55: 191–207

Menary R. (2006) Attacking the bounds of cognition. Philosophical Psychology 19: 329–344

Menary R. (2007) Cognitive integration: Mind and cognition unbounded. Palgrave Macmillan, London

Menary, R. (eds) (2008) The extended mind. Ashgate Press, Aldershot

Plantinga A. (1993a) Warrant and proper function. Oxford University Press, Oxford

Plantinga A. (1993b) Warrant: The current debate. Oxford University Press, Oxford

Pritchard D.H. (2005) Epistemic luck. Oxford University Press, Oxford

Pritchard D.H. (2008a) Greco on knowledge: Virtues, contexts, achievements. The Philosophical Quarterly 58: 437–447

Pritchard D.H. (2008b) Sensitivity, safety, and anti-luck epistemology. In: Greco J. (eds) The Oxford handbook of skepticism. Oxford University Press, Oxford, pp 437–455

Chapter   Google Scholar  

Pritchard D.H. (2008c) Virtue epistemology and epistemic luck, revisited. Metaphilosophy 39: 66–88

Pritchard, D. H. (2009a). Anti-luck virtue epistemology. Manuscript.

Pritchard D.H. (2009) Apt performance and epistemic value. Philosophical Studies 143: 407–416

Rowlands M. (1999) The body in mind: Understanding cognitive processes. Cambridge University Press, Cambridge

Rowlands M. (2009) Extended cognition and the mark of the cognitive. Philosophical Psychology 22: 1–19

Rupert R. (2004) Challenges to the hypothesis of extended cognition. Journal of Philosophy 101: 389–428

Sosa E. (1988) Beyond skepticism, to the best of our knowledge. Mind 97: 153–189

Sosa E. (1991) Knowledge in perspective: Selected essays in epistemology. Cambridge University Press, Cambridge

Sosa E. (2007) Apt belief and reflective knowledge. Oxford University Press, Oxford

Vaesen, K. (2009). Knowledge without credit. Manuscript.

Zagzebski L. (1996) Virtues of the mind: An inquiry into the nature of virtue and the ethical foundations of knowledge. Cambridge University Press, Cambridge

Zagzebski L. (1999) What is knowledge?. In: Greco J., Sosa E. (eds) Epistemology. Blackwell, Oxford, pp 92–116

Download references

Author information

Authors and affiliations.

Department of Philosophy, University of Edinburgh, Edinburgh, Scotland, UK

Duncan Pritchard

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Duncan Pritchard .

Rights and permissions

Reprints and permissions

About this article

Pritchard, D. Cognitive ability and the extended cognition thesis. Synthese 175 (Suppl 1), 133–151 (2010). https://doi.org/10.1007/s11229-010-9738-y

Download citation

Received : 28 February 2010

Accepted : 02 March 2010

Published : 12 March 2010

Issue Date : August 2010

DOI : https://doi.org/10.1007/s11229-010-9738-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Cognitive ability
• Epistemology
• Epistemic luck
• Epistemic virtue
• Extended cognition
• Find a journal
• Publish with us
• Track your research

What is Cognition? Extended Cognition and the Criterion of the Cognitive

According to the thesis of the extended mind, at least some cognitive processes extend into the cognizing subject's environment in the sense that they are composed of processes of manipulation, exploitation, and transformation performed by that subject on suitable environmental structures. In contrast, according to the thesis of the embedded mind, the manipulation, exploitation, and transformation of information-bearing structures provides a useful scaffolding which facilitates cognitive processes but does not, even in part, constitute them. The two theses are distinct but often confused. The extended mind has attracted three ostensibly distinct kinds of objection, all of which on further analysis reduce to the idea that the arguments for the extended mind in fact only establish the thesis of the embedded mind. This chapter argues that these three objections can all be resolved by the provision of an adequate and properly motivated criterion of the cognitive.

• Related Documents

Extended Cognition and Extended Consciousness

Chapter 1 discusses two questions about the extended mind. First, what is the extended mind thesis? Second, can there be extended consciousness, and if not, why not? The chapter answers the first question by arguing that the thesis should be formulated in terms of perception and action: a subject’s cognitive processes and mental states can be partly constituted by entities that are external to the subject, in virtue of the subject’s interacting with these entities via perception and action. The second question is answered by appealing to direct availability for global control as the physical correlate of consciousness: extended processes always involve indirect availability for global control, mediated by perception and action, so there is no extended consciousness.

Embodying algorithms, enactive artificial intelligence and the extended cognition: You can see as much as you know about algorithm

The recent proliferation of artificial intelligence (AI) gives rise to questions on how users interact with AI services and how algorithms embody the values of users. Despite the surging popularity of AI, how users evaluate algorithms, how people perceive algorithmic decisions, and how they relate to algorithmic functions remain largely unexplored. Invoking the idea of embodied cognition, we characterize core constructs of algorithms that drive the value of embodiment and conceptualizes these factors in reference to trust by examining how they influence the user experience of personalized recommendation algorithms. The findings elucidate the embodied cognitive processes involved in reasoning algorithmic characteristics – fairness, accountability, transparency, and explainability – with regard to their fundamental linkages with trust and ensuing behaviors. Users use a dual-process model, whereby a sense of trust built on a combination of normative values and performance-related qualities of algorithms. Embodied algorithmic characteristics are significantly linked to trust and performance expectancy. Heuristic and systematic processes through embodied cognition provide a concise guide to its conceptualization of AI experiences and interaction. The identified user cognitive processes provide information on a user’s cognitive functioning and patterns of behavior as well as a basis for subsequent metacognitive processes.

Going wide: extended mind and Wittgenstein

Extended mind remains a provocative approach to cognition and mentality. However, both those for and against this approach have tacitly accepted that cognition or mentality can be understood in terms of those sub-personal processes ongoing during some task. I label this a process view of cognition (PV). Using Wittgenstein’s philosophical approach, I argue that proponents of extended mind should reject PV and instead endorse a ‘wide view’ of mentality. This wide view clarifies why the hypothesis of extended mind (HEM) is incoherent. However, this view also indicates why the hypothesis of extended cognition (HEC) could be true.

The Network-Extended Mind

Whereas the traditional view in cognitive science has been to view mind and cognition as something that is the result of essentially inner, neural processes, the extended cognition perspective claims that at least some human mental states and processes stem from complex webs of causal influence involving extra-neural resources, most notably the resources of our social and technological environments. In this chapter the authors explore the possibility that contemporary and near-future network systems are poised to extend and perhaps transform our human cognitive potential. They also examine the extent to which the information and network sciences are relevant to their understanding of various forms of cognitive extension, particularly with respect to the formation, maintenance and functioning of extended cognitive systems in network-enabled environments. Their claim is that the information and network sciences are relevant on two counts: firstly, they support an understanding of the mechanisms underpinning socially- and technologically-mediated forms of cognitive extension; secondly, they serve to guide and inform engineering efforts that strive to enhance and expand our cognitive capabilities. The authors discuss the relevance and applicability of these conclusions to current and future research exploring the contribution of network technologies to military coalition operations.

Telling Actions: Institutions, Collective Agency, and Critical Narratives

Concepts of socially extended cognition and cognitive institution have relevance for understanding how social and cultural practices shape not only our cognitive processes, but also our actions and interactions. Cognitive institutions are not only institutions that support cognitive processes, but are also such that without them these specific cognitive processes would not exist. Examples include things like legal systems, schools, universities, and cultural institutions. Narrative practices can establish and support such institutions, but critical narratives can operate as the basis for a critique of such institutions. Narratives not only serve to mediate intersubjective relations, they can map out the immediate and deeper contexts of action and understanding, provide detailed descriptions of events, objects and persons, help to coordinate complex tasks, define the identities and roles of individuals and groups, and express agreement among individuals. Critical narrative practice can also generate a reflective attitude about new or different possibilities.

Economic cognitive institutions

AbstractThis paper introduces the notion of ‘cognitive’ institution and discusses its relevance to institutional economics. Cognitive institutions are conceptually founded on the philosophy of mind notion of extended mind, broadened to also include the distinctly social, institutional, and normative dimensions. Cognitive institutions are defined as institutions that not just allow agents to perform certain cognitive processes in the social domain but, more importantly, without which some of the agents' cognitive processes would not exist or even be possible. The externalist point of view of the extended mind has already had some influence in institutional economics: Arthur Denzau and Douglass North first introduced the notion of institution understood in terms of ‘shared mental models’, and relatedly philosopher Andy Clark introduced the notion of ‘scaffolding institution’. We discuss shared mental models and scaffolding institutions and go a step further by showing that the notion of cognitive institution can capture more fundamental and salient aspects of economic institutions. In particular, we focus on the market as an economic cognitive institution.

The Elusive Extended Mind

For nearly twenty years Andy Clark has been the chief architect and proponent of the thesis of extended mind. But it is only the cognitive processes in the mind that extend, according to Clark (not consciousness itself). However, when it comes to saying what a cognitive process is such that one can determine whether it does or does not extend, Clark is less forthcoming. He has offered a Dennettian “cognition is as cognition does.” He has offered that cognition is “what supports intelligent behavior.” In some cases he comes very close to asserting that we don’t really need to say what cognition is. This chapter explains why this all matters and why the failure to be more forthcoming makes the extended mind an elusive entity.

Extended Cognition and The Innovation Process

Innovations advance into the ‘adjacent possible’, enabled and constrained by the current state of the world, in a way that is unpredictable and not law-entailed. Unpredictability is the hallmark of the idea that innovation processes are contingent and embodied in the interaction between individuals and artefacts in the environment. In this chapter, we explore the cognitive and behavioural factors involved in exaptive innovation processes by using the notion of ‘extended cognition’. Extended cognition builds on the hypothesis that cognitive processes are not limited to the brain but also extend into the physical world as the objects of the environment facilitate, integrate with, and even constitute specific cognitive processes. We argue that exaptive innovations can be better understood by focusing on practicality and procedural knowledge from an extended cognition perspective. Artefact manipulation is not merely pragmatic but also epistemic as it enables specific reasoning processes that lead to the discovery of new uses.

Is Having Your Computer Compromised a Personal Assault? The Ethics of Extended Cognition

ABSTRACT:Philosophy of mind and cognitive science (e.g., Clark and Chalmers 1998; Clark 2010; Palermos 2014) have recently become increasingly receptive to the hypothesis of extended cognition, according to which external artifacts such as our laptops and smartphones can—under appropriate circumstances—feature as material realizers of a person's cognitive processes. We argue that, to the extent that the hypothesis of extended cognition is correct, our legal and ethical theorizing and practice must be updated by broadening our conception of personal assault so as to include intentional harm toward gadgets that have been appropriately integrated. We next situate the theoretical case for extended personal assault within the context of some recent ethical and legal cases and close with critical discussion.

4E’s Are Too Many Why Enactive World-Making Does Not Need The Extended Mind Thesis

4E’s cognition – embodied, embedded, enacted, extended – replaces the cognitivist notion of world-mirroring with an active process of world-making: cognition needs no mental representation and is distributed over body, brain and environment. In recent years, the remark that extended cognition is not enactive and that the embodied approach to cognition fails to provide a definition of body raise the question of whether a postcognitivist approach to experience needs 4E’s. This contribution suggests that it does not. The enactive body as a moving sense-making-system informed by phenomenology and pragmatism and its role in the constitution of the distinctive quality of an experience are discussed.

Export Citation Format

Share document.

• Search Menu
• Browse content in Arts and Humanities
• Browse content in Archaeology
• Anglo-Saxon and Medieval Archaeology
• Archaeological Methodology and Techniques
• Archaeology by Region
• Archaeology of Religion
• Archaeology of Trade and Exchange
• Biblical Archaeology
• Contemporary and Public Archaeology
• Environmental Archaeology
• Historical Archaeology
• History and Theory of Archaeology
• Industrial Archaeology
• Landscape Archaeology
• Mortuary Archaeology
• Prehistoric Archaeology
• Underwater Archaeology
• Urban Archaeology
• Zooarchaeology
• Browse content in Architecture
• Architectural Structure and Design
• History of Architecture
• Residential and Domestic Buildings
• Theory of Architecture
• Browse content in Art
• Art Subjects and Themes
• History of Art
• Industrial and Commercial Art
• Theory of Art
• Biographical Studies
• Byzantine Studies
• Browse content in Classical Studies
• Classical Literature
• Classical Reception
• Classical History
• Classical Philosophy
• Classical Mythology
• Classical Art and Architecture
• Classical Oratory and Rhetoric
• Greek and Roman Papyrology
• Greek and Roman Archaeology
• Greek and Roman Epigraphy
• Greek and Roman Law
• Late Antiquity
• Religion in the Ancient World
• Digital Humanities
• Browse content in History
• Colonialism and Imperialism
• Diplomatic History
• Environmental History
• Genealogy, Heraldry, Names, and Honours
• Genocide and Ethnic Cleansing
• Historical Geography
• History by Period
• History of Emotions
• History of Agriculture
• History of Education
• History of Gender and Sexuality
• Industrial History
• Intellectual History
• International History
• Labour History
• Legal and Constitutional History
• Local and Family History
• Maritime History
• Military History
• National Liberation and Post-Colonialism
• Oral History
• Political History
• Public History
• Regional and National History
• Revolutions and Rebellions
• Slavery and Abolition of Slavery
• Social and Cultural History
• Theory, Methods, and Historiography
• Urban History
• World History
• Browse content in Language Teaching and Learning
• Language Learning (Specific Skills)
• Language Teaching Theory and Methods
• Browse content in Linguistics
• Applied Linguistics
• Cognitive Linguistics
• Computational Linguistics
• Forensic Linguistics
• Grammar, Syntax and Morphology
• Historical and Diachronic Linguistics
• History of English
• Language Evolution
• Language Reference
• Language Variation
• Language Families
• Language Acquisition
• Lexicography
• Linguistic Anthropology
• Linguistic Theories
• Linguistic Typology
• Phonetics and Phonology
• Psycholinguistics
• Sociolinguistics
• Translation and Interpretation
• Writing Systems
• Browse content in Literature
• Bibliography
• Children's Literature Studies
• Literary Studies (Romanticism)
• Literary Studies (American)
• Literary Studies (Modernism)
• Literary Studies (Asian)
• Literary Studies (European)
• Literary Studies (Eco-criticism)
• Literary Studies - World
• Literary Studies (1500 to 1800)
• Literary Studies (19th Century)
• Literary Studies (20th Century onwards)
• Literary Studies (African American Literature)
• Literary Studies (British and Irish)
• Literary Studies (Early and Medieval)
• Literary Studies (Fiction, Novelists, and Prose Writers)
• Literary Studies (Gender Studies)
• Literary Studies (Graphic Novels)
• Literary Studies (History of the Book)
• Literary Studies (Plays and Playwrights)
• Literary Studies (Poetry and Poets)
• Literary Studies (Postcolonial Literature)
• Literary Studies (Queer Studies)
• Literary Studies (Science Fiction)
• Literary Studies (Travel Literature)
• Literary Studies (War Literature)
• Literary Studies (Women's Writing)
• Literary Theory and Cultural Studies
• Mythology and Folklore
• Shakespeare Studies and Criticism
• Browse content in Media Studies
• Browse content in Music
• Applied Music
• Dance and Music
• Ethics in Music
• Ethnomusicology
• Gender and Sexuality in Music
• Medicine and Music
• Music Cultures
• Music and Media
• Music and Culture
• Music and Religion
• Music Education and Pedagogy
• Music Theory and Analysis
• Musical Scores, Lyrics, and Libretti
• Musical Structures, Styles, and Techniques
• Musicology and Music History
• Performance Practice and Studies
• Race and Ethnicity in Music
• Sound Studies
• Browse content in Performing Arts
• Browse content in Philosophy
• Aesthetics and Philosophy of Art
• Epistemology
• Feminist Philosophy
• History of Western Philosophy
• Metaphysics
• Moral Philosophy
• Non-Western Philosophy
• Philosophy of Language
• Philosophy of Mind
• Philosophy of Perception
• Philosophy of Action
• Philosophy of Law
• Philosophy of Religion
• Philosophy of Science
• Philosophy of Mathematics and Logic
• Practical Ethics
• Social and Political Philosophy
• Browse content in Religion
• Biblical Studies
• Christianity
• East Asian Religions
• History of Religion
• Judaism and Jewish Studies
• Qumran Studies
• Religion and Education
• Religion and Health
• Religion and Politics
• Religion and Science
• Religion and Law
• Religion and Art, Literature, and Music
• Religious Studies
• Browse content in Society and Culture
• Cookery, Food, and Drink
• Cultural Studies
• Customs and Traditions
• Ethical Issues and Debates
• Hobbies, Games, Arts and Crafts
• Lifestyle, Home, and Garden
• Natural world, Country Life, and Pets
• Popular Beliefs and Controversial Knowledge
• Sports and Outdoor Recreation
• Technology and Society
• Travel and Holiday
• Visual Culture
• Browse content in Law
• Arbitration
• Browse content in Company and Commercial Law
• Commercial Law
• Company Law
• Browse content in Comparative Law
• Systems of Law
• Competition Law
• Browse content in Constitutional and Administrative Law
• Government Powers
• Judicial Review
• Local Government Law
• Military and Defence Law
• Parliamentary and Legislative Practice
• Construction Law
• Contract Law
• Browse content in Criminal Law
• Criminal Procedure
• Criminal Evidence Law
• Sentencing and Punishment
• Employment and Labour Law
• Environment and Energy Law
• Browse content in Financial Law
• Banking Law
• Insolvency Law
• History of Law
• Human Rights and Immigration
• Intellectual Property Law
• Browse content in International Law
• Private International Law and Conflict of Laws
• Public International Law
• IT and Communications Law
• Jurisprudence and Philosophy of Law
• Law and Society
• Law and Politics
• Browse content in Legal System and Practice
• Courts and Procedure
• Legal Skills and Practice
• Primary Sources of Law
• Regulation of Legal Profession
• Medical and Healthcare Law
• Browse content in Policing
• Criminal Investigation and Detection
• Police and Security Services
• Police Procedure and Law
• Police Regional Planning
• Browse content in Property Law
• Personal Property Law
• Study and Revision
• Terrorism and National Security Law
• Browse content in Trusts Law
• Wills and Probate or Succession
• Browse content in Medicine and Health
• Browse content in Allied Health Professions
• Arts Therapies
• Clinical Science
• Dietetics and Nutrition
• Occupational Therapy
• Operating Department Practice
• Physiotherapy
• Radiography
• Speech and Language Therapy
• Browse content in Anaesthetics
• General Anaesthesia
• Neuroanaesthesia
• Clinical Neuroscience
• Browse content in Clinical Medicine
• Acute Medicine
• Cardiovascular Medicine
• Clinical Genetics
• Clinical Pharmacology and Therapeutics
• Dermatology
• Endocrinology and Diabetes
• Gastroenterology
• Genito-urinary Medicine
• Geriatric Medicine
• Infectious Diseases
• Medical Toxicology
• Medical Oncology
• Pain Medicine
• Palliative Medicine
• Rehabilitation Medicine
• Respiratory Medicine and Pulmonology
• Rheumatology
• Sleep Medicine
• Sports and Exercise Medicine
• Community Medical Services
• Critical Care
• Emergency Medicine
• Forensic Medicine
• Haematology
• History of Medicine
• Browse content in Medical Skills
• Clinical Skills
• Communication Skills
• Nursing Skills
• Surgical Skills
• Medical Ethics
• Browse content in Medical Dentistry
• Oral and Maxillofacial Surgery
• Paediatric Dentistry
• Restorative Dentistry and Orthodontics
• Surgical Dentistry
• Medical Statistics and Methodology
• Browse content in Neurology
• Clinical Neurophysiology
• Neuropathology
• Nursing Studies
• Browse content in Obstetrics and Gynaecology
• Gynaecology
• Occupational Medicine
• Ophthalmology
• Otolaryngology (ENT)
• Browse content in Paediatrics
• Neonatology
• Browse content in Pathology
• Chemical Pathology
• Clinical Cytogenetics and Molecular Genetics
• Histopathology
• Medical Microbiology and Virology
• Patient Education and Information
• Browse content in Pharmacology
• Psychopharmacology
• Browse content in Popular Health
• Caring for Others
• Complementary and Alternative Medicine
• Self-help and Personal Development
• Browse content in Preclinical Medicine
• Cell Biology
• Molecular Biology and Genetics
• Reproduction, Growth and Development
• Primary Care
• Professional Development in Medicine
• Browse content in Psychiatry
• Addiction Medicine
• Child and Adolescent Psychiatry
• Forensic Psychiatry
• Learning Disabilities
• Old Age Psychiatry
• Psychotherapy
• Browse content in Public Health and Epidemiology
• Epidemiology
• Public Health
• Browse content in Radiology
• Clinical Radiology
• Interventional Radiology
• Nuclear Medicine
• Radiation Oncology
• Reproductive Medicine
• Browse content in Surgery
• Cardiothoracic Surgery
• Gastro-intestinal and Colorectal Surgery
• General Surgery
• Neurosurgery
• Paediatric Surgery
• Peri-operative Care
• Plastic and Reconstructive Surgery
• Surgical Oncology
• Transplant Surgery
• Trauma and Orthopaedic Surgery
• Vascular Surgery
• Browse content in Science and Mathematics
• Browse content in Biological Sciences
• Aquatic Biology
• Biochemistry
• Bioinformatics and Computational Biology
• Developmental Biology
• Ecology and Conservation
• Evolutionary Biology
• Genetics and Genomics
• Microbiology
• Molecular and Cell Biology
• Natural History
• Plant Sciences and Forestry
• Research Methods in Life Sciences
• Structural Biology
• Systems Biology
• Zoology and Animal Sciences
• Browse content in Chemistry
• Analytical Chemistry
• Computational Chemistry
• Crystallography
• Environmental Chemistry
• Industrial Chemistry
• Inorganic Chemistry
• Materials Chemistry
• Medicinal Chemistry
• Mineralogy and Gems
• Organic Chemistry
• Physical Chemistry
• Polymer Chemistry
• Study and Communication Skills in Chemistry
• Theoretical Chemistry
• Browse content in Computer Science
• Artificial Intelligence
• Computer Architecture and Logic Design
• Game Studies
• Human-Computer Interaction
• Mathematical Theory of Computation
• Programming Languages
• Software Engineering
• Systems Analysis and Design
• Virtual Reality
• Browse content in Computing
• Business Applications
• Computer Games
• Computer Security
• Computer Networking and Communications
• Digital Lifestyle
• Graphical and Digital Media Applications
• Operating Systems
• Browse content in Earth Sciences and Geography
• Atmospheric Sciences
• Environmental Geography
• Geology and the Lithosphere
• Maps and Map-making
• Meteorology and Climatology
• Oceanography and Hydrology
• Palaeontology
• Physical Geography and Topography
• Regional Geography
• Soil Science
• Urban Geography
• Browse content in Engineering and Technology
• Agriculture and Farming
• Biological Engineering
• Civil Engineering, Surveying, and Building
• Electronics and Communications Engineering
• Energy Technology
• Engineering (General)
• Environmental Science, Engineering, and Technology
• History of Engineering and Technology
• Mechanical Engineering and Materials
• Technology of Industrial Chemistry
• Transport Technology and Trades
• Browse content in Environmental Science
• Applied Ecology (Environmental Science)
• Conservation of the Environment (Environmental Science)
• Environmental Sustainability
• Environmentalist Thought and Ideology (Environmental Science)
• Management of Land and Natural Resources (Environmental Science)
• Natural Disasters (Environmental Science)
• Nuclear Issues (Environmental Science)
• Pollution and Threats to the Environment (Environmental Science)
• Social Impact of Environmental Issues (Environmental Science)
• History of Science and Technology
• Browse content in Materials Science
• Ceramics and Glasses
• Composite Materials
• Metals, Alloying, and Corrosion
• Nanotechnology
• Browse content in Mathematics
• Applied Mathematics
• Biomathematics and Statistics
• History of Mathematics
• Mathematical Education
• Mathematical Finance
• Mathematical Analysis
• Numerical and Computational Mathematics
• Probability and Statistics
• Pure Mathematics
• Browse content in Neuroscience
• Cognition and Behavioural Neuroscience
• Development of the Nervous System
• Disorders of the Nervous System
• History of Neuroscience
• Invertebrate Neurobiology
• Molecular and Cellular Systems
• Neuroendocrinology and Autonomic Nervous System
• Neuroscientific Techniques
• Sensory and Motor Systems
• Browse content in Physics
• Astronomy and Astrophysics
• Atomic, Molecular, and Optical Physics
• Biological and Medical Physics
• Classical Mechanics
• Computational Physics
• Condensed Matter Physics
• Electromagnetism, Optics, and Acoustics
• History of Physics
• Mathematical and Statistical Physics
• Measurement Science
• Nuclear Physics
• Particles and Fields
• Plasma Physics
• Quantum Physics
• Relativity and Gravitation
• Semiconductor and Mesoscopic Physics
• Browse content in Psychology
• Affective Sciences
• Clinical Psychology
• Cognitive Psychology
• Cognitive Neuroscience
• Criminal and Forensic Psychology
• Developmental Psychology
• Educational Psychology
• Evolutionary Psychology
• Health Psychology
• History and Systems in Psychology
• Music Psychology
• Neuropsychology
• Organizational Psychology
• Psychological Assessment and Testing
• Psychology of Human-Technology Interaction
• Psychology Professional Development and Training
• Research Methods in Psychology
• Social Psychology
• Browse content in Social Sciences
• Browse content in Anthropology
• Anthropology of Religion
• Human Evolution
• Medical Anthropology
• Physical Anthropology
• Regional Anthropology
• Social and Cultural Anthropology
• Theory and Practice of Anthropology
• Browse content in Business and Management
• Business Ethics
• Business History
• Business Strategy
• Business and Technology
• Business and Government
• Business and the Environment
• Comparative Management
• Corporate Governance
• Corporate Social Responsibility
• Entrepreneurship
• Health Management
• Human Resource Management
• Industrial and Employment Relations
• Industry Studies
• Information and Communication Technologies
• International Business
• Knowledge Management
• Management and Management Techniques
• Operations Management
• Organizational Theory and Behaviour
• Pensions and Pension Management
• Public and Nonprofit Management
• Strategic Management
• Supply Chain Management
• Browse content in Criminology and Criminal Justice
• Criminal Justice
• Criminology
• Forms of Crime
• International and Comparative Criminology
• Youth Violence and Juvenile Justice
• Development Studies
• Browse content in Economics
• Agricultural, Environmental, and Natural Resource Economics
• Asian Economics
• Behavioural Finance
• Behavioural Economics and Neuroeconomics
• Econometrics and Mathematical Economics
• Economic History
• Economic Methodology
• Economic Systems
• Economic Development and Growth
• Financial Markets
• Financial Institutions and Services
• General Economics and Teaching
• Health, Education, and Welfare
• History of Economic Thought
• International Economics
• Labour and Demographic Economics
• Law and Economics
• Macroeconomics and Monetary Economics
• Microeconomics
• Public Economics
• Urban, Rural, and Regional Economics
• Welfare Economics
• Browse content in Education
• Adult Education and Continuous Learning
• Care and Counselling of Students
• Early Childhood and Elementary Education
• Educational Equipment and Technology
• Educational Strategies and Policy
• Higher and Further Education
• Organization and Management of Education
• Philosophy and Theory of Education
• Schools Studies
• Secondary Education
• Teaching of a Specific Subject
• Teaching of Specific Groups and Special Educational Needs
• Teaching Skills and Techniques
• Browse content in Environment
• Applied Ecology (Social Science)
• Climate Change
• Conservation of the Environment (Social Science)
• Environmentalist Thought and Ideology (Social Science)
• Natural Disasters (Environment)
• Social Impact of Environmental Issues (Social Science)
• Browse content in Human Geography
• Cultural Geography
• Economic Geography
• Political Geography
• Browse content in Interdisciplinary Studies
• Communication Studies
• Museums, Libraries, and Information Sciences
• Browse content in Politics
• African Politics
• Asian Politics
• Chinese Politics
• Comparative Politics
• Conflict Politics
• Elections and Electoral Studies
• Environmental Politics
• European Union
• Foreign Policy
• Gender and Politics
• Human Rights and Politics
• Indian Politics
• International Relations
• International Organization (Politics)
• International Political Economy
• Irish Politics
• Latin American Politics
• Middle Eastern Politics
• Political Behaviour
• Political Economy
• Political Institutions
• Political Theory
• Political Methodology
• Political Communication
• Political Philosophy
• Political Sociology
• Politics and Law
• Public Policy
• Public Administration
• Quantitative Political Methodology
• Regional Political Studies
• Russian Politics
• Security Studies
• State and Local Government
• UK Politics
• US Politics
• Browse content in Regional and Area Studies
• African Studies
• Asian Studies
• East Asian Studies
• Japanese Studies
• Latin American Studies
• Middle Eastern Studies
• Native American Studies
• Scottish Studies
• Browse content in Research and Information
• Research Methods
• Browse content in Social Work
• Addictions and Substance Misuse
• Adoption and Fostering
• Care of the Elderly
• Child and Adolescent Social Work
• Couple and Family Social Work
• Developmental and Physical Disabilities Social Work
• Direct Practice and Clinical Social Work
• Emergency Services
• Human Behaviour and the Social Environment
• International and Global Issues in Social Work
• Mental and Behavioural Health
• Social Justice and Human Rights
• Social Policy and Advocacy
• Social Work and Crime and Justice
• Social Work Macro Practice
• Social Work Practice Settings
• Social Work Research and Evidence-based Practice
• Welfare and Benefit Systems
• Browse content in Sociology
• Childhood Studies
• Community Development
• Comparative and Historical Sociology
• Economic Sociology
• Gender and Sexuality
• Gerontology and Ageing
• Health, Illness, and Medicine
• Marriage and the Family
• Migration Studies
• Occupations, Professions, and Work
• Organizations
• Population and Demography
• Race and Ethnicity
• Social Theory
• Social Movements and Social Change
• Social Research and Statistics
• Social Stratification, Inequality, and Mobility
• Sociology of Religion
• Sociology of Education
• Sport and Leisure
• Urban and Rural Studies
• Browse content in Warfare and Defence
• Defence Strategy, Planning, and Research
• Land Forces and Warfare
• Military Administration
• Military Life and Institutions
• Naval Forces and Warfare
• Other Warfare and Defence Issues
• Peace Studies and Conflict Resolution
• Weapons and Equipment

• < Previous chapter
• Next chapter >

2 Extended Cognition

Julian Kiverstein Academic Medical Center, Department of Psychiatry, University of Amsterdam, The Netherlands

• Published: 09 October 2018
• Cite Icon Cite
• Permissions Icon Permissions

The debates within 4E cognitive science surrounding extended cognition turn on competing ontological conceptions of cognitive processes. The embedded theory (henceforth EMT) and the family of extended theories of cognition (henceforth EXT) disagree about what it is for a state or process to count as cognitive. Advocates of EMT continue to interpret the concept of cognition along more or less traditional lines as being constituted by computational, rule-based operations carried out on internal representational structures that carry information about the world. EXT by contrast argues that bodily actions, and the environmental resources that agents act upon, can under certain conditions count as constituent parts of a cognitive process. I show how the debate between functionalist EXT and EMT ends in deadlock without any clear winner. I finish up by looking to radical embodied cognitive science for an alternative ontology of cognition that can provide grounds for favoring EXT over EMT.

