. 30
( 62 .)


¬eld in such a way as to suggest new directions for the computational
theory of mind. I have suggested that one could interpret Marr™s theory
of vision as invoking the strategy of integrative synthesis (over Marr™s own
likely objections), and identi¬ed three paradigms of research in cognitive
science “ in distributed cognition (Hutchins), in animate vision (Ballard),
and in reactive robotics (Brooks) “ that prima facie also exemplify this
Individualism and Externalism in Cognitive Sciences

Individualists, of course, are unlikely to take the appearances here at
face value. Indeed, they should challenge the claim that any of these re-
search paradigms are properly construed as invoking wide computational
systems. There are various ways of doing so: from questioning the con-
ceptual integrity of the very idea of wide computationalism, to arguing
that individuals are the largest cognitive units to which we in fact apply
computational analyses of cognition, to pointing to inadequacies of wide
computational interpretations of speci¬c theories and explanations.39
Rather than continuing this debate here, I want to move on beyond
computation. In the next chapter, I shall generalize from wide computa-
tionalism to wide psychology, and in so doing discuss a variety of noncom-
putational approaches to cognition that also exemplify the externalist
view of cognition that I have been developing in Part Two. The argument
will be that both taxonomic and locational externalism, whether compu-
tational or not, provide the theoretical backdrop for redirecting some of
the individualistic research on core topics, such as memory and cognitive
development, in ways that draw on some largely forgotten insights within
psychology itself.
part three


The Embedded Mind and Cognition

1 representation and psychology
The chapters in Part Two have laid the foundation for a general exter-
nalist view of cognition and the mind. For the most part, the concepts of
realization, representation, and computation have been developed and
deployed within individualistic frameworks. In broadening and refash-
ioning these concepts I have created a space for an externalist psychol-
ogy, and illustrated ways in which some of that space has already been
occupied within computational cognitive science.
The most important of these concepts for cognitive psychology is that
of representation. Cognitive psychology explores the nature and struc-
ture of mental representations and how they are processed: how they
are stored, retrieved, transformed, and related to one another. In the
last chapter we saw that a representational view of cognition need not
be individualistic. In this chapter, I move beyond foundation laying for
externalism to show the place that exploitative representation has within
cognitive psychology. I shall focus on areas of psychology in which rep-
resentation has played a central role “ on memory (section 4), develop-
mental psychology (section 5), and folk psychology and the theory of
mind (section 6).
Representation is not simply a form of encoding but more generally
a form of informational exploitation of which encoding is a special case.
Representations need not be thought of as internal copies of or codes for
worldly structures. Rather, representation is an activity that individuals
perform in extracting and deploying information that is used in their
further actions. It involves an agent enmeshed with the world not prior

Thinking Through and Beyond the Body

to or following but in the very act of representing. On the traditional view
of representation, cognition is wedged between perception and action,
implying, in the philosopher Susan Hurley™s words, that “[t]he mind is
a kind of sandwich, and cognition is the ¬lling.” The shift in perspective
that the concept of exploitative representation introduces opens the way
to developing a view of cognition that treats what is inside the head and
what is beyond it in a symmetrical fashion.1
One form that this symmetry takes, exempli¬ed by the multiplication
example depicted in Figures 7.1 and 7.2, is the identi¬cation of explicit
symbolic structures in a cognizer™s environment that, together with ex-
plicit symbolic structures in its head, constitute the cognitive system rel-
evant for performing some given task. Such symbols in the world can
be exploited rather than encoded by individuals and their in-the-head
computational systems.
The same is true when information in the world does not take this
explicit symbolic form. To use the psychologist George Miller™s apt term,
we are informavores and can exploit causal and probabilistic dependencies
in the world in generating in-the-head structures that, in part, direct
our behavior. In the previous chapter, I developed this idea in terms
of computation extending beyond the boundary of the individual, but
we could express this more generally in terms of information-processing
systems that do so. Again, there is no metaphysical signi¬cance to the
boundary of an individual™s body for individuating where that individual™s
mind begins and ends.2
The exploitative view of representation is one way to develop the
idea that cognition is situated, embedded, and embodied. One often-
expressed concern about this idea is that it involves an unacceptably
de¬‚ationary understanding of what cognition is. Proponents of the em-
bedded mind either focus exclusively on aspects of cognition that involve
the organism™s interface with the world (for example, perceptual and mo-
tor capacities), or they offer thin behavioristic or functionalist construals
of intuitively more central cognitive abilities. By showing how exploita-
tive representation applies to paradigm cases of core cognitive capacities,
and by suggesting extensions of existing paradigms of representational
psychology, I hope to preempt both of these standard criticisms.

