The empirical evidence is accumulating that simulations, situations, and even bodily states play ubiquitous roles in perception. (1) We simulate the visual trajectory of an object and falsely remember anticipated motion (Freyd ). (2) Preparatory simulation of a grasping action under perception of a cup affects motor response on an unrelated task (Tucker and Ellis ). (3) Such simulations occur even when an object is merely named (“grape,” Tucker and Ellis ), and fMRI evidence shows grasping circuit activation (Chao and Martin ). (4) Being tired from a run makes a hill look steeper; carrying a heavy pack makes a path look longer (Proffitt ). (5) Finally, people's perception of near space extends further outward as arm length increases (Longo and Laurenco ), an effect sustained after arm‐length is extended artificially with a tool (like underestimating walking distance for roads we have driven).

(1) If memory's role is in service of situated action, then it should include bodily actions and their mesh with situations, triggering Gibsonian “affordances” for action (Glenberg ). (2) Just as reasoning about future action should require suppressing current perception (Glenberg et al. ), averting the gaze disengages the environment and facilitates memory. (3) Working memory consists of neurons in the frontal cortex sustaining simulations of absent stimuli in the original modal system, and different regions maintain different modal content, including objects, spatial locations, motion in different directions, and different spatial frequencies (Pasternak and Greenlee ). (4) The neural pattern associated with studying faces reappears when remembering them (Polyn et al. ). (5) Actions relevant to visual imagery are also simulated, bodily constraints shaping visual rotation (Parsons ), which is accompanied by motor area simulation (Richter et al. ).

(1) It appears that readers will not only confuse pictures with text (Intraub and Hoffman ), but that sentence processing is not disrupted by replacing words with pictures (Potter et al. ). (2) Verb retrieval activates motor control areas and nouns activate visual areas (Damasio et al. ; Martin et al. ; Pulvermüller ). (3) Perceiving manipulable objects, or even just seeing their names, activates brain regions for grasping (Chao and Martin ) and manipulating objects can improve reading comprehension in schoolchildren (Glenberg at al. ). (4) Attentional orientation is also affected by the valence of words (positive/up vs. negative/down; Meier and Robinson ). (5) Most interestingly, the gestures that accompany speech may have cognitive functions (McNeill ). Gestures help speakers retrieve related words (Krauss ), help listeners comprehend the speaker (Alibali et al. ), and help children learn words (Goldin‐Meadow ; Kelly ).

(1) Simulations play important roles in physical reasoning, about a static configuration of gears, a configuration of pulleys, or the tipping of a glass, which can also produce associated gestures (Hegarty ). (2) The time to make an inference may be correlated to an event's duration (Schwartz and Black ) and carrying out an action can improve inference (Schwartz ). (3) Even more abstractly, people use spatial metaphors to reason about time (Boroditsky ), and one's actual spatial trajectory can influence the interpretation of sentences like “Next Wednesday's meeting has been moved forward two days.” (4) Abstract planning can activate motor areas even when the task involves no motor activity (Dagher et al. ). (5) Number processing appears to involve the circuits for hand motion (Andres et al. ; Zago et al. ).

Traditional cognitivism presumes that our understanding of the internal states of others depends upon a theory of mind (Premack and Woodruff ). Simulation theory proposes that we simulate our own minds to understand those of others, particularly their interior states, their pleasures and pains, their emotions, and their intentions (Goldman ). (1) Neuroscience may be more consistent with seeing social cognition as a form of sensorimotor simulation in which intersubjectivity is built from felt bodily states (Gallese and Goldman ; Oberman and Ramachandran ; Umiltá et al. ). (2). Social simulation theories are supported heavily by the existence and use of mirror neurons. Such neurons, a subset of the circuits for manipulating objects, are also active when observing someone else perform a goal‐directed action (Rizzolatti and Craighero ), giving them a role in the perception of intent. Such neurons represent a general mechanism for understanding the mental states of another, including emotional responses, via simulating our own feelings under similar motions (Gallese et al. ). (3) Certainly bodily states can be effects of social cognitions; for example, as activating an “elderly” stereotype can result in slower pace and slower lexical decisions (Dijksterhuis and Bargh ). (4) Bodily states can also have causal effects of their own. Engaging smiling musculature can produce positive affect (e.g., Strack et al. ; pencil in teeth does a better job than pencil in lips), slumping negative (Stepper and Strack ). Nodding produces positive affect (Wells and Petty ), pushing away negative (Cacioppo et al. ).

