This paper offers critical commentaries on two arguments in support of the “embodied” view of mind and cognition as presented in the philosophical literature of Cognitive Linguistics (eg Johnson, 1987; Lakoff, 1987; Lakoff and Johnson, 1980, 1999). These commentaries take their inspiration from the dialectical materialist philosophical perspective which underpinned the Vygotskian and Activity Theory (or “CHAT”, cf Cole, 1996) traditions (henceforth just “Materialism” for convenience) and especially from the works of Ilyenkov (eg 1977, 1982, 1997; cf Bakhurst, 1991; Jones, 1994, 1998) in which that perspective receives its “clearest articulation” (Bakhurst, 1995: 156).

So far there has been little dialogue about or between CHAT and the philosophy of “embodiment” (but cf Sinha 1997, Lemke, 1996, Jones 1997). It is certainly rather surprising, to say the least, that an approach can advertise itself as a “challenge to western thought” (Lakoff and Johnson, 1999) without paying any attention at all to the entire CHAT tradition or its Materialist roots. Nevertheless, it is clear that both Materialism and CHAT are also being challenged, albeit indirectly. This challenge certainly needs to be met, despite the attractions that “embodiment” and, more particularly, the Cognitive Linguistics paradigm may hold for those working within the CHAT tradition (cf Sinha, 1997). For one thing, the elaboration of a theoretical perspective which is systematically opposed to the extreme biological reductionism of Chomskyan linguistics is a welcome, and potentially exciting, development. But there is also much in the “embodied mind” approach which is directly at odds with the CHAT tradition and the work of its founders. However, the intention here is not to provide a rounded CHAT response to “embodiment”[ii]  which would be difficult, given that CHAT is today a broader church, philosophically speaking, than in the days of Vygotsky and A N Leont’ev. The aim, rather, is to introduce and to clarify some of the key philosophical issues at stake between Materialism and “embodiment” with the help of a couple of case studies. The first involves reasoning from “typical examples” from Lakoff (1987) and the second a metaphor for cognition from Thelen and Smith (1994).

The “embodied mind”: some general principles

A view of mind as “embodied” is central to the philosophy of “experiential realism” developed within Cognitive Linguistics which, as its name suggests, is concerned with both language and cognition, focussing specifically on the ways in which human cognitive abilities and mechanisms are expressed in language as well as with the role language itself plays in cognition. In terms of its ontological commitments, the “embodied” view considers itself to be a form of philosophical realism - “internal realism” following Putnam (Lakoff, 1987: 261) - due to its “commitment to the existence of the real world” (Lakoff, 1987: xv). It also, however, has deep epistemological differences with other  forms of realism. Specifically, the notion of  “embodiment” expresses opposition to positions which are considered to constitute the dominant trend within western thought and which are referred to as “objectivist”. The key epistemological principle of “objectivism” to which “embodied” philosophy objects, is as follows: 

"Knowledge consists in correctly conceptualizing and categorizing things in the world and grasping the objective connections among those things and those categories” (Lakoff, 1987: 163).

This view is rejected because: 

“What the human body does not do, on the objectivist account, is add anything essential to concepts that does not correspond to what is objectively present in the structure of the world. The body does not play an essential role in giving concepts meaning. And the body plays no role in characterizing the nature of reason” (Lakoff, 1987: 174).

“Experiential realists” prefer a view of human reason as “embodied in the sense that the very structures on which reason is based emerge from our bodily experiences” (Lakoff, 1987: 386). These experiences are one of the sources of “basic-level categories” (Lakoff, 1987: 13) or “basic-level-concepts” (Lakoff, 1987: 302) which also manifest the workings of various “human imaginative processes”, such as metaphor and metonymy, which “do not mirror nature” (Lakoff, 1987: 371). Consequently, on their view: 

“Human conceptual categories have properties that are, at least in part, determined by the bodily nature of the people doing the categorizing rather than solely by the properties of the category members” (Lakoff, 1987: 371).

