AVANT Vol. V, No. 2/2014 www.avant.edu.

pl/en

AVANT, Vol. V, No. 2/2014

ISSN: 2082-6710 avant.edu.pl/en
DOI: 10.12849/50202014.0109.0002

The Enactivist Revolution

Kenneth Aizawa
Department of Philosophy
Rutgers University, Newark
ken.aizawa[]gmail.com
Received December 2013; accepted June 2014; published Autumn 2014.

Abstract
Among the many ideas that go by the name of “enactivism” there is the idea
that by “cognition” we should understand what is more commonly taken to be
behavior. For clarity, label such forms of enactivism “enactivismb.” This ter-
minology requires some care in evaluating enactivistb claims. There is a genu-
ine risk of enactivist and non-enactivist cognitive scientists talking past one
another. So, for example, when enactivistsb write that “cognition does not
require representations” they are not necessarily denying what cognitivists
claim when they write that “cognition requires representations.” This paper
will draw attention to instances of some of these unnecessary confusions.
Keywords: enactivism; enaction; cognition; behavior; autopoiesis.

In soliciting contributions to this special issue of Avant, the editors asked

whether enactivism fits cognition. This question, however, may well misin-
terpret what at least some forms of enactivism are about. It may underesti-
mate the breadth of the revolution that at least some strains of enactivism are
championing. For some in the enactivist movement, it appears that the goal is
not merely to provide a revolutionary new account of what cognition is. We
need no more of that. What is needed, instead, is a cognitive science that
studies something else. What is needed is a cognitive science that does not
study cognition! That would be a real revolution.
But, what, one might ask, would such a really revolutionary cognitive science
study, if not cognition? One popular proposal is that it should study what has
been, and generally continues to be, known as (a type of) behavior.3 For con-

3
Another way of making the present point might be to distinguish cognitione (for enactivist cogni-
tion) and cognitionc (for cognitivist cognition). This might make it easier to see that this issue is

19
 The Enactivist Revolution

venience, and in order to distinguish this form of enactivism from other

forms, let us label it “enactivismb,” where the subscript indicates the focus on
(a type of) behavior. To be sure, not all those who think of themselves as en-
activists are enactivistsb. Nevertheless, there are prominent representatives
of enactivismb. To take one salient example, many enactivistsb have been in-
spired, at least to some degree, by Humberto Maturana and Francisco Varela’s
Autopoiesis and Cognition. Maturana is relatively explicit about understand-
ing a cognitive system as a (self-maintaining) behaving system. According to
Maturana, “A cognitive system is a system whose organization defines a do-
main of interactions in which it can act with relevance to the maintenance of
itself, and the process of cognition is the actual (inductive) acting or behaving
in this domain.” (Maturana and Varela 1980: 13). To take a more recent ex-
ample, Tony Chemero proposes, “cognitive scientists ought to try to under-
stand cognition as intelligent behavior” (Chemero 2009: 25). Further, he be-
lieves that “radical embodied cognitive science can explain cognition as the
unfolding of a brain-body-environment system” (Chemero 2009: 43). But,
“the unfolding of a brain-body-environment system” sounds like a metaphor
for behavior.
The foregoing point might be made in another way. Notice that enactivistsb
often propose their theory as an alternative to cognitivism, but they are not
always explicit about what in the cognitivist view they reject. As one example,
Stewart, Gapenne, and Di Paolo write,
The aim of this book is to present the paradigm of enaction as a framework for
a far-reaching renewal of cognitive science as a whole. There have been many
critiques of classical, first-generation cognitivism based on the Computational
Theory of Mind. A distinctive feature of this book is a deliberate choice not to
go over that old ground yet again, but to reserve the energy for positive explo-
ration of new paths. (Stewart, Gapenne, and Di Paolo 2010: vii).

Di Paolo, Rohde, and De Jaegher 2010, provide a similarly opened-ended rejec-

tion of cognitivism:
Almost two decades since the publication of The Embodied Mind (Varela,
Thompson, and Rosch 1991), the term enactive has moved out of relative ob-
scurity to become a fashionable banner in many regions of cognitive science.
… Theirs was not only an achieved synthesis of existing criticisms to a predom-
inantly computationalist paradigm, but also the articulation of a set of postu-
lates to move these ideas forward. Indeed, the increasing use of enactive ter-
minology serves as an indication that the time is ripe for a new era in cogni-
tive science. To a great extent, we believe this to be so. (Di Paolo, Rohde, and
De Jaegher 2010: 33)

not about the “right” way to use “cognition” or who gets to use it how. Those who prefer to use
this terminology are free to use it, but this paper will adhere to the more mainstream usage. The
important point, of course, is not ultimately about the terminology, but the fact that cognitivists
and some enactivists are talking about two different things.

20
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

One might well think that enactivismb means to displace the cognitivist’s com-
putation theoretic apparatus of rules and (especially) representations with
another sort of apparatus. This alternative apparatus might be the mathemat-
ics of dynamical systems theory or one or another definition of autopoiesis.
In such a vision, enactivismb and cognitivism are competing theories of the
same thing in the way that Newton’s theory of gravitation and Einstein’s theo-
ry of general relativity were competing theories of a single putative force in
nature, namely, gravity. This, however, apparently underestimates just how
sweeping a change the enactivistsb wish to make in cognitive science. Enactiv-
istsb generally propose to walk away from the issues and concerns of main-
stream cognitive science to focus on what is commonly understood as behav-
ior. In practice, therefore, enactivistsb use different tools to study diffe-
rent issues.
While there are times when enactivistsb enthusiastically embrace the dra-
matic changes implicit in their work, they also tend to paper over the signifi-
cance of the proposal to study (a type of) behavior by calling behavior “cogni-
tion” or “lower-level cognition” or “basic cognition” or “minimal cognition.”
(See, for example, Calvo and Keijzer 2009, Chemero 2009, Stewart 2010, Di
Paolo, Rohde, and De Jaegher 2010, and Hutto and Myin 2013.) This terminol-
ogy is likely to be misleading to mainstream cognitive scientists, but a more
serious problem is that it seems to mislead even some enactivistsb. There are
times when they write as if they intend to address traditional, sometimes long-
standing, problems surrounding cognition. Yet, because they use “cognition”
as a term for (a type of) behavior, they are thereby not talking about the same
thing as are the traditional cognitivists. Thus, they sometimes fail to come to
grips with traditional issues in cognitive science.
This paper will begin, in section 1, with a brief review of the distinction be-
tween cognition and behavior as it has formerly been used in cognitive sci-
ence, namely, that cognitive processes have been thought to be among the
many endogenous factors that contribute to the production of behavior. The
point here is not to offer definitions of “cognition” or “behavior” or to offer
much in the way of clarification of what each of these is, but simply to draw
attention to what has been a widely held understanding of the difference be-
tween the two. Successive sections (sections 2-5) will then review ways in
which Stewart 2010, Froese, Gershenson, and Rosenblueth 2013, and Hutto
and Myin 2013, seem not to appreciate the significance of their departure
from traditional problems of cognition. Section 6 will emphasize the fact that
not all enactivists are enactivistsb by providing clear examples of enactivists
who offer a more traditional conception of cognition as a species of endoge-
nous cause of behavior.