Individual differences in abilities to simulate the mental states of others (like pain) correlate with differences in capacities for empathy (e.g., Jackson et al. ). Indeed, such differences are behind Simon Baron‐Cohen's newest work () on The Science of Evil, though our capacities to respond more strongly to immediate others may actually produce difficulties in moral decision‐making affecting larger populations and less immediate events (cf. Staub ). Directing this attention to wider social contexts, and to human persons who are not intimates or nearby neighbors, is certainly one of the important functions that religious traditions may perform.

Modal simulations can implement core cognitive operations like type‐token binding, inference, productivity, recursion, and propositions, whose existence is not in question (Barsalou ). It is not clear how the brain actually does the job. Simulations and embodiments do play causal rather than simply epiphenomenal roles, as the effect of manipulations on motor areas show, as do the effects of bodily manipulations upon social cognition. Unlike amodal representations, such simulations are not autonomous from perceptual systems, bodily action, or their operational details, and meanings are not restricted to their connections to internal symbols. Such motor programs are not separate and independent from cognition, but are part of its composition, with cognitive activity reusing the processes and operations used in perceiving and acting. Such grounding is central to language comprehension, and even abstract concepts seem to depend upon situations and situated action. Moreover, the evidence from embodied cognitive science suggests that nonneural structures are not merely peripheral, but actually contribute, cause, and even constitute the development of specific cognitive capacities, including those necessary for language use. Finally the pervasiveness of a system of mirror neurons may be behind the perception of intentions, the mimicry of action, and inferences about mental states, though it is not clear how human abilities so surpass those of other primates, nor how the compromise of such a system may lead to psychopathologies born from a lack of intersubjectivity. Nevertheless, it is readily apparent why religious practices and traditions which specifically underpin bodily engagement and intersubjectivity may be important to the development of cognitive and moral capacities, as well as to healing social relationships and institutions.

In robotics Rodney Brooks () proposed an embodied approach, which he characterized as “intelligence without representation,” which, instead of heavily computation‐ and representation‐based systems, builds robots in which control is bottom‐up, governed by behavioral engagement with the world rather than by complicated internal algorithms. Andy Clark () provided a theoretical integration for work on embodiment in cognitive science which argues that minds are not centrally “thinking things,” but are primarily in service of doing things in the world in real time. This idea has become central to embodied cognition, and a growing understanding of cognition as scaffolded, embedded, and extended. Robotics research continues to offer a fascinating area in which these insights are actually being used to construct robots which accomplish tasks in real time, and which can also be used to explore human reactivity, for example, by experimentally separating behavioral components which would be more difficult to tease apart in human beings, especially socially interacting ones (cf. Breazeal on designing sociable robots).

Even in the case of human beings, the relationship between conscious experience and neural processes is controversial, despite the neural correlates of such experiences (Noë and Thompson ). Particular experiences may be multiply realizable, even within the same person. As I happily understood from Nancey Murphy (), different contexts might mean the same brain state might constitute multiple mental states; we will return to the argument for externalism and extended cognition later. Even if one assumes that such brain states provide a minimal substrate for a conscious experience, it is not clear whether the neural contents match the content of conscious experience, since they appear to be incommensurable. Unlike the properties of a neural system, experiential content always has a point of view and is active and attentional, able to be revealed and explored via movements of head and body. Noë () argues that the sense of our conscious experience depends on our mastery of sensorimotor contingencies, and is therefore a temporary pattern of skilled activity, something we do rather than something we contain. Even a neuroscientist like Antonio Damasio () presents evidence that our consciousness consists in somatic marking, experiences being tied to bodily sensation.