Introduction

4E cognitive science is a broad church housing a number of theoretical perspectives that to varying degrees conflict with each other ( Shapiro 2010 ). In this chapter I will argue that the debates within 4E cognitive science surrounding extended cognition boil down to competing ontological conceptions of cognitive processes. The embedded theory (henceforth EMT) and the family of extended theories of cognition (henceforth EXT) disagree about what it is for a state or process to count as cognitive. EMT holds that cognitive processes are deeply dependent on bodily interactions with the environment in ways that more traditionally minded cognitive scientists might find surprising. The strong dependence of some cognitive processes on bodily engagements with the world notwithstanding, EMT claims that cognitive processes are nevertheless wholly realized by systems and mechanisms located inside of the brain. Thus advocates of EMT continue to interpret the concept of cognition along more or less traditional lines ( Adams and Aizawa 2008 ; Rupert 2009 ). That is to say, they think of cognitive processes as being constituted by computational, rule-based operations carried out on internal representational structures that carry information about the world.

EXT by contrast argues that bodily actions and the environmental resources that agents act upon can, under certain conditions, count as constituent parts of a cognitive process. Consider, for example, how thoroughly integrated mobile phones have become in those moments in our lives when we are left with our own thoughts. Chalmers describes how he uses his iPhone to daydream, “idly calling up words and images when my concentration slips” ( Chalmers 2008 , ix). Smartphones and other mobile technologies are so thoroughly interwoven in our everyday lives that according to EXT they might be now thought of as parts of our minds.

The debate between EMT and EXT is often taken to turn on the cognitive status or otherwise of bodily actions in which agents exploit the material and technological resources of their environments for cognitive purposes (see Rowlands 2010 , ch. 3; Wheeler 2014). Does thinking always take place entirely inside the head of individuals? Does it sometimes constitutively depend upon an agent’s coupled interactions with structures and resources found in the environment? I shall argue that to resolve this issue we need a mark of the cognitive ( Adams and Aizawa 2008 ; Wheeler 2010 ; Rowlands 2009 ). We need a theory of what makes a state or process a state or process of a particular cognitive kind.

The mark of the cognitive consists of properties a system must possess if it is to count as cognitive. Not everyone is agreed that there is any such well-defined set of properties. Clark (2008) has argued, for instance, that the processes and mechanisms that fall under the category of the cognitive are too disunified for there to be any distinguishing properties they share in common. Yet there remains a question to be settled about whether instances of cognitive processes, which seem to work in very different ways, count as instances of the same kind of process. We might try to answer this question by comparison with prototypes of a given cognitive process such as learning, memory, categorization, decision-making, and so on. We might appeal to folk intuition as Clark and Chalmers (1998) propose. Either way, we are assuming a position on what makes a process count as a process of a particular cognitive kind. We might be drawing our standard from folk psychology or relying on some other standard to identify prototypical examples of cognition. In either case, we are relying at least tacitly on a mark of the cognitive.

The mark of the cognitive is also at the heart of a debate within EXT about the nature of extended cognition. One side in this debate makes the case for EXT on the basis of considerations drawn from functionalism in the philosophy of mind. I will label this position “extended functionalism” (abbreviated as FEX). FEX is in agreement with the cognitive science orthodoxy that cognitive processes are essentially computational in nature ( Clark 2008 ; Wheeler 2011a ). FEX departs from the cognitive science orthodoxy in arguing that some of the relevant computations take place in the world, through bodily actions on information-bearing structures located in the environment. 1

The self-declared “radical” theorists of extended cognition (henceforth REX) propose an alternative explanatory framework to that of classical cognitive science drawn from dynamical systems theory and ecological psychology ( Chemero 2009 ; Silberstein and Chemero 2012 ; Hutto and Myin 2013 ). REX claims that we find extended cognitive processes whenever the variables that describe one system are also the parameters that determine change in the other system, and vice versa. In such a system, it is only as a matter of explanatory convenience that we treat the agent and its environment as separately functioning systems. In reality the dynamics of the two systems are so tightly correlated and integrated that they are best thought of as forming a single extended brain–body–world system ( Silberstein and Chemero 2012 ).

REX claims that basic forms of cognitive processes are essentially extended. Basic cognition is the type of cognition found in non-language-using creatures ( Hutto and Myin 2013 ). It is nonrepresentational and unfolds over time through the skilled bodily engagements of agents with the affordances of the environment. The terminology of “extended” cognition is thus potentially misleading insofar as it seems to imply that cognitive processes have their home inside of the heads of individuals, and occasionally reach out into the world. The extendedness of cognitive processes is to be understood not only in a spatial sense as a claim about the location of the boundaries of the mind. It also refers to the relational character of basic cognitive processes. Basic cognition is relational in the sense of being constituted by an agent’s skilled activity in relation to its environment ( Hutto et al. 2014 ).

My argument will proceed in two stages. In the first stage (first and second sections), I outline the debate between EXT and EMT. In the first section I show how there is substantial agreement in both camps about how cognitive science is to proceed. Both sides agree that the best explanation of human problem-solving will often make reference to bodily actions carried out on externally located information-bearing structures. The debate is not about how to do cognitive science. It is instead, to repeat, a debate about the mark of the cognitive: the properties that make a state or process count as being of a particular cognitive kind. In the second section, I then turn to functionalist formulations of EXT that make appeal to what has come to be called the parity principle. The third section shows how after many twists and turns the debate has reached a deadlock, notwithstanding arguments to the contrary recently developed on either side of the EXT-EMT divide. I then turn my attention to REX and argue that EXT would fare better were it to drop its commitment to a representationalist mark of the cognitive.

The EMT and EXT Debate

Since Descartes’s skeptical arguments in the Meditations it has seemed natural to many philosophers to think of mind and cognition as essentially inner phenomena. The mind partakes in causal transactions with the world by means of epistemically, more or less reliable sensory channels. The mind can likewise produce effects in the world by sending commands to the muscle systems in the body to move in particular ways. However, mental and cognitive processes such as perceiving, remembering, thinking, and reasoning take place within the minds of individuals. EMT departs from this Cartesian tradition in stressing the ways in which mental processes causally depend on the environment in which the agent is embedded in deep and surprising ways. 2 When, for instance, we use a calculator to divide a bill in a restaurant, the calculator is an essential part of how we successfully compute a solution to an otherwise computationally challenging arithmetical problem. The calculator “scaffolds” mathematical thinking that is fully constituted and realized by causal mechanisms found within a person’s brain. 3

EXT is even more thoroughgoing in its rejection of the Cartesian legacy. EXT claims that cognitive processes can, under certain conditions, extend or spread across the boundary separating the agent’s body from the rest of the world. Consider, by way of illustration, the expert bartender who lines up different glasses in a particular spatial order as he prepares a drinks order ( Beach 1988 ). 4 This simple trick makes the bartender’s task of remembering which drink to serve next far easier than it would otherwise be. Instead of needing to store all of this information and keep it in mind, some of the work of remembering is offloaded onto the environment in the line of glasses. The environment now functions as an external store of information, and performs the role of a stand-in for the drinks order. To work out which drink to serve next, the bartender need only look and reach for the next glass in the line. The information-bearing load on his working memory is thereby significantly lightened. Part of this work is delegated to the representational structure temporarily assembled in the world, which can then be used to control and guide action so as to bring the task at hand to successful completion.

The bartender’s initial action of arranging the line of glasses is what David Kirsh has called an “epistemic action.” It is an action that gives structure to the information processed by internal cognitive systems in ways that fit with the goals of the system. The result of this active structuring of information is that now the agent can couple with the external structure (the line of glasses), and through this coupling gather the information needed about the next drink to be served. When the bartender generates a plan to serve the next drink, he does so on the basis of his coupling with this external structure. It is on the basis of such a coupling that the information is generated necessary for successfully planning and accomplishing his task. 5

It might be naturally objected that this is just another example of the scaffolding of internal cognitive processes by the environment. EXT and EMT can both agree that the line of glasses functions as an external store of information that is tightly integrated with inner perceptual, working memory, and attentional processes in such a way as to guide and control action. EXT claims that the external structure works together with inner cognitive processes to form a softly assembled cognitive system that brings about the bartender’s behavior. 6 Describing the system as “softly assembled” marks a contrast with systems made up of component parts, each of which has a pre-specified and fixed function ( Anderson et al. 2012 ). Softly assembled cognitive systems are characterized by an “interaction-dominant dynamics,” which makes it difficult, if not impossible, to assign specific functions to specific component parts. For instance, each time the bartender looks to the row of the glasses, this delivers the systems responsible for planning the next action with just the information they need. The systems that are planning the bartender’s actions are simultaneously influencing and being influenced by the perception-action systems that are sustaining the coupling with the environment. These systems stand in a relation of mutual and continuous causal influence on each other.

As already noted in my introduction, the dispute between EMT and EXT is typically taken to concern the cognitive status of external information-bearing structures and the bodily actions that are performed on those structures. EXT claims that environmentally located structures and resources, and the operations that are carried out on them form constitutive parts of a cognitive process. EMT agrees that coupling to the environment can contribute in an ongoing and interactive way in the production of a cognitive phenomenon of interest. However, proponents of this theory argue that the contribution of such couplings to cognitive processes is best understood as causal, not as constitutive.

This disagreement notwithstanding, there is, however, much that EMT and EXT are agreed upon. Both theories agree that the explanation of how internal cognitive processes give rise to some behavior of interest will often need to advert to bodily actions on external, environmentally located structures. Embedded and extended theorists therefore agree that internal cognitive processes will often not be sufficient for explaining cognitive behaviors. Given the extent of this agreement about how to go about explaining many of our problem-solving behaviors, one might be forgiven for wondering what is really at stake in this debate.

Sprevak (2010) has argued, for instance, that the dispute is unlikely to make a difference to how cognitive scientists go about their everyday business. The two theories can both equally well accommodate the available experimental evidence. 7 Cognitive scientists could frame their theories either in embedded terms or in extended terms. It would make little or no difference when it comes to the explanatory value of the resulting theories. Thus EMT and EXT do not seem to be genuinely competing theories when judged from an empirical perspective by the experimental data each theory can explain. Moreover, it is far from clear that any explanatory advantage is really gained from labeling the construction and manipulation of environmental structures as parts of a “cognitive” process. If this is what the debate is about, it has all the hallmarks of being a merely semantic disagreement.

I agree with Sprevak, however, that the debate between EXT and EMT isn’t about the best conceptual framework for interpreting findings in cognitive science. It is a debate in metaphysics about “what makes a state or process count as mental or non-mental” ( Sprevak 2010 , p. 361). 8 For instance, the two theories fundamentally disagree about the body and world and their role in cognitive processes. EXT casts the body in the role of a tool for mediating between neural processes and the intelligent use of the environment. The body of the agent is what Clark describes as “a bridging instrument” that enables “the emergence of new kinds of distributed information-processing organization” ( Clark 2008 , p. 207). EMT argues by contrast that it is the body and world as represented in the brain that plays a necessary part in problem-solving behavior.

In the next section I explain how EXT has been defended by appeal to the so-called parity principle by defenders of extended functionalism (FEX for short). FEX claims that bodily action on external information-bearing structures is one of the many ways in which the computational processes that underpin cognitive processes can be implemented. The human brain has a rich variety of modes of encoding and processing information, some of which involve constructing and acting on information-bearing structures located in the environment. 9 We shouldn’t treat instances of problem-solving differently when the agent makes active use of resources located in the environment. We will see, however, that the appeal to parity and the equality of treatment for the inner and outer have been found to be less than persuasive by advocates of EMT. In the third section, I will argue this has led to stalemate in the debate, suggesting that EXT might be in need of new, more radical ideas.

The Parity-Based Defense of EXT

The parity principle was first formulated by Clark and Chalmers in a paper that initiated the debate about the extended mind as we know it today ( Clark and Chalmers 1998 ). Here is how they formulated the parity principle:

If as we confront some task, a part of the world functions as a process which, were it to go on in the head, we would have no hesitation in accepting as part of the cognitive process, then that part of the world is (for that time) part of the cognitive process. (p. 8)

Consider how the parity principle might apply to the bartender example discussed in the previous section. Instead of physically arranging the glasses in the world, suppose instead that the bartender visually imagines the same line of glasses. He then keeps in mind this visual image, accessing it when he needs to, until the order is completed. Now, most of us would, I guess, be willing to say that such a visual image would count as a part of the cognitive process that causes the bartender’s behavior. However, if we say this of the visual image of the line of glasses, then surely we ought to say the same of the actually existing line of glasses in the world. The visual image is nothing but an inner reconstruction of the same physical structure in the world. All I have done in constructing this example is transpose a process that in the original example takes place partly in the world into one that instead takes place wholly inside the head. The parity principle says that if we count a process as cognitive when it takes place inside the head, we should also count it as cognitive when it extends into the world. A cognitive process ought to be counted as cognitive regardless of where (inside the head or out in the world) it takes place.

Taken as a self-standing principle, the parity principle doesn’t settle anything. It works as an argument for EXT only when taken in conjunction with some pre-existing conception of when a process counts as a cognitive process ( Adams and Aizawa 2001 ; Rupert 2009 ; Wheeler 2011b ; Walter and Kästner 2012). After all, it is the cognitive status of a process that partly takes place in the world that we are using the parity principle to try to settle. In order to apply the parity principle, we must therefore have some pre-existing standards for making judgments about which processes are cognitive and which are not. We must have some pre-existing philosophical theory of what makes a state or process count as a state or process of a particular cognitive kind. 10

Clark and Chalmers answer this question in part on the basis of considerations drawn from commonsense psychology, and in part by reference to cognitive science. Consider once again Clark and Chalmers’s infamous case of Otto who because of his Alzheimer’s relies on a notebook to remember the location of the Museum of Modern Art (MOMA) in New York ( Clark and Chalmers 1998 ). The notebook can do the work of storing information just as well as the brain can. It doesn’t matter that information is encoded, stored, and recalled very differently in the Otto notebook system as compared with Inga, who recalls the location of MOMA using her biological memory. The entries in Otto’s notebook causally interact with his perception, beliefs, desires, and behavior in many of the same sorts of ways as the memory states of Inga. The Otto notebook system and the declarative memory systems in Inga’s brain can thus be seen as different physical realizations of functionally equivalent dispositional belief states. They are instances of the same type or kind of mental state because they play the same coarse-grained, action-guiding role.

Proponents of EMT have, however, queried this last move by pointing to significant fine-grained functional differences in how memory works in Inga and Otto. Otto, for instance, would likely show no difference when it comes to remembering items at the beginning or end of a list of items as compared with those that occur in the middle of a list ( Adams and Aizawa 2008 ). 11 Functional differences like these, and there are many others, give us grounds for doubting that extended and inner cognitive processes really do count as functionally equivalent.

This objection is not based on the claim that fine-grained similarity of internal and extended cognitive processes is necessary for the application of the parity principle. 12 The objection is rather based on a claim about what makes a state or process count as an instance of a particular cognitive kind. EMT takes cognitive processes to have a nature that is determined by causally explanatory properties, identified by our best theories in cognitive science. It is for this reason that they pay close attention to the details of the causal roles played by cognitive processes in humans.

Clark and Chalmers applied the parity principle according to standards that were in part based on common sense. They take ordinary folk to identify the causal properties that determine the nature of a given cognitive state in the types of explanations they give when making sense of one another as rational agents. Proponents of FEX more generally promote an attitude of maximal inclusiveness in the range of non-standard realizers we count as possible realizers of the mental. They encourage us to be as liberal as possible in the creatures we count as being minded like us ( Clark 2008 ; Sprevak 2009 ; Wheeler 2010 ). AIs, robots, and other creatures of science fiction all count as having states that mediate between inputs and outputs in ways that are similar enough to the states appealed to in folk psychological explanations. All of these systems have states that guide action in roughly the same type of way as the states picked out in folk psychological explanations.

The debate between EMT and EXT (in its functionalist formulations) is at least in part a debate about which theories we appeal to in fixing the reference of our cognitive concepts. These theories tell us whether there is sufficient similarity between two instances of a cognitive process for both to count as tokens of the same type of process. Some have allowed for folk psychological explanation to do the work of fixing the reference of our cognitive concepts. Others have argued that folk psychology is irrelevant and have instead made appeal to scientific theories and explanations to identify causally explanatory properties. It is to these questions that I turn in the next section.

The Varieties of (Extended) Functionalism

Clark and Chalmers suggested we look to folk psychology to decide whether an inner (e.g., Inga) and extended (e.g., Otto) cognitive process count as tokens of the same type of cognitive process. Folk psychology, as has often been noted, doesn’t guarantee an answer to this question that favors EXT. As Chalmers (2008) has noted, folk psychology gives us some reason to treat perception and action as marking the boundaries of the mind. The only way Otto has of retrieving information from his notebook is by means of perception and action. It might then be argued based on folk psychology, that Otto does not have any beliefs before he checks to see what is written in his notebook. The notebook functions as at best an environmental cue for the formation of internal mental representations. Sure, Otto depends on his notebook for the guidance of his behavior, but conceding this is consistent with his behavior being in part the outcome of beliefs located entirely inside of his head about the contents of his notebook. These are beliefs that Otto can form only by reading what is written in his notebook. 13 Folk psychology therefore gives us some reason to treat the notebook as lying outside of the boundaries of the cognitive system. 14

One might wonder in any case how much trust one should put in intuitions drawn from folk psychology. Folk psychology may invite us to count the states of Otto and Inga as states of the same kind by virtue of their action-guiding role (though there is room for debate on this point). Playing a similar role in guiding action is, however, not a scientifically illuminating causal property. We want to know more precisely what the action-guiding causal role is in virtue of which Otto and Inga count as sharing a state or process of the same kind. There is little reason to think folk psychology will prove a useful guide when it comes to identifying causally explanatory properties. Indeed the judgments that folk psychology yields may well encourage us to mistakenly lump together states and processes that empirical science distinguishes, and conversely to make distinctions where none are found in nature.

It is this type of reasoning that has led parties on both sides of the debate to look instead to cognitive science to identify the causally explanatory properties that make a state or process the type of state or process it is. What verdict would we reach if we individuated functional roles instead on the basis of our best empirical scientific theories? Do such theories allow for kinds of cognitive processes that sometimes extend into the world beyond the boundaries of the organism? Once again, there is no clear consensus.

Wheeler has argued in a series of papers for an affirmative answer to this question. He agrees that our empirical theories may well point to many functional differences between extended and internal memory processes. However, he asks us to imagine that cognitive psychologists found people whose internal memory processes exhibited the same functional differences as the Otto notebook system ( Wheeler 2010 ). Wheeler thinks the psychologists would still classify the people in question as having declarative memory processes so long as they exhibited, for instance, “context-sensitive storage and retrieval of information” ( Wheeler 2011b ). He thus doubts that the existence of functional differences speak against treating extended and inner memory processes as processes of the same generic kind.

Why should we count extended and inner memory processes as instances of a generic kind of declarative memory? There is some debate within EXT about whether we must do so because generic cognitive processes share common underlying mechanisms. As we saw in the introduction, Clark argues that extended cognitive processes may prove to be fundamentally disunified, a motley of different mechanisms. The methodological moral of EXT for cognitive science would therefore be to let a thousand flowers blossom. Although Sutton (2010) presents himself as making the case for extended cognition on the basis of computational cognitive science, he can also be read as falling within this first camp. Sutton has argued for a second wave in EXT that stresses the functional differences between inner and outer problem-solving resources. Second-wave EXT argues that external structures like Otto’s notebook complement the cognitive capacities of the biological brain, joining forces with them to deliver new hybrid cognitive systems that are part biological and part cultural. 15 A key claim of second-wave EXT is that the scientific study of intracranial, or inner cognitive processes is just one part of cognitive science. The scientific study of extended cognitive systems is, Sutton argues, also a central research question in cognitive science, often undertaken in collaboration with the social sciences.

The complementarity of heterogeneous inner and outer elements that is emphasized by the second wave seems to have the consequence that extended and intracranial cognitive systems are unlikely to be built on the basis of the same mechanistic principles. On the contrary, the point of stressing complementarity is that “in extended cognitive systems, external states and processes need not mimic or replicate the formats, dynamics, or functions of inner states and processes. Rather, different components of the overall (enduring or temporary) system can play quite different roles and have different properties” ( Sutton 2010 , p. 194).

The alternative empirical functionalist view defended by Wheeler requires that the underlying mechanisms that support generic memory must at least share in common some family resemblance ( Wheeler 2011b ). 16 Wheeler allows for a variety of mechanisms, some of which are dynamical, nonrepresentational, and noncomputational, others of which are computational and representational ( Wheeler 2005 ; Wheeler 2011b ). However, he argues that these mechanisms are nevertheless partially unified: either as computational and representational processes on the one hand, or as dynamical and nonrepresentational on the other.

Trouble lies in waiting for FEX in playing this empirical functionalist card. First, cognitive science licenses a number of different theory-loaded accounts of the cognitive, not all of which support EXT. Wheeler proposes the physical symbol systems hypothesis of Newell and Simon (1976) as an uncontroversial account of the nature of cognitive processes that he thinks every party in the EXT-EMT debate ought to be able to accept. 17 We can think of Otto together with his notebook as forming “a sufficiently complex and suitably organized physical symbol system” ( Wheeler 2011a , p. 236). 18 Both the Otto notebook system and Inga can thus be interpreted as physical symbol systems. While there are no doubt significant differences in the mechanisms that realize declarative memory in Otto and Inga, there is nevertheless a family resemblance. Both can be described as physical symbol systems.

Conceding this much doesn’t settle the issue of whether to describe the notebook as making a cognitive as opposed to a merely causal contribution to Otto’s memory processes. Take Robert Rupert as an example of a proponent of EMT; 19 he could no doubt agree with Wheeler’s proposal to use Newell and Simon as a scientifically informed account of the cognitive. However, he is well known for his skepticism about the concept of generic memory as a scientifically well-formed category (Rupert 2004 , 2009 , 2013 ). Rupert argues that science would only treat Otto and Inga as instantiating a generic kind of declarative memory if memory in Otto and Inga was brought about by the same cluster of integrated and persisting mechanisms. The Otto notebook system arguably does form a cluster of integrated and persisting mechanisms. Thus at first glance it seems to pass one of Rupert’s tests. However, before we celebrate the victory of EXT, we should note that the clusters of mechanisms that are the basis for remembering in Otto are very different to those of Inga. Rupert argues that we would only be warranted in attributing to Otto states of the same kind as Inga on the following condition. Our model of how Otto’s behavior is produced would have to overlap significantly with our model of the mechanisms that bring about Inga’s behavior. Rupert argues that there is no reason to think this will be the case.

This conclusion doesn’t settle the empirical functionalist case in Rupert’s favor. For it should be noted that any mechanistic system can be described at many levels of organization ( Craver 2007 ). Wheeler could argue that extensive lower-level implementational differences may disguise from view significant higher-level functional similarities. However, such a reply doesn’t take us far. Suppose a proponent of EMT were to concede, as I think they should, that declarative memory processes allow for information to be stored either internally (as in Inga) or externally (as in Otto). 20 Consistent with this concession, they could nevertheless insist against EXT that the contribution of the external elements is only causal, and not cognitive. Memory processes so conceived can take the form of hybrid processes made up of causally interacting elements, some of which are cognitive and some of which are noncognitive. 21 Such a version of EMT wouldn’t yield a genuinely competing explanation to EXT ( Sprevak 2010 ). It would agree with EXT that memory processes can take a variety of forms, some of which are wholly internal and others of which are environment-involving. Disagreement would persist over whether to count the environmental components as constituent parts of the memory process.

The upshot of all this is that empirical functionalism leaves us pretty much where we started. In order to bring science to bear on this debate, we must settle the philosophically prior issue of what makes a state or process count as a cognitive state or process of a particular kind. The empirical functionalist answers this question by looking to the causal properties of a state or process as identified by our best scientific theories. However, we have just seen that cognitive science might be taken to yield an answer to this question that doesn’t decide between EXT and EMT. The causal properties in putative cases of extended cognition can be interpreted as hybrid: part cognitive, and part noncognitive. We thus have two possible descriptions. One favors EMT (the description of the system as a hybrid of cognitive and noncognitive elements); the other favors EXT. Science on its own doesn’t seem to allow us to decide which is the better description. To settle the matter we need a mark of the cognitive: a philosophical theory of what it is that makes a state or process a cognitive state or process.

Should we conclude then that we have no alternative but to rely on folk psychology? This would seem to follow on the assumption that our philosophical intuitions about the nature of the mind have their basis in folk psychology. We’ve seen, however, that folk psychology will prove to be of only limited help in making the case for EXT. It can help us to form a pre-theoretical sense of what states and processes stand in need of explanation. However, if we assume it is a state or process’ causally explanatory properties that make it the state or process that it is, folk psychology cannot help us with our original question. It is through scientific investigation that we will learn about causally explanatory properties, not by recourse to folk psychological intuition.

This all leaves us in a rather unsatisfying place. Scientific findings need to be given a philosophical interpretation if they are to settle the issue of which causally explanatory properties are constitutive of a given cognitive state or process. We need a theory that tells us which causal properties count as cognitive, and which do not. However, if a theory is to do this work for us it must be based on scientific findings that identify causally explanatory properties. Philosophical theorizing needs scientific grounding, but scientific theorizing needs philosophical interpretation. Neither empirical functionalism nor commonsense functionalism succeeds in providing us with the philosophical account of the cognitive we need to settle the debate between EXT and EMT. It is time to try something different.

Taking the Radical Option

Chemero (2009) writes that radical embodied cognitive science (henceforth RECS) is “a variety of extended cognitive science” (p. 31). He characterizes this branch of cognitive science as having its roots in the American pragmatist tradition of William James, John Dewey, George Mead, and Charles Peirce. 22 A theme that looms large in the work of these philosophers (and also in RECS) is the mutuality of animal and environment. Here is Dewey explaining the central idea:

To see the organism in nature, the nervous system in the organism, the brain in the nervous system, the cortex in the brain is the answer to the problems which haunt philosophy. And when seen thus they will be seen to be in, not as marbles are in a box but as events are in history, in a moving, growing, never finished process. ( Dewey 1958 , p. 295)

To say that animal and environment stand in a relation of mutual dependence is to claim that animal and environment are interdependent in the sense of together forming a “moving, growing, never finished process.” The connection of RECS to this older tradition in naturalistic thinking comes from the thesis that if we model the agent and environment as coupled dynamical systems, then it is only as a matter of convenience that we treat them as separate systems. Instead of describing how external environmental factors cause changes in the agent’s behavior, we instead model how the whole agent-environment system as a single process changes over time.