21
 The Enactivist Revolution

As a final preliminary note, it is important to bear in mind that the goal of this
paper is clarification, not criticism. Enactivismb clearly represents a dramatic
break with tradition. Enactivists have made that abundantly clear. What
bears greater attention, however, is the character of this break. Rather than
examine endogenous causes of behavior, such as cognition, enactivismb pro-
poses to focus on (a type of) behavior. What also bears attention are some of
the ramifications of this break. Insofar as enactivismb no longer addresses
cognition as it has formerly been understood, it just so far threatens to ignore
cognition. Enactivismb, thus, does not so much solve traditional problems, as
merely walks away from them. This, of course, does not bear directly on the
truth of enactivismb. It only suggests that enactivistsb need to be more careful
in how they deal with traditional problems. If they want to talk about tradi-
tional cognition, they apparently need an account of endogenous influences
on behavior. Alternatively, if they wish to break with tradition, then they
must be careful to make a cleaner break. So, to repeat, the goal of this paper
is not so much criticism of enactivismb as clarification.

1. Cognition and Behavior

One way in which to appreciate the core theoretical commitments of tradi-
tional cognitive science might be to revisit some of its founding documents,
wherein the original commitments are articulated. As an illustration of this
method, Aizawa 2014, describes a bit of common ground between B. F. Skin-
ner and Noam Chomsky, namely, both believed that cognition was a putative-
ly explanatory causal factor in the production of behavior. Where Skinner
and Chomsky differed, of course, was in their assessment of the genuine ex-
planatory value of the cognitive. (See, for example, Skinner 1957, and Chom-
sky 1959.) Another paper that illustrates the core theoretical commitments of
traditional cognitive science is the seminal 1958 paper by Alan Newell,
J. C. Shaw, and Herbert Simon, “Elements of a Theory of Human Problem Solv-
ing.” They propose that
Questions about problem-solving behavior can be answered at various levels
and in varying degrees of detail. The theory to be described here explains
problem-solving behavior in terms of what we shall call information processes.
If one considers the organism to consist of effectors, receptors, and a control
system for joining these, then this theory is mostly a theory of the control sys-
tem. It avoids most questions of sensory and motor activities. (Newell, Shaw,
and Simon 1958: 151).

This brief passage contains a number of ideas that are relevant to understand-
ing the differences between the traditional information processing approach
in cognitive science and enactivismb. These features are worth reviewing
in detail.

22
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

The very first sentence proposes to treat problem solving as a behavior. Prob-
lem solving is, thus, not itself information processing or cognitive processing.
In the literature on embodied and enactive cognition, one sometimes encoun-
ters the idea that problem solving is not behavior, but is instead cognitive pro-
cessing. This claim is more complicated than one might expect. It contains
a subtle ambiguity. There is one sense of this claim that is entirely uncontro-
versial and consistent with the Newell, Shaw, and Simon perspective. It seems
perfectly reasonable to claim that the entire process of, say, physically manip-
ulating pencil and paper to solve a cryptarithmetic problem counts as prob-
lem solving and cognitive processing. This is the sense in which the whole of
the process is cognitive processing in virtue of the fact that an important or
salient component of the process is cognitive processing. The whole of the
manipulative process is cognitive processing, even though strictly speaking
only a proper part of the process is cognitive processing. The idea here might
be understood through an analogy. The whole of the process of baking a cake,
one might say, is not strictly speaking a matter of baking a cake. The process
of baking a cake might include breaking some eggs and the mixing of ingredi-
ents, processes that are not themselves baking processes strictly speaking.
Similarly, the whole of the process of filling up one’s car is not strictly speak-
ing a matter of pumping gasoline into the tank. It includes such things as
slowing the car, pulling it into the station, and shutting off the engine. In con-
trast to this unproblematic claim there is the idea that, strictly speaking, the
whole of the process of manipulating the pencil and paper is cognitive pro-
cessing. This would be the sense in which the whole of the process of baking
a cake is literally the baking of a cake or the whole of the process of filling
one’s gas tank is pumping gasoline into the tank. What probably obscures the
ambiguity in the claim that problem solving is cognitive processing is the rela-
tive lack of clarity about the character of the component processes. There is
a relatively clear distinction between slowing the car to pull it into the gas
station and pumping the gasoline into the tank, but it is less clear how to dis-
tinguish the information processing that might take place only in the brain
and what might be called the information processing that takes place in the
pencil and paper. It is, therefore, useful in discussing such cases to be clear on
the strength of the claim that problem solving is cognitive processing. The
claim is subtler than one might have expected.
Second, the passage from Newell, Shaw, and Simon treats behavior as distinct
from cognitive or information processing. Information processing is taken to
be a mechanism realized in the brain. This flatly contradicts the enactivistb
idea that cognition is (a type of) behavior.
Third, Newell, Shaw, and Simon propose that problem-solving behavior might
be explained, in part, by appeal to information processing. On this model,
behavior is the thing to be explained, whereas information processing is
among the factors that do the explaining. They repeat this idea more emphat-

23
 The Enactivist Revolution

ically a bit later in their paper: “At this level of theorizing, an explanation of an
observed behavior of the organism is provided by a program of primitive infor-
mation processes that generates this behavior” (ibid.) This is a conception of
information processing/cognition they shared with Chomsky. Further, it is
a conception Skinner recognized as the mainstream conception, arguing,
however, that it is misguided.4 Notice that, by proposing that behavior is dis-
tinct from information processing and that information processing is realized
in the brain, Newell, Shaw, and Simon implicitly adopt what is sometimes de-
scribed as the framework of “mechanistic explanation.” They propose to ex-
plain the behavior of a whole organism primarily by appeal to the behavior of
one of its components. This picture might be illustrated with a well-known
image from Craver 2007. (See Figure 1.) In this scheme, S ψ-ing would be
something like a participant in an experiment solving a problem, whereas,
say, x3 ϕ3-ing would be the brain processing information.

Figure 1. Schema for mechanistic explanation.

Redrawn from Craver 2007: 8, Figure 1.1.

Fourth, and finally, Newell, Shaw, and Simon embrace the traditional cogni-
tive science focus on the role of information processing in the production of
behavior, but they do not deny that there can be a role for sensory and motor
activities in the production of behavior. We can describe this view by refer-
ence to the Craver schema. In the figure above, x1 φ1-ing might be the eye
performing a saccade, where x4 φ4-ing might be writing with a pencil. Thus,
they recognize that there are many component processes that conspire in the
production of behavior, but indicate that they will limit their attention to
a subset of these factors. The focus of their attention is methodological, not
theoretical. In other words, even some of the earliest advocates of infor-
mation processing psychology anticipated a day when psychologists might

4
For a contemporary articulation of this picture, there is the Stanford Encyclopedia of Philosophy
entry on cognitive science. (Thagard 2010).

24
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

take up the issues concerning the contributions of sensory and motor factors
to the production of behavior.
This traditional picture of the relationship between cognition and behavior—
that cognition is among the factors that might explain behavior—is rarely, if
ever, explicitly discussed in the enactivist literature. Nevertheless, if one
maintains that cognition is (a type of) behavior, one seems to be walking away
from much of what cognitive science has been up to, namely, the study of pu-
tative endogenous contributions to the production of behavior. Nevertheless,
enactivistsb have often seemed willing to do this, suggesting that a fresh start
for cognitive science is in order. Setting aside questions about the wisdom of
walking away from so much of the work that has been done in cognitive sci-
ence, this paper will show some of the missteps this has engendered.

