CEPA eprint 3949

What is this cognition that is supposed to be embodied?

Aizawa K. (2015) What is this cognition that is supposed to be embodied? Philosophical Psychology 28(6): 755–775. Available at http://cepa.info/3949
Table of Contents
1. Introduction
2. What is “Embodiment”?
3. Cognition as a Putative Cause of Behavior
4. Maturana on Cognition and Behavior
5. Chemero on Cognition and Behavior
6. Dennett on Cognition and Behavior
6.1. Of Human Intelligence
6.2. Dennett’s Argument for Embodied Cognition
7. Cognition Operationalized
7.1. Clark: Operationalist and Non-Operationalist
7.2. Haugeland: Operationalist and Non-Operationalist
7.3. Rowlands: Operationalist and Non-Operationalist
8. Conclusion
Many cognitive scientists have recently championed the thesis that cognition is embodied. In principle, explicating this thesis should be relatively simple. There are, essentially, only two concepts involved: cognition and embodiment. After articulating what will here be meant by ‘embodiment’, this paper will draw attention to cases in which some advocates of embodied cognition apparently do not mean by ‘cognition’ what has typically been meant by ‘cognition. Some advocates apparently mean to use ‘cognition’ not as a term for one, among many, causes of behavior, but for what has more often been called “behavior.” Some consequences for this proposal are considered.
Key words: Behavior, Chemero, Chomsky, Clark, Cognition, Dennett, Embodied Cognition, Haugeland, Maturana, Skinner.
1. Introduction
Many cognitive scientists have recently argued that cognition is embodied. In principle, explicating this thesis should be relatively simple. There are, evidently, only two concepts involved: cognition and embodiment. Despite the modesty of this expository burden, there remains a striking ambiguity on these topics. Many of the ambiguities concerning ‘embodiment’ and ‘cognition’ are likely to be familiar to those working in the area, hence not surprising at all. Nevertheless, there is one bold proposal that has surfaced with considerable regularity that has essentially passed without comment in the literature. This is the proposal that we understand cognition as a type of behavior.[Note 1] What is surprising about this proposal is that cognition has widely been supposed to be different from behavior. It has typically been assumed to be among the causes of behavior.
The proposal merits attention for three reasons. First, it is sometimes unclear what advocates of embodied cognition take to be the relationship between cognition and behavior. Insofar as there is confusion about this relationship, it will be helpful to dispel it. Insofar as the relationship is only implicit, it will be helpful to make it explicit. Second, insofar as one really does mean to identify cognition with a type of behavior, we will be in a better position to appreciate what one means by the phrase ‘cognition is embodied. Perhaps there is less to this surprising claim than one might expect. Third, insofar as one does not really wish to identify cognition with a type of behavior, one can then be clear about that. Moreover, there will be an additional constraint on what one takes cognition to be. If cognition is not a type of behavior, but, say, a cause of behavior, then this limits the sort of thing that cognition might be. If cognition is a cause of behavior, one can better appreciate why it might be something realized in the brain alone. If cognitive processes were among the causes of behavior, then they would be among the many other causal factors occurring within an organism – processes such as digestion and respiration – that manifest themselves in the behavior of the organism.
It should be emphasized that this examination of the proposal that cognition is a type of behavior is not intended to discredit research on embodied cognition. The scope and diversity of hypotheses that go under the heading of “embodied cognition” is far too great for any so simplistic an attempt at a “refutation.” Further, although there will be some indications of the negative consequences of the identification or conflation of cognition and behavior, the principal goal here is not even critical. Instead, the principal point is to get the proposal that cognition is behavior out in the open for critical examination. Thus, rather than offering a challenge to embodied cognition or even one form of embodied cognition, the overarching goal is clarity about what cognition is hypothesized to be in the context of embodied cognition.
The paper will begin with a brief exposition of one sense of ‘embodiment’ that is to be found in the literature. This exposition, in section 2, is not an attempt to give the “correct” usage of ‘embodiment’; it is merely an attempt to fix terminology. Section 3 will recall one long-standing assumption about the relationship between cognition and behavior, namely, that cognition has been supposed to be one, among many, putative causes of behavior. While this assumption is far less than a worked out theory of cognition or of behavior, it does have two virtues: it provides a relatively clear basis for a distinction between cognition and behavior and it avoids the complexities in the current debate on the “mark of the cognitive.”[Note 2] Sections 4-7 will review cases in which cognition has apparently been treated as a type of behavior, along with some of the consequences of this identification. They will consider proposals by Maturana (1970/1980), Chemero (2009), Dennett (1997), Clark (2008, 2010), Haugeland (1998), and Rowlands (2009a, 2009b).
2. What is “Embodiment”?
In “Mind embodied and embedded,” Haugeland rejects what he takes to be a residual Cartesianism found in much of contemporary cognitive science. Haugeland maintains that cognitive states and processes are not realized merely by the brain, but by the brain, body, and world. In the place of this residual Cartesianism, he proposes an “intimacy of the mind’s embodiment and embeddedness in the world” (1998, p. 208). By this “intimacy” he intends “a kind of commingling or integralness of mind, body, and world” (1998, p. 208).
In Out of our heads, Noë identifies the same Neo-Cartesianism as does Haugeland and also rejects it:
It isn’t surprising to be told that there is a thing inside each of us that thinks and feels and wants and decides. This was the view of the seventeenth-century philosopher Rene Descartes, who held that each of us is identical to an interior something whose essence is consciousness; each of us, really, is an internal res cogitans, or thinking thing. … Of course, … Descartes himself, didn’t teach that that thing inside us that thinks and feels is a part of our body, a bit of flesh, such as the brain. [He] supposed that it was something immaterial or spiritual, and so, in that sense, that it was something unnatural. How could mere matter – mere meat – achieve the powers of thought and feeling? Such a possibility boggles the mind. (2009, p. 5)
Like Haugeland, Noë maintains that one’s mental life is not realized in one’s brain. It is realized in one’s physical activities in the world: “consciousness is not something the brain achieves on its own. Consciousness requires the joint operation of brain, body, and world. Indeed, consciousness is an achievement of the whole animal in its environmental context” (Noë, 2009, p. 10) and “my consciousness now … depends not only on what is happening in my brain but also on my history and my current position in and interaction with the wider world” (Noë, 2009, p. 4).