Several experimental findings show surprising limits to our explicit memory which are consistent with more embodied views: (1) change‐blindness to repeated presentations of a visual scene (originally during visual saccades, Levin and Simons ) and (2) inattentional blindness, when even rather large changes to a scene during attention‐intensive tasks go unreported (such as a dancing gorilla appearing in the middle of a scene; Mack and Rock ; Simons and Chabris ). Both suggest that vision is not about the construction of a mental representation, but a skill of an agent whose temporally extended movements of eyes, head, and body are part of the experience, necessary to direct attention to the environment. If nonneural substrates are necessary for the enactment of a conscious state, then consciousness is bodily distributed. This also means that the same brain states could constitute different experiences in differing contexts (multiple constitutability), and that a brain without a body (“a brain in a vat”) would not have such conscious experience. It also suggests that our sensorimotor coupling with the environment is crucial, as it also provides the proprioceptive and kinesthetic feedback that “somatically marks” an experience as our own. We experience the feeling of our own bodies in action by feeling the moment‐by‐moment control of our actions, including the feedback from touching and being touched by external objects and events. Our agency originates in our sensorimotor engagements with the environment, the “neural signature” of our bodily self‐consciousness (Tsakiris et al. ). Conscious intention may be a form of motor cognition (Haggard ), motor awareness and motor control share the same neuroanatomy (Berti et al. ) and are the neural correlates of experiencing oneself versus another person as the cause of an action (Farrer and Frith ; Farrer et al. ).

Multiple realizability is also shown in evidence that the construction of concepts is context‐dependent, varying across individuals, and within individuals across different occasions (Medin and Shoben ; Solomon and Barsalou ). Only 44% of the features in one person's definition, even of simple categories like bird and chair, are found in that of another, and there is even within‐individual flexibility across a period as short as two weeks (Barsalou ). Patterns of interaction with particular objects can produce distinct patterns of categorization with different kinds of expertise, and in ways different from nonspecialists (Medin et al. ).

Empirical evidence suggests that the retrieval of memories is not independent from sensorimotor mechanisms. For example, in the case of remembering the tools and ingredients for baking a cake, the kitchen location can serve as an external aid to memory, and imagining embodied actions affords the retrieval of information (Cole et al. ). The imagined spatial layout with reference to the observer's body has been shown to affect the recall time (Bryant and Wright ; Waller et al. , ). Memory traces also include body posture (Barsalou et al. ). The body itself appears to contain autobiographical links, as memories of past experiences are facilitated if the body posture is reassumed (Dijkstra et al. ). The most fascinating evidence of a direct bodily role in memory is provided in a study by Presson and Montello () in which blindfolded subjects could readily point to the location of objects in a room; no deficit in performance was produced by subjects actual rotation of 90°, but an imagined rotation produced slow and inaccurate performance. Sadly, the literature on trauma amply documents the existence of trauma‐related sensorimotor connections experienced involuntarily, and producing traumatic responses in the present (Herman ; Van der Kolk , ). Successful therapeutic treatment is greatly facilitated by directly addressing and manipulating sensorimotor states of the body, including agency (Ogden et al. ).

Paula Niedenthal () summarizes the evidence on embodied emotion, suggesting that perceiving and thinking about emotion involves a reexperiencing of perceptual, somatovisceral, and motoric responses. Our mimicry of other's facial expressions, for example, may be what enables us to share feelings and to experience and understand what someone else is experiencing. The tendency to mimic facial expressions has been documented widely (Bush et al. ; Dimberg ). As Ohman and Mineka () show, some of this facial processing occurs at sufficiently early stages of limbic system processing that it can have its bodily effects even when our ability to identify both facial identity and emotional response has been masked by subsequent stimuli; people even mimic faces presented subliminally (Dimberg et al. ). Across a number of levels of analysis mimicry helps interlocutors to establish rapport, empathy, and cooperation (Berneiri ; LaFrance ). Even neonate movement is synchronized with adult speech (Condon and Sander ), and adult rapport is tied to postural synchrony (LaFrance and Mayo ).