On the “embodied” view: “Human concepts do not correspond to inherent properties of things but only to interactional properties” (Lakoff and Johnson, 1980: 181) which are “the result of our interactions as part of our physical and cultural environments given our bodies and cognitive apparatus” (Lakoff, 1987: 51).

In fact, “experiential realism”, far from being a challenge to western thought, is a rather typical neo-Kantian creation. It wants to accept the existence of the world independent of our minds while denying that we can have objective knowledge of it, ie knowledge of a kind in which “the mind reproduces the logical relations that exist objectively among the entities and categories in the world” (Lakoff, 1987: 163) - an “objectivist” position disparagingly referred to as a “God’s eye view”. In contrast, the experientialist position is held to motivate a form of relativism, in which truth is not “absolute, objective truth” but is truth “relative to understanding” (Lakoff, 1987: 294) and in which “the existence of alternative, incompatible conceptual schemes” is permitted (Lakoff, 1987: 264). On the other hand, “total relativism” is rejected, since “experience of the real world” sets limits to the way we can think about it (Lakoff, 1987: 264), and, furthermore, “reasonable standards for stable scientific knowledge” are provided by “standards of objectivity and correctness” supplied by the “scientific community” (Lakoff, 1987: 265).

In the terms established by the “embodied” view, Materialism is undoubtedly a species of “objectivism” since it is a form of realism which is also committed to the objective knowability of the world in principle (cf Tolman, 1987). On the other hand, being based on the primacy of human practice it has little in common with the bloodless and abstract logical empiricist realism targeted by Lakoff and Johnson in which a direct and timeless correspondence between categories of the mind and categories of the world is assumed.  But a rejection of such “objectivist” views, contrary to the whole line of argument of the philosophy of “embodiment”, does not in fact entail the rejection in principle of the possibility of objective knowledge and objective truth. Materialism can therefore accept the experientialist characterization of meaning, for example, in terms of “our collective biological capacities and our physical and social experiences as beings functioning in our environment” (Lakoff, 1987: 267) while sticking to its epistemological guns. This may mean that there is more in common between the two traditions than there is dividing them (cf Sinha, 1987, for an upbeat view of the relationship) or it may mean that the philosophical differences are too deep to allow much scope for meaningful dialogue. The commentaries below will, it is to be hoped, help to clarify some of the contrasts and congruences between Materialism and the philosophy of “embodiment”.

Case 1: “Typical examples” and categories in Cognitive Linguistics

Cognitive Linguistics has a special interest in the study of “categorization” which it sees as “a key to the study of reason” (Lakoff, 1987: 368). Categorization here means seeing something “as a kind of thing, for example, a tree” (Lakoff, 1987: 6) and when “we reason about kinds of things - chairs, nations, illnesses, emotions, any kind of thing at all - we are employing categories” (Lakoff, 1987: 5-6). Categorization based on “basic-level experience” or “basic-level objects in our immediate environment” such as chairs, tables, trees, and rocks (Lakoff, 1987: 297) also provides the securest forms of knowledge:

 “The best examples of knowledge are things that we know about basic-level objects, actions and relations in the physical domain - what might be called our cat-on-the-mat-knowledge” (Lakoff, 1987: 297).

Knowledge of “basic-level objects” is expressed in “concrete categories” (Lakoff, 1987: 6, my emphasis) or “physical concepts” (Lakoff, 1987: 267) from which “abstract categories” dealing with “abstract entities” such as events, actions, emotions, governments, illnesses, and “entities in both scientific and folk theories” (Lakoff, 1987: 6, my emphasis) are derived by “metaphorical projection” (Lakoff, 1987: 268) from basic-level categories.

Lakoff argues that categorization provides evidence against “objectivism”. Some of this evidence comes from “prototype effects” which involve “scalar goodness-of-example judgements for categories” (Lakoff, 1987: 136). “Prototype effects” arise from the “normal activities involving the use of human reason” (Lakoff, 1987: 367) and some are due to “metonymically based reasoning” (Lakoff, 1987: 152) as in the use of “typical category members” in inferencing (Lakoff, 1987: 86). The following involve “typical examples”:

“Robins and sparrows are typical birds.
Apples and oranges are typical fruits.
Saws and hammers are typical tools” (Lakoff, 1987: 86).