The word ‘depends’ – used in the last passage by Noë – is, of course, ambiguous. There are many sorts of dependency relations. Most notably, there are causal dependencies and various constitutive or compositional dependencies. Essentially everyone in contemporary cognitive science accepts that what one is conscious of at time t is causally dependent upon, or causally influenced by, one’s history and current position and (causal) interaction with the wider world. One may be conscious of an afterimage at time t due to the flash of a camera in one’s visual field at a prior moment and a current absence of other bright lights. Noë, however, envisions a much more radical dependency between mind and world. His view is evidently that one’s mind – what one is conscious of at a time t – is realized by, constituted by, or supervenes upon one’s history and current position and causal interactions with the world. This is what he means when he claims that you are not your brain.[Note 3]
So, for the present, let ‘embodiment’ mean something like the view that cognition is realized by, constituted by, or supervenes upon parts of the body other than just the brain.[Note 4] This is clearly not the only sense of ‘embodiment’ to be found in the literature. To reinforce this point, there is the important strand of thought found in Lakoff and Johnson’s 1999 book, Philosophy in the flesh. Early on, they write:
Reason is not disembodied, as the tradition has largely held, but arises from the nature of our brains, bodies, and bodily experience. This is not just the innocuous and obvious claim that we need a body to reason; rather, it is the striking claim that the very structure of reason itself comes from the details of our embodiment. The same neural and cognitive mechanisms that allow us to perceive and move around also create our conceptual systems and modes of reason. Thus, to understand reason we must understand the details of our visual system, our motor system, and the general mechanisms of neural binding. In summary, reason is not, in any way, a transcendent feature of the universe or of disembodied mind. Instead, it is shaped crucially by the peculiarities of our human bodies, by the remarkable details of the neural structure of our brains, and by the specifics of our everyday functioning in the world. (Lakoff & Johnson, 1999, p. 5)
In this passage, the embodiment of reason, or cognition, is not a matter of the realization base, or the supervenience base, of cognition.[Note 5] Instead, cognition is embodied in the sense that reason is shaped by – is causally influenced by – the nature of one’s brain, body, and experience. Cognition is embodied in the sense that the mechanisms for perception and action are the same as the mechanisms for concept manipulation and reasoning. These are important themes in the embodied cognition literature, but they just happen not to be the themes to be examined here.[Note 6]
3. Cognition as a Putative Cause of Behavior
While philosophy and psychology have witnessed abundant disagreement about the nature of cognitive processes, there has nevertheless remained at least one simple, widely shared presupposition about them, namely, that they are different from behaviors. More specifically, cognitive processes have been supposed to be among the mechanisms that drive behavior. This is something about which Cognitivists and Behaviorists can agree. Where they will disagree, of course, is whether cognition in fact plays the causal role it is traditionally supposed to have. Cognitivists believe that it does; Behaviorists believe that it does not.
At the risk of a bit of anachronism, we might see that even B. E Skinner and Noam Chomsky shared a view of the putative role of cognition in behavior, although they differed in their views of the actual role of cognition. Recall that, in Verbal behavior, Skinner wrote:
It has generally been assumed that to explain behavior, or any aspect of it, one must attribute it to events taking place inside the organism. In the field of verbal behavior this practice was once represented by the doctrine of the expression of ideas. An utterance was felt to be explained by setting forth the ideas which it expressed. If the speaker had had a different idea, he would have uttered different words or words in a different arrangement. (1957, p. 5)
In proximity to these claims, Skinner labeled ideas ‘explanatory fictions. He claimed that “the difficulty is that ideas for which sounds are said to stand cannot be independently observed” (1957, p. 6). Further:
There is obviously something suspicious in the ease with which we discover in a set of ideas precisely those properties needed to account for the behavior which expresses them. We evidently construct the ideas at will from the behavior to be explained. There is, of course, no real explanation. (Skinner, 1957, p. 6)
Chomsky’s review of Verbal behavior makes clear that he shares with Skinner the picture of the cognitive as among the putative causes of behavior. Chomsky writes:
It is important to see clearly just what it is in Skinner’s program and claims that makes them appear so bold and remarkable. It is not primarily the fact that he has set functional analysis as his problem, or that he limits himself to study of ‘observables, i.e., input-output relations. What is so surprising is the particular limitations he has imposed on the way in which the observables of behavior are to be studied, and, above all, the particularly simple nature of the ‘function’ which, he claims, describes the causation of behavior. One would naturally expect that prediction of the behavior of a complex organism (or machine) would require, in addition to information about external stimulation, knowledge of the internal structure of the organism, the ways in which it processes input information and organizes its own behavior. These characteristics of the organism are in general a complicated product of inborn structure, the genetically determined course of maturation, and past experience. (1959, p. 27)
Here Chomsky does not use the word ‘cognition, but cognitive processing has standardly been thought to be a form of information processing that depends on the internal structure of the organism. Later in his review, he adds a bit more:
It is not easy to accept the view that a child is capable of constructing an extremely complex mechanism for generating a set of sentences, some of which he has heard, or that an adult can instantaneously determine whether (and if so, how) a particular item is generated by this mechanism, which has many of the properties of an abstract deductive theory. Yet this appears to be a fair description of the performance of the speaker, listener, and learner. If this is correct, we can predict that a direct attempt to account for the actual behavior of speaker, listener, and learner, not based on a prior understanding of the structure of grammars, will achieve very limited success. The grammar must be regarded as a component in the behavior of the speaker and listener which can only be inferred, as Lashley has put it, from the resulting physical acts. (Chomsky, 1959, p. 57)
Here again, Chomsky does not use the word ‘cognition, but cognition has often been thought to have an inferential character that figures into the behavior of both speakers and listeners. Insofar as this mechanism or grammar is what we have come to understand as cognitive machinery, we can see this as an instance of the view that cognitive mechanisms are among the causes of behavior. So, evidently, Skinner and Chomsky shared the view of cognition as among the putative causes of behavior. Skinner rejected the need for cognitive causes of behavior, whereas Chomsky embraced them. It was the acceptance of the need for cognitive causes that was among the defining features of the cognitive revolution during the second half of the twentieth century.[Note 7]
While the distinction between cognition and behavior – with cognition being among the putative causes of behavior – has been a central feature of cognitive science, there are certain parts of the embodied cognition literature in which the distinction has not so much been rejected as it has been lost. It is not as though the recent critics of mainstream cognitive science we will discuss have drawn attention to the distinction
between cognition and behavior in order to argue that the distinction is untenable. They have not noted the idea that cognition has generally been proposed to be a cause of cognition, then provided reasons or evidence against it. Instead, many discussions simply ignore the long-standing distinction.
4. Maturana on Cognition and Behavior
The current interest in embodied cognition has many diverse sources. Among these is “The biology of cognition,” first published by Humberto Maturana in 1970. In it, Maturana claims that:
A cognitive system is a system whose organization 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. Living systems are cognitive systems, and living as a process is a process of cognition. This statement is valid for all organisms, with and without a nervous system. (1970/1980, p. 13)
Many will find it surprising to be told that all organisms are cognitive systems, that living systems are cognitive systems. This surprise, however, stems from a likely misunderstanding of Maturana’s view. Maturana does not mean by this what one might think he means. Maturana does not mean by a “cognitive system” what Cognitivists mean by a “cognitive system.” Maturana explicitly equates cognition and behavior: the process of cognition is acting or behaving in a domain.