The argument for this claim is based in part on the mathematics of dynamical systems theory. Dynamical systems theory models change over time in complex systems using the mathematics of differential equations. Examples of complex systems are the solar system, weather systems, the diffusion of ink in water, interaction of populations of predator and prey, and so on. 23 The key concept we will need for the arguments that follow is that of the “coupling” of the agent and environment. Two systems S 1 and S 2 are said to be coupled when the equations describing one system S 1 contain variables whose value is a function of the variables in the equations describing S 2 , and vice versa. Thus take the example of dynamical systems description of rate of change in a population of predators and prey. The equations describing a population of predators will include variables for prey, and the equations describing change in the population of prey will include variables for predators. 24 For example, if the number of predators steadily increases, the number of prey will steadily decrease, thereby putting pressure on the predators. As the predators begin to die off, the prey can begin to recover, and this dynamic will continue until the two populations reach some sort of equilibrium, and the size of each population remains relatively stable.

Now if we apply the concept of coupling to agent–environment interactions, we get the following result. We have two equations: one describing the changes that take place in the agent, and the second describing the changes that take place in the environment. The variables in the respective equations describe how components of the agent and the environment change in relation to each other. For the agent, some of the components are located in the brain, others in the rest of its body relating, for instance, to its bodily movements or affective states. On the environment side, the components will in cases of extended cognition be the information-bearing structures that the agent makes use of in the performance of a cognitive task. The agent and environment are dynamically coupled in cases of extended cognition because the equation describing change in the environment contains variables whose values are determined by the changes taking place in the agent. Similarly, the equations describing change in the agent contain variables whose values are determined by change in the environment.

Two systems that are coupled resist decomposition into separately functioning systems. We cannot model the behavior of the system as the additive product or sum of the interactions of separate structures and components, some on the side of the agent, and others on the side of the environment. Such is the degree of continuous, integrated, and coordinated mutual influence between the two systems that we can’t solve the equations describing the behavior of each system separately. 25 There are really two claims here that combine to yield the result that the agent–environment system is best described as a single system. First, the components that make up the agent system exhibit fluctuating rates of change that depend on components belonging to the environment system, and vice versa. Second, if we look at the behavior of the agent–environment as a whole, this behavior isn’t the product of the behavior of each of the components. We must also look at the interaction of the components and the nonlinear effects that arise from those interactions due to the continuous causal influence of the components on each other.

Clark has also made extensive use of these types of considerations in arguing for extended cognition. He characterizes the interactions between internal and external resources as “highly complex nested and nonlinear.” He continues:

As a result, there may, in some cases, be no viable means of understanding the behavior and potential of the extended cognitive ensembles by piecemeal decomposition and additive reassembly. To understand the integrated operation of the extended thinking system created, for example, by combining pen, paper, graphics programs, and a trained mathematical brain, it may be quite insufficient to attempt to understand and then combine (!) the properties of pens, papers, graphics programs and brains. ( Clark 2008 , p. 116)

Clark retains a commitment to representational and computationalist explanation. This renders his appeal to dynamic coupling vulnerable to Rupert’s objections ( Rupert 2009 , ch. 7). Rupert argues that in a range of cases in which dynamical systems theory is used to model cognitive behaviors, we do not find coupling. We don’t find variables for environmental elements showing up in the equations describing the agent’s behavior. Instead we find the environment causing variation “among a small number of dimensions (e.g., input units) of the organismic system” (2009, p. 136). It is the value of these internal organismic systems that then determine how the agent behaves, not the states of the external environment.

Rupert’s objection depends for its success on his conception of cognitive processes as persisting and integrated sets of mechanisms that causally contribute in the production of a wide range of cognitive phenomena ( Rupert 2009 , ch. 3). Armed with such a conception of cognitive systems, he can argue that agent–environment interaction leads to changes in the agent’s internal representational states. The agent is sensitive to such changes in its internal representational states, and on the basis of this sensitivity it adapts its behavioral outputs so as to accomplish its tasks. The agent’s interaction with the environment is causally relevant only through the changes it brings about in the agent’s internal representational states.

REX argues by contrast that the agent doesn’t interact with the environment through the intermediary of internal representations. It is not only dynamical systems theory and the concept of coupling that does the argumentative work but also crucially ideas drawn from ecological psychology, or so I shall propose. Ecological psychologists show how the layout of the ecological niche of a given species of animal is rich with higher-order, structural invariants that specify affordances, and which the mobile animal is able to immediately and directly detect. Warren summarizes the idea well:

The perceptual system simply becomes attuned to information that, within its niche, reliably specifies the environmental situation and enables the organism to act effectively. ( Warren 2005 , p. 358)

Perceptual systems function first and foremost to guide action. The perceiving animal is immediately and directly sensitive to higher-order invariants or patterns in sensory stimulation that specify affordances, the possibilities for action provided to an animal by its surrounding environment. Interaction with the environment produces patterns of energetic stimulation, which form the basis for directly and immediately detecting higher-order invariants that specify affordances. As an agent approaches the edge of a precipice there is an immediate shearing off of the texture of the ground of the supporting surface ( Gibson 1969 ) and the perceiver immediately detects that here is a place that offers the potential to fall. This is a meaning that is carried in the light that reflects from this place. This type of informational regularity can be thought of as an ecological constraint under which the perceptual systems of animals evolved. Thus the tusk of the narwhal is “tuned to the salinity differentials that specify the freezing of the water’s surface overhead,” information that is critical to its survival ( Warren 2005 , p. 341).

With the ecological context in place, the argument from coupling looks a little different. It can be argued that the environment doesn’t causally influence behavior only by means of internal representations. Interaction with the environment isn’t only about the delivery of afferent stimuli that can be used by the brain to construct internal representations. It can be argued instead that the agent dynamically couples with information-bearing structures located in the environment. Agent and environment exert continuous and mutual causal influence on each other making it the case that agent and environment cannot be modeled as separate systems. They are instead best modeled as a single extended cognitive system.

Rupert, however, has another argument up his sleeve. He claims that even if we grant that the external environment can causally influence behavior, this still doesn’t suffice for coupling. For the direction of causal influence is only one-way: from environment to agent and not the other way around ( Rupert 2009 , p. 136). The environment may make a difference to behavior sometimes, but the agent makes no difference to the environment. For example, when in a game of Tetris the subject rotates the zoid in order to see into which space it might fit, “the fundamental dynamics of the object are not changed: its evolution in state space from any given point remains the same as it was before the rotation” ( Rupert 2009 , p. 136).

What exactly does Rupert mean when he claim that the dynamics of the zoid remain the same before and after rotation? The player’s rotation of a zoid is a now-classic example of an epistemic action. Players perform this action in order to recognize the shape of a zoid, and to verify whether a given orientation will help to fill a line or not. Rotation therefore influences the spatial path of the falling zoid and the place it comes to rest. The path the zoid traverses and its orientation when it finds its resting place all clearly influence play and the arrangement of the pieces on the board. Thus we have what looks like a case of coupling: the equations describing the game will include variables for the player’s action of rotation. The equations describing the player’s actions will include variables for the rotation of the zoid. Recognition of the shape is facilitated by the act of rotation.

Perhaps, however, Rupert has something more demanding in mind when he talks of coupling. He describes the coupling relation as holding when “an order parameter of one subsystem acts as control parameter of the other, and vice versa; as a result, one subsystem’s evolution can change the very character of the evolution of the other” ( Rupert 2009 , p. 133). A control parameter is a value whose continuous quantitative change leads to qualitative change in the behavior of the system. An example is the temperature of a fluid: a difference in this parameter can change dramatically how the fluid changes over time from an initial state. An order parameter is a composite or macroscopic state of a system such as the convection patterns or Bénard rolls that can be seen in a viscous fluid, such as oil when it reaches a certain temperature. In the Tetris example, it is the player’s action of rotating the zoid that is the candidate for the control parameter. It is tricky to say what would count as the order parameter of the game. Let us suppose that it is the overall configuration of the pieces on the board at a particular moment in the game. We can see why Rupert would think the order parameter so conceived isn’t affected by rotation. The pieces do not suddenly change their position when one rotates.

Notice, however, that this is to describe the game at a single point in time. Kirsh’s classic research on Tetris shows that rotation when done early enough in the game does have a sizable influence on the player’s success. Without rotation the layers of the board would fill up much sooner. With rotation the player succeeds in filling more lines and thus playing longer. Suppose we are given the task of predicting how a given game is going to play out. We would need to take into account whether the player used the strategy of rotation, and when they chose to do so. Compare this with the earlier example of populations of predators and prey. The growth in the population of prey, for instance, is dependent on the effects of predation. Similarly, how fast the layers of the board fill up in part depends on the performance of the action of rotation.

Both of Rupert’s arguments fail. An obvious further objection, however, appeals to the concept of hybrid cognitive systems used so effectively against EXT earlier in the chapter. Why doesn’t an extended cognitive system count as a hybrid cognitive system composed of cognitive elements inside of the head of the agent, and noncognitive elements in the environment? ( Adams and Aizawa 2008 , ch. 7).

This objection assumes precisely the kind of decomposition that has been called into question earlier. The elements inside the head could realize cognitive processes only by representing what is outside the head. The debate between EXT and EMT (with its commitment to hybrid cognitive systems) thus turns on whether one takes representation to be the mark of the cognitive. I’ve argued this is a mark of the cognitive that EXT must reject.

The central claim of this chapter has been that to resolve the debate about extended cognition we will need to come up with a mark of the cognitive. We will need to say what makes a state or process count as a state or process of a particular cognitive kind. All sides in the EXT-EMT debate have supposed that we must answer this question by appeal to the causally explanatory properties of a state or process. However, of the causally explanatory properties, some may be only causally relevant and not constitutively relevant. 26 Externally located structures might be argued to fall in the category of causally relevant but not constitutively relevant properties. In order to come down on one side or the other in this debate, we will need to have some basis for deciding whether a causally explanatory property is constitutively relevant. This requires us to have a theory of which causal processes count as cognitive.

RECS may hold the key to breaking the stalemate that has been reached in the debate between EMT and EXT. Interaction with the environment cannot be argued to be of only causal relevance so long as agent and environment are exerting continuous mutual causal influence on each other. This mutual causal influence stands in the way of modeling agent and environment as separate, independently functioning systems.

Doesn’t the success of this argument depend on the empirical functionalist claim that it is causal explanatory properties that make a state or process an instance of a particular cognitive kind? Empirical functionalism has traditionally been aligned with the computational and representational theories of mind, a connection that REX seeks to break. Empirical functionalists take the states that mediate between inputs and outputs to be representational states with an internal structure to which computational processes are sensitive. This commitment to computational and representational explanation is arguably essential to empirical functionalism. One of the main selling points of the computer theory of mind was supposed to be that it can make it intelligible how a causal and mechanistic process can also be sensitive to semantic properties of thinking ( Crane 1995 ; Fodor 2000 ). The computer theory of mind was supposed to help us to understand how causal properties can constitute a cognitive process. REX, however, rejects the computer theory of mind, and stripped of this theoretical commitment, empirical functionalism has little to recommend it.

REX draws its mark of the cognitive from a variety of sources that spans the phenomenological and American naturalist tradition. REX takes extended cognitive systems to be perception-action systems on the basis of which the person or animal is adapted to its environment and so able to deal adequately with its affordances. REX is thus committed to a pragmatist interpretation of what cognition is, inspired by the mutual fit and complementarity of the animal and its environment. The argument I gave earlier for EXT stressed the importance of nonlinear causality and the dynamical properties of interaction-dominant systems. However, what makes all of this relevant to cognition in the end is the way in which dynamical properties of this kind relate to the mutuality of animal and its environment. It is this mutuality that grounds the mark of the cognitive needed to make a successful case for extended cognition.

Adams. F. and Aizawa, K. ( 2001 ). The bounds of cognition.   Philosophical Psychology , 14(1), 43–64.

Google Scholar

Adams, F. and Aizawa, K. ( 2008 ). The bounds of cognition. Oxford: Blackwell.

Google Preview

Adams, F. and Aizawa, K. ( 2010 ). Defending the bounds of cognition. In: R. Menary (ed.), The extended mind. Cambridge, MA: MIT Press, pp. 67–80.

Anderson, M.L. , Richardson, M.J. , and Chemero, A. ( 2012 ). Eroding the boundaries of cognition: implications of embodiment.   Topics in Cognitive Science , 4(4), 717–30.

Beach, K. ( 1988 ). The role of external mnemonic symbols in acquiring an occupation. In: M.M. Gruneberg and R.N. Sykes (eds.), Practical agency of memory. New York: Wiley, pp. 342–6.

Chalmers, D. ( 2008 ). Forword. In: A. Clark Supersizing the mind: embodiment, action and cognitive extension. Oxford: Oxford University Press, pp. ix–xvi.

Chemero, A. ( 2009 ). Radical embodied cognitive science. Cambridge, MA: MIT Press.

Clark, A. ( 2001 ). Reasons, robots and the extended mind.   Mind and Language , 16(2), 121–45.

Clark, A. ( 2008 ). Supersizing the mind: embodiment, action and cognitive extension. Oxford: Oxford University Press.

Clark, A. ( 2011 ). Finding the mind.   Philosophical Studies , 152, 447–61.

Clark, A. ( 2012 ). Spreading the joy? Why the machinery of consciousness is (probably) still in the head.   Mind , 118(472), 963–93.

Clark, A. ( 2015 ). Surfing uncertainty: prediction, action and the embodied mind. Oxford: Oxford University Press.

Clark, A. and Chalmers, D. ( 1998 ). The extended mind.   Analysis , 58, 7–19.

Colombetti, G. and Krueger, J. ( 2015 ). Scaffoldings of the affective mind.   Philosophical Psychology , 28(8), 1157–76.

Crane, T. ( 1995 ). The mechanical mind: a philosophical introduction to minds, machines and mental representation. London: Routledge, Taylor & Francis.

Craver, C. ( 2007 ). Explaining the brain: mechanisms and the mosaic unity of neuroscience. Oxford: Oxford University Press.

Dewey, J. ( 1958 ). Experience and nature. New York: Dover Books.

Dotov, D. , Nie, L., and Chemero, A. ( 2010 ). A demonstration of the transition from readiness to hand to unreadiness to hand.   PLoS ONE, 5, e9433. doi:10. 1371/journal.pone.0009433

Fodor, J.   2000 . The mind doesn’t work that way: the scope and limits of computational psychology. Cambridge, MA: MIT Press.

Gallagher, S. ( 2014 ). Pragmatic interventions into enactive and extended conceptions of cognition.   Philosophical Issues , 24(1), 110–26.

Gibson, E.J. ( 1969 ). Principles of perceptual learning and development . New York: Appleton Century Crofts.

Heft, H. ( 2001 ). Ecological psychology in context: James Gibson, Roger Barker and the legacy of William James’s radical empiricism. Mahwah, NJ: Lawrence Erlbaum Associates.

Hutto, D. , Kirchhoff, M. , and Myin, E. ( 2014 ). Extensive enactivism: why keep it all in?   Frontiers in Human Neuroscience , 8, 706.

Hutto, D. and Myin, E. ( 2013 ). Radicalising enactivism: basic minds without content. Cambridge, MA: MIT Press.

Kelso, J.A.S. ( 1995 ). Dynamic patterns: the self-organisation of brain and behaviour. Cambridge, MA: MIT Press.

Kirsh, D. ( 1995 ). The intelligent use of space.   Artificial Intelligence , 72(1–2), 31–68.

Kirsh, D. ( 2009 ). Problem solving and situated cognition. In: P. Robbins and M. Aydede (eds.), The Cambridge handbook of situated cognition. Cambridge: Cambridge University Press, pp. 264–306.

Menary, R. ( 2007 ). Cognitive integration: mind and cognition unbounded. Basingstoke, UK: Palgrave MacMillan.

Menary, R. ( 2010 ). Cognitive integration and the extended mind. In: R. Menary (ed.), The extended mind. Cambridge, MA: MIT Press, pp. 227–43.

Newell, A. and Simon, H. ( 1976 ). Computer science as empirical enquiry: symbols and search.   Communication of the Association for Computing Machinery , 19(3), 113–26.

Rowlands, M. ( 2009 ). Extended cognition and the mark of the cognitive.   Philosophical Psychology , 22(1), 1–19.

Rowlands, M. ( 2010 ). The new science of the mind: from extended mind to embodied phenomenology. Cambridge, MA: MIT Press.

Rupert, R. ( 2004 ). Challenges to the hypothesis of extended cognition.   Journal of Philosophy , 101(8), 389–428.

Rupert, R. ( 2009 ). Cognitive systems and the extended mind. Oxford: Oxford University Press.

Rupert, R. ( 2013 ). Memory, natural kinds, and cognitive extensions; or, Martians don’t remember and cognitive science is not about cognition.   Review of Philosophy and Psychology, 4, 25–47.

Shapiro, L. ( 2010 ). Embodied cognition. London: Routledge, Taylor & Francis.

Silberstein, M. and Chemero, A. ( 2012 ). Complexity and extended phenomenological cognitive systems.   Topics in Cognitive Science , 4(1), 35–50.

Sprevak, M. ( 2009 ). Extended cognition and functionalism.   Journal of Philosophy , 106, 503–27.

Sprevak, M. ( 2010 ). Inference to the hypothesis of extended cognition.   Studies in History and Philosophy of Science , 41, 353–62.

Sterelny, K. ( 2010 ). Minds: extended or scaffolded?   Phenomenology and the Cognitive Sciences , 9(4), 465–81.

Sutton, J. ( 2010 ). Exograms and interdisciplinarity: history, the extended mind and the civilizing process. In: R. Menary (ed.), The extended mind. Cambridge, MA: MIT Press, pp. 189–225.

Sutton, J. , Harris, C.B. , Keil, P.G. , and Barnier, A.J. ( 2010 ). The psychology of memory, extended cognition and socially distributed remembering.   Phenomenology and Cognitive Sciences , 9(4), 521–60.

Van Orden, G.C. , Holden, J.G. , and Turvey, M.T. ( 2003 ). Self-organization of cognitive performance.   Journal of Experimental Psychology: General , 132(3), 331–50.

Walter, S. and Kästner, L. ( 2012 ). The where and what of cognition: the untenability of cognitive agnosticism and the limits of the motley crew argument.   Cognitive Systems Research , 13, 12–23.

Ward, L.M. ( 2001 ). Dynamical cognitive science . Cambridge, MA: MIT Press.

Warren, W.H. ( 2005 ). Direct perception: the view from here.   Philosophical Topics , 33(1), 335–61.

Wheeler, M. ( 2005 ). Reconstructing the cognitive world: the next step. Cambridge, MA: MIT Press.

Wheeler, M. ( 2010 ). In defence of extended functionalism. In: R. Menary (ed.), The extended mind. Cambridge, MA: MIT Press, pp. 245–70.

Wheeler, M. ( 2011 a). Embodied cognition and the extended mind. In: J. Garvey (ed.), The Continuum companion to philosophy of mind. London: Continuum Press, pp. 220–38.

Wheeler, M. ( 2011 b). In search of clarity about parity.   Philosophical Studies , 152(3), 417–25.

Wheeler, M. ( 2014 ). Revolution, reform, or business as usual? The future of embodied cognition. In: L. Shapiro (ed.), The Routledge handbook of embodied cognition. London: Routledge, Taylor & Francis, pp. 374–84.

Wheeler, M. ( 2015 ). Extended consciousness: an interim report.   The Southern Journal of Philosophy , 53, 155–75.

Wilson, R.A. and Clark, A. ( 2009 ). How to situate cognition: letting nature take its course. In: P. Robbins and M. Aydede (eds.), The Cambridge handbook of situated cognition. Cambridge: Cambridge University Press, pp. 55–77.

Whether or not a cognitive process counts as an extended cognitive process is to be settled on a case-by-case basis. We see this policy at work in the debate about the realizers of phenomenal conscious experience. Prominent defenders of extended cognition deny that the material vehicles of phenomenal experience ever extend into the world (see, e.g., Clark 2012 ; Wheeler 2015 ). These authors argue that the computational processing that forms the basis for phenomenal experience is firmly encased within the heads of individuals.

The exact balance of internal and external resources recruited to solve a problem is negotiated on a case-by-case basis, in ways that are constrained by the task at hand. Whether the cognitive agent makes use of structures located in the environment in problem-solving will depend on the costs and benefits of doing so, as evaluated, for instance, in terms of energy expenditure, risks, and uncertainties ( Clark 2008 , ch. 7; Rowlands 2010 ). There are interesting connections here to Clark’s recent work on predictive processing ( Clark 2015 ).

See, e.g., Rupert 2009 .

Sutton et al. (2010) argue for a distinction between embedded and scaffolded theories of cognition. The former position, which they attribute to Adams and Aizawa, holds that cognition is fundamentally intracranial but may causally depend upon interactions with external resources located in the environment. Theories of scaffolded cognition by contrast argue that cognitive processes can unfold through couplings between heterogeneous internal and external resources. Distributed cognitive processes made up of heterogeneous elements, some inner and some outer, should figure among the processes that are investigated in cognitive science. Sutton and colleagues distinguish scaffolded cognition from extended cognition, arguing that the former comes in degrees and often falls short of satisfying the degree of integration required for some external resource to count as a constitutive part of a cognitive process. (See also Sterelny 2010 .) In what follows I will for the most part treat embedded and scaffolded cognition as coextensive, but it should be noted that some cases of scaffolded cognition may also qualify as cases of extended cognition. Thus, the category of scaffolding cognition may cross-cut the distinction between extended and embedded cognition.

See Kirsh (1995) for a classic treatment of more cases of problem-solving of this flavor in which actions are performed that structure the spatial environment in ways that simplify the reasoning the subject engages in to solve a problem. For more recent update, see Kirsh (2009) .

There is an important question in the literature about the nature of the extra conditions coupling with an external resource has to meet in order for some external resource to count as a part of a cognitive process. Everyone is agreed that causal coupling on its own is not sufficient. (See Adams and Aizawa 2008 , 2010 , on what they call the coupling-constitution fallacy.) Some degree of functional integration of the external resource is necessary. (See Menary 2007 for an account of cognitive integration.) What are the additional conditions that need to be satisfied in order for the external resource to be integrated in the right way? Clark and Chalmers (1998) make some tentative suggestions, specifying what have come to be called “conditions of glue and trust” (see also Wilson and Clark 2009 ). The glue conditions relate to the availability and accessibility of the information that an external resource provides. The trust conditions concern the reliability of this information and the degree to which the individual accepts it without question or critical scrutiny. Sterelny (2010) describes a spectrum of possible cases of cognitive integration of an external resource, identifying three key dimensions—trust, entrenchment, and individualization. For further discussion of these dimensions, see Sutton et al. (2010) . Colombetti and Krueger (2015) make use of Sterelny’s dimensional analysis of integration to argue for extended and scaffolded affectivity.

Not everyone in the EMT camp would reject this claim. Adams and Aizawa (2008 , ch. 7) make a distinction between extended cognitive systems and extended cognition. They allow that softly assembled systems like the ones I’ve just been describing count as examples of extended cognitive systems. They deny, however, that the extension of a cognitive system suffices for extended cognition on the basis that not every part of a cognitive process itself counts as cognitive. Rupert (2009) , by contrast, does deny that softly assembled systems count as cognitive systems. They fail to meet Rupert’s integration condition, according to which systems count as cognitive only when made up of persisting mechanisms, the integrated functioning of which is explanatory of intelligent behavior. My thanks to an anonymous reviewer for reminding me of the lack of consensus in the EMT camp about this issue.

Also see Rupert (2009 , ch. 5) for an argument to this effect. Clark echoes Sprevak’s worry when he writes that the debate “though scientifically important, and able to be scientifically informed, looks increasingly unlikely to admit of straightforward scientific resolution” ( Clark 2011 , p. 454).

This is a question whose answer I have said will come from providing a mark of the cognitive. I depart from Sprevak in framing the debate as being about cognition rather than the mental.

We will eventually see how this question in turn depends on one’s preferred mark of the cognitive. Adams and Aizawa (2001) also make this point (p. 46). Thus I am in effect repeating their claim here that the deployment of the parity principle to motivate EXT will depend on a prior commitment to a mark of the cognitive.

For this line of argument, also see Rupert (2004) .

Criticisms of EMT along these lines can be found in Clark (2008 , ch. 7), Sprevak (2009) , and Wheeler (2010) .

Clark (2008) calls this move “the Otto 2-step” and offers a brief rebuttal (p. 80).

Chalmers responds to this objection by arguing that folk psychological explanation is context-sensitive. Sometimes our explanatory interests concern an individual’s large-scale behavior, in which case it makes sense to look at the larger system of agent together with environmental resource. On other occasions our explanatory interests might be more local, relating to Otto and the interactions with his notebook, in which case it makes sense to treat perception and action as marking the boundaries of the cognitive system.

How does the second wave in EXT differ from EMT? Sutton et al. (2010) make a distinction between embedded and scaffolded cognition (discussed in fn. 5 ). Scaffolded cognition can allow for a spectrum of cases, some of which fit the description of extended cognitive processes in which cognitive processing is partially externally constituted.

It is not entirely clear to me whether the second-wave EXT would disagree with Wheeler on this point. Indeed I suspect there may be no agreement on this point within the second-wave camp. Sutton and colleagues, for instance, stress that the interdisciplinary science of biotechnological minds is one that still works within the classical framework of cognitive science, the only difference being that the object of study is “cognitive and computational architectures whose bounds far exceed those of skin and skull.” ( Sutton 2010 , p. 191, quoting Clark 2001 , p. 138). Menary (2010) by contrast presents the second wave as being aligned with enactive cognitive science, viewing cognition as “constituted by our bodily activities in the world in conjunction with neural processes and vehicles” (p. 227).

It should be noted, however, that the effect of this stipulation might be to exclude REX from the debate. For suppose we go along with Wheeler and agree that the physical symbol systems hypothesis does tell us what it is for a state or process to count as cognitive. Either it will follow that REX is wrong to claim that dynamical and ecological cognitive processes are nonrepresentational and non-computational, or it will follow that the processes REX investigates are not cognitive at all.

This quote is actually taken from Wheeler’s discussion of a neural network model that is able to do pattern completion and recognition for external symbol systems, such as systems of mathematical notation or written symbols. I take it the conclusion Wheeler wants to draw for this particular example generalizes to the Otto notebook system.

I choose Rupert here because he has written extensively against generic memory. Related arguments to those of Rupert can also be found in Adams and Aizawa (2008 , ch. 4)

Again see Sutton et al. (2010) for detailed arguments that this is actually the case based on empirical research concerned with transactional memory.

Recall Adams and Aizawa’s (2008 , ch. 7) distinction between extended cognition and extended cognitive systems (see fn. 6 ): elements can be a part of an extended cognitive system while not themselves counting as cognitive.

Here is not the place to enter into the historical details, but for excellent accounts, see Heft (2001) and Gallagher (2014) .

For useful entry points see Kelso (1995) , Ward (2001) , and Chemero (2009) .

The Lotka–Volterra equations describe variation in population size in predators and prey using two equations. The first describes how the prey population changes over time as a function of growth minus the effect of predation. The second describes change in predator population as a function of size of prey population minus natural loss of predators.

Van Orden et al. (2003) ; Silberstein and Chemero (2012) ; Anderson et al. (2012) . The type of nonlinear ongoing causal influence between coupled systems or components we have described previously is sometimes described as “interaction dominance” ( Anderson et al. 2012 ). Van Orden et al. (2003) show how 1/f scaling, also known as “pink noise,” is a “signature” of interaction dominance. Pink noise has also been found in agent–environment interaction. For instance, Dotov et al. (2010) found 1/f scaling when subjects were playing a video game, controlling an object on a monitor using a mouse. When the mouse connection was temporarily disturbed, however, 1/f scaling decreased, indicating that “during normal operation, the computer mouse is part of the smoothly functioning interaction-dominant system engaged in the task” ( Silberstein and Chemero 2012 , p. 45).

I borrow this distinction between causal relevance and constitutive relevance from Craver (2007) .

• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

• Reference Manager
• Simple TEXT file

People also looked at

Hypothesis and theory article, ariadne’s thread and the extension of cognition: a common but overlooked phenomenon in nature.