Lakoff cites a study by Rips (1975) which showed that “subjects inferred that if the robins on a certain island got a disease, then the ducks would, but not the converse” (Lakoff, 1987: 86). Subjects generalized from the typical case of the “bird” category (the robin) to the nontypical case (the duck). Since a part of the category (the typical case) is made to stand in for the whole category in the inferencing process then it is metonymy, “one of the basic characteristics of cognition”, which underlies this particular prototype effect (Lakoff, 1987: 77). Lakoff is not arguing that the inference is correct. He notes, for example, that the closely related phenomena of “social stereotypes” are used in “what is called ‘jumping to conclusions’” and that these stereotypes “are usually recognized as not being accurate” (Lakoff, 1987: 85). He is arguing that these phenomena show that cognition - ie  “how human beings categorize” (Lakoff, 1987: 160) - is not as it is painted by “objectivism”.  The one-way inference from robins to ducks is not consistent with there being a strict correspondence between categories of mind and categories of the world but suggests, instead, that the content of categories and concepts is “motivated by bodily or social experience” (Lakoff, 1987: 154); categories of mind are, therefore, “not ‘in the world’, external to human beings” (Lakoff, 1987: 56).

A Materialist point of view would require us to be more circumspect about the relation beween typical examples reasoning in the Rips case and cognition. After all, only someone who knew absolutely nothing about the subject would reason in this way, and the resulting inference would be wrong. It is, in fact, a very good example of jumping to conclusions in the absence of any relevant knowledge or serious engagement with the problem scenario. Indeed, one could say with some justification that it looks more like unreason than reason, bearing the same relation to cognition as hitting one’s thumb with a hammer bears to carpentry. But this point simply emphasises the fact that the “embodied” view of cognition (“how human beings categorize”) differs quite sharply from the Materialist view of cognition as the process of knowing the object (as it is in itself).

To clarify the differences, contrast what happens in Rips’ study with the case where we put the same question to someone who actually knows about birds, or diseases, or both. How would a zoologist, or an ornithologist, or a specialist in viruses respond? It is safe to assume that reasoning from typical examples would play no part in their thinking, which shows not just that the metonymically based inference happens to be wrong, but that this very way of reasoning is not going to help us find a solution. Alternatively, imagine that the Rips’ scenario was not hypothetical but a real world problem requiring practical intervention. How would a person who reasons from typical examples of birds fare in this situation? It is safe to assume that either more serious reflection on the problem or practical engagement with it would show that this form of reasoning was leading up the garden path. Maybe this person would admit defeat (and bring the experts in) or start thinking on different lines. In either case there would be a sharp break with the previous form of reasoning and the “categories” used in that reasoning would have to be ditched.

So two contrasting logics are evident here: the logic of reasoning from “typical examples” which must lead to error, and the logic of attacking the problem armed with specialist knowledge which (at least in principle) may lead to truth. A Materialist can happily accept that one person might think in the first way and another person in the second way. If “objectivism” cannot accept this, then “embodied” philosophers are right to attack it. But at the same time, from a Materialist perspective, it would be a mistake to consider both logics indifferently to be instances of cognition since in this particular case at least new knowledge begins to take shape only from the point at which typical example reasoning is left behind and we head towards, and build on, the “stable scientific knowledge” (Lakoff, 1987: 265) worked out by the scientific community. Only the logic of that process in which knowledge actually develops would be called “cognition” otherwise the term loses its specifically philosophical content and simply means any and all psychological or linguistic processes (cf Ilyenkov, 1997: 75-76), as indeed is the case in Cognitive Linguistics. The metonymically based inference in the Rips study is actually more a display of prejudices than “cognition” in the sense of working out something new. In fact, the relationship of this unreflecting response to the question asked is tautological in character as all that it really tells us is what the subjects mean by the words “bird”, “robin”, “duck” and so on. For Materialism, only “an analysis of the abstraction in terms of its real objective content” (Ilyenkov, 1997: 75) can show whether we are dealing with cognition and concepts rather than semantic processes. This requires a more complex view of the relationship between language and thinking, word meaning and concept, concept and category, concrete and abstract.