2. Bourgine and Stewart 2004

Recall Maturana’s claim that “A cognitive system is a system whose organiza-
tion defines a domain of interactions in which it can act with relevance to the
maintenance of itself, and the process of cognition is the actual (inductive)
acting or behaving in this domain.” (Maturana and Varela 1980: 13). It is not
entirely clear what is going on in this brief passage. Maturana apparently
claims that the process of cognition is behaving in a domain—that cognition is
a type of behavior—but one might have one’s doubts that he genuinely pro-
poses to identify cognition and behavior. How is this conception of cognition
supposed to relate to the traditional conception, if at all? And, what are we to
make of the character of this claim? Is it supposed to be a definition, a con-
ceptual analysis, a theoretical hypothesis, or something else? It might well be
read as a stipulative definition, but then again there are times when Maturana
is prone to forceful pronouncements about empirical matters. Matters here
are not that clear.
Some of the ambiguity in Maturana’s text is eliminated from the account in
Bourgine and Stewart 2004. The latter presentation is more deliberate and
explicit about defining cognition as behavior and that this definition does not
capture what is “ordinarily” meant by “cognition.” Their proposal, therefore,
at least looks more like a stipulative definition of “cognition.” To provide their
definition, they, first, define A interactions as those system-environment in-
teractions that have consequences for the internal state of an organism and
B interactions as those system-environment interactions that have conse-
quences for an organism’s immediate environment or modify the relation of
the organism to its environment. These terms then figure into a definition
of “cognition”:

25
 The Enactivist Revolution

D-C1: A system is cognitive if and only if type A interactions serve to

trigger type B interactions in a specific way, so as to satisfy a viabi-
lity constraint.
Bourgine and Stewart explicitly decline to define what a “viability constraint”
is, but the rough idea is that A interactions must trigger B interactions that
are “good for” the system. This proposal seems to imply that systems are cog-
nitive when stimuli provoke them to behave in ways that they are “good
for” the system. And Bourgine and Stewart subsequently substantiate this
interpretation:
It may be useful to illustrate this by examples of interactions such as falling
down stairs, eating, or breathing (including the breathing of a poisonous but
odorless gas). Ordinarily, such interactions are not considered as “cognitive.”
On the definition proposed here, they will not be cognitive unless the conse-
quences for the internal state of the system are employed to trigger specific ac-
tions that promote the viability of the system. Thus, falling down stairs will be
cognitive if but only if the fall triggers reactions such as a modification of mus-
cle tone that limits the damage; and this does require specific sensory and mo-
tor organs. Similarly, eating is cognitive if but only if it triggers a reaction of
satiety that prevents damage from overeating; breathing a poisonous gas is
cognitive if but only if it triggers evasive action, which will require a specific
sensory organ to detect the poison, and the resulting sensation to trigger an
appropriate, coordinated motor response. (Bourgine and Stewart 2010: 338).

Bourgine and Stewart evidently concur with the cognitivist view that falling
down stairs, eating, and breathing are not ordinarily considered to be cogni-
tive (processes). Yet, they differ from cognitivists in their rationale for this
assessment. For Bourgine and Stewart, it is only some instances of falling
down stairs, eating, and breathing that are not to count as cognitive, namely,
those in which there are no prophylactic effects, such as changes in muscle
tone or satiety. For cognitivists, however, falling down, eating, and breathing
are not, strictly speaking, cognitive processes at all;5 they are, at most, behav-
ior. Moreover, they are likely to receive distinct behavioral analyses. By cog-
nitivist lights, many instances of eating may count as cognitive behaviors inso-
far as they require cognitive processes in order to do things such as recognize
food or to plan how to use knife and fork to obtain bite-sized pieces of food.
By contrast, instances of breathing may not count as cognitive behaviors inso-
far as they do not involve cognitive processing. Humans typically breathe
without thinking about it. Non-cognitive autonomic processes generally suf-
fice for breathing. Finally, relatively few cases of falling down stairs will
count as cognitive behaviors. Usually, gravity can do most of the work
of tumbling someone down stairs without their really thinking about it.

5
For clarification of the qualifier, “strictly speaking,” recall the first point in the discussion of the
passage from Newell, Shaw, and Simon 1958, in section 1 above.

26
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

The point here, of course, is not to pin down precise frequencies or anal-
yses of these cases, but merely to give a nod to the sorts of factors that will
enter into cognitivist analyses and to contrast them with Bourgine and Ste-
wart’s analysis.
As we have just seen, the traditionalist might well claim that Bourgine and
Stewart’s theory is subject to counterexamples in which they call things that
are not cognitive, cognitive. Falling down stairs is not a cognitive process,
even though Bourgine and Stewart’s theory says it is. A traditionalist can also
easily imagine cases in which Bourgine and Stewart’s theory would call cogni-
tive processes “non-cognitive.” So, imagine a person, Jane, looking out the
window and seeing a cloudy day and thinking, “It looks like rain.” This looks
to be what Bourgine and Stewart would call an A interaction. Jane might then
think, “Maybe I should take my umbrella. But, then again, I have a lot to carry
today. Maybe I should just chance it and leave my umbrella at home,” before
she finally walks out the door.6 Jane’s interior monologue and the walking out
the door looks to be what Bourgine and Stewart would call a B interaction: it is
a system-environment interaction that modifies the relation of Jane to her
environment. But, now, suppose that a dramatic cloudburst drenches Jane
and that this is not good for her. By Bourgine and Stewart’s enactivismb,
Jane’s interior monologue was not a cognitive process, nor was there anything
like a thought process underlying her interior monologue. This looks, to
a cognitivist, like a counterexample to Bourgine and Stewart’s theory in
which a cognitive process is mistakenly labeled “non-cognitive.”7
At this point, one might propose that Bourgine and Stewart can simply stand
by their stipulative definition of what a cognitive process is. Their theory does
not capture traditional “intuitions,” “theories,” or—one might say, “false
leads,” but that is not what it is meant to do. Bourgine and Stewart, therefore,
have a kind of theoretical “safety” in offering a stipulative definition of what

6
Benny Shannon is an enactivist (though not an enactivistb) for whom the study of such thought
sequences is crucially important. (Cf., Shannon 2010.) As will emerge, Bourgine and Stewart’s
theory of cognition appears to have the consequence that some of the thought sequences that
Shannon has studied will not be cognitive. Or, if one prefers, Bourgine and Stewart’s theory of
cognitionb appears to have the consequence that some of the thought sequences that Shannon has
studied will not be cognitiveb.
7
This sort of scenario can be used to draw attention to another feature of Bourgine and Stewart’s
account. We cannot tell just from the occurrence of the A interaction and the B interaction
whether or not a process is cognitive in Bourgine and Stewart’s sense. Whether Jane’s interior
monologue was (indicative of) a cognitive process or not apparently depends on whether or not it
actually rains and what impact this has on her. So, if it does not rain and this turns out to be good
for Jane, say by sparing her the burden of carrying an unnecessary umbrella, then we have
a cognitive process. Alternatively, if a downpour drenches Jane and this is not good for her, then
the interior monologue would not be (indicative of) a cognitive process. Bourgine and Stewart’s
theory makes a process cognitive or non-cognitive based on contingent events that take place
after the putative thought process.

27
 The Enactivist Revolution

they mean by “cognition.” The problem with this, however, is that when
Stewart tries to use the theory to address traditional problems, he thereby
misses his target.
Stewart claims that “There are two basic requirements for any paradigm in
cognitive science: it must provide a genuine resolution of the mind-body prob-
lem, and it must provide for a genuine core articulation between a multiplici-
ty of disciplines” (Stewart, 2010: 1). He then proposes a solution to the mind-
body problem:
As discussed in Bourgine and Stewart 2004, we may define a system as "cogni-
tive" if and only if it generates its actions, and the feedback sensations serve to
guide actions, in a very specific way so as to maintain its autopoiesis and
hence its very existence. With these definitions, "cognition" and "life" are fun-
damentally the same phenomena; and, in principle, the mind-matter problem
is solved. (Stewart 2010: 1-3).