2 life and mind: from reaction to thought
Our paradigm of a living thing is an organism. All organisms react to
occurrences in their environments, and use those reactions to control
The Embedded Mind and Cognition 185

their bodies in some way. I am inclined to think that much the same is
true of all thinking things, and to view this as a simple but deep fact about
the nature of cognition, certainly for animate creatures like us whose
cognitive functioning is tied to their continued existence as living beings.
Cognition allows us to register what is in our here and now environment,
and to adapt our bodies to what we register. But this truth should not
overshadow the fact that cognition also allows us to do much more, to
go beyond our immediate environments, back to the experienced past
through memory and ingrained habit, and forward to the distant future
through planning and imagination.
The parallel between the embeddedness and embodiment of living
and thinking things, together with the recognition that thought is some-
thing more than stimulus driven and response driving, invite the ques-
tion of what more there is to cognition than simple registration and
reaction. We can think of this in terms of types of representational sys-
tem that organisms possess, each with a distinctive locus of control for
First, a creature might simply have a reactive representational system. It
registers one or more states of its environment, and this guides its behav-
ior, but the connections between registration and reaction are simple and
¬xed. The behavior of the creature is thus effectively under control of the
stimuli in its environment. Vary the position of an intense light source
within certain parameters and you cause a variation in the position of the
¬‚ower in a sun¬‚ower. Adjust the magnetic ¬eld in the liquid in which
a paramecium ¬‚oats, and you change the direction in which it moves.
Although the internal structure of creatures with such representational
systems plays a crucial role in the mediation of stimuli and response, the
locus of control for their behavior is external. It is environmental.
But many creatures are able to exercise more control over their own
reactions. Their representational systems are not simply reactive but en-
active: They endow those blessed with them some control over the nature
and strength of the behaviors enacted. A registration of the environment
is made, but it does not automatically generate a reaction in what is done.
Rather, it is combined with other registrations, and these together gen-
erate a bodily output. I would hazard the guess that most of the animal
behavior that we are familiar with in everyday life “ from interacting with
our domestic pets, to watching squirrels or birds in a yard, or vertebrates in
general in the wild “ involves enactive representational systems. It is with
enactive representational systems that most of us are ¬rst comfortable
in invoking distinctly psychological language in a literal sense. Enactive
Thinking Through and Beyond the Body

creatures perceive and decide, whereas reactive creatures do so only in a
metaphorical sense.
Yet even if an enactive representational system introduces a distinc-
tive psychology, it does not suf¬ce, one might think, for the full range of
psychological capacities that creatures have. In particular, enactive rep-
resentation is still closely tied to what one does with one™s body, and does
not yet give us higher cognition with a completely internal locus of con-
trol. That is what a symbolic representational system creates, the capacity
to divorce cognition from its bodily origins. It creates genuine thinkers,
creatures who can use representations to generate other representations,
and so whose cognition may have a high level of autonomy from the here
and now. It is here that we have thought, inference, reasoning, plan-
ning, wishful thinking, and re¬‚ection. The heart of cognition. Cognition
Human beings manifest all three types of representational systems: in
re¬‚exes (reactive), in bodily skills (enactive), and in higher cognition
(symbolic). Although there is a physiology that underlies any biological
re¬‚ex, it is plausible to view the re¬‚ex as under the control of the stimulus
that elicits it. (In fact, if standard stimuli do not elicit the re¬‚exive reac-
tion, then we view the system responsible for it as having broken down,
as not functioning as it is supposed to.) It is in this sense that re¬‚exes are
not options exercised by the organisms that have them, and the external
locus of their control is one reason not to think of them as cognitive
in nature. Bodily skills, such as riding a bicycle, introduce such options,
given any particular stimulus. Yet there remains a sense in which they
are not controlled “centrally” in that they do not require “ in fact, often
require the absence of “ direction from paradigmatic cognitive states,
such as beliefs and desires. This is why they can be performed, as we say,
“without thinking about it,” and why thinking about it sometimes gets
in the way of successful performance. Once you know how to perform a
bodily skill “ from bicycle riding, to typing, to gymnastics, to catching a
ball “ it is a matter of letting the body do what it knows to do, rather than
trying to control what it does consciously. One may concentrate in per-
forming a bodily skill, but not on the sequence of actions that constitute
that performance.
In contrast to both these cases, higher cognitive capacities have an in-
ternal locus of control: What directs them, what leads them ultimately to
govern our behavior, lies within us. This control is not always conscious
but it is neither environmental nor bodily and is mediated by “what is in
The Embedded Mind and Cognition 187

