Key Takeaways
Treat the working system, not the skull, as cognition's possible boundary
The central claim. Cognition may include processes outside the brain when they are actively integrated with the person, rather than merely influencing them from the sidelines. In Tetris, players can rotate shapes mentally or use a button. Physical rotation can take about 300 milliseconds, compared with roughly 1,000 milliseconds for mental rotation. If an external operation performs part of the same problem-solving work, its location alone may not exclude it from cognition.
A stronger test. Clark and Chalmers illustrate the claim with Otto, who relies on a notebook after developing Alzheimer’s disease. The notebook guides his actions much as Inga’s biological memory guides hers. The example asks readers to consider the whole working arrangement, not assume in advance that thought stops at the skin.
What's striking is that the claim concerns the unit of explanation, not a denial that brains matter. A scientist can study neural activity while also asking whether the task is best explained by brain, body, and tool together. The caution is equally important: an aid that improves performance does not automatically become part of cognition. The argument needs a principled account of integration, not just a vivid demonstration of usefulness.
Require reciprocal integration before calling a tool part of thought
Influence is not inclusion. A diary can prompt recall without thereby becoming part of memory. The volume distinguishes a one-way causal influence from a coupled process in which internal and external activity continuously affect one another. In the stronger case, both sides contribute to behavior as one organized process, and removing the external element impairs the task in a way that resembles removing a functioning part of the person’s cognitive setup.
Integration matters. Menary calls the coordination of bodily and environmental processes cognitive integration. Otto’s notebook is a more convincing candidate when it is reliably available, easy to consult, and ordinarily trusted. These conditions are meant to separate a dependable working resource from a book encountered once or a stranger whose advice must be checked.
The distinction helps prevent an overly expansive view in which sunlight, furniture, or every conversation becomes part of a mind simply because it affects behavior. Reciprocity offers a more demanding test, although it is not a complete boundary rule. A system can be tightly coupled without being cognitive, and temporary tools can be indispensable without becoming parts of a persisting person. Function, organization, and the research question still matter.
Let external memory complement biology instead of impersonating it
Different formats, different strengths. The collection contrasts engrams, biological memory traces, with exograms, external records such as notes and diagrams. External records can last longer, hold more information, and be manipulated in ways biological memory cannot. Biological memory, meanwhile, is dynamic: information can blend, interfere, and change during recall. The point is not that a notebook duplicates the brain, but that both can contribute distinct operations to a larger remembering process.
Follow the partnership. Sutton argues that the external and internal contributions may be complementary rather than equivalent. A written record can preserve a fixed detail while a person interprets, updates, or uses it. The scientific task is to study how the parts work together, including how their differences affect what can be remembered and done.
This account avoids a false choice between treating artifacts as inert props and pretending that paper remembers exactly as a brain does. It also shifts attention toward the design of the partnership: what is stored outside, what is reconstructed inside, and where errors can enter. One limitation is that different media do not automatically form one system. Their reliability, access, and role in a particular task must be established case by case.
Study cognitive activity in motion, not only stored mental objects
Thinking can be an activity. Menary’s cognitive integration approach emphasizes bodily manipulation of external representations, not simply the presence of helpful objects. A person writing may reread a sentence, revise it, move it, and use the changed page to shape the next thought. Those manipulations can be part of completing the cognitive task, because they provide operations that are difficult to reproduce mentally.
Practices have rules. Writing, calculation, and diagram use depend on learned ways of handling symbols. Menary calls these cognitive practices: activities governed by norms for manipulating and interpreting representations toward an intellectual goal. The pen is a tool, but the coordinated work of producing and revising marks is where the cognitive contribution lies.
The activity-based view connects cognition to skilled practice rather than treating thought as a hidden inventory of inner objects. It explains why a notation can change the problem itself, not just make an already formed answer easier to retrieve. Still, not every bodily movement in a task is cognitive. A useful analysis must identify which actions transform information or organize the work, rather than simply accompany it.
Keep the boundary debate empirical, not settled by a slogan
Critics demand a criterion. Adams and Aizawa argue that causal coupling alone cannot establish constitution. They also claim that cognition involves distinctive causal processes and nonderived content, and that external inscriptions differ too much from neural activity to count as the same cognitive kind. Rupert adds that fleeting organism-tool pairings may not explain stable capacities, such as using language across many contexts.