Thus, Stewart has proposed to solve the mind-body problem by a form of type
identity theory: cognitive processes are biological processes (life processes).8
Then biological/life processes are, in turn, identified with physico-chemical
processes (Cf., Stewart 2010: 203). So, it looks like Stewart and his enactivismb
strike squarely at one of the central issues in the philosophy of mind.
Appearances here are deceiving. If Stewart maintains that by “cognition” he
does not mean what has traditionally been meant by cognition—if he does not
correctly characterize what has traditionally been meant by “cognition” or
“the mind,” then he is not addressing the traditional mind-body problem. The
traditional mind-body problem has not been concerned with how to relate
falling down, eating, or breathing to biological or physico-chemical processes.
Instead, the traditional mind-body problem concerns what is perhaps a clus-
ter of problems, none of which centers on behaviors.
Notice that the mind-body problem as found in Descartes’ philosophy is
a question of how there could be causal interactions between an immaterial
soul or mind and a material body. But, this is a question of how cognition, as
traditionally construed, can interact with bodily processes. It is not a question
of how cognition, construed as behavior, can interact with bodily processes.
Stewart’s “cognition” does not speak to the Cartesian version of the mind-
body problem.
Kim 2005, however, observes that the Cartesian mind-body problem is not the
contemporary mind-body problem found in the philosophy of mind and cog-

8
One might well make the case that Stewart does not have a type identity theory solution but
a functionalist solution. The difference probably does not make a difference here, since the
weakness in Stewart’s purported solution lies in his view that cognition is viable behavior. Maybe
there are other enactivistb tools for dealing with this portion of the traditional mind-body prob-
lem, but the Bourgine-Stewart theory of cognition alone will not suffice.

28
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

nitive science. Instead, the contemporary mind-body problem appears to be

a cluster of problems. One of these is a problem of mental causation. This
problem, however, is not (at least in the first instance) about how bodily
movements can be causally efficacious; it is about how internal states and
processes, such as beliefs, desires, intentions, and decisions, could be among
the causes of behavior. (Cf, Kim 2005: 7f). If Stewart is not thinking of cogni-
tive states as internal causes of behavior, then he is not grappling with the
mental causation portion of the mind-body problem. A second problem in
the cluster of mind-body problems is the problem of consciousness. By con-
sciousness, Kim has in mind qualitative features such as “the smell of the sea
in a cool morning breeze, the lambent play of sunlight on brilliant autumn
foliage, the fragrance of a field of lavender in bloom, and the vibrant, layered
soundscape projected by a string quartet” (Kim 2005: 11). David Chalmers
seems to have a similar thing in mind, though described with different exam-
ples: “the quality of deep blue, the sensation of middle C” (Chalmers 1995:
201). Insofar as Stewart’s theory of “cognition” (or mind) does not capture
what is traditionally meant, but constitutes a stipulative definition not tied to
these cases, Stewart evidently fails to address the traditional mind-body prob-
lem. We might concede, if only for the sake of being agreeable, that Bourgine
and Stewart do solve a version of the mind-body problem, namely, the version
that shows how cognition defined in D-C1 can be related to the body. But,
even with that concession, it remains true that Stewart has not resolved the
traditional mind-body problem(s). Instead, enactivismb threatens to leave
aside old issues for new issues.
So, to summarize this section, one might begin with the possibility that Bour-
gine and Stewart have proposed a definition of “cognition” that is meant to
capture what has traditionally been thought of as cognition. Two sorts of
counterexamples, however, challenge this view. 1) Falling down stairs is not
a cognitive process; it is (at best) a cognitive behavior. 2) Interior monologues
constitute, or are indicative of, cognitive processes even in cases where these
interior monologues (or the processes underlying them) lead to bad outcomes
for their possessors. In light of these considerations, it is perhaps better to
interpret Bourgine and Stewart as not merely overthrowing the theoretical
apparatus of cognitivism with its computational rules and representations.
Instead, they also propose to overthrow the putative exemplars of cognition.
Cognition is not one among many possible endogenous causal contributors to
behavior; it is, instead, viable behavior. If, however, this is what Stewart is up
to, then he appears to be walking away from many of the problems that cogni-
tivism faced. For example, Stewart’s identification of cognition and life does
not solve the traditional mind-body problem. It solves, at best, a novel version
of the mind-body problem.

29
 The Enactivist Revolution

3. Froese, Gershenson, and Rosenblueth 2013

The hypothesis of extended cognition, in at least some of its earliest articula-
tions, maintains that, while brains realize some cognitive processes, some-
times, under certain conditions, larger configurations of brain, body, and
world also realize cognitive processes. This is the conception that was implicit
in parts of Andy Clark and David Chalmers’ seminal paper. (See Clark and
Chalmers 1998.) It was the conception in play in the two “cognitive equiva-
lence” arguments based on the play of the video game Tetris and the Inga-Otto
thought experiment. Recall the description of three modes of Tetris play:

(1) A person sits in front of a computer screen which displays images of vari-
ous two-dimensional geometric shapes and is asked to answer questions con-
cerning the potential fit of such shapes into depicted "sockets". To assess fit,
the person must mentally rotate the shapes to align them with the sockets.
(2) A person sits in front of a similar computer screen, but this time can choose
either to physically rotate the image on the screen, by pressing a rotate button,
or to mentally rotate the image as before. We can also suppose, not unrealisti-
cally, that some speed advantage accrues to the physical rotation operation.
(3) Sometime in the cyberpunk future, a person sits in front of a similar com-
puter screen. This agent, however, has the benefit of a neural implant which
can perform the rotation operation as fast as the computer in the previous ex-
ample. The agent must still choose which internal resource to use (the implant
or the good old fashioned mental rotation), as each resource makes different
demands on attention and other concurrent brain activity. (Clark and
Chalmers 1998: 7).

Case (1) looks to be a case of someone playing the video game using old-
fashioned, brain-internal information processing. By contrast, (2) and (3) are
supposed to be cognitively the same as (1) with the only difference between
them being the material substrates that realize cognition. Next recall the
Inga-Otto thought experiment. Inga has normal human memory, reads that
there is an interesting exhibit at the Museum of Modern Art, and decides that
she would like to see it. She thinks for a moment, recalls that it is on 53rd St.,
then heads on her way. By contrast, Otto is suffering from the early stages of
Alzheimer’s Disease, so he has developed a system for maintaining infor-
mation in a notebook. In this notebook, he has the address of the Museum of
Modern Art. When he reads of the exhibit at the museum, he decides that he
would like to see it. He then picks up his notebook, flips through it until he
finds the address, then heads on his way. Clark and Chalmers claim that “in
relevant respects the cases are entirely analogous: the notebook plays for Otto
the same role that memory plays for Inga. The information in the notebook
functions just like the information constituting an ordinary non-occurrent
belief; it just happens that this information lies beyond the skin” (Clark and
Chalmers 1998: 13).