There are details of the interpretation of Maturana’s view that remain unclear. For one thing, the foregoing passage may be meant merely to offer a stipulation of what he means by ‘cognitive system’ and by ‘cognitive process. (This stipulation will be misleading to many, since many cognitive psychologists suppose that cognition is distinct from behavior.) Alternatively, the foregoing passage may be meant to offer some sort of theoretical identity between two prima facie distinct kinds of things, namely, cognition and behavior. Here the model might be something like the proposal that water is H2O or that electrical phenomena and magnetic phenomena are distinct manifestations of a single underlying force, namely, electromagnetism. In favor of the idea that Maturana is offering a stipulative definition of what he means, there is the fact that Maturana offers no argument in support of the identification. If he were reporting an empirical conjecture, one might expect some argumentation in support of the conjecture. Yet, Maturana offers none. This observation, however, is not above challenge. Much of what Maturana writes has an oracular tone which simply tells the readers how specific empirical things are. The identification of cognition and behavior could be another of Maturana’s pronouncements on the nature of the world.
Suppose, however, that we grant Maturana his (misleading) stipulation about how to use the terms ‘cognitive system’ and ‘cognitive process, or that we accept (merely for the sake of argument) the idea that cognitive processes and living processes are one and the same thing in the actual world. If we adopt Maturana’s way of thinking, we can see that “all organisms are cognitive systems” is much less bold when Maturana says it than when Cognitivists say it. Maturana means only that all organisms are behaving systems. Maturana is not claiming that all organisms have cognitive processes that contribute to the production of their behavior. Moreover, once we see clearly what Maturana is claiming, we can also see that cognition, construed as behavior, is typically embodied. Perhaps not all behaviors are embodied, but the paradigmatic cases of behavior certainly are. Once we adopt Maturana’s understanding of ‘cognition’ or his theoretical identification of cognition and behavior, then it falls out quite trivially that cognition is embodied.
5. Chemero on Cognition and Behavior
One might suggest that Maturana did not mean to challenge mainstream Cognitivist and Behaviorist views. His work was neither directed toward nor informed by that tradition. So, it is, perhaps, understandable that his terminology or his theoretical proposals do not mesh well with the assumptions of Cognitivism or Behaviorism. Nevertheless, other philosophers, such as Chemero, do wish to exploit at least some of the ideas advanced by Maturana in an effort to offer a scientific rival to Cognitivism. Moreover, Chemero encounters much the same problems regarding cognition and behavior.
In a footnote dedicated primarily to the question of what cognition is supposed to be, Chemero writes, “I take it that cognition is the ongoing, active maintenance of a robust animal-environment system, achieved by closely co-ordinated perception and action” (2009, p. 212, note 8). Note that, here, Chemero seems to be doing something like “stating his terms,” telling us what he means by ‘cognition. But, notice as well that this proposal sounds like a decision to use ‘cognition’ to mean what many cognitive scientists have meant by ‘behavior. Here is why it at least sounds this way. Presumably, mere thinking doesn’t typically maintain a robust animal-environment system. Merely thinking doesn’t get an organism food, mates, freedom from predation, etc. Thinking must typically be linked to behavior to achieve these ends. As Cognitivists would describe things, cognition is among the causal factors that might guide behavior in ways that (stereotypically?) help sustain an organism, but bare cognitive processing rarely does this.
Insofar as Chemero is simply introducing a new way to use ‘cognition, he will mislead any number of his readers. The confusion is likely to be greater, since he does not make it all that clear that this is his proposal and that he is knowingly and willingly diverging from common usage. But, worse than that, the usage marks no theoretical advance. Suppose we follow Chemero in deciding to call behavior ‘cognition. This simply reformulates one of the core issues separating Behaviorists and Cognitivists, namely, the extent to which “unobservable” structures internal to an organism contribute to the production of behavior. Chemero’s proposal would reformulate the issue to be one of a disagreement over the causal contribution that structures internal to an organism make to the production of “cognition.” This, however, leaves the substance of the issue just as it was. Grant Chemero his (misleading) terminology, so that we should no longer claim that cognitive processes realized in the brain contribute to the production of behavior. Nevertheless, following Chemero, we can still ask how the processes that are realized in the brain contribute to the production of “cognition.” Presumably, the brain has something to contribute to the production of what Chemero calls “cognition,” but what is this? Perhaps the brain processes information, manipulates symbols, or transforms representations. If it does that, then very little indeed has been changed. On the other hand, maybe the brain does not contribute information processing or symbol manipulation or the transformation of representations. But, if not, then what does it do? What emerges is that we have a (misleading) terminological shift that threatens simply to reformulate the debates of the last half-century without advancing them.
As a second problem for Chemero’s terminological shift, notice that it threatens to trivialize the thesis that cognition is embodied. If one understands ‘cognitive processes’ as behavioral processes, then of course, “cognitive processes” are typically realized in the brain, body, and world. Behavioral processes are typically realized in the brain, body, and world. That is just the consensus twentieth-century view. It is quite far from offering a radical embodied cognitive science; by itself, it is completely pedestrian twentieth-century cognitive science. What would be radical would be the conclusion that cognition understood as a particular kind of computation over representations is embodied.[Note 8] What would be surprising would be to find that what has commonly been thought to occur only within the brain in fact occurs in an unexpectedly larger space.[Note 9]
Yet, perhaps all of this is too unsympathetic a reading of Chemero. Perhaps, what he is offering is not a fixing of terms or a definition of what he means by ‘cognition. Indeed, in the footnote cited above, Chemero writes:
I disagree that proponents of radical embodied cognitive science actually require a definition of ‘cognition. That aside, I will say a few things about what I mean by ‘cognition. I take it that cognition is the ongoing, active maintenance of a robust animal-environment system, achieved by closely co-ordinated perception and action. This understanding of the nature of cognition is intended to reflect claims by radical embodied cognitive scientists in philosophy, psychology, AI, and artificial life. (2009, p. 212, note 8)
Perhaps, therefore, he intends to make an empirical claim. After all, Chemero later claims that the models found 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, p. 43). It appears that these models are supposed to provide evidence for the view that “cognitive scientists ought to try to understand cognition as intelligent behavior” (Chemero, 2009, p. 25). Notice that by “explain[ing] cognition as the unfolding of a brain-body-environment system,” Chemero seems to mean that cognition is the unfolding of a brain-body-environment system. Further, we should reject the idea that cognition is (to be explained as) “mental gymnastics.” But, the unfolding of a brain-body-environment system sounds like a metaphorical description of what cognitive scientists have agreed to call “behavior.” Moreover, “mental gymnastics” would seem to be the kind of thing that cognitive scientists would have called “cognition.” [Note 10]
So, suppose that Chemero is offering an empirical proposal that cognition is behavior. The problem for this interpretation, surprisingly, is that Chemero does not give any arguments to support the conclusion. Chemero describes the Beer model and the van Rooij, Bongers, and Haselager model, but he says nothing that ties these descriptions to the issue of understanding cognition as behavior. Consider these models and their descriptions in turn. Chemero reports that:
Randy Beer’s 2003 target article in Adaptive Behavior gives a good sense of what radical embodied cognitive science is all about: it utilizes dynamical systems theory to describe and explain behavior of a simulated robot controlled by an evolved, artificial neural network. (2009, p. 33)
Notice that, by Chemero’s own reporting, Beer’s model describes and explains the behavior of a simulated robot.[Note 11] Nor does the foregoing description seem to represent a mere infelicity on Chemero’s part. In later describing the model, he reports that:
It is important to realize three things about this [model’s] behavior: First, the behavior on each trial is both a discrimination and an action, but these are not separate Finally, it is important to notice that the model is of an individual agent, not of a collection of agents. This focus on individuals is a common feature of dynamical analyses, which take behavior and, especially, development to be the unfolding of a particular brain in a particular body in a particular environment, and not the playing out of a neural or genetic program. (Chemero, 2009, pp. 34-35)
Again, Chemero reports that Beer’s model is a model of behavior. Moreover, note the last sentence very carefully. It claims that behavior is the unfolding of a particular brain in a particular body in a particular environment. There is nothing in this discussion of Beer’s model that supports the view that cognition is behavior. Chemero simply seems to see no difference between the claim that cognition is the “unfolding” of a brainbody-world system and the claim that behavior is the “unfolding” of a brain-bodyworld system.