To begin with, it is fundamental to CHAT, and clear and explicit in Vygotsky as well as Ilyenkov, that someone can understand what a word means and use that word successfully in everyday communication without having acquired a concept or category. Ilyenkov illustrates this with an example not too unlike the Rips case, although he draws very different conclusions: 

“In order to check if someone has really grasped [usvoil] a category (and not just a word or term referring to it) there is no more reliable way than to invite him to consider a concrete fact from the point of view of that category.

A child who has learnt the word ‘cause’ [‘prichina’] (in the form of the word    ‘why?’) will answer the question ‘why does a car go?’ immediately without thinking - ‘because its wheels are going round’, ‘because someone is driving it’ or something like that.

Someone who understands the meaning of the category [soznaiushchii smysl’ kategorii] will not answer straightaway. First he will ‘think it over’ [‘podumaet’]; he will carry out a series of mental actions. Either he will ‘remember’, or he will look at the thing again, trying to find the real cause, or he will say that he cannot             answer this question. To him, the question about ‘the cause’ is a question which orientates him towards very complex cognitive actions and which outlines the general contour of  the way in which he may arrive at a satisfactory answer - at the correct consciousness of the thing. The category for him is first and foremost the forms of objective cognition, of concrete cognition of things given in contemplation” (Ilyenkov, 1997: 64)[iii].

From this point of view, not all forms of mental or linguistic activity manifest the action of categories: 

“For the adult, the primary significance of categories is that they express the totality of means with which he may work out a correct consciousness of the thing, a consciousness verified by the practice of contemporary society. They are forms of thinking, forms without which thinking itself is impossible. And if   there are only words in someone’s head but no categories, there is no thinking but only the verbal expression of sensuously perceived phenomena” (Ilyenkov, 1997: 64-65).

Going back to the Rips case, Ilyenkov would see our second logic, but not the first, as thinking involving categories. In Lakoff’s basic-level categories used in typical example reasoning, on the other hand, there is only the ability “to express one’s sense impressions in speech, in words, in utterances, in the totality of utterances”, an ability which is “more elementary and which arises earlier than the ability to think, the ability to reflect reality in concepts” (Ilyenkov, 1997: 79). Ilyenkov sees basic-level categories as expressing the form in which “a person becomes conscious of things but does not yet think them, does not yet ‘reflect’ on them” (Ilyenkov, 1997: 65) with this first form being “a necessary premiss and condition for the development of the second” (Ilyenkov, 1997: 79).

On the Materialist view, the basic-level categories of the “embodied” view are, therefore, neither the “best examples of knowledge” per se, nor examples of “concrete concepts”. They are the best examples of the most abstract rather than the most concrete knowledge. Instead of expressing things in their concreteness, that is in their objectively conditioned place within a system or systems of interconnected and interacting phenomena functioning independently of our minds, such categories express them in terms of how we see, touch, hear, taste or smell them, ie as things “abstracted” from that system. No form of reasoning applied to such uncritically accepted “categories” can result in genuine theoretical or conceptual knowledge however indispensable they are in everday life. But someone who remains under the sway of these categories and accompanying prototype effects “will not be an active, self-acting subject of social activity, but never anything more than an obedient tool of somebody else’s will” (Ilyenkov, 1997: 60-61) as indeed is the case with social stereotypes.

So categories, in the Materialist view, are only actively employed when “the process of abstraction begins to be guided by such categories as the ‘essential’ and ‘inessential’” (Ilyenkov, 1997: 79). Thus, for the person who really thinks about Rips’ question, the “asymmetry between typical and nontypical cases” (Lakoff, 1987: 87) proper to the basic-level categories of robin, duck, bird etc must be separated out as inessential to the cognitive task along with the metonymically based inference, otherwise no correct solution can be obtained. Indeed, the very categories of robin, sparrow, duck etc may well prove irrelevant, and therefore an obstacle to objective understanding in so far as these “categories” are not respected by the real world processes under investigation.