• 1 School of Biological Sciences, University of Reading, Reading, United Kingdom
• 2 Laboratory of Immunoregulation (iREG), Department of Microbiology, Immunology and Parasitology, Biological Sciences Centre, Federal University of Santa Catarina, Florianopolis, SC, Brazil
• 3 Faculty of Environmental Science and Engineering, Southern Cross University, Lismore, NSW, Australia
• 4 The Biological Intelligence (BI) Lab, Faculty of Science, Southern Cross University, Lismore, NSW, Australia

Over recent decades, our philosophical and scientific understanding of cognition has changed dramatically. We went from conceiving humans as the sole truly cognitive species on the planet to endowing several organisms with cognitive capacities, from considering brains as the exclusive seat of cognition to extending cognitive faculties to the entire physical body and beyond. That cognition could extend beyond the organism’s body is no doubt one of the most controversial of the recent hypotheses. Extended cognition (ExC) has been discussed not only to explain aspects of the human cognitive process, but also of other species such as spiders and more recently, plants. It has been suggested that ExC could offer insights for the grounding of environmentally extended cognitive traits in evolved ecological functions. Here, we reviewed the ecological literature for possible ExC examples that satisfy the mutual manipulability criterion, which can be used to establish experimentally the boundaries of cognitive systems. Our conclusion is that ExC might be far more common than previously thought, and present in organisms as diverse as plants, fungi, termites, spiders, mammals, and slime moulds. Experimental investigation is needed to clarify this idea which, if proven correct, could illuminate a new path into understanding the origins and evolution of cognition.

1. Introduction

Perception and communication of information as well as learning, remembering, problem-solving and decision-making, are faculties routinely considered cognitive. But what is cognition? Relatively easy to observe in most living organisms, the term is hard to define and lacks a widely accepted consensus on its meaning ( Bayne et al., 2019 ). Here, we take an all-inclusive definition to describe cognition as the process by which living organisms perceive and process information, and value, store and use this information to increase their chances of survival ( Neisser, 1976 ; Shettleworth, 2010 ; Bayne et al., 2019 ; Lyon et al., 2021 ). And what do we mean by this tricky word: information? This ambiguous term can be understood in different ways. A broadly accepted interpretation is that of ‘statistical information,’ formalised by Shannon (1948) and commonly employed in physics, which refers to decrease in uncertainty of a system. While this definition can also be useful in the life sciences, information sensu Shannon becomes highly problematic when we talk about biological systems, where information is context-dependent (e.g. the same object can convey different information depending on the context; see Witzany and Baluška, 2012 ; Lyon et al., 2021 ). Hence, we believe it is more appropriate to talk about semantic information , which refers to the ‘meaning’ attributed by a biological system to an object, thus changing the behaviour or functioning of this system ( Harms, 2006 ). It is not the aim of this paper to debate the different concepts of information, but for a detailed discussion on the three broad ‘classes’ of information, see Harms (2006) .

Cognition refers to a particular interaction of the components of a biological system that makes the whole system flexibly adaptable and responsive to the environmental conditions ( Smart et al., 2017 ). As a basic requirement for life to occur, cognition co-emerged with (rather than after ) life and is present in all living beings with different degrees of complexity ( Lyon et al., 2021 ). Departing from a very simple form of cognition present in our last common ancestor, cognition evolved differently in each branch of the tree of life, depending on the evolutionary trajectory, the perceptual apparatus and body structure and the needs and problems of each lineage. As a result, different organisms from bacteria to plants to animals have different cognitive systems that serve the purpose of keeping them alive ( Lyon et al., 2021 ). We have long considered cognition as a feature possessed by humans and other primates, and understood most other animals to lack higher cognitive faculties (or to perform higher cognitive faculties in less complex ways than humans). Many cognitive scientists might not agree that cognitive faculties can be found beyond a handful of animal species and non-animal organisms like plants or slime moulds, partly because it is assumed that a large brain and a complex neural system are necessary to support cognitive faculties. Nevertheless, more recent research has included studies on cognitive processes of invertebrates ( Mathel and Dickel, 2017 ; Perry et al., 2017 ), and non-traditional kingdoms, such as plants ( Trewavas, 2016 ; Souza et al., 2018 ; Castiello, 2021 ), fungi ( Aleklett and Boddy, 2021 ), slime moulds ( Boussard et al., 2019 ), and bacteria ( Shapiro, 2007 , 2020 ). These new studies have forced us to reassess what cognition is, what is it for and where it occurs in the living world.

The recent inclusion of non-traditional model systems in the field of cognitive science testifies to an ongoing shift in our philosophical and scientific understanding of cognition. For example, let us take the notion of cognition as entirely contingent on specific structures of cognitive systems, such as the brain, and the accompanying computational metaphor of the brain as an extremely efficient information-processing computer, a metaphor we inherited from the early cybernetics/information theorist movement ( Shannon, 1948 ; Weiner, 1948 ). These ideas have deeply guided our understanding of cognition and the brain most prominently since the 1960s with the emergence of cognitive psychology ( Neisser, 1967 ; Figure 1 ). However, despite being highly developed, our brain is not an efficient computer and simply does not have the capacity to process every piece of information we encounter. Instead, we use ‘heuristics,’ mental shortcuts that allow us to make this information overload manageable so that we can arrive at solutions for given situations ( Simon, 1955 ; Golstein and Gigerenzer, 2002 ), a phenomenon we share with numerous other species ( Marsh, 2002 ; Hobson et al., 2021 ) with and without brains ( Latty and Beekman, 2015 ). Crucially, this means that cognition does not work simply as computation and, contrary to what is commonly assumed, its substrate does not need to be a brain, given that information-processing and decision-making can happen even in its absence. Cognition, thus, operates differently from what our conventional understanding would suggest. Moreover, recent research shows that even in neural organisms, not all information is stored in synaptic connections. For example, non-associative long-term memory was found to be stored in the RNA of the sea snail Aplysia californica . Naïve sea snails became sensitised to a stimulus they never encountered before when the RNA of sensitised sea snails was transferred to them via injection in their haemocoels ( Bedecarrats et al., 2018 ). Interestingly, transmission of a learned behaviour was also found in the aneural organisms Physarum polycephalum through the fusion of their cytoplasms, in the plural ( Vogel and Dussutour, 2016 ). The fact that brains or neurons are not always necessary for information-processing and learning strongly suggests that the basis of information-processing and storage can be non-neural, and cognitive processes do not, necessarily, require a brain and central nervous system (CNS). If brains and CNS are dispensable for cognition to exist (although, of course, crucial for proper cognitive functioning in the organisms that have them), where else is cognition?

Figure 1 . A cognitivist modular and linear view of cognition. Perception builds a discrete and abstract representation of the local environment which is, in turn, transformed by the cognitive processes of the brain into actions. Within this model, decision-making resides and is confined within the middle box of the brain. This model is also known as the ‘Classic Sandwich Model’ where the outer slices of perception and action are peripheral to the inner filling of cognition and separate from one another. They are also separate from cognition, which interfaces between perception and action (adapted from Burr, 2017 ).

2. Cognition within and beyond the body

The notion of cognition as an input–output mechanism tied to a CNS performing computations is typical of the cognitivist approach and has guided much of the cognitive research since the 1950s. However, another understanding of cognition, known as post-cognitivism, has been developed since the 1970s (e.g. Bateson, 1972 ; Dreyfus, 1972 ; Maturana and Varela, 1980 ) with the contribution of many different scholars with quite differentiated set of views. Among the range of views and research programmes within post-cognitivism, one particularly influential is the Santiago theory of cognition, initiated by Humberto Maturana and Francisco Varela ( Varela et al., 2016 ). From this perspective, the processing of information is based on dynamic and reciprocal interactions of perception and action in the environment because living organisms are embedded in the environment and cannot be studied separated from it ( Figure 2 ). Furthermore, cognition is an essential feature of life itself, and has evolved before the emergence of the first metacelluar organisms ( Lyon et al., 2021 ). Cognition is what ensures the ability of organisms to manipulate and exploit the environment around them to secure their survival. As a process evolved billions of years before brains, the realisation of cognition requires, first and foremost, a living body, and not a CNS. Hence, cognition is an embodied and embedded process, present in all living organisms, occurring in the entire body (embodied), and entangled with the environment where the organisms are embedded in ( Clark, 1997 ; Varela et al., 2016 ; Cazalis et al., 2017 ; Souza et al., 2018 ). For example, the way human cognition is altered and modulated by the gut through the gut-brain axis ( Tillisch et al., 2013 ; Bagga et al., 2018 ), and by the skeletal muscles (e.g. Coles et al., 2019 ; Onishi et al., 2022 ), indicates that our own cognitive processes are not restricted to the CNS but span throughout our entire body, involving digestive and muscular systems. This does not lessen the importance of the CNS in the cognitive process, but rather suggests that human cognition is not restricted to it.

Figure 2 . A post-cognitivist model of cognition. Within this model, the brain and nervous system (or other internal systems, such as the vascular system in plants) are embodied in the biological body of the organism, the body is situated in a local environment and behaviour is not a property of the organism, but an emergent property of the interacting dynamical system of nervous system-body-environment (adapted from Beer, 2000 ).

Clark and Chalmers (1998) proposed the idea of extended cognition in humans where cognition extends outside the boundaries of the physical body to objects in our immediate environment which we manipulate and transform into integral parts of our cognitive process. A functional coupling between the internal and external elements would underpin an extended cognition ( Clark and Chalmers, 1998 ; Menary, 2010a ). This coupling of the organism with environmental elements can be advantageous to the organism because the extension of the cognitive process beyond the boundary of the physical body can increase cognitive capacity. For example, it may allow the brain to offload the cognitive burden of processing some environmental information ( Armitage et al., 2020 ), making certain cognitive tasks easier to perform.

An example of this cognitive burden offload is the use of pen and paper to solve a mathematical formula. By writing down on paper symbols and marks that aid the process, we externalise part of our short-term memory and information-processing ability. Without it, this cognitive task is significantly harder for most people. Today, our dependence on electronic gadgets for a myriad of tasks makes explicit more than ever how our cognition is extended to these objects ( Armitage et al., 2020 ). More recently, it has also been proposed that even our gut microbiota, being able to modulate our cognitive abilities, composes our extended minds ( Boem et al., 2021 ).

The extended cognition hypothesis has provided insight into the functioning of cognitive processes previously unknown in some organisms like spiders and slime moulds ( Japyassú and Laland, 2017 ; Sims and Kiverstein, 2022 ). The possibility that organisms from different kingdoms (and two different phyla in the case of humans and spiders) extend their cognition suggests that the ability of extending cognition might be much more common and present in a wider range of organisms than previously thought.

In this article, we review the literature looking for organisms that could likely extend their cognition, focusing on niche-constructing organisms and how they manipulate their environment. Some studies did not directly test the extended cognition hypothesis but, we suspect, may at least partially satisfy the demarcating criterion adopted here (and described in the next section) and could provide insights into what extended cognition is for in those systems. We aim to provide an updated overview on where extended cognition occurs in the living world, proposing that this phenomenon could be far more common than expected.

3. Demarcating criteria for extended cognition

One of the difficulties to investigate extended cognition is using a clear demarcating criterion for what can be considered cognitive (see Menary, 2010b ). When Clark and Chalmers presented the ExC hypothesis, they proposed a demarcating criterion for extended cognitive systems called the parity principle (PP). Accordingly, if some element of the world functions as part of a process that, were it occurring in the brain (or body, see Sims and Kiverstein, 2022 ), we would not hesitate in dubbing it as part of the cognitive process, then it is part of the cognitive process while being used for that aim. However, the PP is problematic as a criterion and considered weak ( Sutton, 2010 ; Smart, 2022 ). According to its critics, the PP fails to evidence differences between inner and outer (i.e. extended) components of the cognitive system by giving similar weights to the importance of what happens inside the body and outside it ( Sutton, 2010 ; Smart, 2022 ). Besides, it is not a good criterion to limit which elements of the world can be considered part of a cognitive process, and it relies on a theoretical assumption of the mind based on functionalism, which overlooks how psychological states are physically sustained ( Sutton, 2010 ; Kaplan, 2012 ). In other words, the substrate for cognition to happen would be irrelevant. Furthermore, the PP fails to prevent what is known as the ‘coupling-constitution fallacy’, when every object could be considered constitutive to the cognitive process just because it is coupled to the cognitive agent ( Adams and Aizawa, 2001 ). Because something is coupled to a system showing causal dependence, it does not mean that it constitutes the system. This problem can further lead to what is known as the ‘cognitive bloat’: everything that is readily accessible to the cognitive agent would be considered as part of its cognitive system ( Rupert, 2004 ; Clark, 2008 ; Allen-Hermanson, 2013 ). Consequently, Wikipedia or Google should be considered as parts of our cognitive process when we access these websites, meaning that our memory extends to all of these databases ( Ludwig, 2015 ).

To solve problems related to the coupling-constitution fallacy and cognitive bloat, Kaplan (2012) proposed the mutual manipulability (MM) criterion of Craver (2007) to demarcate the boundaries of cognition. Kaplan noted how neurobiologists already demarcate cognitive processes within the body by mutually manipulating the components of the system to establish relations of constitutive relevance between them ( Kaplan, 2012 ). Putting it very simply, to establish these relations, there are two conditions of MM that must be satisfied with interventions occurring at different levels of the phenomenon to be explained. When testing whether an element is part of a system in a particular way, an intervention in that system’s behaviour should cause a corresponding alteration in the element (this is known as top-down intervention). Similarly, an alteration in the element also must cause an alteration in the system’s functioning (bottom-up intervention). If both interventions are successful in establishing these causal relations, one can then say that the element is part of the mechanism, and both constitute the same system ( Craver, 2007 ; Kaplan, 2012 ).

As an example, let us suppose that a scientist suspects that the cognitive process of an organism X involves the secretion and sensing of slime into the environment. According to the MM criterion, the scientist should submit X to a cognitive task that requires the use of slime. Depending on the challenge presented, the scientist observes that the organism X changes the chemical composition of the slime, which helps it to accomplish the task. To satisfy the first condition of the MM criterion, she or he administers a drug that interferes with the production of slime, observing that it affects the slime chemical composition and the organism’s response. Subsequently, in another set of experiments, the scientist applies a synthetic slime in the environment, confusing the organism X, which can no longer perform its task properly. The second condition of the MM criterion is satisfied, and the scientist concludes that the slime is part of the cognitive process of the organism X.

The MM criterion is a powerful tool to ensure that simple causal background conditions are not confused with extended cognition; in other words, the criterion prevents the ‘cognitive bloat’. For example, when we are performing a cognitive task Y, areas in the brain related to this task are known to be more activated, which increases its oxygen consumption. Someone could argue that oxygen, therefore, is part of our cognitive processing. However, this scenario does not satisfy both conditions of the MM criterion; supposing that it is possible to enhance the presence of oxygen in the same brain areas, hardly anyone would be inclined to think that it would make us spontaneously engage in cognitive task Y. Oxygen is a background condition as necessary as glucose or water to the normal functioning of the brain, but not an active part of the cognitive process ( Kaplan, 2012 ; Parise et al., 2020 ).

The usefulness of MM to establish constitutive relevance has been questioned but also elaborated since it was proposed ( Krickel, 2018 ; Craver et al., 2021 ; Smart, 2022 ). For example, critics to this method say that it is inadequate because it is spelled out in terms of interventions, which is an approach to establish causal relations, whereas constitutive relevance implies non-causal relations ( Krickel, 2018 ). Consequently, when employing the MM, some additional care must be taken to consider, for example, the importance of time in the manipulations ( Krickel, 2018 ) and the kind of interventions made ( Craver et al., 2021 ). It will depend on the kind of experiment being performed and the system under study. Nevertheless, given the potential of MM to underpin empirical experiments on extended cognition, it was adopted by behavioural ecologists to study this phenomenon at least in spiders ( Japyassú and Laland, 2017 ), and recently proposed for plants ( Parise et al., 2020 ).

In an excellent review, Japyassú and Laland (2017) applied the MM criterion to analyse the behaviour of spiders and their webs, and concluded that spiders have their cognition extended to their webs. Both the spider and the web are part of the same cognitive system. The authors corroborate their hypothesis by presenting a series of experiments that satisfy the MM criterion. For example, spiders that weave orb-webs learn which sector of their web captures more prey. Hence, when they are hungry, they tend to direct their attention to the most profitable section of their webs by pulling the radial threads of that section. The tension in the threads makes them more accurate in perceiving the vibrations caused by even small prey, enhancing the chances of success in procuring food ( Watanabe, 2000 ; Nakata, 2010 ; Mortimer et al., 2018 ). However, if the tension of another area of the web is experimentally increased by pulling its threads, the spider automatically directs its attention to that section, even though this is not the section the spider has learned to be most rewarding ( Nakata, 2010 , 2013 ). By satisfying both conditions of the MM criterion, we can conclude that the spider’s ability to process information and act on it is highly dependent on the web, and the web is part of the spider’s cognitive system. Other examples for the case of spider cognition extending to their webs can be found in Japyassú and Laland (2017) .

Similarly, Parise et al. (2020) proposed that plants could also extend their cognitive abilities to the environment they manipulate. For example, plants could use the substances they exude through their roots, the soil microbiota, and the common mycorrhizal network as an extension of their cognitive system. Below, we provide some specific plant examples, and suggest that a model of plant extended cognition can be applied to the behaviour of other brainless, sessile, modular and niche-constructing organisms, like sponges and corals.

In summary, the MM is a promising criterion to explore extended cognition. However, depending on the species studied, the experimental conditions, and how the resulting data will be interpreted, the employment of the manipulations and the kind of interventions will vary. Krickel (2018) and Craver et al. (2021) are good references to help planning these experiments. For the purpose of this review, we will adopt a loose interpretation of the MM to make our point, also because none of the studies that will be discussed here were designed to test for extended cognition. We hope that our analysis below serves as an inspiration and a starting point for research looking for other cases of extended cognition in nature, where experiments will have to take into consideration the subtleties of the MM account.

4. Ariadne’s thread and the external memory

According to the legend, the Greek hero Theseus entered Mino’s labyrinth, where the monstruous Minotaur lived, with a ball of thread famously given to him by Ariadne in the hope that he could find his way back out of the deadly trap. Theseus unspooled the ball of thread as he entered the labyrinth; after defeating the beast, he was able to find his way out by following the thread.

Mechanisms to externalise memory such as the one employed by Theseus are common. We can find them in our daily life from the much less epic post-its glued over our computer or fridge to appointments noted in a diary or cell phone ( Figure 3A ). What these mechanisms have in common is the fact that they are ways of relieving us from the burden of remembering certain things, a mechanism that is relatively common across the domains of life. From the MM perspective, we could say that the thread is an extension of Theseus’ cognition. His mind is unable to remember all the turns taken inside the labyrinth; however, by actively unspooling the thread along the way, he successfully manipulates his environment (which satisfies the first condition of the MM criterion). In this case, we understand that a change in Theseus’ cognitive state, i.e. the one elicited by the challenge of entering the labyrinth, changes his interaction with the environment, i.e. unspooling the ball of thread. Since he is a human subject, we take his cognition for granted. However, if one wants to be sure of this top-down relation, one could experimentally drug Theseus or damage part of his brain, making him to forget Ariadne’s instructions, not unspooling the ball of thread, or even letting the thread go. He would never complete the navigational task in this case. To satisfy the second condition of the MM criterion (bottom-up intervention), we would need to manipulate the thread and observe Theseus’ behaviour. If someone maliciously cuts the thread or stretches it through another path, Theseus will never find his way back and die inside the labyrinth, to Ariadne’s grief. In other words, Theseus’ ability to navigate a maze is compromised if someone messes with the thread. Both conditions of the MM criterion are satisfied and we can say that, in that moment, Ariadne’s thread was part of Theseus’ cognitive system.

Figure 3 . Representation of three different biological systems (two neural and one aneural) that may extend their cognitive process to the environment. Arrows represent the relationships between the organism and the objects that sustain the extended cognition (ExC). By experimentally manipulating these relations, ExC could be tested empirically. (A) Humans, the best-studied example, have a plethora of mechanisms for extending their cognitive process by modifying their environment, from ‘to-do lists’ to computers and cell phones. Gut microbiota could also be involved in human ExC ( Boem et al., 2021 ), but is not represented here. (B) Spiders could extend their cognition to their webs; manipulations in the web lead to changes in spider cognitive states, and vice versa. (C) Plants are an example of aneural organism that could use ExC to support its cognitive process. In this case, through the manipulation of VOCs, root exudates and soil microbiota, and mycorrhizal symbionts. Similar processes may occur in other organisms without neurons or central nervous system like fungi, slime moulds and sea sponges.

Spiders also use threads to remember the path they have taken, especially when spinning their webs. This is a complex behaviour and the spider cannot remember all the steps all the time, especially in the earlier stage of web-spinning ( Japyassú and Laland, 2017 ). Therefore, they rely on external cues perceived sequentially, in particular the position of the spiral thread that was previously glued to the radial threads ( Eberhard and Hesselberg, 2012 ; Japyassú and Laland, 2017 ). Before moving from one radial thread to the other, the spider measures its distance from the spiral thread to know how to glue the new spiral thread. The experimental removal of a segment of the spiral thread creates conflicting cues to the spider, and usually, it ignores one cue in favour of another by skipping the measurement and simply glueing the new thread in a displaced spot. In the next round, it will use the newly glued thread as a guide, but an irregularity will remain in the web ( Eberhard and Hesselberg, 2012 ). According to Japyassú and Laland (2017) , much of the information to build the web is present in the web itself, and not within the spider’s CNS—just like the case of Theseus—and the information can be altered if we alter the shape of the web ( Figure 3B ). We think most researchers would also take for granted that spiders are cognitive organisms, which means that it is obvious that alterations in their cognitive states would change how the webs are spined and used. But as examples of top-down manipulations in the case of spiders, studies have demonstrated that interventions in the spider’s CNS regions related to web-building change dramatically the structure of their webs. These interventions include laser-induced brain damage ( Witt, 1969 ), drug administration ( Witt, 1971 ; Hesselberg and Vollrath, 2004 ; Albín et al., 2014 ), starvation ( Watanabe, 2000 ), and even parasitoid wasps ( Eberhard, 2000 ).

Interestingly, organisms with no brain also leave cues to help navigation. The slime mould Physarum polycephalum releases a slime composed of glycoproteins that function as an external memory of the places where it has already been ( Reid et al., 2012 ). The slime is repulsive to the P. polycephalum ; given the option, the slime mould chooses not to crawl over it. In an experiment with a U-shaped obstacle, the researchers observed that P. polycephalum uses the slime as a cue to navigate better and exit the maze more efficiently. When they manipulated the slime effectiveness by putting the P. polycephalum in a substrate entirely covered by slime, they masked its effect and the exit from the U-shaped obstacle was significantly impaired ( Reid et al., 2012 ). The manipulation of an object interfered with the cognition of the slime mould, which could be considered a bottom-up intervention and suggests that P. polycephalum extends its cognition to the slime. To complete the MM criterion, researchers will have to think on experiments that manipulate the slime through interventions in the cognitive system of the slime mould. Such experiment would add robustness to claim of extended cognition of Sims and Kiverstein (2022) in slime moulds.

Plants are another group of brainless organisms that use external memory to solve their problems ( Figure 3C ). For example, Arabidopsis thaliana and Cucurbitaceae plants can actively manipulate the soil microbiota community in the rhizosphere. By altering the chemical composition of its root exudates, these plants inhibit the growth of some taxonomic groups of microbes and favour the development of others, leading to increased pathogen resistance and fitness ( Huang et al., 2019 ; Zhong et al., 2022 ). This creates a microbiological community that is very specific to the exudates released, and this alteration can not only help to tackle pathogens, but also encode and store the memory of past events. For instance, Yuan et al. (2018) grew five generations of A. thaliana infected by Pseudomonas syringae in the same pot and substrate. When they planted a sixth, naïve plant in the same pot, and later infected it with P. syringae , the plant was significantly more resistant to this disease than the control, even though it never experienced the disease before ( Yuan et al., 2018 ). This means that alterations in the plant (disease) change the microbial rhizosphere community, and alterations in the rhizosphere community change the behaviour of the plant. Besides, in another set of experiments, the authors applied artificial exudates that mimicked the chemical composition of A. thaliana exudates in naïve plants, which altered the microbiological community of the substrate, and the plants became more resistant to P. syringae than the control group ( Yuan et al., 2018 ). This means that the manipulation of the soil or exudates by growing infected plants in the substrate or applying artificial exudates creates an immunological memory that primes the plants against the disease, which satisfies the second condition (bottom-up intervention) of MM. The phenomenon studied by Yuan et al. (2018) is called ‘soil memory,’ or ‘soil-borne legacy’, and has received interest because of its potential applications to agriculture. Although here we are talking about immunological memory, the fact that plants can manipulate the soil microbiota depending on their needs should be taken into account when studying cognitive traits of plants.

5. Exudates: The liquid border of cognition?

Root exudates are one of the most important means by which plants interact and manipulate their underground world. Beyond the probiotic and mnemonic functions mentioned above, it seems that they are also important for the plant to synchronise its flowering. Brassica rapa is a plant that preferentially blooms when daylight time reaches or passes a certain threshold of hours. However, it can also bloom under the non-inductive condition of short days, when daylight duration is below that threshold. Falik et al. (2014) observed that when grown under long-day conditions and blooming, this plant likely alters the composition of its root exudates (as suggested by the results of the experiment—the authors did not analyse exudate chemical composition) which possibly could be considered as satisfying the first condition of the MM. Subsequently, the researchers manipulated the exudates of B. rapa grown under short-day conditions by watering it with leachates of blooming plants exposed to long-days (a manipulation of the exudates that satisfies the second condition of the MM). The modification of exudates changed the information available to the plants, but this was aboveground-borne information previously processed and encoded by other plants. As a result, even under the non-inductive condition, the plants accelerated its bloom and increased fruit production ( Falik et al., 2014 ).

Furthermore, many allelopathic plants—plants that exude toxic substances through their roots, in general gaining competitive advantage over other plants—are sensitive to the exudates they release. One important hypothesis to explain this self-toxicity proposes that this effect enables the plant to perceive and explore its three-dimensional environment around it. For example, Falik et al. (2005) demonstrated that the accumulation of root exudates between Pisum sativum roots and an obstacle in the soil promotes growth inhibition and even root withering. Consequently, the plant can stop its growth towards an obstacle even before its roots get close to it ( Falik et al., 2005 ). When the researchers manipulated the root exudates by adding activated carbon or potassium permanganate, which inactivated the exudates, it supressed the root obstacle avoidance mechanism. As a result, the roots grew towards the obstacle as if it was not present ( Falik et al., 2005 ).

Mahall and Callaway (1991 , 1992) studied two species of desert plants: Larrea tridentata and Ambrosia dumosa . The first one is an allelopathic plant, while the latter is not. To understand the distribution of these plants in the field ( L. tridentata , randomly scattered; A. dumosa , clumped), Mahall and Callaway studied their root interactions in a set of experiments. They discovered that L. tridentata uses allelopathic exudates to interact with roots of the same plant or of other plants. The accumulation of allelopathic exudates between two roots of L. tridentata leads to the inhibition of its growth, and this interaction structures the distribution of the roots in the soil and the distribution of the plants in the field. However, when the exudates were experimentally removed from the soil with activated carbon, this inhibitory effect ceased, and the plant grew its roots regardless of the presence of other roots.

These experiments ( Mahall and Callaway, 1991 , 1992 ; Falik et al., 2005 ) suggest that exudate accumulation provides the plant information about the three-dimensional environment around belowground organs. This process is almost entirely external to the plant, but leads to an optimised placement of roots, likely with effects on plant distribution in the environment ( Mahall and Callaway, 1991 , 1992 ) and fitness. Besides, the experiments suggest that activated carbon is a useful tool to test the hypothesis of root exudates as part of the plant’s cognitive process. It is a promising, relatively simple and cheap technique to investigate extended cognition in plants. The experimental removal of root exudates followed by the observation of plant behaviour could lead to insights on the contribution exudates make to the plant cognitive process.