There is a further paradigm here, summarized by De Jaegher et al. () which argues that social interaction may actually constitute social cognition. Empirical results do imply that social cognition is not reducible to the cognitive mechanisms of individuals, and that interaction is more than a context, but may complement and even replace individual mechanisms. The notion of engagement is used to capture the qualitative aspect of a social interaction in a complex coregulated pattern once it takes on a “life of its own.” An example is the case of Murray and Trevarthen's () study in which infants and mothers interact via a television link. When presented with a recorded (rather than “live”) display of their mothers, infants disengage, become distracted and upset. This could be because of an individual mechanism that, via relative timing of actions, might distinguish contingency and noncontingency. However, research suggests an enabling role for prior engagement (DiPaolo et al. ). Nadel et al. () fail to replicate the results of Murray and Trevarthen when mothers and infants are not allowed to develop sufficient engagement before the recording is presented, suggesting that the enabling condition is prior engagement, not merely contingency.

Social interaction is not merely enabling but a constitutive element in an experiment by Auvray et al. (). Subjects attempt to detect each other's movable sensor on a virtual line where there is also a static object and a “shadow” object copying the movements of the sensor at a fixed distance, of which subjects are unaware. Subjects do learn to concentrate their mouse clicks on each other's sensors rather than the shadow (66% vs. 23%). But this is not based on contingency; individuals cannot distinguish between detecting the others sensor or the shadow. They use back and forth patterns to help distinguish moving from nonmoving objects, but the situation only stabilizes when they scan each other's sensors, scanning the others shadow is a disengaged one‐way coupling, and the other participant will move away because she is still searching. Hence the stability of sensor detection is produced by the stability of the coupling and not to individual strategies. The interaction process is constitutive; the detection task could not be accomplished otherwise. Characteristics of social interactions external to individual participants play explanatory roles in a cognitive task.

Kohlberg's developmental work on moral reasoning () remains the gold standard, but not until recently has it asked questions about embodiment. The work of Jonathan Haidt and others (Greene and Haidt ; Haidt ; Haidt et al. ) has been challenging this view, and his research suggests that moral judgment may be driven by bodily affect, the rational tail being wagged by the emotional dog. Harmless actions which produce strong affect (eating a dead pet, incest between consenting adults) are often judged to be morally wrong; subjects provide justifications based on nonexistent harms, rationalizations after the fact which may mask the affective origins of the judgment. There is a relationship between moral cognition and disgust (Wheatley and Haidt ), disgust also being an emotion of social rejection (Niedenthal et al. ). Feelings of disgust induced by exposure to a bad smell or a dirty room can make moral judgments more severe (Schnall et al. ), and subjects given a cleanliness manipulation find certain immoral actions to be less wrong (Schnall et al. ). All in all, we see an essential role for the embodied aspects of cognition in much of moral judgment (Haidt ).

The research literature suggests that social and moral cognition may be guided or constrained by specific bodily and affective reactions, and states of bodily excitation readily transfer across situations (Bryant and Miron ). Damasio's somatic marker hypothesis (, ) suggests that the reenaction of bodily states triggered during emotional experiences provides a source of information centrally important to prioritizing in decision‐making, and organizing our courses of action. When the capacity to integrate such feelings with one's knowledge is compromised, as it is in patients with damage to the ventro‐medial prefrontal cortex (e.g., Phineas Gage), their decision‐making is impaired. Missing an embodied state, like a galvanic skin response in anticipation of possible loss, such patients miss information essential to making a less risky choice (Bechara et al. ). This suggests that affective and bodily feedback is a necessary part of the normal physical implementation for realizing cognitive processes.