What, then, of the van Rooij, Bongers, and Haselager model? Their goal is to develop a model with two features. First, it should engage in some behavior or perform some task that prima facie requires representations. (Such a task is one that Clark & Toribio, 1994 would describe as “representation-hungry.”) Second, it should not in fact use representations. Now concede, if only for the sake of argument, that van Rooij, Bongers, and Haselager succeed in developing a model with both of these features. How does that “show how radical embodied cognitive science can explain cognition as the unfolding of a brain-body-environment system, and not as mental gymnastics”? How is this model even relevant? We might put a sharper point on these questions by asking why we should reject van Rooij, Bongers, and Haselager’s interpretation of their model as a model of behavior in favor of Chemero’s interpretation of their model as a model of cognition. Why, more pointedly still, should we reject van Rooij, Bongers, and Haselager’s interpretation of much of dynamical system theory (DST) in favor of Chemero’s interpretation? van Rooij, Bongers, and Haselager offer the following analysis:
According to some (but not all) proponents of DST, a non-representational account of behavior is possible and in many cases more fruitful for understanding the underlying causal processes … Within DST, the behavior of a system is analyzed as an emergent property of the interactions between its subsystems. During the last decade, the tools of DST have proven to be valuable assets for understanding behavior emerging out of multiple interacting components. This behavior ranges from rhythmic movements of fingers and hands … to the temporal patterns of social interaction. (2002, pp. 345-346)
The issue is, thus, not merely a matter of critics of dynamical systems theory having one understanding of it versus supporters having another understanding. The matter is one that looks to divide proponents of dynamical systems models. Advocates of dynamical systems theory differ among themselves over the goals and interpretation of dynamical systems theory. van Rooij, Bongers, and Haselager seem to have the standard cognitive science conception of behavior being the product of a number of underlying causal processes and that among them are cognitive processes, where Chemero does not. Moreover, Chemero gives no reason in support of his revisionary interpretation.
The reader who is sympathetic to Chemero may well believe that something has been overlooked here. There must be arguments for the view that cognition is behavior somewhere in the text. Unfortunately, aside from the analysis just provided, the sympathetic reader will simply have to review that whole of the relevant text to see that no argument has been omitted. There really is no argument there. What may have happened is that Chemero became focused on other features of the Beer and van Rooij, Bongers, and Haselager models and lost sight of the thesis that cognition is behavior. Regarding Beer’s model, Chemero was concerned to point to behaviors that are both discriminations and actions and that the model is of an individual agent, not of a collection of agents. Regarding van Rooij, Bongers, and Haselager’s model, he was interested in its putative capacity to solve a “representation-hungry” task without using representations. The models might well have these features, but those are irrelevant to the thesis that cognition is behavior. That thesis and any arguments for it seem to have been lost.
The upshot of this discussion is that there appear to be two ways to read Chemero’s claim that cognition is the “unfolding” of a brain-body-world system, namely, as something like a stipulative definition or as something like an empirically motivated identification. If we read Chemero as offering a stipulative definition, then his account is misleading, it marks no theoretical advance, and it trivializes the hypothesis that cognition is embodied. If we read Chemero as offering a theoretical identification, then we have been given no reason in its support. Clearly this is a case in which greater attention to the distinction between cognition and behavior is in order.
6. Dennett on Cognition and Behavior
Maturana is not, we have presumed, concerned to address the mainstream cognitive science picture of cognition and behavior. By contrast, Chemero is concerned to do this. He wishes, in some fashion, to displace the common view of cognition as “mental gymnastics.” Neither of these projects, however, is on Dennett’s agenda in Kinds of minds. Dennett is not stalking what either Maturana or Chemero is. Dennett does not so much propose to identify cognition and behavior as he simply conflates them. This manifests itself in two discussions: Dennett’s treatment of the putative intellectual superiority of humans over other organisms and an argument for embodied cognition.
6.1. Of Human Intelligence
Regarding the roots of greater human intelligence, Dennett offers the following:
Our brains are modestly larger than the brains of our nearest relatives (although not larger than the brains of some dolphins and whales), but this is almost certainly not the source of our greater intelligence. The primary source, I want to suggest, is our habit of off-loading as much as possible of our cognitive tasks into the environment itself – extruding our minds (that is, our mental projects and activities) into the surrounding world, where a host of peripheral devices we construct can store, process, and re-represent our meanings, streamlining, enhancing, and protecting the processes of transformation that are our thinking. This widespread practice of off-loading releases us from the limitations of our animal brains. (1996, pp. 134-135)
In this passage, and through much of the chapter that contains it, Dennett runs roughshod over the distinction between intelligence or cognitive capacity, on the one hand, and behavioral performance, on the other. He begins with the view that we have greater cognitive capacity (greater intelligence) than other organisms, but immediately turns to the idea that we intellectually accomplish more than other organisms by means of our use of tools.