For Ilyenkov, thinking “in the strict sense of the word” (Ilyenkov, 1997: 74) goes beyond the ability to use the type of generalized abstractions found in word meanings expressing “basic-level categories” and involves the concept understood as an abstraction expressing the “essential general” [sushchestvennoe obshchee], which “reflects the essential side, the essential characteristics of the object” (Ilyenkov, 1997: 74). The concept reflects the thing from the side of its objective interrelations with the concrete whole to which it belongs and it is therefore the concept, and not the “basic-level category” which has the property of concreteness. The concept takes us closer to an understanding of the thing as it is in itself even though the content of the concept may completely contradict the immediate evidence of our senses.

Empirical, abstract generalizations of the kind expressed in basic-level categories are referred to as “notions” and not “concepts” by Ilyenkov. In his treatment of the relations between the two he is not intending to “belittle the significance and cognitive role of elementary, ‘intellectual’ general abstractions. Their role is great: no concrete universal concept would be possible without them” (1982: 85). However, the notion is an “abstraction, whose formation is affected by a great number of factors, and first of all the direct practical interest, man’s need and the purpose reflecting the need ideally” (Ilyenkov, 1982: 51-52) whereas “the links between the concept - a theoretical abstraction expressing the objective essence of the thing - and practice is [sic] much broader, deeper and more complicated” (Ilyenkov, 1982: 52). He explains: 

“In the concept, the object is comprehended from the standpoint of mankind’s practice in its entire volume throughout the history of world development, rather   than from the standpoint of the particular, narrow pragmatic objective and need” (Ilyenkov, 1982: 52).

Materialism, like “embodiment”, rejects the view of cognition as a disembodied process and wholeheartedly embraces the view that human experience - the source of all knowledge - is the product of an interaction between humanity and the world. It also follows that “in the ideas we directly have of the external world, two quite dissimilar things are muddled and mixed up: the form of our own body and the form of the bodies outside it”, as Ilyenkov puts it in his discussion of Spinoza’s philosophy (Ilyenkov, 1977: 67), a situation giving rise to the types of “experiential” reasoning processes discussed above:

“The naive person immediately and uncritically takes this hybrid for an external thing, and therefore judges things in conformity with the specific state evoked in his brain and sense organs by an external effect in no way resembling that state” (Ilyenkov 1977: 67).

Ilyenkov’s naive thinker is just like the subjects in Rips’ study who show, by reasoning from typical examples that their ideas “have properties that are, at least in part, determined by the bodily nature of the people doing the categorizing rather than solely by the properties of the category members” (Lakoff, 1987: 371, quoted above). This is also exactly why this kind of reasoning is of no earthly good when the bodily natures of the robin, the sparrow, and the duck are at issue rather than the bodily nature of the thinker. The crucial issue here is not whether two “quite dissimilar things” may be mixed up in one idea, but whether it is in principle possible to begin to separate them out and to distinguish the states and dispositions of one’s own body from those of the bodies one is seeking to understand. The “embodied” view is premissed on the impossibility in principle of unravelling this experiential complex (which would entail a “God’s eye view”), while Materialism is based not just on the possibility of doing so, but on the theoretically and empirically verifiable fact of so doing in the course of history.