Organism‐environment couplings can be described by dynamic systems theory rather than broken down into stimulus inputs and internal computations. States of such systems are coupled to reflect interaction‐dominant dynamics between each other, with the environment, or with each other, in the couplings between perception and action. One particular program of research in nonlinear dynamic modeling has shown that 1/f noise (sometimes called “pink noise” or “long memory;” intermediate between white noise, with no correlation in time S(f) = k, and random walk Brownian motion with no correlation between increments S(f) = 1/f²) is ubiquitous in cognitive activity. This shows that the connections among the cognitive system's components are highly nonlinear, which indicates that they are not modular (Holden et al. ; Van Orden et al. , ), since the operations of nonlinear systems are not easily localizable but distributed. “Parts” of the system cannot be treated as functionally or structurally separate: they are synergies. Van Orden et al. () argue that this is a signature of a soft assembly, sustained not by component‐dominant dynamics, but interaction‐dominant dynamics (where parts alter the dynamics of other parts, with complex interactions extending to the body's periphery). Soft assembly, as the product of strongly nonlinear interactions can account for the 1/f character of behavioral data; assembly by virtue of components with predetermined roles and communications cannot. Interaction‐dominant systems cannot be modular (Anderson et al. ).

This might all seem rather arcane, except that 1/f noise is found in a lot of cognitive and behavioral tasks (Van Orden et al. ), showing that task‐specific, softly assembled systems including brain and body were responsible for performance. Hence, even in tasks like tapping, key pressing, word naming and others, the cognitive system is not encapsulated in the brain, and includes the motor component of the body itself, though 1/f noise is also found in purely cognitive phenomena, modeling insight in problem solving. Stephen et al. () found that finding a new strategy for problem solving coincides with the appearance of 1/f noise in eye‐movements, suggesting that eye‐movements are part of the cognition.

As Watts pointed out, the Judeo‐Christian tradition has taken a renewed emphasis on the importance of embodiment, the Hebrew Bible viewing the human person as an “ensouled body,” St. Paul being better understood as assuming a complex, nondualist view of the person, and Aquinas seeing the soul as the form of the body. But whether a religious tradition refers to the resurrection of the body, or to its reincarnation, the belief from the book of Job that “in my flesh shall I see God,” receives direct empirical support from research on cognition in the flesh; without my flesh I cannot see anything at all. What it suggests is that while flesh alone may be insufficient to see God, it is a necessary mediator. Processes beyond the brain and even the nervous system are involved in the guidance of bodily action, in intelligence, in the “somatic marking” of consciousness, in bodily memory, in moral judgment, and even in social engagement, in ways that may even step beyond the individual body, emerging from the interaction between embodied persons.

If the boundary between the cognitive agent and her environment is malleable, blurry, or indeterminate, so might cognition be extended to the inclusion of other human agents. The belief (at least in this culture) that individuals are independent, functionally autonomous, and bounded from each other is often accepted as self‐evident. But there are research examples in which this is belied.

(1) The rhythmic movement coordination of two interacting individuals is unintentionally and spontaneously constrained to an inphase or antiphase relationship. The inphase is more stable, but stability is lessened by increases in either the movement frequencies themselves, or the degree of difference between the natural frequencies of components. The stability and patterning of the coordination doesn't depend upon the specific movements or parts involved, but on the strength of the (visual or auditory coupling) between the two. The same dynamics occur in intrapersonal interlimb coordinations of finger, wrist, arm, leg, torso, or even the interpersonal coordination of two individuals in rocking chairs. The order and regularity depends more on the nonlinear relations that couple the movements than any part of the human perceptual‐motor system, a soft‐assembled synergy (Richardson et al. ).

(2) Research by Harrison and Richardson () provides a compelling example. Pairs of subjects are instructed to walk and jog together tethered by a 75 cm long foam pole. As in the previous research, the leg movements became spontaneously phase‐locked. Surprisingly, they also exhibited a preference for particular four‐legged gait patterns, like pace or trot, preferences determined by differences in gait stability. Apparently stable, multilegged patterns can emerge without direct neuromuscular coupling, showing that the organization of stable interpersonal motor control need not require any centralized mental or neural structure.

(3) Other research also shows the formation of coherent perception‐action synergies, Chang et al. () showed that the perception of the passability of an opening by an adult perceiver with a child companion was perceived on the basis of body‐scaled information defined by the dyad rather than by either alone. Isenhower et al. () showed similar findings for paired subjects in a graduated plank‐lifting task, showing that the transition from solo to joint activity bifurcated at a ratio of their collective action capabilities, a point of transition dependent on their previous history, transitioning later from joint to solo lifting when the pair began with descending presentation rather than ascending. (This is called hysteresis, a dynamic “memory” process inherent to interaction‐dominant, softly assembled system), implicitly committing to being a “plural subject” when beginning together, a decision to cooperate made without planning or prior expectation occurring as a dynamic response.