It is, of course, plausible to suppose that using tools does make us smarter, that it might make us more intelligent, that it might increase our cognitive capacities. Learning to use a personal computer with a mouse, for example, might enhance certain cognitive capacities. Figuring out how to use an iPhone might require a flexibility and adaptability of thought that fosters creativity. But, that is not how Dennett argues that tools “make us smarter.” Instead, he notes how the use of tools facilitates performance. So, for example, he notes how marking the boundaries of one’s territory enables one to reduce one’s cognitive load on perception and memory. This is a kind of case where performance is improved, not where one is smarter in the sense of having one’s cognitive capacities enhanced. As a second example, he considers the task of finding a key hidden in one of many shoeboxes. One way in which to avoid looking in the same box twice, hence searching inefficiently, is to have two stacks of boxes, the “searched” stack and the “unsearched” stack. After searching a box, one simply moves the searched box from one stack to the other. Clearly this method enables one to accomplish the task more efficiently, namely, by eliminating redundant examinations of individual boxes, but this is far from being a matter of improving the quality of the cognitive processes that cause behavior. Instead, it is a matter of improving the efficacy of behavior.
Here is perhaps another way of viewing the issue.[Note 12] We might take guarding the boundaries of one’s territory and finding a key in a shoebox to be problems to be solved. Dennett apparently takes problem solving to be a kind of cognition, so improved problem solving (guarding the boundaries more effectively, finding the key more reliably) is improved cognition. The reply to this is that the phrase ‘problem solving’ is ambiguous between engaging in cognitive processes and engaging in behavior. This ambiguity masks the difference between improving in cognitive or intellectual capacities, on the one hand, and improving in behavior or performance, on the other. We can see this, perhaps, by noting that marking one’s territory is a means of helping guard one’s territory that reduces cognitive load. It is a way of guarding one’s territory that does not involve thinking about where the boundaries are. One can simply, say, smell them as they are approached. Turning to Dennett’s other example, we can see this by noting that using the stacking procedure, one can find the key without thinking about whether one has already searched a given box or not. One might even be tempted to say that one can solve the problem without even really thinking about it. So, to repeat the moral above, it is one thing to improve behavior or performance; it is another to improve cognitive capacities.
Notice that once one becomes aware of Dennett’s conflation of cognition and behavior, then one can see, first of all, that it leaves untouched the initial question about why humans are cognitively superior to other organisms. Why are we smarter than chimpanzees? Why are our cognitive capacities superior to those of chimpanzees? Granting that tools enable us to perform better leaves untouched the question of our cognitive superiority. We could perform better without increased cognitive capacities. We could, in theory, perform better even with decreasing cognitive capacities. Maybe using a calculator enables us to perform better (solve arithmetical problems more quickly and more reliably), while diminishing our cognitive capacities through lack of practice. Second of all, notice that it is unsurprising that tools often enable us to accomplish things that we could not otherwise accomplish or that they generally enable us to accomplish things more efficiently or reliably than we could otherwise. That’s an entirely pedestrian observation. So, rather than offering a bold conjecture about what makes us smarter, Dennett ends up reminding us why tools are useful. Once we note the long-standing cognition/behavior distinction, the apparent interest of Dennett’s analysis fades.
6.2. Dennett’s Argument for Embodied Cognition
Noticing the cognition/behavior conflation highlights another problem for Dennett. In the same chapter in which he makes the case that tool use makes us smarter, he also argues that cognitive processes are realized in the body and in one’s environment. He observes that many elderly people appear to be quite demented in hospital settings away from the usual environmental cues they find in their familiar homes, but that, in truth, their minds are not as disturbed as their behavior might lead us to believe. What happens in these cases is that the elderly perform less well in hospital settings than they do in their home settings, because the cognitive routines on which they previously relied involved environmental cues that are not available in the hospital. All this is unassailable, but from it Dennett concludes: “taking them out of their homes is literally separating them from large parts of their minds – potentially just as devastating a development as undergoing brain surgery” (1996, pp. 138-139). From the case, Dennett concludes that we have an example of embodied cognition.
Here again, Dennett conflates cognition and behavior. Recall Dennett’s claim that many elderly people in nursing homes are not as disturbed as their behavior might lead one to believe. Couch this with explicit reference to the distinction between cognition and behavior. Dennett’s claim is that these elderly patients are not as cognitively impaired as they might appear to be on the basis of their behavior. The capacities realized in their brains are not as diminished as their behavior suggests. Up to this point, Dennett’s analysis relies on the familiar twentieth-century picture, shared by Behaviorists and Cognitivists, according to which cognition and behavior are distinct and cognition is a putatively brain-bound cause of behavior. But, the adoption of this picture contradicts the embodied/extended view of cognition to which Dennett moves. By adopting the view that environmental tool manipulation is cognitive processing, he contradicts the view that environmental tool manipulation is merely behavior that is sometimes a misleading indicator of cognitive capacities.[Note 13]
The morals from our review of Dennett’s two discussions might be organized differently. Suppose, on the one hand, that we accept the familiar understanding of intelligence/cognition as a cause of behavior. First, recall the discussion of human versus non-human animal intelligence. The improvement of human performance with tool use does not, by itself, suffice to explain the putative superiority in cognitive capacities. In principle, human performance can be improved without cognitive capacities being modified. Second, recall Dennett’s argument for embodied cognition. The fact that elderly people with Alzheimer’s disease can cope better with the familiar tools of their home does not provide a reason for thinking that their cognition is embodied in those tools. Again, behavior and performance can be constituted by tool use and cognitive processes can be causally influenced by tool use, without cognitive processing being embodied in those tools. The mainstream cognitive science view is that cognitive and non-cognitive factors can contribute to performance or the production of behavior. Dennett gives no reason to suppose that the causal contribution of tool use is cognitive.[Note 14]
Suppose, on the other hand, that we understand intelligence/cognition as behavior or performance. First, recall the discussion of greater human intelligence. Now, the observation that tool use makes humans more intelligent amounts to the claim that tool use enables humans to perform tasks that they could not perform otherwise or enables us to perform more reliably or efficiently. That is not a bold conjecture; it is a pedestrian observation. Moreover, it is one that leaves untouched the question of why humans have what the Cognitivists have taken to be greater human cognitive capacities. More concretely (and simplistically), what are the (synchronic) capacities that enable humans to use a greater variety of tools so much more effectively than can non-human animals?[Note 15] Second, recall the discussion of embodied cognition. If one agrees with Maturana and Chemero (on at least one understanding of their views) that ‘cognition’ should be used to speak of behavior, then it is trivial to propose that behavior is embodied and extended. No one has ever doubted that.
7. Cognition Operationalized
In the preceding sections, we have seen how embodied cognition is sometimes tacitly understood to be what Skinnerian Behaviorists and Chomskyan Cognitivists would typically understand as embodied behavior. One variant on this idea, however, is to propose that it is “operationalized cognition” that is embodied. Like behavior, operationalized cognition is also obviously embodied.