There is another way by which plants could use exudates to understand their three-dimensional environment. When a seed germinates, it is important for the young plant to be aware of the soil volume around it, for it is a limiting resource to be exploited. Therefore, it is imperative the germinating seed acquires non-nutritional information about the immediate substrate to match its growth to the available soil volume. Restrains in growth are due to forecasted limitations, occurring irrespective of nutrient or water availability ( Wheeldon et al., 2021 ). To explain this phenomenon, Wheeldon et al. (2021) carried out a series of experiments on the growth and development of wheat ( Triticum aestivum cv. Mulika) and found that this plant actively assesses the substrate volume in a two-step process. Wheeldon and colleagues proposed that the plant first releases a soluble, rapidly diffusible substance that spreads through the substrate (‘substrate volume-sensing signal’, SVS). Its concentration will depend on the substrate volume and modulate the shoot growth to match the substrate volume. However, the roots also secrete another, less mobile substance (‘root density-sensing signal’, RDS) that will signal to the roots its density, inhibiting its growth. These substances do not diffuse across the entire substrate but remain restricted to the roots’ influence zone and modulate rooting patterns, preventing the roots of growing too much and, perhaps, competing with each other.

This idea is very intriguing despite the unknown nature of the SVS and RDS. The authors do not discuss the possibility of the SVS and RDS signals being of other kind like sound, which is another possible mechanism for the plants to sense their three-dimensional environment ( Gagliano, 2013 , 2015 ). Regardless of their nature (chemical or acoustic), SVS and RDS are strong candidates to compose the extended cognitive system of wheat. In that case, one condition of the MM criterion is secured: the plant actively produces these chemicals or sounds and uses them to sense the environment. It is worthy to explore how manipulating SVS and RDS influences plant growth. If this second condition of the MM is satisfied, then we would have another example of extended cognition in plants.

The hypothesis of cognitive processes extending to exudates could be applied to fungi as well, for they also exude components in the soil that could help in their navigation and foraging behaviour in their three-dimensional environment. For example, Aleklett et al. (2021) observed that hyphae of the fungus Gymnopus confluens matched the speed of its growth with the width of the channel it is growing in. The wider the channel, the faster it grows. The authors also observed fungi’s exudates released in the environment, which could be regulating the reproductive activity of the fungi themselves, and also helping to process information about the environment around the hyphae. In this case, fungal exudates could play an analogous role as plant exudates in regulating reproduction ( Falik et al., 2014 ) and root navigation ( Mahall and Callaway, 1992 ; Falik et al., 2005 ). It is tempting to hypothesise that this is a form of extended cognition in fungi, but this possibility awaits rigorous experimental testing.

6. Self-generated chemotactic gradients

The modification of the organism’s surroundings to aid navigation in complex environments has been demonstrated to be employed even by single migrating cells. Chemotactic gradients are known for driving the steering of cells when they displace themselves. This phenomenon is present in cells migrating during embryonic development, the displacement of protists in the environment, immune responses and the spread of tumorous cells during metastasis ( Donà et al., 2013 ; Evans and Wood, 2014 ; Tweedy et al., 2016 , 2020 ). However, there is evidence that the gradients that guide these cells are not imposed externally, but generated by the cells themselves when they metabolise attractant substances in the medium ( Donà et al., 2013 ; Tweedy et al., 2016 ). As the cells modify the environment around them, they guide their own migration. This allows single cells to explore the environment around by sensing it from a distance, sampling it and using this information to decide the path to take. With mathematical modelling simulations confirmed by experiments with the protist Dictyostelium discoideum and mouse pancreatic cancer cells navigating complex mazes, Tweedy et al. (2020) demonstrated that the breakdown of, respectively, cAMP and lysophosphatidic acid from the environment creates self-generated gradients that guide the navigation of these cells in complex labyrinths. Remarkably, the cells use this gradient to anticipate dead ends from a distance, before entering them, and choosing the path that leads to the source of attractant. The authors observed that the capacity to generate such gradients and the structure of the environment are important to this navigation. When they manipulated the capacity of D. discoideum to break down cAMP, enhancing it by mutation, or preventing attractant breakdown by delivering to the wild type D. discoideum the non-degradable attractant Sp-cAMPS instead of cAMP, they observed an outperform in the maze (compared to the control) in the mutated D. discoideum and almost no progress in the migration of the group exposed to Sp-cAMPS ( Tweedy et al., 2020 ). In the same study, the authors tested different labyrinth shapes, some of them designed to confound the cell’s self-generated gradient. In these cases, both D. discoideum and cancer cells choose the dead end instead of the ‘right’ path, indicating that the manipulation of the self-generated gradient influences the cell’s ability to migrate in complex environments.

The study of Tweedy et al. (2020) seems to have satisfied both conditions of the MM criterion by manipulating the cells’ capacity to perform a cognitive task and the environment modified by the cells, with clear effect in their performance. Through intervention in the cell, they manipulated the chemotactic gradients, and through interventions in the chemotactic gradients, they manipulated cell cognitive behaviour. These results suggest that even single cells, be they part of a higher organism or not, are able to manipulate the environment to increase their ability of problem-solving—which leads to the appealing idea that even cells extend their cognitive process.

7. Volatile organic compounds: The ‘mind’ is in the air

The most well-documented form of plant–plant communication is through volatile organic compounds (VOCs). These are substances released in the air that can be sensed by other plants and modulate their behaviour. Notoriously, they can be used by plants to communicate herbivory attack, and induce the transcription of defence-related genes in other plants, stimulate the production of other VOCs, induce the production of extrafloral nectar—which attracts predators of herbivores—activate direct defences and the release of hormones, trigger calcium depolarisation and also activate induced systemic resistance (reviewed in Heil and Karban, 2010 ).

A particular bouquet of VOCs is released depending on the stimulus a plant receives, and the different chemical compositions encode different messages ( Heil and Karban, 2010 ). This mechanism probably evolved as a fast communication mechanism within individual plants. Without VOCs, communication between branch apexes may take several minutes or hours because this process would rely on electrical or chemical signals transported through the vascular bundles ( Heil and Silva Bueno, 2007 ). Volatiles are ‘shortcuts’ for the information traffic in a plant crown. In that sense, we could say they have an importance for fast integration in the crown comparable to certain pathways for information in the animal brain, like the dopaminergic pathways or the corpus callosum connecting both brain hemispheres, which allow rapid convey of information to distant parts of the brain. The dopaminergic pathways, like the mesolimbic and mesocortical pathways, connect relatively distant areas of the brain, for instance, the ventral tegmental area to the ventral striatum and the prefrontal cortex, respectively. Disruptions on these connections lead to neurological problems related to motivation, reward, memory, learning and movement ( Engert and Pruessner, 2008 ; Yamagushi et al., 2011 ). Similarly, human corpus callosum ensures that both brain hemispheres are connected and in interaction with one another for proper functioning of the brain as a whole. If this structure is severed, cognitive impairments arise. For example, the pathway that connects speech ability to the senses of the left side of the body is interrupted, and the subject can no longer describe verbally which object the left hand is holding, or what was seen in the left side of the visual field ( Gazzaniga, 1967 ). This demonstrates that long-distance pathways for exchange of information are important for proper cognitive functioning, both in animals and plants.

Therefore, VOCs constitute part of the cognitive process of plants, as they evolved primarily as a system of internal communication. Plants alter its VOCs depending on what they are experiencing, normally when stimuli are related to predation ( Heil and Silva Bueno, 2007 ; Heil and Karban, 2010 ), which could be considered as the first condition of the MM criterion. Conversely, if we manipulate the VOCs, we also manipulate the plant’s cognitive state. For example, when leaves of Lima bean plants ( Phaseolus lunatus ) undergo herbivory, they release in the air a bouquet of VOCs that communicate to other parts of the plant (and sometimes other plants nearby) the information of the attack, enabling the priming of the plant’s defences ( Heil and Silva Bueno, 2007 ). In that case, the plants more promptly secrete extrafloral nectar, and in higher amounts, which attract the predators of their herbivores, causing the plants to be less damaged by herbivory. The VOCs are actively produced by plants to convey information throughout the crown and are related to what the plant is experiencing ( Karban, 2015 ). Not surprisingly, the manipulation of these VOCs, such as their removal from the atmosphere ( Heil and Silva Bueno, 2007 ) or the application of synthetic VOCs to undamaged plants ( Heil and Kost, 2006 ), changes the plant strategy to fight herbivores, respectively, preventing its response or causing it to activate its defences following misleading information. This can be interpreted as the second condition of the MM being satisfied.

Sometimes, due to intraspecific variations in different populations, the same stimulus creates different VOC bouquets. In a study conducted with two populations of Artemisia tridentata , emitter plants received the same damage in the leaves, but the chemical composition of their VOCs was population-specific. Plants exposed to VOCs from the same population responded more effectively to the stimulus than plants that received VOCs from a different population. As a result, the latter plants underwent higher herbivory, as assessed 3 months later ( Karban, 2015 ). By altering the VOCs in a plant headspace—which ‘meaning’ was presumably the same—the researchers altered the plant’s capacity to fight predators.

In summary, VOCs are part of the extended cognitive process of plants, ensuring rapid integration of information in the aboveground organs as an external resource that fills with relevant infochemicals the air around the plant’s crown.

8. Organism(s)-environment feedbacks

Many niche-constructing organisms are known for building complex structures out of their collective work. The behaviour of ants and termites is an important example of swarm intelligence, when organisms follow a small set of simple rules or have limited information-processing, but their interaction leads to the emergence of complex collective behaviours ( Krause et al., 2010 ).

For example, it was previously thought that termites build their nests based on simple sets of pheromones released by their peers on soil pellets, in a process of indirect communication called stigmergy ( Grassé, 1959 ). In this process, a worker would release in the environment a pheromone that would be perceived by another worker, which would also release more pheromones, and so on. In a positive feedback loop cycle, places with stronger pheromones would present more labour activity, and this would regulate overall nest behaviour ( Calovi et al., 2019 ). Nevertheless, the pheromone aspect of stigmergy has been challenged since other mechanisms have been discovered, while no pheromone involved in nest construction has been identified to date ( Camazine et al., 2003 ; Calovi et al., 2019 ). The instructions for nest-building seem to be elsewhere.

The construction of termite nests was demonstrated to be a complex, self-organised process, built upon a set of simple rules that each termite follows individually, which ultimately leads to the architectural complexity of the nest ( Camazine et al., 2003 ; Facchini et al., 2020 ). Nevertheless, when building nests, termites are known for relying more on environmental cues than their own memory for making decisions ( Fouquet et al., 2014 ). This characteristic, together with the fact that they are prime examples of niche-constructing organisms, makes termites promising candidates to extend their cognition.

Accordingly, Fouquet et al. (2014) demonstrated that termites rely more strongly on the surface structure of a site than on chemical marking to make decisions on where to dig and where to deposit soil pellets. In a set of experiments, they exposed the termites to structurally different environments and observed that the structure orients the deposition of pellets, which feedbacks-forwards to the building of the nest. For example, when they were exploring the territory, they do not wander randomly, but form trails. Because of their walking, the trails are more depressed than the environment around, which stimulates the termites to deposit pellets on the edges of the trails. Consequently, the trails become some sort of template for the building of the new nest and a kind of ‘collective memory’ for the new, emerging structure ( Fouquet et al., 2014 ).

Calovi et al. (2019) hypothesised that the soil displacement activity of Macrotermes michaelseni termites would be determined by the inclination of the surface where they are working. To test it, they manipulated the surface by building a support with different, controlled inclinations. They covered it with nest sediments, placed termites on it and observed their activity. Their results demonstrate that termites are particularly sensitive to the surface curvature, which was correlated with soil displacement and building activity (in accordance with findings of Fouquet et al., 2014 ). The soil geography is the main cue that drives the termites’ constructing behaviour, and not necessarily the cues released in the environment. It is clear that the information-processing to build a nest does not occur solely inside the termites’ brains ( Calovi et al., 2019 ). Based on their findings, the authors proposed that both the termites and the nest would be part of a same cognitive entity, where the nest serves as long-term memory repository and the guiding force to the construction of the nest itself. The termites are part of this cognitive process by ‘reading’ the forms of their own labour and continuing this work. In short, there is a cognitive process that involves each individual termite and their own, physical nest ( Calovi et al., 2019 ).

The studies above suggest the possibility that termites may offload the cognitive burden of knowing how to build a complex nest not only to their own partners but also to the nest itself, which is a non-living component of their cognitive process 1 . The manipulation of the environment structure affects the building and structure of the nest because what guides the nest-building is the environment itself. This is different from humans adjusting the project of a house to the terrain where they are building because in this case the cognitive process of projecting the house is extrinsic to the environment, and not driven by it. Building a suspended house in an inclined terrain, for example, or dividing it in two floors, may be clever solutions, but not the only possible ones to build on this terrain. If the only way of building the house was tied to the interaction of the workers with the environmental characteristics of the building place, the idea of an extended cognitive process could be raised. Finally, other social and nest-building insects like ants, wasps and bees could be good candidates for examining extended cognition in these invertebrate systems in experiments specifically designed for testing this possibility.

9. Gut microbiota as part of cognition

Boem et al. (2021) proposed that the gut microbiota can act as a cognition element by modulating bidirectional neural pathways, neuroendocrine signalling, as well as its impact on the immune system. Boem and colleagues did not mention the MM criterion for demarcating extended cognition; but other researchers partially employed the MM criterion by altering the gut microbiota element and observing brain connectivity changes. For instance, Tillisch et al. (2013) demonstrated that chronic intake of fermented milk with living probiotic bacteria modulates brain activity related to emotional attention tasks in humans. In a more recent study, Bagga et al. (2018) used a 4-week treatment of a multi-strain oral probiotic supplement to evaluate changes in human brain activity using functional magnetic resonance imaging. After treatment, changes in decision-making and memory processes were observed, especially related to unpleasant stimuli when compared to neutral ones. It suggests that the impact of gut microbiota on human brain activity is even bigger in the context of emotions. Unfortunately, both experimental designs only satisfy one condition of the MM criterion (i.e. altering the ‘environment’ and observing its impacts on the cognitive system). If we are to say that gut microbiota is part of the human extended cognition process, it is critical that both conditions of the MM criterion are met. If the cognitive state of the subjects is proven to alter the gut microbiota just like gut microbiota alters the cognitive state of the subjects, we will have empirical evidence for another way by which humans and other mammals extend their cognition.

Moving from human to rodents, studies with mice have yielded more interesting results. Bottom-up manipulations have been made, for example, by transferring the gut microbiota of human patients with depression to rats pre-treated with antibiotics, causing them to develop depressive symptoms and anxiety-like behaviours ( Kelly et al., 2016 ). Other manipulations of mice gut microbiota with probiotics or transfer of microbiota caused enhanced exploratory behaviour and reduced anxiety ( Bercik et al., 2011 ; Bravo et al., 2011 ), while the absence of microbiota colonisation was demonstrated to change neurogenesis, the density of network connections in the brain, and alter behaviour in germ-free mice ( Scott et al., 2020 ).

Finally, a study particularly interesting to our case was carried out by Chevalier et al. (2020) . The researchers stressed mice using an unpredictable chronic mild stress (UCMS) protocol, which is a mouse model for depression. They observed that mice who had been through UCMS had altered gut microbiota characterised by losses in Lactobacillaceae , decreased endocannabinoid (eCB) signalling and as a consequence, presented reduced adult neurogenesis and depressive-like behaviours. This is an interesting example of top-down intervention that causes an alteration in gut microbiota. Then, Chevalier et al. (2020) orally transferred the faecal microbiota of UCMS and unstressed control mice to germ-free and pathogen-free naïve mice (treated with antibiotics). The researchers observed that recipients colonised with UCMS microbiota developed the same depressive-like symptoms and neurological disorders, an effect not observed in the mice that received control microbiota. This can be considered a bottom-up manipulation where an alteration in the gut microbiota (the putative element of the extended cognitive system) resulted in a corresponding alteration in the explanandum phenomenon (mouse cognition). Finally, Chevalier et al. (2020) did a further bottom-up manipulation when they supplemented UCMS-recipient mice with probiotic diet enriched with Lactobacillus plantarum . Consequently, this treatment reversed the depressive-like behaviour of mice and partially restored hippocampal neurogenesis. The authors concluded that dysbiosis disrupted fatty acid metabolism, reducing the availability of eCB signalling precursors in the mice gut, which caused the behavioural and neurological effects observed. This is a beautiful example of the MM criterion being employed—perhaps, inadvertently—to establish constitutive relevance of mice gut microbiota for mice (extended) cognition.

10. Conclusion

Here, we reviewed studies on a broad range of organisms that could likely extend their cognitions. This review is by no means exhaustive; rather, it provides a starting point for considering how common extended cognition might be in nature. We propose that extended cognition may be present in organisms as diverse as humans, mice, spiders, plants, fungi, termites, protists and even cancer cells, and may occur in many different manners. When we observe organisms manipulating their environment and behaving in ways that seem far more complex than it would be expected, we can suspect that there might be an extended cognitive process taking place. Consequently, extended cognition might be much more common than we currently think, especially if we consider that basal organisms also possess this ability. The need for designing studies that attend to the mutual manipulability criterion as a form of clearly demarcating the boundaries of cognitive systems and creating a common base for comparisons is essential. Experimentally testing for extended cognition in as many organisms as possible, could help shedding some light on the evolution of cognition, a process that appeared in the planet billions of years ago, well before the evolution of brains. Cognition emerges through the interaction between the cognitive agent and the environment ( Clark, 1997 ; Lyon, 2006 ; Varela et al., 2016 ). In some cases, this interaction could have led to a tighter coupling between organisms and environment that facilitated the organisms’ cognitive process, thus increasing their adaptability. As an evolutionary consequence, some external elements became an integral part of the organisms’ cognitive system, a possibility that blurs even more the boundaries of cognitive processes. Studies on extended cognition may prove to be necessary for a better understanding of what cognition is, how it works and how it evolved.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

AP and MG conceived the manuscript. AP, GG, SW, and MG contributed to writing and discussions. All authors contributed to the article and approved the submitted version.

This work was partly funded by the Templeton World Charity Foundation (Grant TWCF-2020-20541).

Acknowledgments

The authors thank Madeleine Beekman, Pamela Lyon, Paco Calvo, and Gustavo Maia Souza for early discussions on non-neural cognition, and the reviewers for their comments. The authors also thank Raffaele Parise for the design of Figure 3 .

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

1. ^ We acknowledge that the MM is a debated criterion for the identification of extended cognitive systems. For more information on this debate see Krickel (2018) , Smart (2022) , and references therein.

Adams, F., and Aizawa, K. (2001). The bounds of cognition. Philos. Psychol. 14, 43–64. doi: 10.1080/09515080120033571

CrossRef Full Text | Google Scholar

Albín, A., Lacava, M., and Viera, C. (2014). Effects of gonyleptidine on the orb-web spider Araneus lathyrinus (Holmberg, 1875). Arachnology 16, 154–156. doi: 10.13156/arac.2012.16.5.154

Aleklett, K., and Boddy, L. (2021). Fungal behaviour: a new frontier in behavioural ecology. Trends Ecol. Evol. 36, 787–796. doi: 10.1016/j.tree.2021.05.006

PubMed Abstract | CrossRef Full Text | Google Scholar

Aleklett, K., Ohlsson, P., Bengtsson, M., and Hammer, E. C. (2021). Fungal foraging behaviour and hyphal space exploration in micro-structured soil chips. ISME J. 15, 1782–1793. doi: 10.1038/s41396-020-00886-7

Allen-Hermanson, S. (2013). Superdupersizing the mind: extended cognition and the persistence of cognitive bloat. Philos. Stud. 164, 791–806. doi: 10.1007/s11098-012-9914-7

Armitage, K. L., Bulley, A., and Redshaw, J. (2020). Developmental origins of cognitive offloading. Proc. R. Soc. B 287:20192927. doi: 10.1098/rspb.2019.2927

Bagga, D., Reichert, J. L., Koschutnig, K., Aigner, C. S., Holzer, P., Koskinen, K., et al. (2018). Probiotics drive gut microbiome triggering emotional brain signatures. Gut Microbes 9, 1–11. doi: 10.1080/19490976.2018.1460015

Bateson, G. (1972). Steps to an Ecology of Mind . London: Jason Aronson Inc

Google Scholar

Bayne, T., Brainard, D., Byrne, R. W., Chittka, L., Clayton, N., Heyes, C., et al. (2019). What is cognition? Curr. Biol. 29, R608–R615. doi: 10.1016/j.cub.2019.05.044

Bedecarrats, A., Chen, S., Pearce, K., Cai, D., and Glanzman, D. (2018). RNA from trained Aplysia can induce an epigenetic engram for long-term sensitization in untrained Aplysia . eNeuro. 5:e0038. doi: 10.1523/ENEURO.0038-18.2018

Beer, R. D. (2000). Dynamical approaches to cognitive science. Trends Cogn. Sci. 4, 91–99. doi: 10.1016/S1364-6613(99)01440-0

Bercik, P., Denou, E., Collins, J., Jackson, W., Lu, J., Jury, J., et al. (2011). The intestinal microbiota affect central levels of brain-derived neurotropic factor and behavior in mice. Gastroenterology 141, 599–609.e3. doi: 10.1053/j.gastro.2011.04.052

Boem, F., Ferretti, G., and Caiani, S. Z. (2021). Out of our skull, in our skin: the microbiota-gut-brain axis and the extended cognition thesis. Biol. Philos. 36:14. doi: 10.1007/s10539-021-09790-6

Boussard, A., Delescluse, J., Pérez-Escudero, A., and Dussutour, A. (2019). Memory inception and preservation in slime moulds: the quest for a common mechanism. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 374:20180368. doi: 10.1098/rstb.2018.0368

Bravo, J. A., Forsythe, P., Chew, M. V., Escaravage, E., Savignac, H. M., Dinan, T. G., et al. (2011). Ingestion of lactobacillus strain regulates emotional behavior and central GABA receptor expression in a mouse via the vagus nerve. Proc. Natl. Acad. Sci. U. S. A. 108, 16050–16055. doi: 10.1073/pnas.1102999108

Burr, C. (2017). “Embodied decisions and the predictive brain” in Philosophy and Predictive Processing . eds. T. Metzinger and W. Wiese, vol. 7 (Frankfurt am Main, Germany: MIND Group), 1–23.

Calovi, D. S., Bardunias, P., Carey, N., Turner, J. S., Nagpal, R., and Werfel, J. (2019). Surface curvature guides early construction activity in mound-building termites. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 374:20180374. doi: 10.1098/rstb.2018.0374

Camazine, S., Deneubourg, J.-L., Franks, N. R., Sneyd, J., Theraulaz, G., and Bonabeau, E. (2003). Self-Organization in Biological Systems . Princeton, USA: Princeton University Press

Castiello, U. (2021). (re)claiming plants in comparative psychology. J. Comp. Psychol. 135, 127–141. doi: 10.1037/com0000239

Cazalis, R., Carletti, T., and Cottam, R. (2017). The living organism: strengthening the basis. Biosystems 158, 10–16. doi: 10.1016/j.biosystems.2017.04.007

Chevalier, G., Siopi, E., Guenin-Macé, L., Pascal, M., Laval, T., Rifflet, A., et al. (2020). Effect of gut microbiota on depressive-like behaviors in mice is mediated by the endocannabinoid system. Nat. Commun. 11:6363. doi: 10.1038/s41467-020-19931-2

Clark, A. (1997). Being There: Putting Brain, Body and World Together Again . Cambridge, MA: MIT Press

Clark, A. (2008). Supersizing the Mind . Oxford, UK: Oxford University Press

Clark, A., and Chalmers, D. (1998). The extended mind. Analysis 58, 7–19. doi: 10.1093/analys/58.1.7

Coles, N. A., Larsen, J. T., and Lench, H. C. (2019). A meta-analysis of the facial feedback literature: effects of facial feedback on emotional experience are small and variable. Psychol. Bull. 145, 610–651. doi: 10.1037/bul0000194

Craver, C. F. (2007). Explaining the Brain . Oxford, UK: Oxford University Press

Craver, C. F., Glennan, S., and Povich, M. (2021). Constitutive relevance & mutual manipulability revisited. Synthese 199, 8807–8828. doi: 10.1007/s11229-021-03183-8

Donà, E., Barry, J. D., Valentin, G., Quirin, C., Khmelinskii, A., Kunze, A., et al. (2013). Directional tissue migration through a self-generated chemokine gradient. Nature 503, 285–289. doi: 10.1038/nature12635

Dreyfus, H. L. (1972). What Computers Can’t Do , New York: MIT Press

Eberhard, W. G. (2000). Spider manipulation by a wasp larva. Nature 406, 255–256. doi: 10.1038/35018636

Eberhard, W. G., and Hesselberg, T. (2012). Cues that spiders (Araneae: Araneidae, Tetragnathidae) use to build orbs: lapses in attention to one set of cues because of dissonance with others? Ethology 118, 610–620. doi: 10.1111/j.1439-0310.2012.02048.x

Engert, V., and Pruessner, J. C. (2008). Dopaminergic and noradrenergic contributions to functionality in ADHD: the role of methylphenidate. Curr. Neuropharmacol. 6, 322–328. doi: 10.2174/157015908787386069

Evans, I. R., and Wood, W. (2014). Drosophila blood cell chemotaxis. Curr. Opin. Cell Biol. 30, 1–8. doi: 10.1016/j.ceb.2014.04.002

Facchini, G., Lazarescu, A., Perna, A., and Douady, S. (2020). A growth model driven by curvature reproduces geometric features of arboreal termite nests. J. R. Soc. Interface 17:20200093. doi: 10.1098/rsif.2020.0093

Falik, O., Hoffmann, I., and Novoplansky, A. (2014). Say it with flowers: flowering acceleration by root communication. Plant Signal. Behav. 9:e28258. doi: 10.4161/psb.28258

Falik, O., Reides, P., Gersani, M., and Novoplansky, A. (2005). Root navigation by self inhibition. Plant Cell Environ. 28, 562–569. doi: 10.1111/j.1365-3040.2005.01304.x

Fouquet, D., Costa-Leonardo, A. M., Fournier, R., Blanco, S., and Jost, C. (2014). Coordination of construction behavior in the termite Procornitermes araujoi : structure is a stronger stimulus than volatile marking. Insect. Soc. 61, 253–264. doi: 10.1007/s00040-014-0350-x

Gagliano, M. (2013). Green symphonies: a call for studies on acoustic communication in plants. Behav. Ecol. 24, 789–796. doi: 10.1093/beheco/ars206

Gagliano, M. (2015). In a green frame of mind: perspectives on the behavioural ecology and cognitive nature of plants. AoB Plants 7:plu075. doi: 10.1093/aobpla/plu075

Gazzaniga, M. S. (1967). The split brain in man. Sci. Am. 217, 24–29. doi: 10.1038/scientificamerican0867-24

Golstein, D. G., and Gigerenzer, G. (2002). Models of ecological rationality: the recognition heuristic. Psychol. Rev. 109, 75–90. doi: 10.1037/0033-295X.109.1.75

Grassé, P.-P. (1959). La théorie de la stigmergie: essai d’interprétation du comportement des termites constructeurs. Insect. Soc. 6, 41–80. doi: 10.1007/BF02223791

Harms, W. F. (2006). What is information? Three concepts. Biol. Theory 1, 230–242. doi: 10.1162/biot.2006.1.3.230

Heil, M., and Karban, R. (2010). Explaining evolution of plant communication by airborne signals. Trends Ecol. Evol. 25, 137–144. doi: 10.1016/j.tree.2009.09.010

Heil, M., and Kost, C. (2006). Priming of indirect defences. Ecol. Lett. 9, 813–817. doi: 10.1111/j.1461-0248.2006.00932.x

Heil, M., and Silva Bueno, J. C. (2007). Within-plant signaling by volatiles leads to induction and priming of an indirect plant defense in nature. Proc. Natl. Acad. Sci. U. S. A. 104, 5467–5472. doi: 10.1073/pnas.0610266104

Hesselberg, T., and Vollrath, F. (2004). The effects of neurotoxins on web geometry and web-building behaviour in Araneus diadematus cl. Physiol. Behav. 82, 519–529. doi: 10.1016/j.physbeh.2004.04.058

Hobson, E. A., Mønster, D., and DeDeo, S. (2021). Aggression heuristics underlie animal dominance hierarchies and provide evidence of group-level social information. Proc. Natl. Acad. Sci. U. S. A. 118:e2022912118. doi: 10.1073/pnas.2022912118

Huang, A. C., Jiang, T., Liu, Y.-X., Bai, Y.-C., Reed, J., Qu, B., et al. (2019). A specialized metabolic network selectively modulates Arabidopsis root microbiota. Science 364:eaau6389. doi: 10.1126/science.aau6389

Japyassú, H. F., and Laland, K. N. (2017). Extended spider cognition. Anim. Cogn. 20, 375–395. doi: 10.1007/s10071-017-1069-7

Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biol. Philos. 27, 545–570. doi: 10.1007/s10539-012-9308-4

Karban, R. (2015). Plant Sensing and Communication . Chicago: The University of Chicago Press.