Replies shift the question. Clark argues that the parts of a cognitive system need not each be cognitive on their own. Hurley and Ross and Ladyman question whether a simple causal-versus-constitutive distinction guides mature science. The debate remains unsettled: advocates must show that studying extended systems yields genuine explanatory gains, while critics must avoid defining cognition as internal merely because familiar examples are internal.
This disagreement is valuable because it exposes two different questions: what belongs to a working system, and what counts as cognitive in the first place. The anthology’s strongest move is to resist solving either by intuition alone. Yet appeals to scientific practice cut both ways. A new explanatory framework earns standing through productive research, not merely by describing ordinary tool use in a more expansive vocabulary.
Language is a cognitive habitat, learned through bodies and relationships
Words grow from interaction. Spurrett and Cowley use utterance-activity to mean the full range of vocal, bodily, and expressive behavior in human interaction, not words alone. Infants coordinate with caregivers through gaze, gesture, rhythm, facial expression, and sound before they can use ordinary language. These exchanges can scaffold attention, learning, and the development of symbolic communication.
Social support can do cognitive work. Their account describes caregivers helping infants regulate attention and behavior through ongoing interaction. The chapter also discusses Kanzi, a bonobo who learned to use symbol boards in a highly supportive setting. Such examples challenge the image of a solitary learner extracting language from words alone, while leaving open how much of human language depends on specifically human capacities.
The emphasis on interaction broadens language research beyond grammar and vocabulary to the conditions that make learning possible. It also suggests that cognition can be socially distributed without implying that people literally share one mind. The infant examples support a role for coordinated attention and affect, but the interpretation of particular behaviors is difficult: an adult may read intention into actions that the infant has not yet organized as a deliberate request.
Perception may depend on the actions that disclose the world
Seeing involves active sampling. Rowlands argues that conscious perception is not simply an inner image. Eye movements, exploratory actions, and bodily interaction with the environment can help disclose what objects are like. Yarbus’s eye-movement studies found that people’s visual scan paths changed with the question they were asked about a painting. The task shaped where they looked, and looking helped determine what they noticed.
The tools can become transparent. Merleau-Ponty’s blind cane example makes the point vividly: a practiced user can experience the world through the cane rather than attend to the cane as an object. Rowlands extends this kind of analysis to perception, arguing that the processes enabling experience can reach beyond neural activity. This is a contested account, not a settled claim about every conscious experience.
The account is compelling where perception depends on repeated movement and feedback, because it treats seeing as an embodied skill rather than a passive receipt of images. Its broader claim about consciousness is more ambitious. The fact that action shapes perception does not by itself prove that action constitutes experience. Distinguishing causal preparation from the process that makes an experience what it is remains a central challenge.
Trace cognition through institutions, artifacts, and history
A performance can remember. John Sutton applies extended cognition to Shakespearean theater. Actors in the period learned many roles with limited rehearsal, while cue-scripts, plots, stage layout, and company routines helped coordinate performances. These artifacts did not simply duplicate what each actor knew. Their sparse information worked alongside individual skills and social organization to guide action.
The past shapes present minds. Sutton also examines medieval and Renaissance memory techniques, in which practitioners built imagined places to organize material for recall. These methods show how cultural tools and disciplines can reshape internal habits. Together, the cases support a historical cognitive science that studies brains, bodies, objects, and practices in combination, rather than treating cultural history as background decoration.
The historical cases make the theory more than a claim about notebooks or modern devices. They show how cognitive abilities can be cultivated by durable institutions and shared practices. They also expose an important risk: describing a whole group as one cognitive system can obscure who contributes what, who controls the records, and who receives credit. A strong account should preserve both collective organization and individual agency.
Do not confuse useful external support with a person's belief
Otto is the hard case. Clark and Chalmers say he believes the museum is on 53rd Street before opening his notebook, because its contents guide him as reliably as Inga’s memory guides her. Preston objects that Otto must consult the notebook to discover the address, while Inga can simply avow what she remembers. The two cases may produce similar behavior without sharing the same kind of belief.
Credit has limits. A person may achieve something by using a calculator or record without the tool itself knowing or believing anything. Preston argues that attributing belief to an extended system risks erasing distinctions between remembering, looking something up, and finding out. The volume leaves the issue open, but makes clear that external support and personal belief are not interchangeable descriptions.