30
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

Whereas Clark and Chalmers sometimes offer “cognitive equivalence argu-

ments” for extended cognition, Froese, Gershenson, and Rosenblueth 2013,
propose a different path. They reject the idea that there is ever any brain-
bound, information processing type cognition of the sort that Inga was said to
have. Instead, they adopt the enactivistb view that cognition is (a type of) be-
havior: “cognition is primarily conceived of as a form of viable conduct by an
agent in an environment” (Froese, Gershenson, and Rosenblueth 2013: 1).
Thus, they take it that “[the Dynamical Systems Hypothesis] takes the notion of
an extended mind as its starting point, rather than as a curious exception”
(ibid.) Froese, Gershenson, and Rosenblueth, therefore, defend extended
“cognition” in the sense of extended adaptive behavior. This path, however,
does nothing to help the version of extended cognition in the Tetris and Inga-
Otto examples. Through those examples, Clark and Chalmers maintained the
bold conclusion that what was once thought to have been realized only in the
brain—a type of information processing—is, in fact, also sometimes realized
in the brain, body, and world. Thus, Froese, Gershenson, and Rosenblueth, do
not so much defend the hypothesis of extended cognition as leave it in favor of
the hypothesis of extended viable conduct.9
Suppose, then, for the sake of argument that we follow Froese, Gershenson,
and Rosenblueth and think about viable conduct. If so, then the further step
to the view that such “cognition” is extended is trivial. “Conduct” is essentially
another word for behavior and behavior typically is realized by brain, body,
and world. In cases in which hammering in a nail is a viable behavior, it is
apparently realized by cognitive, attentional, and motivational processes in
the brain, along with the propagation of light in the eye, along with muscular
processes in the arms, and contact between the hammer and a nail. Who
would doubt that? Why would anyone doubt that? So, where the hypothesis
of extended cognition is a controversial hypothesis, the hypothesis of extend-
ed viable conduct appears to be widely if not universally accepted. What
makes the latter appear interesting is the enactivistb terminology of using
“cognition” as a term for viable conduct, where most cognitive scientists think
of “cognition” as a term for cognition.
Froese, Gershenson, and Rosenblueth appear to appreciate that they cannot
rely on the view that cognition is a form of viable conduct in order to argue
for the extended cognition hypothesis. Therefore, they argue that “even if we
accept [the] internalist starting point, a proper understanding of neuronal
activity will force us to accept an extended view of the mind nonetheless.”
(Froese, Gershenson, and Rosenblueth 2013: 2). Given this concession, they
propose two distinct arguments for extended cognition. There is an extensive

9
If one prefers, one might say that Clark and Chalmers defend something more like extended
cognitionc, where Froese, Gershenson, and Rosenblueth defend extended cognitionb. Again, the
body of the text uses more standard terminology.

31
 The Enactivist Revolution

discussion of an evolutionary robotics simulation, but the details of that are

irrelevant to their arguments, so will not be reviewed.
Consider their two arguments in reverse order of their appearance. Their
second argument is relatively simple:
6. The non-isolated [Continuous-Time Recurrent Neural Network]’s output is
determined by its input, albeit mediated by its internal activity, while this in-
put is determined by its motor output, albeit mediated by bodily and environ-
mental (including social) activity.
7. It logically follows from the above that the non-isolated CTRNN’s additional
neural complexity is partially constituted by its own sensorimotor and social
10
coupling. (Froese, Gershenson, and Rosenblueth 2013: 6).

This is a relatively simple version of the infamous “coupling-constitution fal-

lacy”.11 Froese, Gershenson, and Rosenblueth’s contention notwithstanding,
the second sentence above does not logically follow from the first. In the most
simplistic of coupling arguments, one might observe that a cognitive process X
is causally influenced by a prima facie non-cognitive process Y, then infer that
the appearances of these processes are misleading. Instead, the entire Y-X
process is a cognitive process. The Froese, Gershenson, and Rosenblueth ex-
ample, however, is a bit more complicated, since there are supposed to be
reciprocal causal connections wherein Y causally influences X and X causally
influences Y.
Reciprocal causal connections, however, do not suffice to close the gap be-
tween causation and constitution. Suppose we accept the internalist view that
the non-isolated CTRNN has some “neural complexity.” The premise in 6
notes that the CTRNN output is causally determined by its input and its input
is causally determined by its output. This is a causal premise. The conclusion,
however, is that the “additional neural complexity” is constituted by the pro-
cesses outside of the CTRNN. But, why go beyond thinking that the additional
neural complexity is merely causally influenced by processes outside the
CTRNN? There seems to be no warrant for the conclusion that the additional
neural complexity is constituted by external processes.12 Note that, in a nor-

10
Technically, the argument seems to be for extended “neural complexity,” whatever that is. To
make this relevant to the hypothesis of extended cognition, there would need to be some link
between “neural complexity” and cognition and it is unclear what Froese, Gershenson, and Ros-
enblueth propose this link to be. They think that cognition is viable conduct, not neural complexi-
ty. And, traditional cognitivism takes cognition to be something like rule-governed, symbol ma-
nipulation or information processing, not “neural complexity.” For present purposes, however,
these idiosyncrasies can be set aside.
11
Cf., e.g., Block 2005, Adams & Aizawa 2008, Rupert 2009.
12
Froese, Gershenson, and Rosenblueth seem to think that it matters whether the environment
qualitatively changes the CTRNN implementing an artificial nervous system (ANS):

32
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

mally functional HVAC system, the thermostat’s output is determined by its

input, albeit mediated by its internal activity, while its input is determined by
its output, albeit mediated by environmental activity, such as the burning of
natural gas in the furnace. Nevertheless, the change of shape of the bimetallic
strip in the thermostat is still limited to the bimetallic strip.
Froese, Gershenson, and Rosenblueth’s second argument relies on a distin-
ction between the properties of parts and the properties of wholes and the
idea that these properties are often qualitatively distinct. This is an idea that
is quite familiar from the literature on mechanistic explanation and is illus-
trated in Figure 1, in section 1, above. The idea is that entities do things in
virtue of their parts doing qualitatively distinct things. The idea also appears
in the enactivism literature in the introduction to Hutchins 2010:
Distributed cognition is a framework for exploring the cognitive implications
of the commonsense observation that in systems characterized by multiple
levels of interacting elements, different properties may emerge at different
levels of organization. Thus, a colony of social insects has different properties
than any individual insect in the colony. At the level of organisms, bodies have
different properties than organs, which have different properties than cells. In
the realm of cognition, a neural circuit has different properties than any of the
neurons in the circuit. The same can be said of a brain area with respect to the
neural circuits that compose it, or of an entire brain with respect to the areas
that interact within the brain. This is also true of the body/brain system with
respect to either brain or body, and the world/body/brain system with respect
to any of its parts. A system composed of a person in interaction with a cogni-
tive artifact has different cognitive properties than those of the person alone.
(Hutchins 2010: 425.)

Froese, Gershenson, and Rosenblueth, propose to avoid the coupling-

constitution fallacy by appeal to something like this picture:
This critique is known as the “coupling-constitution fallacy”. In this paper we
respond to this reductionist challenge by using an evolutionary robotics ap-

our aim is to build a model of an embodied agent, whose artificial nervous system (ANS) has
mathematical properties that are in principle impossible for it to have in isolation. The moti-
vation for this criterion is the need to go beyond a demonstration of how an agent’s situated-
ness within a sensorimotor loop modulates the internal activity of the ANS, but can transform
the ANS into a qualitatively different kind of system altogether.

if an ANS with less than 3D is nonlinearly coupled with other non-chaotic systems, and its in-
ternal neural activity spontaneously becomes chaotic, then an explanation of this property as
resulting from an extended process of interaction cannot be accused of committing the cou-
pling-constitution fallacy. (Froese, Gershenson, and Rosenblueth 2013: 2).

So, their idea is that when this ANS interacts with the environment, it undergoes a qualitative
shift. It becomes chaotic. But, what does this have to do with cognition? It isn’t that becoming
chaotic is the same as becoming cognitive, is it? And, what does this have to do with what does, or
does not, constitute a cognitive process? Why not stick with the idea that interaction with the
environment causes the ANS to become chaotic?