PubMed Abstract | Google Scholar

Kelly, J. R., Borre, Y., O’Brian, C., Patterson, E., El Aidy, S., Deane, J., et al. (2016). Transferring the blues: depression-associated gut microbiota induces neurobehavioural changes in the rat. J. Psychiatr. Res. 82, 109–118. doi: 10.1016/j.jpsychires.2016.07.019

Krause, J., Ruxton, G. D., and Krause, S. (2010). Trends Ecol. Trends Ecol. Evol. 25, 28–34. doi: 10.1016/j.tree.2009.06.016

Krickel, B. (2018). Saving the mutual manipulability account of constitutive relevance. Stud. Hist. Phil. Sci. 68, 58–67. doi: 10.1016/j.shpsa.2018.01.003

Latty, T., and Beekman, M. (2015). Slime moulds use heuristics based on within-patch experience to decide when to leave. J. Exp. Biol. 218, 1175–1179. doi: 10.1242/jeb.116533

Ludwig, D. (2015). Extended cognition and the explosion of knowledge. Philos. Psychol. 28, 355–368. doi: 10.1080/09515089.2013.867319

Lyon, P. (2006). The biogenic approach to cognition. Cogn. Process. 7, 11–29. doi: 10.1007/s10339-005-0016-8

Lyon, P., Keijzer, F., Arendt, D., and Levin, M. (2021). Reframing cognition: getting down to biological basics. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 376:20190750. doi: 10.1098/rstb.2019.0750

Mahall, B. E., and Callaway, R. M. (1991). Root communication among desert shrubs. Proc. Natl. Acad. Sci. U. S. A. 88, 874–876. doi: 10.1073/pnas.88.3.874

Mahall, B. E., and Callaway, R. M. (1992). Root communication mechanisms and intracommunity distributions of two Mojave Desert shrubs. Ecology 73, 2145–2151. doi: 10.2307/1941462

Marsh, B. (2002). Do animals use heuristics? J. Bioecon. 4, 49–56. doi: 10.1023/A:1020655022163

Mathel, J. A., and Dickel, L. (2017). Complex cephalopod cognition. Curr. Opin. Behav. Sci. 16, 131–137. doi: 10.1016/j.cobeha.2017.06.008

Maturana, H. R., and Varela, F. J. (1980). Autopoiesis and Cognition: The Realization of the Living . Dordrecht, Holland: D. Reidel Publishing Company

Menary, R. (2010a). “Cognitive integration and the extended mind” in The Extended Mind . ed. R. Menary (Cambridge, MA: The MIT Press), 189–225.

Menary, R. (2010b). The Extended Mind . Cambridge, MA: MIT Press

Mortimer, B., Soler, A., Siviour, C. R., and Vollrath, F. (2018). Remote monitoring of vibrational information in spider webs. Sci. Nat. 105, 37–39. doi: 10.1007/s00114-018-1561-1

Nakata, K. (2010). Attention focusing in a sit-and-wait forager: a spider controls its prey-detection ability in different web sectors by adjusting thread tension. Proc. R. Soc. B Biol. Sci. 277, 29–33. doi: 10.1098/rspb.2009.1583

Nakata, K. (2013). Spatial learning affects thread tension control in orb-web spiders. Biol. Lett. 9, 20130052–20130054. doi: 10.1098/rsbl.2013.0052

Neisser, U. (1967). Cognitive Psychology . New York: Appleton-Century-Crofts

Neisser, U. (1976). Cognition and Reality . San Francisco: W. H. Freeman and Company

Onishi, S., Tobita, K., and Makioka, S. (2022). Hand constraint reduces brain activity and affects the speed of verbal responses on semantic tasks. Sci. Rep. 12:13545. doi: 10.1038/s41598-022-17702-1

Parise, A. G., Gagliano, M., and Souza, G. (2020). Extended cognition in plants: is it possible? Plant Signal. Behav. 15:1710661. doi: 10.1080/15592324.2019.1710661

Perry, C. J., Barron, A. B., and Chittka, L. (2017). The frontiers of insect cognition. Curr. Opin. Behav. Sci. 16, 111–118. doi: 10.1016/j.cobeha.2017.05.011

Reid, C. R., Latty, T., Dussutour, A., and Beekman, M. (2012). Slime mold uses an externalized spatial “memory” to navigate in complex environments. Proc. Natl. Acad. Sci. U. S. A. 109, 17490–17494. doi: 10.1073/pnas.1215037109

Rupert, R. D. (2004). Challenges to the hypothesis of extended cognition. J. Philos. 101, 389–428. doi: 10.5840/jphil2004101826

Scott, G. A., Terstege, D. J., Vu, A. P., Law, S., Evans, A., and Epp, J. R. (2020). Disrupted neurogenesis in germ-free mice: effects of age and sex. Front. Cell Dev. Biol. 8:407. doi: 10.3389/fcell.2020.00407

Shannon, C. E. (1948). A mathematical theory of communication. Bell Syst. Tech. J. 27, 379–423. doi: 10.1002/j.1538-7305.1948.tb01338.x

Shapiro, J. A. (2007). Bacteria are small but not stupid: cognition, natural genetic engineering and socio-bacteriology. Stud. Hist. Phil. Biol. Biomed. Sci. 38, 807–819. doi: 10.1016/j.shpsc.2007.09.010

Shapiro, J. A. (2021). All living cells are cognitive. Biochem. Biophys. Res. Commun. 564, 134–149. doi: 10.1016/j.bbrc.2020.08.120

Shettleworth, S. J. (2010). Cognition, Evolution, and Behavior . Oxford, UK: Oxford University Press

Simon, H. (1955). A behavioral model of rational choice. Q. J. Econ. 69, 99–118. doi: 10.2307/1884852

Sims, M., and Kiverstein, J. (2022). Externalized memory in slime mould and the extended (non-neural) mind. Cogn. Syst. Res. 73, 26–35. doi: 10.1016/j.cogsys.2021.12.001

Smart, P. R. (2022). Toward a mechanistic account of extended cognition. Philos. Psychol. 35, 1107–1135. doi: 10.1080/09515089.2021.2023123

Smart, P., Clowes, R. W., and Heersmink, R. (2017). Minds online: the interface between web science, cognitive science and the philosophy of mind. Foundat. Trends Web Sci. 6, 1–232. doi: 10.1561/1800000026

Souza, G. M., Toledo, G. R. A., and Saraiva, G. F. R. (2018). “Towards systemic view for plant learning: ecophysiological perspective” in Memory and Learning in Plants . eds. F. Baluška, M. Gagliano, and G. Witzany (Cham: Springer), 163–190.

Sutton, J. (2010). “Exograms and interdisciplinarity: history, the extended mind, and the civilizing process” in The Extended Mind . ed. R. Menary (Cambridge, MA: The MIT Press), 189–225.

Tillisch, K., Labus, J., Kilpatrick, L., Jiang, Z., Stains, J., Ebrat, B., et al. (2013). Consumption of fermented milk product with probiotic modulates brain activity. Gastroenterology 144, 1394–1401.e4. doi: 10.1053/j.gastro.2013.02.043

Trewavas, A. (2016). Intelligence, cognition, and language of green plants. Front. Psychol. 7:588. doi: 10.3389/fpsyg.2016.00588

Tweedy, L., Susanto, O., and Insall, R. H. (2016). Self-generated chemotactic gradients—cells steering themselves. Curr. Opin. Cell Biol. 42, 46–51. doi: 10.1016/j.ceb.2016.04.003

Tweedy, L., Thomason, P. A., Paschke, P. I., Martin, K., Machesky, L. M., Zagnoni, M., et al. (2020). Seeing around corners: cells solve mazes and respond at a distance using attractand breakdown. Science 369:eaay9792. doi: 10.1126/science.aay9792

Varela, F., Thompson, E., and Rosch, E. (2016). The Embodied Mind: Cognitive Science and Human Experience . Cambridge, MA: MIT Press.

Vogel, D., and Dussutour, A. (2016). Direct transfer of learned behaviour via cell fusion in non-neural organisms. Proc. R. Soc. Lond. B Biol. Sci. 283:20162382. doi: 10.1098/rspb.2016.2382

Watanabe, T. (2000). Web tuning of an orb-web spider, Octonoba sybotides , regulates prey-catching behaviour. Proc. R. Soc. Lond. B Biol. Sci. 267, 565–569. doi: 10.1098/rspb.2000.1038

Weiner, N. (1948). Cybernetics or the Control and Communication in the Animal and the Machine . Cambridge, USA: MIT Press

Wheeldon, C. D., Walker, C. H., Hamon-Josse, M., and Bennett, T. (2021). Wheat plants sense substrate volume and root density to proactively modulate shoot growth. Plant Cell Environ. 44, 1202–1214. doi: 10.1111/pce.13984

Witt, P. N. (1969). Behavioral consequences of laser lesions in the central nervous system of Araneus diadematus cl. Am. Zool. 9, 121–131. doi: 10.1093/icb/9.1.121

Witt, P. N. (1971). Drugs alter web-building of spiders: a review and evaluation. Behav. Sci. 16, 98–113. doi: 10.1002/bs.3830160109

Witzany, G., and Baluška, F. (2012). Life’s code script does not code itself: the machine metaphor for living organisms is outdated. EMBO Rep. 13, 1054–1056. doi: 10.1038/embor.2012.166

Yamagushi, T., Wang, H.-L., Li, X., Ng, T. H., and Morales, M. (2011). Mesocorticolimbic glutamatergic pathway. J. Neurosci. 31, 8476–8490. doi: 10.1523/JNEUROSCI.1598-11.2011

Yuan, J., Zhao, J., Wen, T., Zhao, M., Li, R., Goossens, P., et al. (2018). Root exudates drive the soil-borne legacy of aboveground pathogen infection. Microbiome 6, 1–12. doi: 10.1186/s40168-018

Zhong, Y., Xun, W., Wang, X., Tian, S., Zhang, Y., Li, D., et al. (2022). Root-secreted bitter triterpene modulates the rhizosphere microbiota to improve plant fitness. Nat. Plants. 8, 887–896. doi: 10.1038/s41477-022-01201-2

Keywords: extended cognition, behavioural ecology, cognitive traits, niche construction, navigation, memory, spider behaviour, plant behaviour

Citation: Parise AG, Gubert GF, Whalan S and Gagliano M (2023) Ariadne’s thread and the extension of cognition: A common but overlooked phenomenon in nature? Front. Ecol. Evol . 10:1069349. doi: 10.3389/fevo.2022.1069349

Received: 13 October 2022; Accepted: 19 December 2022; Published: 09 January 2023.

Reviewed by:

Copyright © 2023 Parise, Gubert, Whalan and Gagliano. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: André Geremia Parise, ✉ [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

What is the extension of the extended mind?

1 MCTS, Technische Universität München, Munich, Germany

2 Department of Philosophy, Alpen-Adria-Universität, Klagenfurt, Austria

Two aspects of cognitive coupling, as brought forward in the Extended Mind Hypothesis, are discussed in this paper: (1) how shall the functional coupling between the organism and some entity in his environment be spelled out in detail? (2) What are the paradigmatic external entities to enter into that coupling? These two related questions are best answered in the light of an aetiological variety of functionalist argument that adds historical depth to the “active externalism” promoted by Clark and Chalmers and helps to counter some of the core criticisms levelled against this view. Under additional reference to conceptual parallels between the Extended Mind Hypothesis and a set of heterodox theories in biology—environmental constructivism, niche construction, developmental systems theory—an argument for the grounding of environmentally extended cognitive traits in evolved biological functions is developed. In a spirit that seeks to integrate extended functionalism with views from cognitive integration and complementarity, it is argued (ad 1) that instances of environmental coupling should be understood as being constitutive to cognitive functions in either of two distinct ways. It is further argued (ad 2) that the historically and systematically prior environmental counterparts in that coupling are features of the natural environment. Language and linguistically imbued artefacts are likely to have descended from more basic relations that have an extension over the environment.

Introduction

In their brief, bold, and controversial manifesto of a thorough, “active” externalism in the philosophy of mind, titled “The Extended Mind”, Clark and Chalmers highlight “the active role of the environment in driving cognitive processes” ( 1998 , p. 7). Not only does the content of mental processes or linguistic expressions depend on what is in the environment—as it does according to externalist views in the philosophy of mind and language, respectively. 1 In some respects at least, the very accomplishment of cognitive processes, and, in a relevant subset of cases, their very existence, depends on what is in the environment. Clark and Chalmers argue that there are situations in which “the human organism is linked with an external entity in a two-way interaction, creating a coupled system that can be seen as a cognitive system in its own right” ( 1998 , p. 8, emphasis in original).

This externalist position, in its general outlook, shall be endorsed here. However, the notion of cognitive extension has become many things to many people in a continuously sprawling debate, where the most forceful critiques appear to frequently miss the most serious points that arguments for extended cognition are capable of making. 2

On pain of adding another specimen to the zoo of views what cognitive extensions really are, I shall carve out an interpretation that strengthens the functionalist credentials of the Extended Mind Hypothesis by rooting it in natural history. My guiding questions will be these: firstly, how shall the functional coupling between the organism and some entity in his environment be spelled out in detail? Secondly, what are the paradigmatic external entities to enter into that coupling? In seeking to answer these questions, I will make a suggestion towards sharpening the notion of “extension” that is (or should be) in play here: the paradigm of cognitive extensions are those features of the environment which play either of two kinds of constitutive role to the function of some cognitive trait, and, in functional-historical terms, these features have been primarily natural rather than artificial. This suggestion shall help preventing the notion of cognitive extension from becoming extensionally bloated to the point of vacuousness. 3

I will begin by placing some of the central claims of the Extended Mind Hypothesis—its notion of “active” versus “passive” externalism and its notion of features of the environment serving cognitive functions—in the context of aetiological theories of function and some positions within evolutionary biology (Sects.  2 and 3 , respectively). The insights from this discussion will then be used to identify a number of possible interpretations of what extensions of the mind are, and what explanatory roles they might play in cognitive inquiries, with particular attention to the specific ways in which features of the environment may be constitutive to cognitive functions (Sect.  4 ). The constitutiveness theme will be further discussed in Sect.  5 , with the aim of providing some common ground for “parity” and “complementarity” views of extendedness. Finally, I will discuss two paradigms of cognitive extension in the light of the aetiological variety of functionalist argument developed in the preceding sections (Sect.  6 ).

The extension of functional histories

When locating themselves within the context of contemporary philosophical debates, Clark and Chalmers ( 1998 , p. 9) characterise active externalism as “active” for its focus on interactions between organism and environment in the here and now. The activity of interest is in the environment and the organism at the same instance, and it is that concurrent activity which serves to make both cognition extended and externalism active. Conversely, Clark and Chalmers characterise traditional semantic externalism as “passive” for its focus on the causal history that supposedly endows linguistic items, and by extension of the argument (McGinn 1977 ), mental events with their meanings. It is this history of interactions that explains any possible difference between the contents of two prima facie identical mental or linguistic tokens, no matter what the current interactions may look like to participants and observers.

Hence, if one takes active externalism by its ahistorical first word, the chemical difference between water and twin water in Putnam’s well-known thought experiment ( 1975a ), contrary to what its author aimed to demonstrate, will not matter as long as my interaction with either substance produces identical effects, cognitively and practically. Conversely, on the “passive” account, the different histories of earthly and twin-earthly water are all that matters in making the difference between the semantic success conditions for “water” tokens being applied to either substance, even if I cannot see, feel or taste any difference between the two.

Although the active/passive terminology is now well-established (see, for example, Carter et al. 2014 ), there remains a certain artificiality to referring to the varieties of externalism in this way. Past interactions have been relevant to shaping the content of some present linguistic or mental token, and there is no reason to assume the past to be in any normative sense remote from the present. What Clark and Chalmers are after is quite different from Putnam-style semantic externalism though: their focus is on the locus of cognitive processes, whereas Putnam, Burge, McGinn and many others are concerned with the external conditions that ground the content of mental or linguistic tokens. The difference at issue here is also known as “vehicle” versus “content” externalism (see Hurley 1998 , 2010 ). On an ahistorical reading of active externalism, the concrete content of what is being processed appears secondary to the issue of the locus of cognitive activity, just as history is deemed secondary to present states of cognitive affairs.

Such a choice of priorities might be unfortunate, for several interrelated reasons—of which, to be sure, theories of extended cognition have become at least partly aware since that initial statement. 4 Nonetheless, it will be worthwhile to spell out why, in terms of a unified rationale, any serious theory of extended cognition will benefit from accounting for the history of the functions of environmental coupling.

Firstly, when referring to the functions of some object or process, one will have to tell the difference between genuine functions and coincidental effects. That difference can only be established on the grounds of a history of how the object or process in question came to display the effects that it displays—and why it fails to do so under some circumstances. For artefacts, their design will provide part of the relevant criteria, whereas all relevant criteria will be historical in the case of natural systems and their functions.

Secondly, the content of cognitive activities—what in the environment they refer to and whether, why and how they succeed in doing so—is relevant to an explanation of the shape, function and extension of these activities, historically and at present, on more than one level. On the one hand, the concrete content of some cognitive process depends on the history of cognisers living in a concrete natural and social environment to which they relate. More fundamentally, on the other hand, the mechanisms for producing and acting upon these contents are rooted in histories of evolution and ontogenetic development, broadly construed.

Thirdly, present interactions dynamically reshape the mode of extension in many cases. There is no ending to history. If the conditions under which some relation of cognitive coupling is realised are altered—either from within or from outside the organism—that change, unless ending in failure of the entire process, will provide some degree of variation that will result in a modified coupling relation further down the line. Variability in environmental conditions will be one key factor in shaping coupling relations.

The common rationale to unify this set of claims is an aetiological, that is, historically based variety of functionalist argument. It departs from the classical Putnamian point of “machine state functionalism”, which is based on the assumption that one and the same function may be realised in a variety of different structures (Putnam 1975b ). This claim of multiple realisability has been central to any argument for the possibility of Artificial Intelligence, and it is equally important to the “parity principle” in the Extended Mind debates. It has its roots in Turing’s observation that any logical or mathematical operation that is computable at all can be realised by a set of elementary formal operations, which may in turn be implemented in a broad, in principle indefinite, variety of physical systems ( 1936 ). The claim of multiple realisability is also implied by observations on neuronal equipotentiality, according to which one cognitive or perceptual function can be realised by a variety of neuroanatomical structures (see, for example, the discussion of Karl Lashley’s work in Proust 1995 ).

While sharing with these types of functionalism the basic notion of a relation of underdetermination between functions and structural properties of some system, an aetiological account is interested in the concrete enabling and constraining conditions under which functions come to be established and realised. Functional aetiologies have their natural home in natural history and the Darwinian theory of evolution. Charles Darwin gave an elegant interpretation to the observation made by many naturalists that commonalities in form (“homologies”) and commonalities in function (“analogies”) might be, partly or wholly, disjunct between species of different taxa. 5 On the one hand, one ancestral form, for example a mammal’s forelegs, may come to serve divergent functions—as wings in bats and arms in human beings. On the other hand, one and the same function may come to be accomplished by structures of various degrees of similarity that have developed along independent pathways in genealogically remote species—as for the structurally similar but evolutionary independent lensed eyes in vertebrates and many Cephalopods, and the structurally dissimilar and equally independent compound eyes in Arthropoda. Accordingly, the concrete histories of convergent and divergent functions within and between populations will matter to an aetiological account.

If cognitive traits have biological functions in the same way as other traits of an organism have biological functions, one will be entitled to analyse the mechanisms that realise these processes in the same functional-historical terms. Like any other trait, they will be subject to processes of variation and natural selection. In turn, the content of these processes can be analysed in analogous fashion, to the extent that a type of cognitive state of one or a number of related individuals is constituted by reproducible tokens that may succeed or fail to map onto some world affair, and hence be selected to accomplish that mapping. These are the basic assumptions of what has become known as “teleosemantics” or “teleofunctionalism”, as first and most influentially introduced by Millikan ( 1984 ). 6

More precisely, a functional-historical account of some cognitive trait will recur to the contributions that the effects of that trait have made with sufficient frequency among its ancestral bearers to their rate of reproduction as compared to other members of the same population who did not possess that trait. By virtue of conferring a reproductive advantage over the course of several generations, the trait will acquire the proper function of producing these effects. 7 The “ancestors” and “generations”, rather than being organisms themselves, may also be component mechanisms, and they may also be reproducible forms that can be iteratively used by some individual or a collective thereof, such as artefacts or linguistic items. The individual tokens of some such reproducible item have proper functions derived from the functions of the mechanisms that produce them. It is the direct proper function of those mechanisms to generate the kind of effects in question if these effects, qua individually bearing certain relations to the environment with sufficient reliability, form part of the necessary conditions for the reproduction of the mechanisms themselves and of the systems that rely on their presence. In turn, the tokens of some reproducible item have the derived proper function of being adapted to some concrete world affair if and when one relevant condition for their reproduction lies in their co-occurrence with that sort of world affair and, in a relevant subset of cases, their role in directing an organism’s behaviours towards it.

Nothing in this account rules out the possibility of mechanisms that are only partly based within an organism. In fact, Millikan ( 1993 , p. 179) herself considers the insight that “the organismic process has no skin” fundamental to the science of psychology. 8 Since all sorts of traits, cognitive and other, are characterised in terms of their historically acquired functions, and since that history incorporates whatever turns out to realise the function in question, the locus of the components involved in realising it is systematically irrelevant, under a two-part proviso: the organismic traits involved are to be coupled with persistent or reproducible features of their environment with a degree of reliability sufficient for meeting the reproduction criterion, where that coupling occurs in accordance with a uniform explanation that accounts for the “normal mechanisms” of coupling. Hence, although the content externalism defended in the aetiological account does not entail a vehicle externalism, it is methodically impartial to the locus of the vehicles involved.

As to the concrete benefits of an aetiological perspective for extended functionalism, accounting for the historical nature of functions will, firstly, provide an argument against the “coupling/constitution” objection brought forward by Adams and Aizawa ( 2001 ): in contrast to what the critics suspect, instances of accidental coupling between an organism and some entities in his environment will not count as coupling proper. Nor will legitimate instances of coupling endow the individual entities thus coupled with the characteristics of the entire coupled system. These entities play constitutive, historically established roles in the establishment or accomplishment of the functions of the entire system, where these roles may widely diverge. I will discuss this point in more detail in Sect.  4 .

Secondly, and for related reasons, accounting for the historical nature of functions will provide an argument against the “fleeting vs. persistence” criticism that has been levelled against extended cognition, first and foremost, by Rupert ( 2010 ). The suspicion here is that environmental extensions, unlike internally based cognitive capacities, are not persistently coupled with the organism. If some such extension is accidentally detached from the cognising organism, it seems, the cognitive system comes apart, only to be re-instantiated if and when that extension is successfully recovered. Clark’s reply to this charge was to emphasise the conditions of persistence and reliability (“glue and trust”), which supposedly rule out overly ephemeral coupling relations (Clark 2010a , p. 83 f). If the coupling in question has a historically established function, the issue of temporary non-occurrence or non-performance of that coupling relation, rather than being resolved by means of such auxiliary conditions, is actually dis solved: it does not matter if the function in question, for temporary loss of one component, turns out not to be performed in many instances, as long as the coupling occurs with a degree of reliability and according to an unequivocal regularity so as to become part of the explanation for the reproduction of the organic and, in some cases, the environmental components involved. To use a non-cognitive case of extension for comparison (see Menary 2010b , p. 14): it does not matter if a spider loses her web once in a while, or perhaps even very often, as long as the, however intermittent, production and use of webs provides part of the explanation for the presence of web-making mechanisms in the spider population.

Thirdly and most profoundly perhaps, a historical account of functions will counter Sprevak’s critique of classical machine-state functionalism ( 2009 ) which targets the “Martian intuition” that an identical or analogous kind of function can be realised by physically or biologically very differently constituted systems. On that intuition, which is basically the Lewisian one ( 1980 ), if we can specify a set of functions for certain traits and behaviours in human beings and Martians, we should be able in principle to identify which concrete structures within or around the human being or Martian will play what sort of causal role in accomplishing those functions. However, Sprevak objects, one could easily imagine the conditions of persistence and reliability of the respective structures and the coupling relations involved to be violated in manifold ways, and hence fail to meet the specifications laid out by extended functionalism. Playing on the Martian intuition in this way will probably kill the Martian—but, if we follow the historical account of functions, it will not kill the intuition. Any coupling with whatever structure that were as unreliable or otherwise insufficient or even freakish in nature as described by Sprevak would militate against the imagined organism’s reproductive chances, and against the factual (rather than conceptual) possibility of his existence. Conversely, Sprevak’s Martian may be imagined to be in command of capabilities and enter into environmental coupling relations so unfamiliar that we could not find any remote analogue in the specifications of human cognition. But then, the entire functional analogy will not hold. After all, there might be hyper-cognitive functions to Martians that human beings are insufficiently equipped to imagine in the first place. At any rate, the historical account of functions serves as a reminder that observations of real properties of real members of real populations provide a more solid foundation to an inquiry into cognitive functions than counterfactual-supporting conditions in some nearby or remote possible worlds.

Evolution, construction, and development

The reservations against history diagnosed in the previous section become explicit when Clark and Chalmers introduce a set of criteria that are supposed to be necessary and sufficient for the function of internal or external resources as components of cognitive processes ( 1998 , p. 17; see also Clark 2010b ):

• The resource should be of reliable availability, and should be used on a frequent basis.
• The information provided by that resource should be directly available.
• This information should be endorsed automatically, i.e., without requiring further reflection on its reliability.
• Present endorsement of this information should be based on conscious endorsement in the past.

The authors add a footnote to this list in which the implication of the constancy (3) and the past-endorsement (4) criteria that history is co-constitutive of cognitive processes is countered by the suggestion that the past-endorsement criterion might be dropped and the constancy criterion be given “a purely dispositional reading”: one is found simply to endorse the information automatically; it is not asked why. While this might be a sensible tactical manoeuvre in terms of catering for the broadest possible range of extensions of the mind, relativising the latter two criteria amounts to skirting an inquiry into the normal mechanisms of cognitive coupling. The past-endorsement criterion in particular would assimilate the argument to a teleofunctional one: the decisive condition for past endorsement being causally relevant to present endorsement is that the former has been successful, although it might also have been conscious. If endorsement had not been successful, although it might still have been conscious, there would be no endorsement at present.

Beyond tactical considerations, one possible systematic reason why the proponents of the Extended Mind Hypothesis have been reluctant to adopt a historical account of cognitive functions is detectable in Wheeler and Clark ( 2008 )—a contribution that actually acknowledges arguments for the rooting of cognitive extensions in evolutionary history: the authors match the dyad of extended/embodied cognition and cultural evolution hypotheses against the tenets of evolutionary psychology. The latter is charged with a view of evolution that takes psychological and physiological adaptations alike to be, temporally remotely anchored, solutions to a predetermined set of problems posed by the environment, where these solutions will be uniformly distributed over a species while being expressly domain-specific. Moreover, Wheeler and Clark continue, the possibility of organisms modifying their environments to fit their specific needs is not considered by evolutionary psychologists. The basic concern here appears to be that an anchoring of cognitive extensions in past states of affairs that are beyond present cognisers’ reach—especially if the past endorsement criterion concerns the evolutionary past—would undermine the notion of a two-way interaction that is central to the coupled system claim (see p. 1 above). Adaptations and the conditions to which they respond might be more malleable and more specific to sub-populations in the here-and-now than evolutionary psychology appears to consider.

In order to get a clearer view of what is at issue here, a brief look at Dawkins’ theory of the extended phenotype ( 1999 ) will help. Notably, it is listed among the biological cousins of extended cognition by Wilson and Clark ( 2009 ). Prima facie, the parallels seem striking: Dawkins not only highlights the importance of an organism’s interaction with the environment but also promotes the notion of biological traits extending into, or incorporating features of, the environment. Dawkins suggests a change of perspective of inquiry under which the organism ultimately becomes “transparent” ( 1999 , p. 4f, 250). What remains visible once the organism has become transparent are replicating gene sequences within a population and their interactions with their intra- or extra-somatic environment. By means of this, explicitly metaphorical, change of perspective, Dawkins argues that the organism might not always be the key unit to consider when explaining patterns of genetic replication. Instead, he suggests, an organism’s interaction with objects or other organisms within his environment will have to be counted into the explanation of whether and how a gene succeeds in replicating. Hence, we might, and often should, treat the organism and his environment in conjunction as the wider environment of some gene sequence in which it acts, and which it manipulates.

Precisely in rendering the organism thus transparent, Dawkins’ “gene’s eye view” re-instantiates the notion of a unitary, individual and individualistic agent who is clearly distinguished from his external environment, albeit with a substitution of the referent of “individual”. In many relevant cases, the manifest aims of individual organisms will end up subordinated to their genes’ hypothetical interests. Moreover, the systematic point of adopting this view within the context of evolutionary theorising is to highlight a possible, and perhaps the most relevant, unit of natural selection—as the main force in evolution and the source of any goal-directedness or progress therein. The effects that the organism’s own traits and behaviours might have on the course of evolutionary events will not figure prominently in such an account.