(4) Roberts and Goldstone () provide a more traditionally cognitive example of a social system that is softly assembled and interaction‐dominant. An Internet connected group was to guess a randomly generated number over a series of rounds, summing the group members’ responses and giving feedback on whether their number was too high or too low. Groups took fewer rounds to reach the target across a series of games. But the reactive strategies taken by members of the group spontaneously differentiated into the role of reactors (always decreasing or increasing estimates) or nonreactors. The faster this differentiation occurred, the more successful was the group, a success not reducible to any member, or the result of a steady state for any individual. The group itself exhibited cognition by being temporarily constrained to act as a synergy, which was softly assembled and exhibited interaction‐dominant dynamics.

Embodied cognition is actually a subset of a broader set of approaches to situated cognition (for which there is a whole Cambridge Handbook), first stepping beyond the brain to other aspects of an agent's body, including those involved in sensory and motor systems in particular (which arguably give thoughts their content), then to its embedding in a natural and social environment, where cognitive activity in the wild, to use the title of Hutchins's book on navigation, may be distributed across the agent‐environment systems, including social ones. The core of the embodied science community still holds most of the nonneural cognitive resources to be bounded by the skin. Nevertheless, research on group‐level effects, like Hutchins (), and the research summarized above from the tradition of ecological psychology (Anderson et al. ), suggest that the expansion of the cognitive system need not stop at the boundaries of the biological body, and that even the boundaries between individuals may open in different ways under interaction‐dominant dynamics, and be replaced by varying levels of synergy. Extended cognition stipulates that external features are more than distributive, but may actually constitute a broader cognitive system (Clark ; Clark and Chalmers ; Wilson ). Might some of our cognitive tasks, including even remembering, involve off‐loading some of their components not only to our bodies, but even to the external world? Is it even possible, as Merlin Donald (, ) suggests, that we live in an era where our externalization of memory has made it all the more obvious that the nature of human beings is as symbolic symbionts? And if we are symbolic symbionts, are not the most important of those symbioses found in diachronic relationalities with other human beings? Here, of course, the argument becomes more philosophical, but I think this is the direction to which embodied cognitive science is going to be inexorably drawn.

Until quite recently, our conceptions of mind, self, and soul have held these to be internal to the central nervous system of our biological organism, an internal/external boundary which has roots as far back as the early modern emergence of science. Rejecting both the individual possession and locational internalism of Descartes, there is a growing externalism within scientific and philosophical studies of mind which view it as embodied, enactive, encultured, and interwoven with a social and technical web, and as a construction not limited to the boundaries of the individual organism (Wilson ). Originating a generation ago in the content externalism of Hilary Putnam () and Tyler Burge (), that the semantic content of mental states (seeing vs. hallucinating a tree) is often dependent on factors external to the subject (the actual presence or absence of a tree), the last decades have seen the emergence of a substantially stronger process or vehicle externalism, that the structures or mechanisms making various mental states possible may themselves extend beyond the skin (e.g., Clark and Chalmers ; Hurley ; cf. Teske for a review of these varieties). Many of our mental states are hybrids, spread across internal and external materials, biological or not.

Externalism is, quite simply, the view that “the mind ain't in the head.” It is the claim that the mind is constituted by the mechanisms and resources that we use to think. It asserts that the constitution of thoughts, beliefs, and desires often includes, even requires, states and processes external to our biological organism. This is not unlike the extended phenotype of evolutionary biology. In the case of human beings, our cognitive niche includes many things by which we extend our minds into the environment, like imitation and symbol use, as well as external artifacts, and even social practices and structures, from which emerge the possibilities of distributing cognitive tasks across individuals or accumulating knowledge across generations. That does not mean that the mind's location is separate from heads and bodies, as these are proper parts of a mind (mereological, part‐whole relationships). Mental phenomena are hybrids that couple events in the world to physical processes in the nervous system. According to Mark Rowlands (), this is the most important development in the philosophy of mind in the latter half of the twentieth century, rooted in the phenomenological philosophy of Husserl, the linguistic philosophy of Wittgenstein, and in the existentialism of Sartre.