What is “operationalizing” cognition? This is a way of “defining” or articulating what is cognitive. This approach begins with a putative cognitive task, and then proposes that anything involved in the accomplishment of that task is cognitive. So, the Turing test is widely treated as an operationalization of the concept of being cognitive or of having human cognition. If a computer cannot be distinguished from a real human being by means of a series of questions and answers, then that computer has (human) cognition. Another famous operationalization maintains that intelligence just is what intelligence tests measure. To give an example from the embodied cognition literature, there is the computer simulation discussed in Beer (2003) (mentioned above in the discussion of Chemero’s views). In this case, the task is to distinguish simulated falling diamonds from simulated falling circles. If a simulated device can perform this task, then that device is capable of categorical perception. So, to repeat, the general strategy of the operational approach is to specify a task, then allow that anything involved in the accomplishment of this task is a cognitive processor. Note well that this sort of operationalizing is not merely a matter of articulating a test or method for finding cognition; instead, performance on a particular task is taken to be definitive of possessing or exercising a particular cognitive capacity.
It is unclear to what extent the advocates of embodied cognition genuinely wish to embrace operationalism, for in several cases we find that those who embrace the operationalist approach at one point will also embrace alternatives at other points. The following subsections will document this with discussions from Clark, Haugeland, and Rowlands. What this shows is that each of these philosophers may intend to offer more than one type of argument for embodied cognition. One argument would be based on operationalized cognition, another on a non-operationalized cognition. So, in principle, the rejection of operationalized cognition need be of only minor importance to their overall view. These philosophers can remain committed to embodied cognition by the force of their other arguments.
7.1. Clark: Operationalist and Non-Operationalist
In one paper, Clark proposes that “what makes a process cognitive, it seems to me, is that it supports genuinely intelligent behavior … To identify cognitive processes as those processes, however many and varied, that support intelligent behavior may be the best we can do” (2010, p. 93). This is clearly the operationalist understanding of cognitive processes. Yet, Clark does not uniformly adhere to this approach. Consider, as one example, his invocation of conditions of “trust and glue” as a basis for claiming that cognition is embodied.
We [Clark and Chalmers] then offered a rough-and-ready set of additional criteria to be met by nonbiological candidates for inclusion into an individual’s cognitive system. They were1. That the resource be reliably available and typically invoked ….2. That any information thus retrieved be more or less automatically endorsed. It should not usually be subject to critical scrutiny (e.g., unlike the opinions of other people). It should be deemed about as trustworthy as something retrieved clearly from biological memory.3. That information contained in the resource should be easily accessible as and when required.4. That the information … has been consciously endorsed at some point in the past and indeed is there as a consequence of this endorsement. (Clark, 2008, p. 79)
In this account, Clark departs from a pure operationalism regarding the cognitive by implicitly limiting embodied cognition to some sort of information processing. So, contrary to what Clark (2010), proposed, it is not that literally any causal contributor to performance realizes cognition. Instead, it is only causally relevant informational contributions to performance that realize cognition.
An example might help to illustrate the point. Consider Nigel who bakes cakes by using recipes he has collected in a book in the way Otto uses his notebook to navigate to the MoMA.[Note 16] Per conditions 1 and 3, Nigel keeps the book in a convenient place in the kitchen where it is easily and reliably available and typically invoked at baking time. Per condition 2, he more or less automatically follows the instructions in his cookbook, not usually subjecting them to critical scrutiny. He deems the information in the recipe book to be about as trustworthy as something retrieved clearly from biological memory. Finally, he consciously endorsed the information at the time in which he entered it into his book. So, Nigel’s use of his cookbook constitutes embodied cognition, by Clark’s analysis. By contrast, consider Nigel’s use of his oven. As spelled out in conditions 1 and 3, his oven is easily and reliably available and typically used when baking cakes. As spelled out in condition 2, it is more or less automatically turned to when it is time to bake. Moreover, Nigel was quite confident of the quality and reliability of the oven when he purchased it two years ago.[Note 17] Nevertheless, Nigel’s use of the oven is not an instance of embodied cognition, by Clark’s lights, since the oven does not constitute an informational resource for Nigel.
So, while Clark at times endorses operationalism, at other times he apparently does not. Given the much greater attention Clark has invested in the “trust and glue” conditions, as opposed to the operationalist condition, it is plausible to read him as more committed to the former than to the latter. So, in principle, he might simply abandon the operationalism and defend embodied cognition on the basis of his conditions of trust and glue.
7.2. Haugeland: Operationalist and Non-Operationalist
Haugeland (1998) also invokes operationalized cognition, although not by name. In a late section of his paper, he invites us to think about the ability to go to San Jose. He notes that going to San Jose is a task that requires coping with something that is out of view, a feature he suggests is indicative of cognition. Surely there are paradigmatic cases where humans travel to San Jose and employ cognitive processes to help them get there. So, this method of delimiting the cognitive enjoys some prima facie plausibility. But, Haugeland also draws attention to more unusual cases, such as retaining a horse that is trained to go to San Jose or picking a road that leads to San Jose. His implicit point is that these uses of environmental resources do not involve data-structure-like representations, but should still count as instances of (embodied) cognition. To get this argument to work, however, he must rely on the operationalist idea that going to a place out of view is a cognitive task and that any way of accomplishing this task is cognitive.[Note 18]
Despite his apparent commitment to operationalism for the space of the “San Jose Argument,” Haugeland also turns to other resources later. His paper ends by claiming that:
As our ability to cope with the absent and covert, human intelligence abides in the meaningful – which, far from being restricted to representations, extends to the entire human world. Mind, therefore, is not incidentally but intimately embodied and intimately embedded in its world. (1998, p. 237)
One way of reading this last part is as offering a theory of what intelligence or cognition is: cognition is trafficking in the meaningful. Such an account, however, is not necessarily an operationalist account. And, indeed, if Haugeland had simply embraced this theory from the outset, he might not have had to appeal to operationalism in the “San Jose Argument.” If cognition really were a matter of abiding in the meaningful, then it would indeed be plausible to conclude that mind is intimately embodied and embedded in its world. As with Clark, it is plausible to read Haugeland as more strongly committed to this latter view of intelligence than to operationalism.
7.3. Rowlands: Operationalist and Non-Operationalist
Rowlands makes an appeal to operationalism:
For the liberal functionalist, if it walks like a duck and talks like a duck, it is a duck. How it manages to walk and talk like a duck is not directly relevant. To this, the extended mind simply adds: Nor does it matter where it walks or talks like a duck. (2009a, pp. 632-633)
Rowlands’ proposal constitutes a very liberal functionalism. Although he couches his discussion in terms of functionalism, one must bear in mind that operationalism is a kind of functionalism wherein functional equivalence is articulated in terms of behavior or input-output functions.[Note 19] What is the standard for being a duck? Walking like a duck and talking like a duck or, more generally, behaving like a duck. It is presumably operationalism to say that if it behaves like a duck, then it is a duck. Nonoperationalis ts are free to say that, even though it behaves like a duck, it is not a duck. Perhaps it is a brilliantly designed robot. So, Rowlands has his operationalist moments.