Wilson and Clark ( 2009 ) expressly endorse a continuum of theories in biology that takes issue with these implications of Dawkinsian view, and that bears more profound analogies to some of the central claims of extended cognition. That continuum encompasses, firstly, environmental constructivism as introduced by Lewontin ( 1982 ), secondly, the development of the latter into theories of niche construction, beginning with West and King ( 1987 ) and Odling-Smee ( 1988 ) and, thirdly, developmental systems theory, as developed by Oyama ( 2000 ). This continuum of theories is unified by its opposition to what has been termed by its critics (first by Gould and Lewontin 1979 ) the “adaptationist programme”, which in turn has Dawkins as one of its main advocates. For its emphasis on the nature of functions as, paradigmatically naturally, selected effects, and for its reliance on the positive selection for some trait, the teleofunctional account should be partly subsumable under the adaptationist rubric, too (these commitments are quite clearly stated in Millikan 2004 , Chaps. 1–2).

Environmental construction, on the account presented in Lewontin ( 1982 , 2000 ), will include any intervention by an organism or population into processes in the external environment that changes conditions therein in such a way that the intervention in question becomes a necessary part of the explanation of the nature of the organism or population and their adaptive success—or failure. The construction of environments is not to be understood in a straightforwardly literal sense, although the building of material structures might be involved. Environmental construction also includes cases where features of the environment remain physically unaltered while being treated in such a way as to affect the attainment of an organism’s or population’s goals. These two kinds of cases are distinguished by Godfrey-Smith ( 1996a , Chap. 5), in his critical reconstruction of Lewontin’s approach, as the construction of features of the environment in a more narrow sense, in terms of creating material structures and artefacts, versus the constitution of features of the environment, in terms of treating them in a certain way or putting oneself in a certain relation to them without materially changing the structure of the environment. (The use of the term “constitutive” by Godfrey-Smith will have to be distinguished from the notion of “constitutiveness” to be employed later in this essay.) Beaver dams and many other kinds of artefacts will be subsumed under the category of construction, activities of seeking out favourable conditions in space and time under the rubric of constitution. In either case, yet in different ways, organisms or populations are actively involved in determining the environmental conditions that are relevant to them.

The “construction” and the “constitution” types of cases, which are best viewed as complementary rather than mutually exclusive, should be distinguished from another type of intervention by organisms into environmental conditions that was also subsumed under the “construction” label by Lewontin: situations where the effects of an organism’s behaviours undermine rather than foster the attainment of his goals or his chances of reproduction. Such accidental side-effects only indirectly, and clearly only negatively, relate to the possible adaptive functions of the traits that produce those effects. If not in treating construction and constitution as equivalent, it is here where Lewontin’s notion of environmental construction becomes too broad to be entirely useful, as Godfrey-Smith observes.

Theories of niche construction, having found their paradigmatic formulations in Odling-Smee et al. ( 1996 , 2003 ), may provide more focus to the notion of construction involved here. Broadly speaking, ecological niches are the specific sets of conditions in an environment that are reproductively relevant to a certain population. Different populations inhabiting one area at one time will depend on different sets of conditions at that place at that time. The make-up of these conditions partly depends on the activities of the organisms themselves. While some of the authors who introduced the concept of ecological niches into biology (especially Elton 1927 ) found the idea of a space of pre-existing ecological niches perfectly permissible, where some niche may be inhabited by different populations over space and time or even left uninhabited, there is no such thing as independently defined ecological niche for the niche constructivist. Niches are both made by and might be undone by the organisms inhabiting them. If there is some adaptive function to the construction and constitution of features of the environment, these features will become part of the necessary conditions in an explanation of an organism’s success on a proximate behavioural or an ultimate selective level, or both. If these features are indeed part of the necessary conditions in question, there are good reasons to view them as coupled with the organism: he has to track the presence of some feature and put himself in the right relation to it, or he has to modify some feature in order to prevail.

An approach that, among other strands of heterodox biological theorising, incorporates environmental and niche constructionism, is to be found in developmental systems theory (Oyama 2000 ). Being a deliberate change of perspective on the animate realm rather than a conventional predictive theory, as Oyama et al. ( 2001 , 1f) admit in the introduction to their developmental systems anthology, it reverses the direction of change taken by Dawkins: a developmental system is the conjunction of organismic and environmental factors that accounts for the presence of certain phenotypic traits within an organism or population, where environmental and non-genetic organismic factors are considered as intrinsic to the development of the organism as genetic ones. On the one hand, an identical set of genes might be found in clearly distinct phenotypes. For example, first-generation worker ants of a newly founded colony, one of whose tasks is to construct many of the standard features of ant colonies, will look and behave differently from genetically identical later generation specimen raised in the fully established colony, and hence in an environment that was shaped by those first generations (Gordon 2001 ). Similarly, the sex of turtles and crocodiles is not genetically determined but depends on environmental temperature during embryonic development (Bateson 2001 ). On the other hand, modification of environmental factors may affect the development of traits that match with environmentally unmodified genetic variants, hence providing for distinct developmental routes to a similar phenotype, as in the phenomenon of “phenocopying” (Goldschmidt 1949 ).

Either way, the focus of inquiry has to be on the combined system, with environmental factors fully integrated but not necessarily interchangeable with genetic ones, and with the key unit of analysis often not being the individual organism but supra-individual entities. The very notion of one central unit of control in development might be misguided to begin with. Even mechanisms of inheritance might be distributed over a variety of factors, including persistent structures in the environment. There is no such thing as genetic information that could be taken by itself and still be informative about what phenotype an organism will develop. Such would be the case only if “strong instructionism” were true, that is, if genetic information were supposed to fully specify phenotypic traits (Wheeler and Clark 1999 , 2008 ). Rather than being relegated to the status of context, however important, to the content of the genetic code, environmental and other non-genetic factors are placed on equal explanatory footing with respect to informing biological development. Some developmental systems theorists even reject the notion of genetic information altogether, and speak of genetic and non-genetic inheritance of developmental resources instead (Griffiths and Gray 2001 ).

The critique of adaptationism common to the biological allies of extended cognition can be unpacked into two more positively stated leitmotifs: a principled openness of both evolutionary and developmental pathways and an emphasis on supra-individual factors as explanantia, rather than explananda, for the respective theories. The principled openness is expressed, on a first level, in the marked abstention from postulating some unifying force that would govern phylogeny and ontogeny. Natural selection counts as just one among other relevant factors in evolution, and it works on various levels, from genes to populations. Genes are just one among other relevant factors in shaping organisms and, possibly, transmitting information between generations of a population. On a second level, there is no reason to assume that there is such a thing as evolutionary progress or adaptive optima. What might seem like a perfect adaptation now may turn out to be a disadvantageous solution later, and there is no way of knowing in advance. Current evolutionary states of affairs are historically contingent facts, providing no guidance as to how things should stand or where they will go next.

The constitution of cognitive extensions

What do the observations in the two preceding sections tell us about the Extended Mind Hypothesis? After all, all parallels to the debates within evolutionary theory were only identified in retrospect. Nor did the Extended Mind Hypothesis, as originally conceived, relate to teleofunctional theories. Still, arguments from either tradition might help to elucidate, and possibly achieve, Clark and Chalmers’ explanatory aims, on two levels: the epistemological status of the hypothesis (discussed under the short form Int ) and the question of the nature of extensions (discussed under the short form Ext ).

The status of epistemic claims

While the theories considered above partly figure as predictive theories in their respective scientific fields, developmental systems theory and the theory of extended phenotypes deliberately assume a different (separate or additional) role, namely as challenges to seemingly commonsensical perspectives on their subject matter. As such, they might ultimately but will not necessarily result in predictive theories. 9 This dual notion of theories may serve as a template for a strong and a weak interpretation of the Extended Mind Hypothesis respectively: if, at least in important subclasses of cases, entities and processes in the environment are indispensable for the accomplishment or presence of some cognitive function, the expectation will be justified that the advocate of extended cognition shall identify the necessary conditions for those objects and processes to become part of that cognitive function while remaining impartial to what common sense might tell us. Some of Clark’s own works (in particular 2007 ) as well as some complementarity-based accounts (most notably Kiverstein and Farina 2011 ; Sutton 2010 ) aim in this direction, in looking for real-world cases where the assumption of environmental extendedness of cognitive processes accomplishes tangible explanatory tasks. If however it is a mere matter of perspective to view some objects in the environment as components of cognitive processes, the claim is considerably weaker: it amounts to a suggestion to temporarily suspend common sense beliefs in order to help us to a better understanding of some aspects of cognition.

The critique in Adams and Aizawa ( 2001 ), although not intending to capture this point, is diagnostic of what is at issue here. The authors explicitly mount a defence of common sense, presuming that common-sense views, qua being commonsensical, provide good guidance to cognitive inquiries. If, as Adams and Aizawa argue in alignment with a long-standing intellectual tradition (which includes, among many others, Popper 1959 ; Quine 1957 ), science is a methodical refinement and extension of common-sense reasoning, mobilising it against overambitious flights of imagination will be useful in itself. As, however, many scientific findings happen to defy common-sense beliefs, and as, arguably, the advancement of science sometimes even requires transgressions of the boundaries of common sense (as has been prominently argued by Einstein 1918 , along with other physicists whose theories indeed defy common sense, as well as by philosophers as different as Nagel 1961 ; Feyerabend 1975 ), the force of this argument will be limited to begin with. If there are defensible reasons for adopting a counterintuitive image of how the world stands, and if evidence collected under the guidance of that image turns out to advance inquiry, the transgression will be vindicated. It will be difficult though to determine in advance whether that vindication is forthcoming. In terms of an epistemological economy, the Extended Mind Hypothesis incurs the risk of investing into a transgression of common sense whose returns are much less than certain.

The nature of extensions

One common denominator of all theories discussed above is that there will be no way of determining the goals and accomplishments of an organism without in some way considering his coupling with the environment, where this coupling will have the nature of a two-way interaction rather than the organism either manipulating or being determined by the environment. Certain entities and conditions in the environment are relevant to the organism in such a way that he could not exist without them and that they would not exist without him, as the very set of relevant conditions that they are. On this set of views, the coupling between organism and environment plays a constitutive role to the trait or behaviour in question in either of two ways: without that coupling, it would not be possible at all, or it would have to rely on different means.

Organisms with a certain degree of adaptive plasticity, that is, the ability to respond in various and partly innovative ways to a given set of problems, can be found to alternate between using internal means and one or a variety of external objects or tools for the same purpose, each involving different organic and behavioural resources and, possibly, differential effects on reproductive chances. For example, geographically separate populations of wild chimpanzees ( Pan troglodytes ) have been observed to vary in using versus not using tools and in the kinds of tools they use for a certain purpose, in spite of identical biological constitution and ecological similarities between the respective populations’ habitats (Boesch and Boesch 1990 ; Boesch 1993 ; Matsuzawa 2001 , Chap. 1). It has been suggested that cultural transmission of tool-using abilities is involved here. 10 While there is an analogy of function in these cases, both the entities involved and the mechanisms by which they are integrated in pursuit of that function will be at variance. The constituents of the external and internal variants of the processes cannot be expected to be mutually substitutable. This is the minimal sense in which extensions may become constitutive to the accomplishment of some biological function. They will be called “constitutive W ” where required for disambiguation.

The situation will be different in cases where there are objects and artefacts in the environment that enter into some coupling without which a particular, and possibly vital, accomplishment for the organism could not be attained at all . Citing a famous example, without felling trees and constructing a dam from timber, stones and mud, so as to impede the flow of a creek and create a water reservoir, a beaver would neither be in the position to secure himself a sufficiently large territory for foraging nor to find a place for building lodges sufficiently protected from terrestrial predators (Dawkins 1999 , pp. 200, 209). There is no ‘internal’ alternative for the beaver to building a dam. He does so because this is the only way available to him to secure the presence of some of the conditions on which he vitally depends. This is the maximal sense in which extensions may become constitutive to some biological function: extendedness will become essential to an explanation of the presence, rather than the mode of accomplishment, of some core functions—and, by implication, of the nature of the organisms to which those functions pertain. Cases of this kind will be labelled “constitutive S ”.

The constitutive S and constitutive W types of case are clearly distinct from an interpretation of cognitive extensions in which they appear as instruments of cognition: if the cognitive function in question could be equally performed by purely internal means, their role would be similar to that of an artefact that produces a behavioural output identical to that of some type of human action, and that does so on the basis of at least similar mechanisms. Internal and external elements would be mutually substitutable without any notable effects on the performance, and the mode of performance, of the cognitive function. If (and only if) extensions of the mind were supposed to be understood in such a straightforwardly instrumentalist fashion—which, to be sure, they are not—a central point of Adams and Aizawa’s critique of extended cognition ( 2001 ) would become tenable. The authors maintain that the causal processes involved in the use of cognitive extensions would have to be identical to those involved in internal or “intra-cranial” cognitive processes in order to count as functionally equivalent. Since the causal processes, as a matter of empirical fact, will be at variance, functional equivalence is not accomplished. If, however, the extensions Clark and Chalmers have in mind are not instrumental in nature but constitutive W to cognitive processes, the mechanisms involved in extended versus internally based cognitive processes can be freely acknowledged to be at variance. Adams and Aizawa hence appear to miss the entire point of functional analogy, which is precisely not determined on the grounds of a homology between the structures that realise some function but on the grounds of what purpose the respective structures serve.

Varieties of extendedness

To sum up, we are presented with four different possible interpretations of the Extended Mind Hypothesis, on two partly interdependent levels:

When introducing their hypothesis, what Clark and Chalmers appear to do is this: they present a set of theoretical claims that speak for an interpretation under which objects and conditions in the environment are part of the necessary conditions for the presence and function of some cognitive traits. Accordingly, the strong interpretation ( Int S ) would seem in place here. However, these claims are quite casually placed alongside remarks on how a view of cognition as extended into the environment allows for a more elegant, unified and simplified explanation of cognitive processes (see Clark and Chalmers 1998 , pp. 10, 14). There is no reference in these places to explanatory problems that would require a view of cognition as being thus extended. Accordingly, the weak interpretation ( Int W ) would seem in place here.

There are two adjacent claims on the first pages of Clark and Chalmers’ essay where that ambiguity becomes manifest:

If, as we confront some task, a part of the world functions as a process which, were it done in the head , we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process. (Clark and Chalmers 1998 , p. 8, emphasis in original)

After having thus introduced the parity principle, the authors present the core of the hypothesis, the coupled system claim (see p. 2), and add:

All the components in the system play an active causal role, and they jointly govern behaviour in the same sort of way that cognition usually does. If we remove the external component the system’s behavioural competence will drop, just as it would if we removed part of its brain. [...] The external features in a coupled system play an ineliminable role – if we retain internal structure but change the external features, behaviour may change completely. The external features here are just as causally relevant as typical internal features of the brain. (Clark and Chalmers 1998 , 8 f)

If, taking the first claim at face value, a part of the world worked as a process that could, in practice or in principle, also occur in the head, then the ‘external’ and the ‘internal’ versions of that process would be equivalent, hence interchangeable without loss, and arguably without a difference beyond the physical location of that process. If however, according to the second claim, we took some environmentally coupled cognitive process and removed the external component, we would not be left with an intact set of internal operations within the person accomplishing the cognitive task—only minus that component. Even less would we be left with a set of internal operations identical to the case of accomplishing the same task by internal means. As Clark and Chalmers implicate in the second passage, very different kinds of mental operations are required for doing a calculation in one’s head and for doing it with a pocket calculator, or for finding one’s way by visual recognition of a scene or by consulting a map or a notebook.

Clark and Chalmers’ choice of examples is not precisely helpful in keeping these points apart. The examples they extensively discuss in order to substantiate their hypothesis—which are the examples that also dominate the Extended Mind debates—concern devices that we are suggested to consider as substitutes for, or supplements to, internally based cognitive processes. Their notebooks, keyboards and pocket calculators are artefacts that can be used for operations that could be, at least prima facie, equally accomplished ‘in the head’, for example making a calculation or remembering an appointment or the directions to a certain place.

This choice of examples not only seems to suggest Int W but also, however inadvertently, creates the impression that the extensions in question might be mere instruments of cognition—which is a logical non sequitur : while Int S stands in a relation of mutual implication with the constitutiveness claim ( Ext C ),  the instrumentalist claim ( Ext I ) and Int W are logically independent. The claim that entities in the environment might be viewed as parts of cognitive processes does allow for the possibility but does not necessarily imply that these entities will work as instruments of a cognitive process, nor does the reverse hold. At the same instance, one will be entitled to accept that Ext C and Ext I might hold at the same time, albeit in different, non-intersecting domains: some extensions might play a constitutive role, and thus imply Int S for the respective class of cases, while others might count as mere instruments of cognition in another class of cases. Still, Int S will be vindicated, since there is a class of cases by which it is implied. On this analysis, the Ext C type of cases provides much better support for the Extended Mind Hypothesis, and it does so for its stronger version.

In conjunction with the aetiological argument, these distinctions will contribute to a reasonably detailed classification of relations of extendedness. Any claim for a functional parity between the internal and external mechanisms involved will have to be cashed out in some, not necessarily directly biological, reproductive advantage determinable for each part of the equation.

If, firstly, I were able to substitute an environmentally coupled for an internal process at will, where that substitution made no difference to the achievement of my goals, in terms of scope, effectivity, efficiency or reliability, the functional equivalence involved would be of the weak Ext I kind. We still would have to explain how the elements involved have come to be thus substitutable.

If, secondly, the members of some population become equally able to mobilise some novel environment-based mechanism in order to supplement an established cognitive function, so as to achieve an increase in scope, effectiveness, efficiency or reliability, functional equivalence of a non-trivial constitutive W kind will be reached, while adaptive fitness will actually be increased across the entire population. This is where some of the standard examples of cognitive extensions will fit in, namely computing devices that increase the amount and complexity of available information, or the speed of retrieval.

If, thirdly, members of some population become able to individually mobilise a novel environment-based mechanism for an established function, so as to outperform other members of that population to whom the mechanism remains unavailable, the condition of functional equivalence will hold with respect to the sameness of function but not with respect to adaptive fitness within the population. Such would be the case for technologies of cognitive enhancement that are only available to some members of a population.

If, fourthly, I use an environmentally coupled process in order to compensate for an internally based one that I am not able to perform for some reason, and if that environmentally based substitute enables me to accomplish the same task with roughly the same degree of effectivity, efficiency and reliability (see Clark’s “glue and trust” criterion, p. 7), functional equivalence will be reached on all levels by means of a mechanism that is clearly distinct from the impaired one. Such will be the case in what has been termed “sensory substitution” (see Farina 2013 ; Kiverstein et al. 2014 ): situations of sensory impairment might be mended by artefacts that compensate for the function that has been lost, and that typically recruit an alternative sensory modality.

If, fifthly, an environmentally coupled process allows for the establishment of novel functions, and hence is constitutive S ,  there will be no reference point for determining functional equivalence, so that the function in question will have to be identified in the first place. It is upwards along this hierarchy of types of functional relations that one will find the most rewarding proving grounds for any claim of extendedness.

Constitutional matters

The distinction between constitutive W and constitutive S types of cases introduced in the previous section partly matches the respective foci of “first wave” parity-oriented views of extended cognition, which highlight functional equivalence between internal and environmentally coupled cognitive processes, and “second wave” complementarity views, which highlight the role of environmentally extended processes as co-constituents of some cognitive abilities. 11 Still, the mapping of these “waves” onto the types of constitutiveness introduced above is not strictly disjunctive, and some of the preceding points might help to partly integrate these arguments: where classical functionalism focused on analogies between input/output relations in different systems on a general level, and where the “microfunctionalism” defended by Wheeler ( 2010 ) focuses on more fine-grained analogies in functions of the structures that realise a superordinate function, a history of selected effects of the traits of the systems under consideration will help to explain the establishment of certain functions in the first place. At the same instance, it is a matter of course to an aetiological view of functions to be impartial towards somatic boundaries, provided that the environmental structures in question form part of the necessary conditions in an explanation of the presence of some function. It is equally a matter of course for an aetiological view, as it is for the developmental systems theorist, to accept that the roles played by the co-constituents of some function are more likely heterogeneous in kind than mutually substitutable.

Still, one might accept either or both types of constitutive roles of environmental factors for some of the adaptive functions of some organism’s traits without necessarily accepting them as being coupled with the organism in the sense suggested by Clark and Chalmers. According to Sterelny ( 2010 ), the Extended Mind Hypothesis overemphasises prima facie functional similarities between internal and artefact-involving cognitive processes while playing down the role of the agents who not only rely on environmental supports but actively mobilise environmental resources, natural and other, so as to construct their own cognitive niches (a concept more richly developed in his 2003 book). Sterelny ( 2010 ) argues that functional couplings of the Extended Mind kind are limiting cases of the more general phenomenon of environmental scaffolding: the reliance of organisms on the presence of some environmental features or artefacts without which some function could not be accomplished. For example, human beings have become reliant on the availability of cooked food or means of cooking food, as their metabolic apparatus has become insufficient for coping with exclusively raw food. Hence, cooking is an environmental scaffolding for the human metabolism, but it will be hardly illuminating to refer to the human metabolism as being extended into the environment.

With respect to constitutive W cases, all hopes for an artefact to pass muster as being part of an extended cognitive processes rather than a mere scaffolding rest on the degree of integration of the artefact into the accomplishment of a cognitive function, such as memorising facts or orienting oneself in space—which are functions that still can be accomplished to some extent by the unaided mind. However, genuinely material “anchors” for certain cognitive tasks may often be both more efficient and historically prior to analogously formed conceptual ones (Hutchins 2005 ). Still, there will always be room for debate as to whether some concrete coupling with some artefact is tight and integrated enough to warrant the bestowal of the “extended” predicate, or to make us, in Clark’s words, “natural-born cyborgs” ( 2003 ). There is no steadfast ontological criterion to make that decision for us, although the evolutionary history of the mechanisms involved will provide some guidance.

Hence, the quest will be for evolved mechanisms for recruiting artefacts for the performance of cognitive functions. Clearly, it is not some concrete type of artefact (the notebook, the smartphone) that we have evolved to be coupled with. What can be credibly argued to have evolved is a general and highly adaptable ability to create and recruit such artefacts—from simple drawings to computing gadgets. In Millikan’s terms (see p. 5 above), it is the direct proper function of the organism-based mechanisms involved to recruit a variety of artefacts for the accomplishment of a cognitive task, and it is the direct proper function of any concrete type of artefact involved to make its contribution to the accomplishment of that task, in co-operation with the organism-based mechanisms. If, though, one takes the lessons to be learned from developmental systems theory seriously, the practice of using artefacts should be expected to feed back into the organisation of the brain, and might do so beyond childhood development (Farina 2014 ).

It will be more difficult to identify the nature and functions of instances of environmental coupling that play a constitutive S role, as the respective cognitive ability supposedly would not exist without the environmental counterpart. It will even be difficult to identify the ability in question as being properly cognitive . If one follows Adams and Aizawa ( 2001 ) in defining cognitive processes ab initio as a subset of the processes within the human nervous system, one has a criterion for what counts as cognitive that rules out “extra-cranial” processes by default, albeit a question-begging one. After all, there are many non-cognitive neuronal processes. It is the additional criterion of the involvement of “non-derived content” recruited by the authors that shall suffice to identify the subset of intra-cranial processes that are properly cognitive. In introducing that second criterion, the authors follow an argument in Searle’s critique of Artificial Intelligence ( 1980 ): while linguistic items or other artefacts have the meanings they have and serve the functions they do because they are endowed, by stipulation or convention, with those meanings and functions, our thoughts have their meanings and serve their functions in at least partial independence from any such external fiat. The nature of the human mind bearing those thoughts is supposedly at least partly sufficient for the presence of that content. That nature is not further explicated though.

On the aetiological view, in contrast, both artefacts of all sorts and our thoughts have the meanings and functions they have because individual artefacts and mental tokens have been reproduced by some mechanism with sufficient frequency because they mapped onto some world affair, in terms of being connected to an appropriate set of behaviours towards it, reliably enough and according to a well-defined set of regularities. On this view, cognition basically is a specific activity of relating to one’s environment, in which such mapping relations are used to produce adaptive behaviours. No principled difference exists between artifactual and internal mechanisms that accomplish this feat, namely producing the appropriate tokens on the appropriate occasions, where the conditions of appropriateness essentially depend on what in the environment the organism relies on. As we will see in more detail in the next section, the human faculty of language is the primary candidate for a properly cognitive trait that may both count as environmentally extended and fits the constitutive S condition, and hence militates against either of Adams and Aizawa’s criteria: its function and contents depend on the operation of mechanisms that partly operate outside the human brain or body.

A more detailed attempt at spelling out what kind of specific activity of relating to one’s environment cognition is has been undertaken by Godfrey-Smith: if “the function of cognition is to enable the agent to deal with environmental complexity” ( 1996a , p. 3, the “environmental complexity hypothesis”), the criterion for a process to count as cognitive is whether it contributes to dealing with such complexity. Cognition has a function if an environment is distally complex and changeable but, at the same instance, correlated with proximal, detectable conditions reliably enough to warrant the effort of tracking the latter, so as to modify one’s behaviour towards the former. If proximal conditions in one’s environment were too chaotic, resilience would be the preferable evolutionary strategy instead. If, in turn, distal conditions were too dull, local confinement to an optimal range of conditions would be the solution of choice. Neither perception nor object recognition nor memory would be required in these latter two types of cases, but they will be essential to the first one. On this analysis, cognition arises with the necessity of either tracking favourable conditions or actively making conditions favourable (see p. 10).

Entities and processes in the environment may contribute to these abilities in constitutive S fashion in either of two ways, both of which include evolved mechanisms of coupling: on the one hand, they might be part of the necessary conditions for the realisation of the direct proper functions of a mechanism that is otherwise based within the organism. The entities and processes in the environment need not be reproducible items with their own direct proper functions, but their presence and use will be essential to the realisation of the functions of the organism-based production mechanisms. On the other hand, there might be mechanisms for the production of tokens of some type that are based in the environment while being co-ordinated with organism-based mechanisms, so as to jointly help him to produce behaviours appropriate to a given world affair in a given situation. It is in the latter, more complex type of case where reproducible artefacts will come to play a central role.

The art of coupling, basic and advanced

Clark and Chalmers only briefly and tentatively refer to relations that would count as constitutive S on the present account ( 1998 , p. 11). However, these references are part of a brief evolutionary excursus and include two suggestions that match the distinction introduced at the end of the preceding section: the evolution of vision as exploiting the structure of local environments, and the evolution of language as a key structuring feature of human interaction. 12 While Clark and Chalmers proceed to using language as their paradigm extension of the mind, it might be worthwhile to pay attention to more basic modes of environmental coupling. The function of vision and language is certainly not one of instruments that could be used or discarded by human beings at will. In the context of human evolution, both are indispensable components of, and hence constitutive S to, human cognition. If an individual is deprived of either faculty, his or her cognitive functions will be impaired to some extent, and some functions could not be accomplished at all.

Still, there is a difference between vision and language, with their respective modes of environmental coupling, in terms of the depth of their evolutionary roots and the levels of cognition they feed into. Vision is not only the older and more common phenomenon in the animal kingdom. It can also be plausibly argued to be more directly tied to objects and structures in the environment, serving as a prerequisite for spatial orientation and object recognition for most animals, including humans. Language, in contrast, serves the coupling between human agents in the first place and presupposes a number of pre-linguistic cognitive capabilities, including mechanisms of orientation and object recognition. Moreover, whereas the environmental coupling in visual perception, in terms of necessary constituents of the visual process, occurs between mechanisms of perception and invariant properties of objects in the environment, language will rightfully count as a production mechanism in its own right.

On this view, the cognitively most foundational set of external entities to be coupled with is to be found in natural objects or processes in the environment. According to the ecological view of visual perception, as developed by Gibson ( 1979 ), features of an organism’s natural environment will guide his activities and be available to him reliably, directly, implicitly and automatically, without the need of reflection upon their operation, and without the need of having been consciously endorsed in the past. Reliably successful endorsement in the past will be necessary and partly sufficient. Thus, all of Clark and Chalmers’s criteria cited on p. 8 above will be met.