The “embodied” view, like Materialism, attempts to study the cognitive process from the point of view of the real, suffering human beings who are responsible for it. Its strength, and its potential, lies in its opposition to the dualistic, formalistic, or crudely biological reductionist tendencies of “objectivism”. Its weakness, in Materialist terms, is that its naturalistic focus on the body and bodily interactions as the source of knowledge entails an idealistic and relativist epistemology and with it a preoccupation with the mere form of the cognitive process rather than its objective content, which means that the “real Logic of thinking” (Ilyenkov, 1997: 76) is left out of account. Within Materialist epistemology, the process of cognition is just the reflection in knowledge of what is above all a practical process in which the social body of humanity must learn to use the forces of nature in accordance with the objective properties of the latter. In the struggle to make these forces serve our vital ends, a process unfolds in which these forces ultimately dictate the necessary lines along which thinking must go to arrive at knowledge adequate to our aims. Practice ultimately reveals what part of our ideas belongs to our bodies and what part belongs to the body external to us. The standpoint of objective truth  - the “God’s eye view” as it were - coincides with the standpoint of purposeful practical transformation of reality.


Case 2: Watt’s centrifugal governor as a metaphor for cognition

Thelen and Smith (1994) present an original and very powerful account of “embodied cognition” as a fluid and dynamic process taking shape through the interactions between human bodies and their environment (cf Lemke, 1996). While they are favourably disposed towards work in the CHAT tradition (eg Thelen and Smith, 1994: 328 -329), they rely heavily on the ideas of Lakoff (1987) and particularly Johnson (1987) to argue against “objectivist philosophy” in which “concepts represent an external reality - an external set of ‘true’ categories that exist independently of the minds that create them” (Thelen and Smith, 1994: 162, emphasis in original). Their own position involves shifting the focus to “how concepts are used in particular tasks... how they make contact with reality” (Thelen and Smith, 1994: 162, emphasis in original) and its “central tenet” is “that behaviour is not symbolically represented in the system or programmed in the absence of the here-and-now context in which it is performed” (Thelen and Smith, 1994: 263, my emphasis).

In illustration of this non-representational, “embodied” view of cognition, Thelen and Smith (1994: 331) re-present a metaphor from van Gelder (1992) involving an account of the invention of a mechanical device to solve “a 19th century engineering problem involving a steam engine that drives a flywheel connected to some machinery” (Thelen and Smith, 1994: 331). The problem is as follows: 

“the speed of the flywheel must remain constant despite irregular and continuous fluctuations in the workload on the engine and the steam pressure. How could one design a device, called a governor, to maintain a constant     flywheel speed?” (Thelen and Smith, 1994: 331).

First, van Gelder considers “a computational solution that contains the following: a tachometer for measuring the speed of the wheel, a device for calculating the throttle valve adjustment, a throttle valve adjustor, and an executive to handle the sequencing of operations” (Thelen and Smith, 1994: 331). In other words, the device contains “representations (measures of the steam pressure, the speed and so forth)” and “works - just like the traditional metaphor of cognition - by the manipulation and passing of representations from one component to the next” (Thelen and Smith, 1994: 331). While acknowledging that a device involving such representations “probably could be built that would work” they argue that “it is unlikely that it would work as well as or adapt as intelligently and fluidly to changes in workload and pressure as does the simple and elegant device invented by James Watt in the early 1800s” (Thelen and Smith, 1994: 331). This device is the “centrifugal governor” which works “in a way that allows for intelligent and continuous context sensitivity and a global order without representations or computations” (Thelen and Smith, 1994: 332). Watt’s device “consists of a verticle spindle geared into the main flywheel so that it rotates at a speed directly dependent upon that of the flywheel itself” (Thelen and Smith, 1994: 331).Van Gelder’s explanation follows (Thelen and Smith, 1994: 332): 

“Attached to the spindle by hinges were two arms, and on the end of each arm was a metal ball. As the spindle turned, centrifugal force drove the balls outwards and hence upwards. By clever arrangment, the arm motion was linked   directly to the throttle valve. The result was that as the speed of the main wheel increased, the arms raised, closing the valve and restricting the flow of steam; as the speed decreased, the arms fell, opening the valve and allowing more steam to flow. The result was that the engine adopted a constant speed,           maintained with extraordinary swiftness and smoothness in the presence of large fluctuations in pressure and load” (Van Gelder,1992: 3).