The manipulation and exploitation of information‐bearing structures is also likely to have been important in the historical development of some of the abilities which they make possible, as in code memorization, or the development of capacities for reading and writing. There is a huge difference in how our brains are shaped, and how we interact with the world, and how we use this important set of external memory structures. Indeed, as documented by Luria () and Vygotsky (), many of our higher cognitive functions have been socially scaffolded in ways that are contingent on historical changes in social life and organization. We think of our higher cognitive functions as being produced by the basic equipment of the brain, but there are historical developments, learned though socialization, which are also necessary to make them possible. We are evolved to have remarkably plastic brains, which are what make us historical beings, including substantial changes in how our brains are shaped developmentally to do what we so take for granted as part of intellectual functioning. Such abilities have been so shaped by the symbolically rich environment around us that we cannot make a principled separation between our ability to remember and our ability to exploit ambient information. This includes the formalisms that reduce complicated arithmetic calculations to an iterated set of simpler steps, as well as the use of technological artifacts which we find increasingly indispensable.

The Parity Principle states that “if something plays a role in cognitive activity, such that, were it internal we would have no difficulty in concluding that it was part of the mind, it should be counted as part of the mind whether it is internal or not” (Clark and Chalmers ). If Inga and Otto both set out for the Museum of Modern Art, Inga by recalling its location on 53rd Street and the Alzheimer‐suffering Otto by consulting his notebook, there is no reason to treat their memories differently, as both are just as accessible, reliable, and transparent in use. Even paradigmatically mental events, like biologically instantiated memory, are not constantly available, not always easily accessible, not automatically endorsed, and can be implicit or procedural. Why not also include the downloading to external artifacts which distributes the cognitive load of a task, from paper and pencil to the electronic prostheses of calculators or cell phones?

Parity may only be a special case of a much wider complementarity principle which argues, on grounds of both individual differences and historical variation that the advantages of external resources, both physical and symbolic, might not merely be in duplicating internal functions, but in developing capacities otherwise unavailable. Different cognitive artifacts do actually have different effects on our brains, as brain changes under literacy show (expansion of the planum temporale), and historical and cultural differences in the uses of such artifacts suggest. As Clark () points out, the questions about the range and variety of cognitive scaffolding and the different ways they can enhance (or damage) performance are empirical questions. The nonlinear, complex, and iterated couplings between the brain and the body, and between the brain and external resources, in which each may have effects on the other, render any boundaries highly permeable. Even the extent of context dependence varies widely over contexts and tasks. Such external couplings may be precisely what make us the kind of creatures we are, easily extending our minds onto the environment, including the shared social space with which we are so mimetically engulfed. Indeed, given the evolution of our extended childhoods (Konner ), and the extensive shaping of our neuroplasticity by socialization and enculturation, they may make human history and civilization possible.

If the mind is extended does the self follow? Much of what matters to our identity are our cognitive capacities, so this would seem to be a natural extension (Clark ). If so, this might have many implications, especially for autonomy and responsibility. Is violating someone's externalizations have comparable moral significance to violating one's body? Anyone who has felt her personal space violated, or been the victim of a theft, can feel invaded and vulnerable, so perhaps we already have a kind of sliding‐scale of ownership and identification. Could a “frail control hypothesis,” that external contingencies partially realize a behavior, mean that human beings have little control and no normative competence, or does it extend it? If agency is distributed, then how should punishment be applied? If agency is not restricted to neural circuits and bodily experience, don't we have to rethink ideas of normative competence, freedom, and personal identity? Or can we still appeal to a difference between agency, identified as the locus of control located in the agent's body, though we allow that cognitive systems can be extended, not unlike what we do when we distinguish between the biological boundaries of an organism, like a spider, and the extended biological system, which includes the webs they spin (Wilson ).