More precisely, the Gibsonian will hold that a perceiving organism is engaged in the activity of picking up information from his environment, where invariants in the structure of that environment are the bearers of that information, and where the organism’s own position, movements and orientation in relation to some object—and, as Lauwereyns ( 2012 ) emphasises in his ecologically minded but neuroscientifically based “intensive” account of vision, his purposes towards that object—will be as essential to the structuring of perception as those invariants. The object will then quite directly offer or “afford” possible behaviours to the organism, depending on how he is disposed and positioned towards it. Visual perception is not to be conceived of as a string of retinal images passively received in order to be processed by the perceiver’s nervous system. It crucially relies on the perceiver’s interaction with his environment. On the ecological view, inference or inner representations are largely irrelevant to perception, whereas none of these higher-order cognitive accomplishments could emerge without the prior establishment of mechanisms of perceptual coupling. Alternatively, between such elementary direct guidance by perception and fully fledged internal representations, one might find representations that detect patterns in one’s environment and direct appropriate behaviours towards them at the same instance. These elementary representations have been introduced as “pushmi-pullyu representations” by Millikan ( 1995 ), and they figure as “action-oriented representations” in Clark ( 1997 ; see also Clark and Toribio 1994 ).

This image of directness has to be extended when moving from perception to higher-order kinds of cognition. On the one hand, some of the guidance that has hitherto been directly provided by perceptual couplings with the environment is now delegated to internal resources, some but not all of which will assume the shape of explicit representations. Thus, the next remarkable development, again in phylogenetic terms, does not as much lie in an extension of cognition into the environment as it lies in the internalisation of some of the guidance that would otherwise be directly provided by the environment. Externalisation reappears on the level of cognitive artefacts, that is, devices that are used for cognitive functions, thus re-instantiating environmental guidance on a higher level. 13

On the other hand, the purposeful creation and appropriation of cognitive resources fosters their (partial) detachment from a present, imminent or otherwise pre-determined practical use. In turn, however, the use of these artificial resources, once it has been properly learned, and if it functions smoothly, normally is quite as direct, implicit and automatic as in the more basic cases. One may pause to reflect on the information taken up from the environment, and one may then either resort to other sources of information or store away that information for later use. However, doing so remains a special case mostly reserved for instances of interference or error, and for (re-) assessing the goals of one’s actions. In a familiar and well-behaved environment, the natural and artificial features of that environment will equally provide direct guidance.

Language has the intriguing property of essentially involving both artefacts and internal mechanisms in order to perform its functions. It is at once rooted in evolution and in the history of artefact use. More specifically, it has been argued, most prominently by Deacon ( 1997 ), that language co-evolved with the human brain and its capability of symbolic reference. Deacon takes language to be a structure whose properties and reproduction within language communities are in a straightforwardly evolutionary way interdependent with the properties particular to the human brain. The complexity and the adaptive functions of the human mind have as one of the necessary conditions of their emergence and present functioning the development and use of linguistic structures. The intra- and extra-somatic mechanisms for the production of linguistic items are tightly integrated with each other. Neither mechanism would be present nor could not function in absence of the other. Individually, they contribute to shaping their counterpart and its functions, historically and at present. Jointly, they enable speakers to create concrete artifactual structures on which their further interaction relies. This claim is to be distinguished from the considerably broader view that human linguistic capacities are a product of natural selection on the one hand, where the possible role of linguistic artefacts in shaping these capacities is not further considered, and Chomskyan linguistics on the other, where linguistic structures proper are supposed to be at once innate and not a product of natural selection. 14

If the human faculty of language and linguistic structures in conjunction are thus correctly described as achieving a complex cognitive coupling between agents and their environments that would not be possible without them, they may actually and at the same instance be the cognitive resource with the highest potential for de coupling, as they allow for public and individual use in a multitude of, partly unanticipated, contexts, and as they allow for degrees of abstraction and detachment from actual matters of fact that would otherwise be hard to come by. Hence, contrary to one concluding observation in Clark and Chalmers’ essay, the human mind, without language, would not be “much more akin to discrete Cartesian ‘inner’ minds, in which high level cognition relies largely on internal resources” ( 1998 , p. 18). If we may call our evolutionary kin as witness, our minds would be coupled with their environments in considerably tighter and more direct fashion, in that there would be little exception to being directly guided in our actions by what we encounter there.

If constitutive modes of cognitive coupling carry the day as the paradigms of extensions of the mind, as I have sought do demonstrate, we might better conceive of the human and other minds as extended not by virtue of being equipped with a number of artificial add-ons that work as external instruments of internally based cognition. Instead, we should conceive of cognition as being extended over traits of the organism and properties of the environment, incorporating and integrating entities, processes and mechanisms on either side of the somatic boundaries. Both sides have evolved, grown and, in part, purposefully created to be so integrated. In conjunction, they will either constitute W alternative mechanisms for the accomplishment of some cognitive functions or constitute S some functions, as necessary conditions for their presence. In either case but to different degrees, the establishment and performance of these functions depends both on evolved mechanisms and on the environment-shaping activities of the cognising organisms themselves.

To end on a historical note: whether developmental systems theory or environmental constructivism or niche constructivism or teleofunctionalism or theories of extended cognition are concerned, they appear to share a leitmotif introduced into modern Anglo-Saxon philosophy by Dewey (especially in 1929 ). 15 In critical reaction to Herbert Spencer’s self-styled Darwinian conception of the organism–environment relation and in accordance with a different lesson drawn from Darwin, Dewey considers the distinction between organism and environment artificial to begin with. Presuming that distinction to be ontologically grounded is likely to create many of the problems that most accounts of the adaptive efforts of an organism keep struggling with. To live plainly is a particular way of being related to one’s environment. It also is an interactive, environment-shaping relation, and it is continuous with being a cognitive system, in terms of the evolved function of cognition to help an organism cope with a complex and continuously changing environment. On this view, it will be difficult to find cognitive processes that do not incorporate features of the environment. At root, all cognition is extended, and only some of the historically most recent cognitive traits allow for a notable degree of decoupling. The mind is not extended into but over the environment.

Acknowledgments

Work on this publication was funded by the Austrian Science Fund (FWF), Grant number J 3448-G15 within the Erwin Schrödinger Programme.

1 For these, historically prior, kinds of externalist view, see Putnam ( 1975a ) and Burge ( 1979 ) in philosophical semantics and McGinn ( 1977 ) in the philosophy of mind.

2 For concise attempts at systematically mapping that debate, see Menary ( 2010b ) and Wilson and Clark ( 2009 ). Hurley ( 2010 ) embeds her account in a broader taxonomy of externalist positions in the philosophy of mind. A useful commented bibliography on extended cognition is provided by Kiverstein et al. ( 2013 ).

3 This concern is shared by the main critics of extended cognition (Adams and Aizawa 2001 ; Rupert 2004 ; Sprevak 2009 ).

4 See, for example, Wheeler and Clark ( 2008 ), where evolutionary and developmental considerations play a major role in an argument for dynamic interactions between genetic, organismic and cultural factors; or Wilson and Clark ( 2009 ), where an explicit and positive reference is made to evolutionary accounts of biological functions; or Rowlands ( 2004 , pp. 224–227) who refers to Millikan-style proper functions and their aetiological credentials.

5 The loci classici for this distinction are Owen ( 1848 , p. 7) and Darwin ( 1859 , p. 434).

6 Other paradigmatic statements of this approach include Papineau ( 1987 ), Neander ( 1991a , b ), Godfrey-Smith ( 1994 ) and Price ( 2001 ); see also the anthology by Macdonald and Papineau ( 2006 ).

7 For the full definition of proper functions, see Millikan ( 1984 , p. 28 and Chaps. 1–2 in general).

8 Remarkably, this link between teleosemantics and extended cognition was first highlighted by one of the main critics of the latter: Rupert ( 2004 , p. 401, n22).

9 At the end of ( 2008 ), Clark asks for a “mental flip” explicitly styled after the one suggested by Dawkins ( 1999 ), whereas much of the preceding text is an attempt at providing concrete evidence for the explanatory value of extended cognition.

10 Among the few non-primate tool-using species, individuals of the New Caledonian crow ( Corvus moneduloides ) were observed to alternate between tool-guided and tool-less techniques of extracting food in the wild (Bluff et al. 2010 ). A recent example of allegedly innovative tool-use and its social transmission in captive Goffin cockatoos ( Cacatua goffini ) is to be found in Auersperg et al. ( 2014 ). For an authoritative account of the varieties and complexities of animal tool use, see Shumaker et al. ( 2011 ).

11 The distinction between waves of arguments for extended cognition has been made by Sutton ( 2010 ). Defences of extended functionalism include Clark ( 2008 , 2010b ) and Wheeler ( 2010 ), while second wave views have been introduced by Menary ( 2010a , integrationism), Sutton ( 2010 ) and Kiverstein and Farina ( 2011 , complementarity), and Rowlands ( 2010 , amalgamationism). It was Clark ( 1997 ) himself who anticipated complementarity views.

12 For extended-mind-based accounts of language, see Clark ( 1998 , 2006 ) and Wheeler ( 2004 ).

13 The term “cognitive artefacts” is not used by Clark and Chalmers ( 1998 ), but can be found in Clark ( 1998 ). Sterelny ( 2004 ) uses the roughly synonymous term of “epistemic artefacts”, which in turn is taken up by Clark ( 2010b ).

14 See Christiansen and Kirby ( 2003 ) for a collection of positions on the evolutionary roots of language and the possibility of brain/language co-evolution. For a classic, anti-Chomskyan account of the evolution of the human faculty of language, see Pinker and Bloom ( 1990 ), and for the position under attack here Chomsky ( 2006 ).

15 For historical and systematic considerations of the Spencer–Dewey axis, see Pearce ( 2014 ) and Godfrey-Smith ( 1996a , b ). Notably, Dawkins’ extended phenotype theory may be considered playing the Spencerian part in this context.

• Adams F, Aizawa K. The bounds of cognition. Philosophical Psychology. 2001; 14 :43–64. doi: 10.1080/09515080120033571. [ CrossRef ] [ Google Scholar ]
• Auersperg A, von Bayern A, Weber S, Szabadvari A, Bugnyar T, Kacelnik A. Social transmission of tool use and tool manufacture in Goffin cockatoos ( Cacatua goffini ) Proceedings of the Royal Society, B. 2014; 281 :20140972. doi: 10.1098/rspb.2014.0972. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Bateson P, et al. Behavioral development and Darwinian evolution. In: Oyama S, et al., editors. Cycles of contingency. Developmental systems and evolution. London: MIT Press; 2001. pp. 149–166. [ Google Scholar ]
• Bluff LA, Troscianko J, Weir AA, Kacelnik A, Rutz C. Tool use by wild New Caledonian crows Corvus moneduloides at natural foraging sites. Proceedings of the Royal Society, B. 2010; 277 :1377–1385. doi: 10.1098/rspb.2009.1953. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Boesch C. Aspects of transmission of tool-use in wild chimpanzees. In: Gibson KR, Ingold T, editors. Tools, language and cognition in human evolution. Cambridge, MA: Cambridge University Press; 1993. pp. 171–183. [ Google Scholar ]
• Boesch C, Boesch H. Tool use and tool making in wild chimpanzees. Folia Primatologica. 1990; 54 :86–99. doi: 10.1159/000156428. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Burge T. Individualism and the mental. Midwest Studies In Philosophy. 1979; 4 (1):73–121. doi: 10.1111/j.1475-4975.1979.tb00374.x. [ CrossRef ] [ Google Scholar ]
• Carter JA, Kallestrup J, Palermos SO, Pritchard D. Varieties of externalism. Philosophical Issues. 2014; 24 :63–109. doi: 10.1111/phis.12026. [ CrossRef ] [ Google Scholar ]
• Chomsky N. Language and mind. 3. New York: Cambridge University Press; 2006. [ Google Scholar ]
• Christiansen MH, Kirby S, editors. The evolution of language. New York: Oxford University Press; 2003. [ Google Scholar ]
• Clark A. Being there: Putting brain, body and world together again. Cambridge, MA: MIT Press; 1997. [ Google Scholar ]
• Clark A. Magic words: How language augments human computation. In: Carruthers P, Boucher J, editors. Language and thought: Interdisciplinary themes. Cambridge, MA: Cambridge University Press; 1998. pp. 162–183. [ Google Scholar ]
• Clark A. Natural-born cyborgs: Minds, technologies, and the future of human intelligence. New York: Oxford University Press; 2003. [ Google Scholar ]
• Clark A. Language, embodiment, and the cognitive niche. Trends in Cognitive Sciences. 2006; 10 (8):370–374. doi: 10.1016/j.tics.2006.06.012. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Clark A. Mikrofunktionalismus: Konnektionismus und die wissenschaftliche Erklärung mentaler Zustände. In: Metzinger T, editor. Grundkurs Philosophie des Geistes. Paderborn: Mentis; 2007. pp. 392–426. [ Google Scholar ]
• Clark A. Supersizing the mind: Embodiment, action, and cognitive extension. New York: Oxford University Press; 2008. [ Google Scholar ]
• Clark A. Coupling, constitution, and the cognitive kind: A reply to Adams and Aizawa. In: Menary R, editor. The extended mind. London: MIT Press; 2010. pp. 81–99. [ Google Scholar ]
• Clark A. Memento’s revenge: The extended mind, extended. In: Menary R, editor. The extended mind. London: MIT Press; 2010. pp. 43–66. [ Google Scholar ]
• Clark A, Chalmers D. The extended mind. Analysis. 1998; 58 (1):7–19. doi: 10.1093/analys/58.1.7. [ CrossRef ] [ Google Scholar ]
• Clark A, Toribio J. Doing without representing? Synthese. 1994; 101 :401–431. doi: 10.1007/BF01063896. [ CrossRef ] [ Google Scholar ]
• Darwin C. On the origin of species by means of natural selection. Or the preservation of favoured races in the struggle for life. 1. London: John Murray; 1859. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Dawkins, R. (1999). The extended phenotype. The long reach of the gene . Revised edition with new afterword and further reading; originally published 1982. New York: Oxford University Press.
• Deacon TW. The symbolic species. The co-evolution of language and the brain. New York: Norton; 1997. [ Google Scholar ]
• Dewey J. Experience and nature. London: George Allen and Unwin; 1929. [ Google Scholar ]
• Einstein, A. (1918). Motive des Forschens. In E. Warburg & M. Planck (Eds.), Zu Max Plancks sechzigstem Geburtstag. Ansprachen, gehalten am 26. April 1918 in der Deutschen Physikalischen Gesellschaft, Berlin, C.F. (pp. 29–32). Karlsruhe: Müllersche Hofbuchhandlung.
• Elton CS. Animal ecology. London: Sidgwick and Jackson; 1927. [ Google Scholar ]
• Farina M. Neither touch nor vision: Sensory substitution as artificial synaesthesia? Biology and Philosophy. 2013; 28 :639–655. doi: 10.1007/s10539-013-9377-z. [ CrossRef ] [ Google Scholar ]
• Farina, M. (2014). Three approaches to human cognitive development: Neo-nativism, neuroconstructivism, and dynamic enskillment. British Journal for the Philosophy of Science . doi: 10.1093/bjps/axu026 .
• Feyerabend PK. Against method: Outline of an anarchist theory of knowledge. New York: New Left Books; 1975. [ Google Scholar ]
• Gibson JJ. The ecological approach to visual perception. Boston: Houghton Mifflin; 1979. [ Google Scholar ]
• Godfrey-Smith P. A modern history theory of functions. Noûs. 1994; 28 :344–362. doi: 10.2307/2216063. [ CrossRef ] [ Google Scholar ]
• Godfrey-Smith P. Complexity and the function of mind in nature. Cambridge, MA: Cambridge University Press; 1996. [ Google Scholar ]
• Godfrey-Smith P. Spencer and Dewey on life and mind. In: Boden MA, editor. The philosophy of artificial life. Oxford: Oxford University Press; 1996. pp. 314–331. [ Google Scholar ]
• Goldschmidt RB. Phenocopies. Scientific American. 1949; 181 (4):46–49. doi: 10.1038/scientificamerican1049-46. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Gordon DM, et al. The development of ant colony behavior. In: Oyama S, et al., editors. Cycles of contingency. Developmental systems and evolution. London: MIT Press; 2001. pp. 141–148. [ Google Scholar ]
• Gould SJ, Lewontin RC. The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist programme. Proceedings of the Royal Society, B. 1979; 205 :581–598. doi: 10.1098/rspb.1979.0086. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Griffiths PE, Gray RD, et al. Darwinism and developmental systems. In: Oyama S, et al., editors. Cycles of contingency. Developmental systems and evolution. London: MIT Press; 2001. pp. 195–218. [ Google Scholar ]
• Hurley SL. Vehicles, contents, conceptual structure, and externalism. Analysis. 1998; 58 (1):1–6. doi: 10.1093/analys/58.1.1. [ CrossRef ] [ Google Scholar ]
• Hurley SL. The varieties of externalism. In: Menary R, editor. The extended mind. London: MIT Press; 2010. pp. 101–153. [ Google Scholar ]
• Hutchins E. Material anchors for conceptual blends. Journal of Pragmatics. 2005; 37 :1555–1577. doi: 10.1016/j.pragma.2004.06.008. [ CrossRef ] [ Google Scholar ]
• Kiverstein J, Farina M. Embraining culture: Leaky minds and spongy brains. Theorema. 2011; 30 (2):35–53. [ Google Scholar ]
• Kiverstein, J., Farina, M., & Clark, A. (2013). The extended mind thesis. http://www.oxfordbibliographies.com/view/document/obo-9780195396577/obo-9780195396577-0099.xml .
• Kiverstein, J., Farina, M., & Clark, A. (2014). Substituting the senses. In M. Matthen (Ed.), The Oxford handbook of the philosophy of perception . Oxford: Oxford University Press (in press).
• Lauwereyns J. Brain and the gaze: On the active boundaries of vision. Cambridge, MA: MIT Press; 2012. [ Google Scholar ]
• Lewis D. Mad pain and Martian pain. In: Block N, editor. Readings in philosophy of psychology. Cambridge, MA: Harvard University Press; 1980. pp. 216–222. [ Google Scholar ]
• Lewontin RC. Organism and environment. In: Plotkin HC, editor. Learning, development, and culture. Chichester: Wiley; 1982. pp. 151–170. [ Google Scholar ]
• Lewontin RC. The triple helix: Gene, organism, and environment. London: Harvard University Press; 2000. [ Google Scholar ]
• Macdonald G, Papineau D, editors. Teleosemantics. Oxford: Oxford University Press; 2006. [ Google Scholar ]
• Matsuzawa T, editor. Primate origins of human cognition and behavior. Berlin: Springer; 2001. [ Google Scholar ]
• McGinn C. Charity, interpretation, and belief. Journal of Philosophy. 1977; 74 (9):521–535. doi: 10.2307/2025795. [ CrossRef ] [ Google Scholar ]
• Menary, R. (2010a). Cognitive integration and the extended mind. In R. Menary (Ed.), The extended mind (pp. 227–243). London: MIT Press.
• Menary, R. (2010b). Introduction: The extended mind in focus. In R. Menary (Ed.), The extended mind (pp. 1–25). London: MIT Press.
• Millikan RG. Language, thought, and other biological categories. London: MIT Press; 1984. [ Google Scholar ]
• Millikan RG. Explanation in biopsychology. In: Heil J, Mele A, editors. Mental causation. Oxford: Oxford University Press; 1993. pp. 211–232. [ Google Scholar ]
• Millikan RG. Pushmi-pullyu representations. Philosophical Perspectives. 1995; 9 :185–200. doi: 10.2307/2214217. [ CrossRef ] [ Google Scholar ]
• Millikan RG. Varieties of meaning. London: MIT Press; 2004. [ Google Scholar ]
• Nagel E. The structure of science. New York: Harcourt, Brace and World; 1961. [ Google Scholar ]
• Neander K. Functions as selected effects: The conceptual analyst’s defense. Philosophy of Science. 1991; 58 (2):168–184. doi: 10.1086/289610. [ CrossRef ] [ Google Scholar ]
• Neander K. The teleological notion of ‘function’ Australasian Journal of Philosophy. 1991; 69 (1):454–468. doi: 10.1080/00048409112344881. [ CrossRef ] [ Google Scholar ]
• Odling-Smee FJ. Niche-constructing phenotypes. In: Plotkin HC, editor. The role of behavior in evolution. Cambridge, MA: MIT Press; 1988. pp. 73–132. [ Google Scholar ]
• Odling-Smee FJ, Laland KN, Feldman MW. Niche construction. The American Naturalist. 1996; 147 (4):641–648. doi: 10.1086/285870. [ CrossRef ] [ Google Scholar ]
• Odling-Smee FJ, Laland KN, Feldman MW. Niche construction: The neglected process in evolution. Princeton, NJ: Princeton University Press; 2003. [ Google Scholar ]
• Owen R. On the archetype and homologies of the vertebrate skeleton. London: John van Voorst; 1848. [ Google Scholar ]
• Oyama S. The ontogeny of information: Developmental systems and evolution. 2. Durham, NC: Duke University Press; 2000. [ Google Scholar ]
• Oyama S, Griffiths PE, Gray RD, editors. Cycles of contingency. Developmental systems and evolution. London: MIT Press; 2001. [ Google Scholar ]
• Papineau D. Reality and representation. Oxford: Blackwell; 1987. [ Google Scholar ]
• Pearce T. The origins and development of the idea of organism–environment interaction. In: Barker G, Desjardins E, Pearce T, editors. Entangled life, organism and environment in the biological and social sciences. Dordrecht: Springer; 2014. [ Google Scholar ]
• Pinker S, Bloom P. Natural language and natural selection. Behavioral and Brain Sciences. 1990; 13 :707–784. doi: 10.1017/S0140525X00081061. [ CrossRef ] [ Google Scholar ]
• Popper KR. The logic of scientific discovery. London: Hutchinson; 1959. [ Google Scholar ]
• Price CS. Functions in mind: A theory of intentional content. Oxford: Oxford University Press; 2001. [ Google Scholar ]
• Proust J. Functionalism and multirealizability: On interaction between structure and function. In: Gavroglu K, Stachel J, Wartofsky MW, editors. Science, mind and art. Dordrecht: Springer; 1995. pp. 169–185. [ Google Scholar ]
• Putnam H. The meaning of ‘meaning’ Minnesota Studies in the Philosophy of Science. 1975; 7 :131–193. [ Google Scholar ]
• Putnam H. Mind, language and reality. Philosophical papers. Cambridge, MA: Cambridge University Press; 1975. [ Google Scholar ]
• Quine WVO. The scope and language of science. British Journal for the Philosophy of Science. 1957; 8 :1–17. doi: 10.1093/bjps/VIII.29.1. [ CrossRef ] [ Google Scholar ]
• Rowlands M. The body in mind: Understanding cognitive processes. Cambridge, MA: Cambridge University Press; 2004. [ Google Scholar ]
• Rowlands M. The new science of the mind: From extended mind to embodied phenomenology. London: MIT Press; 2010. [ Google Scholar ]
• Rupert RD. Challenges to the hypothesis of extended cognition. Journal of Philosophy. 2004; 101 (8):389–428. doi: 10.5840/jphil2004101826. [ CrossRef ] [ Google Scholar ]
• Rupert RD. Representation in extended cognitive systems: Does the scaffolding of language extend the mind? In: Menary R, editor. The extended mind. London: MIT Press; 2010. pp. 325–353. [ Google Scholar ]
• Searle JR. Minds, brains, and programs. The Behavioral and Brain Sciences. 1980; 3 :417–457. doi: 10.1017/S0140525X00005756. [ CrossRef ] [ Google Scholar ]
• Shumaker RW, Walkup KR, Beck BB. Animal tool behavior: The use and manufacture of tools by animals. 2. Baltimore, MD: Johns Hopkins University Press; 2011. [ Google Scholar ]
• Sprevak M. Extended cognition and functionalism. Journal of Philosophy. 2009; 106 (9):503–527. doi: 10.5840/jphil2009106937. [ CrossRef ] [ Google Scholar ]
• Sterelny K. Thought in a hostile world. Oxford: Blackwell; 2003. [ Google Scholar ]
• Sterelny K. Externalism, epistemic artefacts and the extended mind. In: Schantz R, editor. The externalist challenge. New studies on cognition and intentionality. New York: de Gruyter; 2004. pp. 239–254. [ Google Scholar ]
• Sterelny K. Minds: Extended or scaffolded? Phenomenology and the Cognitive Sciences. 2010; 9 :465–481. doi: 10.1007/s11097-010-9174-y. [ CrossRef ] [ Google Scholar ]
• Sutton J. Exograms and interdisciplinarity: History, the extended mind, and the civilizing process. In: Menary R, editor. The extended mind. London: MIT Press; 2010. pp. 189–225. [ Google Scholar ]
• Turing AM. On computable numbers, with an application to the Entscheidungsproblem. Proceedings of the London Mathematical Society. 1936; s2–42 :230–265. [ Google Scholar ]
• West MJ, King AP. Settling nature and nurture into an ontogenetic niche. Developmental Psychobiology. 1987; 20 (5):549–562. doi: 10.1002/dev.420200508. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Wheeler M. Is language the ultimate artefact? Language Sciences. 2004; 26 :693–715. doi: 10.1016/j.langsci.2004.09.011. [ CrossRef ] [ Google Scholar ]
• Wheeler M. In defense of extended functionalism. In: Menary R, editor. The extended mind. London: MIT Press; 2010. pp. 245–270. [ Google Scholar ]
• Wheeler M, Clark A. Genic representation: Reconciling content and causal complexity. British Journal for the Philosophy of Science. 1999; 50 :103–135. doi: 10.1093/bjps/50.1.103. [ CrossRef ] [ Google Scholar ]
• Wheeler M, Clark A. Culture, embodiment and genes: Unravelling the triple helix. Philosophical Transactions of the Royal Society, B. 2008; 363 (1509):3563–3575. doi: 10.1098/rstb.2008.0135. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Wilson RA, Clark A. How to situate cognition. Letting nature take its course. In: Robbins P, Aydede M, editors. The Cambridge Handbook of Situated Cognition. Cambridge: Cambridge University Press; 2009. pp. 55–77. [ Google Scholar ]

Identifying Cognitive Dysfunctions Earlier with 'Fingerprints'

• by Fallon Duffy
• April 10, 2024

Fingerprinting has been used for centuries for identification, but what if a similar concept could help physicians detect specific impressions of diseases in the brain, such as types of dementia? Enter vascular fingerprinting.  

Vascular fingerprinting is a newly developed Magnetic Resonance Imaging, or MRI, technique that compares brain scans against images in a database. Biomedical engineering Ph.D. student Greg Wheeler, in the Fan Lab at the University of California, Davis, leads a project exploring how the method can provide more precise information than traditional MRIs to identify cognitive diseases earlier than ever before.  

Building on Existing Technology  

MRI is used in clinical settings to take images of tissues inside the body. The patient lies in a tube with a strong magnetic field that aligns the water protons throughout one's body. Radio frequencies are then introduced in a prescribed fashion to stimulate these protons and their specific responses is what shows up in the images.   

Traditional MRI generally takes 30 to 60 minutes to generate several scans. These scans are typically qualitative, meaning physicians compare them against images of healthy tissue to detect abnormalities.   

Vascular fingerprinting, however, focuses on the brain's vascular system, including blood flow and vessels. This type of scan is quantitative, collecting numerical measurements of the organ that can be precisely compared against other people and over time. It's also much faster, taking about five minutes to produce multiple images.  

"If I have quantitative numbers," Wheeler said, "I can see exactly how things changed over six months, whereas in these qualitative images of traditional MRI, it's more difficult to make this kind of ​​​​​precise, long-term ​assessment."  

Identifying Matches  

When scanned, a human fingerprint gets compared to a database to find a match. Similarly, a vascular fingerprint MRI is compared to a database of computer-simulated images until the best match is found.   

The computer-simulated images are developed by making a biophysical simulation at the same time as the MRI scan. The simulations involve thousands of virtual brain tissues with varying vascular components, such as blood volumes and oxygen saturations. Researchers can compare the simulated images with the original MRI to find the simulated image that most closely matches the real one. Once this match is made, researchers can derive quantitative information from the known parameters — i.e., those vascular components — in the simulated image.  

"You get unique 'fingerprints' in your database. You reference that to your acquired image, and you can get these multiple vascular factors from a single scan," said Wheeler.  

Currently, Wheeler is studying vascular fingerprinting to develop tools to measure vascular function changes. His work is based on the idea that vascular disease may underpin cognitive disorders like Alzheimer's disease.   

"Some of these vascular factors are treatable and are potentially preventable," he said. "If we saw these functional vascular biomarkers before someone developed cognitive impairment, maybe a doctor can prescribe them something to lower their cholesterol or change their diet."   

While this technology is only in its beginning stages, Wheeler believes this research is a crucial step in establishing similar tools for doctors to find solutions to these difficult diseases.  

Primary Category