33
 The Enactivist Revolution

proach to create a minimal model of two acoustically coupled agents. We

demonstrate how the interaction process as a whole has properties that can-
not be reduced to the contributions of the isolated agents. We also show that
the neural dynamics of the coupled agents has formal properties that are in-
herently impossible for those neural networks in isolation. (Froese, Gershen-
son, and Rosenblueth 2013: 1).
From the perspective of the DSH, which proposes a distributed view of cogni-
tion as the default mode of cognition, there is no coupling-constitution fallacy
because properties of the sensorimotor interaction process cannot be reduced
to that of the isolated components. (Froese, Gershenson, and Rosenblueth
2013: 6).
It is a good strategy for the EMH to highlight that ongoing interaction between
a cognitive agent and environment results in a novel, mutually encompassing
process with new properties of its own. (Ibid.)

With this picture, the work-around for the coupling-constitution fallacy is

easy. There are brain properties, bodily properties, environmental properties,
and brain-body-environment properties. In other words, there are properties
the brain has, properties the body has, properties the environment has, and
properties a brain-body-environment system has. One does not need to say
that it is interaction between brain, body, and world that converts bodily
properties and environmental properties into new properties. Instead,
a causal interaction between brainy, bodily, and environmental processes
gives rise to a new, qualitatively distinct, emergent property that is cognitive.
Problem averted.
Notice that, for this argument, one does not really need a whole lot of evolu-
tionary robotics simulation. It depends primarily on the picture of mechanis-
tic explanation according to which higher level properties are realized by
properties of lower level individuals. All of that is fine. There is, however,
one tacit premise that is the downfall of this reply to the coupling-constitution
fallacy. Recall that Froese, Gershenson, and Rosenblueth proposed to begin
with the internalist assumption that cognitive processes are realized by the
brain. Recall that they wanted to show that “even if we accept [the] internalist
starting point, a proper understanding of neuronal activity will force us to
accept an extended view of the mind nonetheless.” (Froese, Gershenson, and
Rosenblueth 2013: 2). But, if they concede that assumption, namely, that the
brain properties are cognitive, then given their other premise that the proper-
ties of the brain-body-world system are qualitatively distinct from the proper-
ties of the brain, the body, and the world, then this gives us the beginnings of
an argument that the properties of the brain-body-world system are not cogni-
tive! The argument potentially backfires.
Of course, Froese, Gershenson, and Rosenblueth may be too quick to concede
that brain properties are cognitive. Perhaps they should simply stick to their
view that cognitive processes are instances of viable conduct. If they do that,

34
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

however, then there really is no need for the coupling kinds of arguments at
all. Clearly viable conduct is extended. As noted above, no one doubts that
when hammering in a nail is a viable behavior, it is probably realized by cog-
nitive, attentional, and motivational processes in the brain, along with the
propagation of light in the eye, along with muscular processes in the arms,
and physical processes in the hammer, nail, and wood. Once you have the
view that cognition is a type of behavior, it is relatively smooth sailing to the
conclusion that such “cognition” is extended.
So, the upshot of our discussion of Froese, Gershenson, and Rosenblueth’s
paper is three-fold. If they wish to defend the view that viable conduct is ex-
tended, then they are doing nothing to support one of the original versions of
the idea that cognition is extended. They are not at all supporting the idea
that some sort of information processing style of cognition is extended. Sec-
ond, if they wish to defend the view that viable conduct is extended, then they
are not defending a view that it seems anyone has ever doubted. The stand-
ard view in cognitive science is that conduct or behavior is extended. Third,
the appeal to the framework of mechanistic explanation does nothing to avoid
the problems of the coupling-constitution fallacy, unless one begins with the
assumption that cognition is a property of a brain-body-world system. But, if
one has that assumption, there is no need for further argumentation using the
framework of mechanistic explanation. Properties of a brain-body-world sys-
tem are clearly extended. What this suggests is that, by taking “cognition” to
be a term for viable conduct, rather than for some endogenous causal con-
tributor to the production of conduct, Froese, Gershenson, and Rosenblueth
have marginalized their view from the concerns of extended cognition and
the coupling-constitution fallacy.

4. Hutto and Myin 2013

From the earliest pages of their Radicalizing Enactivism, Daniel Hutto and Erik
Myin challenge the view that all cognition requires representations. Their
alternative is Radically Enactive (or Embodied) Cognition (REC):
The most radical versions of these approaches are marked by their uncom-
promising and thoroughgoing rejection of intellectualism about the basic na-
ture of mind, abandoning the idea that all mentality involves or implies con-
tent. Call this—the view we defend—Radically Enactive (or Embodied) Cogni-
tion—REC for short. (Hutto and Myin 2013: 1)
Embodied ways of thinking reject the familiar explanatory framework of or-
thodox cognitive science in favor of alternative platforms. Adherents of such
views deny that the best way to explain cognition is to posit the construction of
internal representational models (ibid.: 2)

35
 The Enactivist Revolution

Defenders of REC argue that the usual suspects—representation and computa-

tion—are not definitive of, and do not form the basis of, all mentality. (ibid.: 3).

Notice in these claims they write of rejecting “the familiar explanatory

framework of cognitive science” and “the usual suspects—representation and
computation.” The familiar view, however, is that all cognition, understood as
one of the internal, brain-realized causes of behavior, involves representation.
The tradition does not, however, maintain that (adaptive) behavior involves
representation. Plants, for example, might display adaptive behaviors, such
as phototropism, without deploying representations to do this. Such cases
would be cases in which plants produce behaviors that are not cognitive be-
haviors. They are not behaviors that are produced, in part, through cogni-
tive processes.
By contrast, when Hutto and Myin stake out their own view about cognition,
they apparently have a different conception of cognition. They use “basic
cognition” as a phrase to describe what sounds like behavior: “We restrict our
ambitions to promoting REC, calling upon strong versions of two theses. We
dub these the Embodiment Thesis and the Developmental-Explanatory Thesis.
The former equates basic cognition with concrete spatio-temporally extended
patterns of dynamic interaction between organisms and their environments.”
(Hutton and Myin 2013: 5). For Hutto and Myin, “basic cognition” is concrete
spatio-temporally extended patterns of dynamic interaction between organ-
isms and their environments. This is a description of what traditional cogni-
tive science would call “behavior.” They also seem to use “cognition” as a term
for behavior. In the early pages of their book, they write that “proponents of
Enactive and Embodied ways of thinking reject the familiar explanatory
framework of orthodox cognitive science in favor of alternative platforms.
Adherents of such views deny that the best way to explain cognition is to posit
the construction of internal representational models built on the basis of re-
trieved informational content.” (Hutto and Myin: 2). It is somewhat odd to say
that traditional approaches explain cognition by positing the construction of
internal representational models. As we saw with Newell, Shaw, and Simon,
the traditional approach is to explain behavior by positing representations,
among other things. This oddity disappears, however, if we understand them
to use “cognition” as a word for behavior. Second, they claim that “Enactivism
is inspired by the insight that the embedded and embodied activity of living
beings provides the right model for understanding minds.” (ibid.: 4). Embed-
ded and embodied activity, which sounds like another description of behav-
ior, probably would be a very good model for the mind, if the mental were
the behavioral.
The stage is now set for traditional cognitive science to go about its business of
studying cognition that purportedly must involve representations, where en-
activismb goes about its business of studying behavior which need not involve