table 8.1 Locus of Control and Representational Type

Type of Organism/ Example in
Locus of Control Representational System Humans
environmental reactive re¬‚exes
bodily enactive mimetic skills
cranial symbolic beliefs, desires

the head.” Symbolic representation makes for truly voluntary behavior,
free action, genuine choice, and rationality. Some would say that symbolic
representational systems are uniquely human, or that they are what make
language, a social life structured by conventions, and rich cultural tradi-
tions possible. Table 8.1 summarizes the three types of representational
system, and some of what I have said about them.
Given this view of higher cognition, it is perhaps natural to think that
cognitive psychology, at least in studying such higher cognitive capacities,
should focus just on what is in the head. It should bracket off what the
head is in and maintain that the symbols that it is concerned to under-
stand supervene on the intrinsic, physical states of not just the individual,
but of the brain. In short, higher cognition is individualistic.
The plaint of this chapter is that this ¬nal conclusion is mistaken.
Before turning to the substantive discussion, a brief sketch of why I think
even so-called higher cognition is best viewed from the externalist point
of view.

3 the embeddedness and embodiment
of higher cognition
All three areas of cognitive psychology that I will discuss in this chapter “
memory, cognitive development, and folk psychology “ have a metarepre-
sentational edge to them in that they involve the representation of represen-
tations. Metarepresentation epitomizes much of what I have said about
symbolic representational systems. It has been thought to characterize,
and even to uniquely identify, human intelligence. It is a prerequisite for
rational re¬‚ection and deliberation, and for living one™s life in accord
with goals, plans, values, and ideals. If the locus of control as we move
from re¬‚exes to bodily skills to symbolic abilities becomes increasingly
internal, increasingly removed from the direct effects of what the head is
Thinking Through and Beyond the Body

in, then surely metarepresentational abilities go one further step in this
direction. And if an internal locus of control makes for an individualistic
perspective, then externalists should face an uphill battle in trying to
make sense of these abilities.
What I would like to do is grant the symbolic nature of cognition and
the idea that cognition has an internal locus of control, and show how nei-
ther leads to individualism. But this simply makes space for an externalist
view of cognitive psychology. What reason is there to occupy this space?
A common theme in my discussion will be that as we move from simpler
to more complicated cognitive processing in our accounts of memory,
cognitive development and folk psychology, the pressure to move from
an individualistic to an externalist psychology increases. Far from being
the province of an inwardly withdrawn mind, metarepresentation and
the levels of cognitive performance that it facilitates belong to the mind
as it is located in the social and physical world.
The strategy of argument that I shall use in making a case for an exter-
nalist cognitive psychology will draw on an inversion of the trichotomy
between reactive, enactive, and symbolic representational capacities. I
will argue that many cognitive capacities in symbol-using creatures, far
from being purely internal, are either enactive bodily capacities, or world-
involving capacities. These capacities are not realized by some internal
arrangement of the brain or central nervous system, but by embodied
states of the whole person, or by the wide system that includes (parts of)
the brain as a proper part (Table 8.2).
Enactive bodily capacities and the embodiment of cognition more gen-
erally are important in their own right. But understanding the ways in
which higher cognition is embodied can also be instrumental in show-
ing the limitations to encoding views of representation that underpin
individualistic views of the mind. Since my primary concern is to show
how higher cognition extends beyond the boundary of the individual, I


. 30
( 62 .)