This dispute keeps the extended-mind proposal honest. The notebook is plainly central to Otto’s practical life, yet practical dependence does not settle whether its contents are his standing beliefs. The distinction matters for responsibility, self-knowledge, and epistemic credit. A useful middle position recognizes external tools as parts of cognitive achievement while reserving belief attribution for cases where the person can genuinely own and avow the commitment.
Analysis
This anthology is best read as a sustained argument about where to draw the explanatory boundary around thinking, not as a single, settled theory. Its recurring image is a person working with a notebook, a screen, a symbol board, or another person. The provocation is that the resource may be more than an influence on thought: it may help constitute the process itself. The collection then makes the claim earn its force by confronting objections about causal coupling, intrinsic content, scientific kinds, stable capacities, and the ordinary meaning of belief.
Its most useful conceptual development is the move from parity to complementarity. Early arguments ask whether an external operation could count as cognitive if performed inside the head. Later contributions recognize that external and neural processes may differ profoundly, yet still cooperate. This turns the question from resemblance into organization: what does each component contribute, how tightly are they coordinated, and what explanatory work is gained by treating them as one system? Menary’s emphasis on bodily manipulation and Sutton’s historical cases make this shift concrete.
The strongest caution comes from the dissenting chapters. Causal dependence is not enough to establish constitution, and practical reliance on a resource does not automatically make its contents a person’s beliefs. Rupert and Preston show that disputes about scientific classification can also alter familiar concepts of memory, agency, and epistemic responsibility. These are not side issues. They mark the difference between saying that tools help people think and saying that tools literally belong to their minds.
The collection’s breadth is both its strength and its difficulty. It moves from philosophy of belief to developmental interaction, perceptual consciousness, history, and scientific methodology. No single boundary rule resolves every domain. The durable lesson is methodological: investigate the organized activity that produces cognition, while resisting both skull-centered prejudice and indiscriminate expansion. The mind may be extended in some tasks, but each proposed extension needs evidence about integration and explanatory payoff.
Glossary
Active externalism
Environment actively participatesThe view that some cognitive processes are constituted partly by active interactions between a person and present environmental resources. It differs from views where the environment merely influences cognition, or where external history determines what a thought means. On this account, the relevant resource must contribute directly to the ongoing process.
Parity principle
Location alone does not excludeA test proposed by Clark and Chalmers: if an external process performs a role that would count as cognitive were it performed inside the head, its external location alone should not disqualify it. The principle motivates the extended-mind hypothesis, though contributors disagree about whether similarity is enough to establish cognitive constitution.
Cognitive integration
Coordinated internal and external workRichard Menary’s account of cognition as a coordinated process involving neural activity, bodily action, and sometimes the manipulation of environmental resources. It emphasizes how these elements work together to complete a cognitive task, rather than claiming that external resources qualify as cognitive simply because they resemble internal processes.
Complementarity principle
Different parts contribute differentlyJohn Sutton’s approach to extended cognition: internal and external resources can have unlike properties and still make complementary contributions to a cognitive process. A notebook need not function like biological memory. Its distinct capacity to preserve and expose information can combine with biological processes in useful ways.
Engram
Biological memory traceA term used for a memory trace held within biological systems, especially the brain. In contrast with external records, engrams can be dynamic and susceptible to blending, interference, and reconstruction. The contrast helps explain why internal and external memory may contribute different strengths to a larger remembering activity.
Exogram
External memory recordA term used for a memory resource outside the brain, such as a written note, diagram, or other durable symbol. Exograms can persist and be manipulated in ways biological memory traces cannot. They may complement internal memory, though their presence alone does not prove that they are part of a person’s cognitive system.
Epistemic action
Action that aids thinkingAn action that changes the environment to make a cognitive task easier, such as physically rotating a shape to determine whether it fits. The action is performed to improve recognition, search, or problem solving, rather than solely to bring about a desired physical result.
Utterance-activity
Whole-body communicative behaviorDavid Spurrett and Stephen Cowley’s term for the full range of vocal, bodily, and expressive behavior involved in human interaction. It includes words, but also timing, gesture, gaze, facial expression, and prosody. They use it to study how communication and language develop within embodied social interaction.
Coupling-constitution fallacy
Mistaking connection for constitutionFred Adams and Ken Aizawa’s charge that extended-mind theorists sometimes infer that an external object is part of cognition merely because it is causally connected to a person. The criticism distinguishes causal influence from genuine constitution and demands an account of what makes a process cognitive.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.