Then again, Rowlands, like Clark, will sometimes throw in some restrictions on what sort of operations will count as cognitive, namely, that they must be operations involving information processing. In another recent paper, Rowlands proposes that
A process P is a cognitive process if and only if:(1) P involves information processing – the manipulation and transformation of information-bearing structures.(2) This information processing has the proper function of making available either to the subject or to subsequent processing operations information that was (or would have been) prior to (or without) this processing, unavailable.(3) This information is made available by way of the production, in the subject of P, of a representational state.(4) P is a process that belongs to the subject of that representational state.This criterion is understood as providing a sufficient condition for a process to count as cognitive: if a process satisfies these conditions, it counts as cognitive. Whether it also provides a necessary condition is an issue that I shall leave aside (although, for what it’s worth, I suspect that it does not). (2009b, p. 8)[Note 20]
So, it is plausible that, while Rowlands has his operationalist moments, he also conjectures a more “mechanical” approach to cognition. He conjectures a theory of cognition that places restrictions on the mechanisms by which behavior is produced.[Note 21]
After drawing attention to these operationalist moments from Clark, Haugeland, and Rowlands, no argument has been given here that operationalism is mistaken.[Note 22] The principal point, however, has not been to discredit operationalism or to challenge the overall views of Clark, Haugeland, or Rowlands on embodied cognition. Instead, the point is to draw attention to the very idea of cognition operationalized and some of its implications. One can see the allure of cognition operationalized for the advocate of embodied cognition. If cognition is operationalized, then such cognition is behavioral, and behavior is obviously embodied. Behavior is obviously realized by brain, body, and world. So, the thesis of embodied operationalized cognition should not be controversial. Who would doubt it? What would surprise most twentieth- century cognitive scientists is the idea that cognition – construed perhaps as a computational cause of behavior – is realized by brain, body, and world.
8. Conclusion
It is important to bear in mind that the embodied cognition movement encompasses many distinct ideas and tendencies. This paper is meant to draw attention to just one theme that emerges from time to time in the embodied cognition literature. This strand proposes – sometimes tentatively and without maximal explicitness – that “cognition” is behavior. This is a strand we have found in Maturana (1970/1980), along with Chemero (2009), Clark (2010), Dennett (1996), Haugeland (1998), and Rowlands (2009a).[Note 23] The primary goal has, therefore, been to urge greater clarity regarding what various approaches to embodied cognition have had to say regarding the relationship between cognition and behavior. This goal, however, is more than an exercise in pedantry. This paper also draws attention to ways in which the relationship between cognition and behavior evidently matters for our evaluation of the hypothesis of embodied cognition and some of the arguments that have been given to support it. Most notably, if we adopt the unorthodox view that ‘cognition’ applies to types of behavior – and it is not clear that many, if any, advocates of embodied cognition really want this – then among the consequences is that such embodied “cognition” is not a proposal that cognitive science should resist. It is, instead, a proposal that has been a largely unchallenged mainstay of cognitive science since its inception.
Adams F. & Aizawa K. (2001) The bounds of cognition. Philosophical Psychology 14: 43–64.
Adams F. & Aizawa K. (2008) The bounds of cognition. Malden MA: Blackwell.
Adams F. & Garrison R. (2013) The mark of the cognitive. Minds and Machines 23: 339–352.
Aizawa K. (2007) Understanding the embodiment of perception. Journal of Philosophy 104: 5–25.
Aizawa K. (2010) Consciousness: Don’t give up on the brain. Royal Institute of Philosophy Supplements 67: 263–284.
Beer R. (2003) The dynamics of active categorical perception in an evolved model agent. Adaptive Behavior 11: 209–243.
Block N. (2005) Review of Alva Noë, Action in perception. Journal of Philosophy 102: 259–272.
Calvo P. & Keijzer, F (2008) Cognition in plants. In: F Baluska (ed.) Plant-Environment interactions: From sensory plant biology to active plant behavior (pp. 247–266) Berlin: Springer Verlag.
Chemero A. (2009) Radical embodied cognitive science. Cambridge MA: MIT Press.
Chomsky N. (1959) A review of Verbal behavior. Language 35: 26–58.
Churchland P. M. (1984) Matter and consciousness: A contemporary introduction to the philosophy of mind. Cambridge MA: MIT Press.
Clark A. (2005) Intrinsic content, active memory and the extended mind. Analysis, 65 (1): 1–11.
Clark A. (2008) Supersizing the mind: Embodiment, action, and cognitive extension. Oxford: Oxford University Press.
Clark A. (2010) Coupling, constitution, and the cognitive kind: A reply to Adams and Aizawa. In: R. Menary (ed.) The extended mind (pp. 81–99) Cambridge MA: MIT Press.
Clark A. & Chalmers D. (1998) The extended mind. Analysis 58: 7–19.
Clark A. & Toribio J. (1994) Doing without representing? Synthese 101: 401–431.
Dennett D. (1996) Kinds of minds. New York: Basic.
Goldman A. & Vignemont, F (2009) Is social cognition embodied? Trends in Cognitive Sciences 13: 154–159.
Haugeland J. (1998) Mind embodied and embedded. In: J. Haugeland (ed.) Having thought (pp. 207–237) Cambridge MA: Harvard University Press.
Lakoff G. & Johnson M. (1999) Philosophy in the flesh: The embodied mind and its challenge to Western thought. New York: Basic.
Maturana H. (1980) The biology of cognition. In: H. Maturana & F. Varela (eds.) Autopoiesis and cognition: The realization of the living (pp. 2–62) Dordrecht: Springer. (Original work published 1970)
Menary R. (2006) Attacking the bounds of cognition. Philosophical Psychology 19: 329–344.
Menary R. (2010) The holy grail of cognitivism: A response to Adams and Aizawa. Phenomenology and the Cognitive Sciences 9: 605–618.
Noë A. (2009) Out of our heads: Why you are not your brain, and other lessons from the biology of consciousness. New York: Hill and Wang.
Rowlands M. (2009a) The extended mind. Zygon 44: 628–641.
Rowlands M. (2009b) Extended cognition and the mark of the cognitive. Philosophical Psychology, 22 (1): 1–19.
Rowlands M. (2010) The new science of the mind: From extended mind to embodied phenomenology. Cambridge MA: MIT Press.
Rupert R. (2009) Cognitive systems and the extended mind. New York: Oxford University Press.
Shapiro L. (2013) Dynamics and cognition. Minds and Machines 23: 353–375.