36
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

representations. Nevertheless, things do not go so smoothly. Hutto and

Myin’s use of the phrase “basic cognition” muddies what might otherwise be
a very simple argument. Hutto and Myin want to challenge the standard view
that all cognition requires representation and one way they wish to do this is
by providing counterexamples. As a point of logic, they evidently want a case,
or two, in which there is cognition without representation. But, instead of
cases in which there is cognition without representation, they only provide
cases of their “basic cognition” without representation. That is, they only pro-
vide cases of behavior without representation.
Hutto and Myin’s two putative counterexamples are Rodney Brooks’ behavior-
based robots and Barbara Webb’s models of cricket phonotaxis. (See Brooks
1997, Webb 1994, 1996.) Even as Hutto and Myin describe the example,
Brooks’ robots show only cases of behavior that does not require representa-
tion: “Brooks’ first-generation behavior-based robots, and those that followed,
succeed precisely because their behaviors are directly guided by continuous,
temporally extended interactions with aspects of their environments, rather
than being based on represented internal knowledge about those domains
(knowledge that would presumably be stored somewhere in the robots’ in-
nards).” (ibid.: 42). Such an analysis is irrelevant to mainstream cognitive
science, since it only shows that behavior does not require representation, not
that cognition does not require representation. Thinking of “cognition” and
“behavior” as referring to the same thing seems to mislead Hutto and Myin.
The same oversight appears in their analysis of Webb’s models of crickets. By
their own analysis, “Webb’s (1994, 1996) work on cricket phonotaxis offers
a vivid example of a model of how successful navigation takes place in the
wild, apparently without the need for representations or their manipulation.
… In other words, the capacity of these animals to adjust their behavior when
successfully locating mates requires them to engage in a continuous interac-
tive process of engagement with the environment.” (ibid.: 43). So, by Hutto
and Myin’s own analysis, these are only cases of behavior without representa-
tion; not cases of cognition without representation.13
What the foregoing suggests is that enactivistb terminology confuses even Hut-
to and Myin. It complicates what should be a simple exercise is trying to gen-
erate a counterexample to the claim that all cognition requires representation.
Rather than offering examples of cognition that do not involve representation,
they only offer examples of behavior that do not involve representation. This,
of course, only shows that one argument for REC has not worked. It does not
show that the argument cannot be fixed, as by revisiting the examples and

13
AUTHOR shows how Chemero 2009, makes much the same sort of mistake in thinking that
models in Beer 2003, and van Rooij, Bongers, and Haselager 2002, “show how radical embodied
cognitive science can explain cognition as the unfolding of a brain-body-environment system, and
not as mental gymnastics” (Chemero 2009: 43).

37
 The Enactivist Revolution

showing how they have internal mechanisms that are plausibly cognitive, but
which nevertheless contain no representations. Nor does it show that there
are no other arguments for REC that might work.14 Much less does it show
that REC is false. As billed at the outset, the goal here is to clarify some of mis-
steps in the enactivistb revolution. The point of clarification here is that Hutto
and Myin’s formulation of an argument based on the Brooks and Webb mod-
els do not work as billed.

5. Not all Enactivists are Enactivistsb

The goal of this paper has not been to critique the whole of the enactivist pro-
gram. As is often noted, enactivism is still in its formative stages and many
ideas remain to be worked out on many different topics. Some authors asso-
ciated with enactivism do not seem to be concerned with cognition at all. So,
for example, Barbaras 2010, which was included in Stewart, Gapenne, and Di
Paolo’s Enaction anthology, wrote about life and metabolism and barely men-
tioned cognition. Moreover, as is to be emphasized now, not all enactivists are
enactivistsb. One can say this, but its force might be better appreciated if we
describe enactivists who are not enactivistsb.
Perhaps one of the more significant examples is that, some years ago, Stewart
seems not to have been an enactivistb.15 In the abstract to a 1996 paper, Stew-
art writes, “In contemporary cognitive science, there are two distinct para-
digms with contrasting definitions of cognition. The computational theory of
mind is based on the syntaxical manipulation of symbolic representations;
this paradigm is objectivist because the postulate of a unique independent
reality is necessary as a referential basis for semantic grounding of the sym-
bols. The alternative ‘constructivist’ paradigm is based on the biological meta-
phor ‘cognition = life’ and programmatically follows the evolution of cognition
from bacteria to civilized humans; it is non-objectivist.” (Stewart 1996: 311).
Rather than articulating two distinct accounts of what cognition is, however,
Stewart appears to be thinking of something along the lines of what philoso-
phers might interpret as a distinction between a form of realism and a form of
anti-realism: “In an objectivist [computationalist] perspective, cognition is the
subjective representation of an ontologically independent objective reality. In
a constructivist perspective, based on the biological metaphor ‘cognition =
life’, the clear point of contrast is that the subject and the object of knowledge

14
In fact, Hutto and Myin spend a lot of time arguing that causal and informational approaches to
naturalizing content have failed, hence that we therefore have some reason to think we should
abandon the hypothesis that all cognition requires representation. This argument is untouched
by the foregoing.
15
Shannon 2010, also takes an enactivist, but not an enactivistb position. Limitations of space
preclude a discussion of this.

38
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

are not independent but are mutually constitutive” (Stewart 1996: 316.)16
Stewart’s “objectivism” and “constructivism” seem to be versions of realism
and anti-realism.
In fact, it appears that at two points in this earlier paper Stewart understands
cognition in a more traditional way, not as behavior, but as an internal mech-
anism that contributes to the production of behavior. As the first point, there
is his commentary regarding what appear to be (cognitive) mechanisms by
which animals can cancel out the effects of bodily movements in order to per-
ceive organism independent objects.
Animals with a central nervous system have the capacity to distinguish within
their own cognitive repertoire between modifications of their sensory input
which are the immediate consequence of their own actions, and modifications
which are not so caused. For example, when an animal moves its eyes, the ret-
inal image (and hence the stimulation of the retinal cones) is modified, but
a mammal does not usually confuse this movement with the movement of an
object in its environment. The construction of perceptual invariants on the ba-
sis of motor-sensory correlations of this sort is thus at the basis of the emer-
gence of a ‘stable external world’ populated by ‘objects’ which exist as such in
the cognitive repertoire of the organism itself. … Bacteria (or trees), for exam-
ple, are quite incapable of cognitive feats of this sort. (Stewart 1996: 320).

Here it is at least possible that Stewart conceives of there being cognitive

mechanisms that enable (some) animals, but not bacteria and trees, to avoid
confusing the effects of self-movement with the effects of object movement.
In fact, these different mechanisms might be just the kinds of differences be-
tween (some) animals, on the one hand, and trees and bacteria, on the other,
that cognitivists would contend mark the difference between cognitive and
non-cognitive processes. As a second point, Stewart notes that

16
It is indicative of the cross-currents in enactivism that Hutchins 2010, was part of the Stewart,
Gapenne, and Di Paolo Enaction anthology, but also apparently endorses the kind of computa-
tional theory of mind that Stewart would reject as “objectivist”. Hutchins describes computation-
al transformations on representations of what appear to be navigator-independent features of
the world:

Two successive positions of a ship are plotted on a three-minute interval. Suppose the distance
between them is 1500 yards. The navigator computes ship's speed to be 15 knots by doing the
following: "The distance between the fix positions on the chart is spanned with the dividers
and transferred to the yard scale. There, with one tip of the divider on 0, the other falls on the
scale at a tick mark labeled 1500. The representation in which the answer is obvious is simply
one in which the navigator looks at the yard-scale label and ignores the two trailing zeros"
(Hutchins 1995…, 151-152). In this analysis, high-level cognitive functions were seen to be real-
ized in the transformation and propagation of representational states. The span between the
fix positions on the chart is a representational state that is transformed into a span on the di-
viders. This representational state is then transformed into a span on the yard scale. Finally,
the span on the yard scale is transformed into the answer by reading the label on the desig-
nated tick mark in a particular way. (Hutchins 2010: 427).

Hutchins, thus, also appears to be an enactivist, but not an enactivistb.