From my own religious tradition, there are explicit references to us all being parts of one body; as Paul puts it in his first epistle to the Corinthians: “But God has so composed the body, giving the greater honor to the inferior part, that there be no discord in the body, but that the members may have the same care for one another. If one member suffers, all suffer together; if one member is honored, all rejoice together” (1 Cor. 12: 24–6). Within the Buddhist tradition, the idea of “interdependent arising” may capture some of this same spirit, and this same kind of bodily grounding.

Personal identity is made possible by the evolution of a human neuropsychology that requires social interdependency for its development. Our neuroplasticity requires shaping over a lifetime, socially scaffolding our neuroregulation, including emotional attachments and dynamics. The evolutionary hypertrophy of our prefrontal cortex leads to a colonization of brain function making possible the social construction of virtual realities, novel forms of socially constituted experience, and the transforming effects of mythic, ideological, and religious systems (Teske ).

A neural affect system is shaped into emotional patterns by the social scripts laid down during our lengthy period of developmental dependency, including second‐order emotions, the development of independence, autonomy, and relations of intimacy and power (Nathanson ; Tomkins ). Pride, guilt, and shame are generally thought to be emotions about other emotions, and involve experienced expansions and contractions of self‐boundaries respectively (e.g., “hiding in shame”). While the affect systems are strictly biological, it is the production of regular patterns of emotion, and their recall, which produce the organizing scenes and scripts that are the basis of our personal dramas. These patterns will not only be heavily dependent upon the domestic or family dynamics of a particular moment in history and culture, but are likely to shape, and necessarily so, our extremely plastic and immature nervous systems during the course of development, in ways that may often be irrevocable (Teske ).

If our neuroplasticity makes it possible for us to be “natural‐born cyborgs” (Clark ) one of the crucial lessons of our extended developmental dependency (Konner ) must certainly be how much our externalism is rooted in biologically embodied relationships with other human beings. We offload memory anytime we ask someone to remind us of something, and we even make “mind” distributed when we distribute memories socially, as when couples specialize, for example, where one remembers birthdays on both sides of the family, the other remembers vacation locales. The externalist position being put forth here is that mental life is both embodied and embedded in the world, not just located within the nervous system: the nervous system is a necessary part, but it is only a part, not the whole construction.

Research on empathy and on the neurophysiology of social interconnectedness suggests the existence of a primary intersubjectivity, out of which experiences of separate autonomies need to be differentiated. There are contemporary cognitive and neuropsychological views that self and other have no independent existence, no intrinsic identity, and our subjectivity is preceded by an intersubjectivity produced by empathies running deeply beneath our embodied and interdependent biological lives (Thompson , ). Empathy exists in our involuntary and sensorimotor coupling, mediated by “mirror neurons” (Iacoboni ) which respond similarly whether preparing one's own or observing the movements of another. There is also an affective resonance resulting from our capacity to read and mimic facial expressions automatically, and by which we feel what someone else feels. The measurable nonverbal duet in empathy includes matched patterns of arousal and even complimentary breathing (Goleman ). Higher levels of empathy include the imaginary transposition to the place of another, and a mutual self and other understanding which involves a reiterated experience of seeing each other as experienced empathically by the other. This is how we come to experience our bodies as objects belonging to an intersubjective world. “In this way, my sense of self‐identity in the world, even at the basic level of embodied agency, is inseparable from recognition by another, and from the ability to grasp that recognition empathically” (Thompson , 268). Human subjectivity is intersubjectivity from the outset, developing from it, “configured by the distributed cognitive web of symbolic culture” (382).

We establish our autonomy, our freedom, and our identity, finally, only on the fragile and vulnerable ground of our intimate interdependencies (Winter ). While there are dangers when our self‐boundaries are overwhelmed, intimacy requires us not only to guard them less zealously, but to open them, both to another, and to ourselves, as the only way we can ever transcend ourselves is by going beyond them. Self‐transcendence is driven by our longing for the kenosis of pouring ourselves into things greater than ourselves. Perhaps we too easily forget the ones closest to us, those who know us best, and in whom we may find the only others who really matter, and with whom we might find the other within, in our anxiety and vulnerability, to step outside of ourselves and genuinely love.