Skinner B. E (1953) Science and human behavior. New York: Macmillan.
Skinner B. E (1957) Verbal behavior. New York: Appleton-Century-Crofts.
Thompson E. (2007) Mind in life. Cambridge MA: Harvard University Press.Philosophical Psychology 21.
van Rooij I., Bongers R. M. & Haselager W. P. E G. (2002) A non-representational approach to imagined action. Cognitive Science 26: 345–375.
Wilson R. A. (1994) Wide computationalism. Mind 103: 351–372.
Wilson R. A. & Clark A. (2009) How to situate cognition: Letting nature take its course. In: M. Aydede & P. Robbins (eds.) The Cambridge handbook of situated cognition (pp. 55–77) Cambridge: Cambridge University Press.
For one recent exception to the silence on this issue, there is Shapiro (2013).
See, for example, Adams and Aizawa (2001, 2008), Adams and Garrison (2013), Clark (2005, 2010), Menary (2006, 2010), Rowlands (2009b), and Rupert (2009).
For additional discussion of Noë’s view and of the issues of causation versus constitution, see, for example, Adams and Aizawa (2008), Aizawa (2007, 2010), Block (2005), and Rupert (2009).
The phrase ‘extended cognition’ is often, or even typically, used to describe this view. For simplicity, however, the discussion here will simply work with ‘embodied.
For present purposes, we may treat ‘reason’ and ‘cognition’ as interchangeable.
For other conceptions of embodiment, see Goldman and de Vignemont (2009).
As noted above, there is some risk of anachronism in interpreting Skinner and Chomsky as I have. Perhaps there is some historical oversimplification as well. So, for one thing, Skinner’s rejection is only of ideas, meanings, and images. For a second thing, the interpretation that Skinner denies the existence of the cognitive is based on the assumption that fictions do not exist. It is also based on the assumption that Skinner adopts the principle that science should not postulate unobservable, non-explanatory entities. Opposing this interpretation, however, is Skinner’s claim that “the objection to inner states is not that they do not exist, but that they are not relevant in a functional analysis” (1953, p. 35). Perhaps it would be better to have a more contemporary statement of the view that cognition is a putative cause, such as, “according to functionalism, the essential or defining feature of any type of mental state is the set of causal relations it bears to (1) environmental effects on the body, (2) other types of mental states, and (3) bodily behavior” (Churchland, 1984, p. 36). Of course, even this is a claim about the mental, rather than the cognitive, and it is a statement of functionalism. Nevertheless, the basic picture of”inner goings on” that are distinct from behavior and among the causes of behavior remains. This is the picture that is relevant for our current discussion.
We find a view like this in Wilson (1994), and Wilson and Clark (2009).
One reviewer for this journal puts the point against Chemero more forcefully. It is not okay for Chemero simply to use the word as he pleases. If an author claims to prove that God exists, but then by ‘God’ means Dubuque, Iowa, then the author does not prove anything interesting. Similarly, if Chemero claims that cognition is embodied, but then by ‘cognition’ he means behavior, then he does not claim anything interesting.
This paragraph papers over some expository issues concerning the “empirical” interpretation of Chemero’s view. It is not entirely clear what it is to understand cognition as behavior, that is, whether Chemero has in mind something like a reduction of cognition to behavior or an identification of cognition and behavior or something else. He does not say. Moreover, the idea is apparently unlike proposing to understand cognition as constraint satisfaction (per some connectionist theories of cognition) or as symbol manipulation (per computationalist theories of cognition). These latter proposals at least appear to be referring to the same thing, namely, cognition, but attributing to it a different character, namely, as either constraint satisfaction or symbol manipulation. By contrast, Chemero’s proposal looks to be offering the view that one thing, cognition, is really another thing, behavior. One of the fundamental assumptions of cognitive science has been that cognition and behavior are distinct and that cognition is a non-behavioral cause of behavior. For present purposes, however, we might bracket these expository issues, since carefully reading through Chemero’s exposition of the Beer and van Rooij, Bongers, and Haselager models reveals no arguments for the view that cognition is behavior.
Note that this description is perfectly consistent with the view that the evolved artificial neural network realizes cognitive processing that makes a causal contribution to the behavior of the robot. So, this does not, by itself, undermine the traditional view of cognition as a brain-realized causal contributor to the production of behavior. One reviewer, however, notes that Chemero believes that nonlinearly coupled dynamical systems cannot be decomposed. They are intrinsically holistic. Chemero does, in fact, claim this sort of thing; however, he and Beer do isolate an internal neural network as a component of the larger system. Evidently, the non-decomposability has limits. Pending some clarification, we should not conclude that this non-decomposability prevents us from maintaining that the evolved artificial neural network realized cognitive processing. For further discussion, see Adams and Aizawa (2008, pp. 107-112).
This paragraph was prompted by comments from one reviewer.
Just to be clear, the point is not that the cognition/behavior conflation is intrinsic to the hypothesis of embodied cognition. One might read much of the literature on embodied cognition and find very few cognitive scientists or philosophers who are guilty of this conflation.
No reason unless, perhaps, one is an operationalist. For more on this, see the next section.
The question here is meant to ask, not for an evolutionary account of our tool using capacities – that we can use tools because our ancestors used tools – but for a synchronic account of this behavioral capacity now.
For the story on Otto, see, of course, Clark and Chalmers (1998).
One might think that this is a strained reading of Clark’s conditions. And, perhaps this is a consequence of the implicit information resource assumption in the conditions of “trust and glue.”
As Adams and Aizawa note: “A train on rails has the ability to go to San Jose from a point out of sight. An ICBM has the ability to go to San Jose from a point out of sight. These abilities require no intelligence, no cognition. There are lots of combinations of cognitive abilities that one might deploy to get to San Jose, but not every way of getting to San Jose involves cognition.” (2008, p. 80)
This sort of functionalism seems to defeat one of the core respects in which functionalism was thought to be superior to behaviorism, namely, that functionalism admitted the importance of internal cognitive states and processes in the etiology of behavior.
Just to be clear, the conditions are laid out with a biconditional, but Rowlands then claims that they are sufficient, but perhaps not necessary conditions on being a cognitive process. This is apparently a typographical error.
The present point is not to defend or critique either of Rowlands’ approaches to cognition. However, the reader interested in the later account might examine Rowlands (2010), where the second approach is developed in greater detail. For a critique of this approach, see Adams and Garrison (2013).
A likely defense of Clark, Haugeland, and Rowlands would be that they are not seriously committed to operationalism, so it is perhaps premature to invest time and effort trying to discredit operationalism.
Other possible cases include Beer (2003), Calvo and Keijzer (2008), and Thompson (2007).
Found a mistake? Contact corrections/at/cepa.infoDownloaded from http://cepa.info/3949 on 2017-01-03 · Publication curated by Alexander Riegler