39
 The Enactivist Revolution

Armed with representations of this sort, an organism can set itself a ‘goal’ (ex-
pressed in terms of a desired perceptual configuration), and then by purely
mental activity (without having to take the risks involved in proceeding by tri-
al and error by actually acting in the world) elaborate a sequence of actions
which, according to these representations, can be expected to achieve that goal
(Stewart 2010: 320).

Here it looks likes there is a “purely mental activity” independent of actually

acting in the world, i.e., independent of physical behavior in the world. So,
the discussion in Stewart, 1996, suggests that one can be an enactivist without
being an enactivistb.

6. Conclusion
At the heart of this paper is the observation that some enactivists do not mean
by “cognition” what traditionalists have meant by “cognition.” There are, if
you will, two concepts of cognition in play, a traditional concept and an enac-
tivist concept. This observation would seem to be entirely unproblematic.
Moreover, it would seem to be entirely unproblematic to note that some enac-
tivists use “cognition” to describe (a kind of) behavior. These enactivists
maintain that cognition is (viable) behavior. These enactivists are enactiv-
istsb. This choice of terminology—or this way of theorizing, if you will—
however, looks to be misleading. Moreover, it is not misleading just for tradi-
tionalists. It is misleading for even some enactivistsb. By adopting a new con-
ception of cognition—by thinking of behavior and cognition as the same
thing—enactivistsb sometimes overlook ways in which they have detached
themselves from the traditions of cognitive science. Enactivists are generally
happy to break with these traditions, but there are also times when this break
is not as complete as it should be. There are times when they try to engage
with mainstream cognitive science, but are hampered by the steps they have
already taken to break with tradition. One cannot solve the traditional mind-
body problem, if one is not dealing with (something near enough to) the tradi-
tional conception of the mind. One might dissolve the problem or abandon
the problem, if one rejects the traditional concept, but one cannot solve it.
One cannot argue that cognition is embodied and extended, by observing that
behavior is embodied or extended. And, one cannot show that not all cogni-
tion involves representation by providing instances of behavior that do not
involve representation. None of these observations undermines the enactiv-
istb approach, much less any other enactivist approaches. They merely draw
attention to some missteps in the evolution of enactivismb. Perhaps the safest
route for enactivistsb is simply to make a clean break with traditional views.
Perhaps enactivistsb should walk away from traditional views and leave them
to their own devices.

40
 AVANT Vol. V, No. 2/2014 www.avant.edu.pl/en

References
Adams, F., and Aizawa, K. 2008. The Bounds of Cognition. Malden. MA: Wiley-Blackwell.

Aizawa, K. 2014. What is this cognition that is supposed to be embodied? Philosophical

Psychology. DOI:10.1080/09515089.2013.875280.
Barbaras, R. 2010. Life and exteriority: The problem of metabolism. E. Di Paolo,
O. Gapenne and J. Stewart, eds. Enaction: Toward a new paradigm for cognitive
science: 89-122). Cambridge, MA: MIT Press.

Block, N. 2005. Review of Alva Noë, Action in Perception. Journal of Philosophy, 102(5):
259-272.
Bourgine, P., and Stewart, J. 2004. Autopoiesis and cognition. Artificial life, 10(3), 327-
345.
Brooks, R. A. 1997. Intelligence without representation. J. Haugeland, ed. Mind design
II: 395-420. Cambridge, MA: MIT Press.
Calvo, P., and Keijzer, F. 2008. Cognition in plants. F. Baluska, ed. Plant-Environment
Interactions: From Sensory Plant Biology to Active Plant Behavior : 247-266. Ber-
lin: Springer Verlag.
Chalmers, D. 1995. Facing up to the problem of consciousness. Journal of Consciousness
Studies, 2(3): 200-219.
Chemero, A. 2009. Radical Embodied Cognitive Science. Cambridge, MA: MIT Press.
Chomsky, N. 1959. A review of Verbal Behavior. Language, 35(1): 26-58.
Clark, A. 2010. Coupling, constitution and the cognitive kind: A reply to Adams and
Aizawa. R. Menary, ed. The extended mind: 81-99. Cambridge, MA: MIT Press.

Clark, A., and Chalmers, D. 1998. The extended mind. Analysis, 58(1): 7-19.
Craver, C. 2007. Explaining the Brain. Oxford: Oxford University Press.
Froese, T., Gershenson, C., and Rosenblueth, D. 2013. The Dynamically Extended Mind--
A Minimal Modeling Case Study. arXiv preprint arXiv:1305.1958.
Di Paolo, E., Rohde, M., and De Jaegher, H. 2010. Horizons for the enactive mind: Val-
ues, social interaction, and play. E. Di Paolo, O. Gapenne and J. Stewart, eds. En-
action: Towards a New Paradigm for Cognitive Science: 33-87.
Haugeland, J. 1998. Mind embodied and embedded. Having Thought: 207-237. Cam-
bridge, MA: Harvard University Press.
Hutchins, E. 2010. Enaction, imagination, and insight. J. Stewart, O. Gapenne and E. Di
Paolo, eds. Enaction: Towards a New Paradigm for Cognitive Science: 425-450.
Cambridge, MA: MIT Press.
Hutchins, E. 1995. Cognition in the Wild. Cambridge, MA: MIT Press.
Hutto, D. D., and Myin, E. 2012. Radicalising Enactivism. Cambridge, MA: The MIT
Press.

41
 The Enactivist Revolution

Kim, J. 2005. Physicalism, or Something Near Enough. Princeton, NJ: Princeton Universi-
ty Press.

Maturana, H. R., and Varela, F. J. 1980. Autopoiesis and Cognition: The Realization of the
Living. Dordrecht: Springer.
Newell, A., Shaw, J. C., and Simon, H. A. 1958. Elements of a theory of human problem
solving. Psychological Review, 65(3): 151-66.
Noë, A. 2004. Action in Perception. Cambridge, MA: MIT Press.

Rupert, R. 2009. Cognitive Systems and the Extended Mind. New York: Oxford University
Press.
Shannon, B. 2010. Toward a phenomenological psychology of the conscious. J. Stewart,
O. Gapenne and E. Di Paolo, eds. Enaction: Toward a New Paradigm for Cognitive
Science: 387-424. Cambridge, MA: MIT Press.

Skinner, B.F. 1957. Verbal Behavior. New York, NY: Appleton-Century-Crofts.

Stewart, J. 1996. Cognition = life: Implications for higher-level cognition. Behavioural
Processes, 35(1): 311-326.
Stewart, J. 2010. Foundational issues in enaction as a paradigm for cognitive science:
From the origin of life to consciousness and writing. J. Stewart, O. Gapenne and
E. Di Paolo, eds. Enaction: Towards a New Paradigm for Cognitive Science: 1-31.
Cambridge, MA: MIT Press.
Stewart, J, Gapenne, O, & Di Paolo, EA. 2010. Introduction. J. Stewart, O. Gapenne
& E. Di Paolo, eds. Enaction: Toward a New Paradigm for Cognitive Science: vii-
xvii.

Thagard: 2010. Cognitive science. Retrieved November 24, 2013, 2013, from
http://www.science.uva.nl/~seop/entries/cognitive-science/
Varela, F., Thompson, E., and Rosch, E. 1991. The Embodied Mind. Cambridge, MA: MIT
Press.
Webb, B. 1994. Robotic experiments in cricket phonotaxis. D. Cliff: Husbands, M. J. A.
and S. W. Wilson, eds. From Animals to Animats 3: Proceedings of the Third In-
ternational Conference on Simulation of Adaptive Behavior: 45-54. Cambridge,
MA: MIT Press.
Webb, B. 1996. A cricket robot. Scientific American, 275(6): 94-99.

42