Hence, Watt’s governor “does not represent anything, it just does the job” (Thelen and Smith, 1994: 332). The device “is smart; it embodies [sic] (although it does not represent) as much knowledge as the computational governor” (Thelen and Smith, 1994: 332). Cognition “is like the smartness of Watt’s centrifugal governor - an activity emergent in the simultaneous and continuous interactions of a myriad of heterogeneous forces” (Thelen and Smith, 1994: 332), with one qualification: 

“the theory of cognition that we have laid out in this book is not like the centrifugal governor in one critical way: cognition develops. We must envision a centrifugal governor that through its own activity changes its very components and the manner of their interaction. We believe that this vision of the developing  centrifugal governor offers a good metaphor for cognition” (Thelen and Smith, 1994: 332).

The centrifugal governor certainly makes for an interesting philosophical case. But from a Materialist point of view, its value as a metaphor for specifically human cognition in general is questionable. Of course, Thelen and Smith’s work is a salutary reminder that cognition must not be reduced to or equated with the manipulation and processing of symbolic representations; thinking is more than just verbal thinking. The kind of spontaneously developing, exquisitely sensitive dialectic of brain and limbs, of eye and brain that we observe and experience in skilled tool use, in the creative work of the craftsman (cf Keller and Keller, 1996), or in the goalkeeper’s dive to save the penalty involves not so much the “‘expression’ of thinking” but “thinking itself, thinking as such”, as Ilyenkov (1974: 76) puts it in his discussion of “the work of the hands” as opposed to verbally expressed thought. On the other hand, the capacity to think in this way is exercised by individuals only within a community, a culture, a life space saturated by, amongst other things, symbols. And consequently, if human cognition - in its most essential or typical form - is viewed as a dimension of social practice, and that practice is necessarily mediated by the production and use of tools (both material and symbolic)[iv], then the case of the governor’s action, metaphorically construed, cannot be the best example of it.

Let us instead, then, indulge in a rather mischievous subversion of Thelen and Smith’s argument. Instead of taking the governor as a metaphor for cognition, why not take it for what it is, namely a genuine instance of actual cognition? If we do so, then suddenly everything looks different. Now the governor is only one component of a larger system of dynamic cognition which also includes the practical and theoretical activity of James Watt, its inventor. It is still true that the governor contains no representations or measures of other processes taking place within the mechanical system; the governor ‘just does the job’. But the job that it does is the job that it is designed to do, a design that exists outside of and independently of the governor. If steam pressure and the speed of the wheel are not represented in the “body” of the governor they are represented in the body and mind of Watt himself in the course of designing the device. Watt’s design activity consists in abstracting all necessary “moments” or interconnections between the governor and the mechanical system and representing them symbolically. Through the “bodies” of these symbolic objects the governor does its job only in representation, or “ideally”, as the Materialist tradition would say (Ilyenkov, 1977). The design idea, though the product of the bodily processes of its inventor, including the action of “neuronal groups” (Thelen and Smith, 1994: 142), becomes something separate from and independent of the inventor and can then be passed to the manufacturer who will produce the real thing in conformity with it.

Consequently, if we look at the system as a whole, we find that the centrifugal governor exists in a number of empirically distinct, and apparently contradictory, forms (or “embodiments”): as a real piece of functioning machinery and as an idea expressed in symbols to name two. The first does the job without representing anything and the second represents without doing anything. What’s more (so to speak), in between there is Watt himself who “does the job” of designing the first by means of the second. This now looks all very confusing from an “embodied cognition” point of view. If it is Watt’s thinking that is really “embodied” in the literal sense, we should concede that it is not actually his body, but that of the governor, which is doing the real work. Alternatively, if we prefer to say that the device “embodies” knowledge, as the authors’ metaphor has it, then this knowledge is “disembodied” in the body of Watt and in the symbolic body of the design. Thus, with respect to this complex system of action and cognition we find, contrary to Thelen and Smith’s central tenet, that the governor’s behaviour is indeed both “symbolically represented in the system” and “programmed in the absence of the here-and-now context in which it is performed” (Thelen and Smith, 1994: 263).

Ilyenkov offers quite a different metaphor for “embodied cognition”: 

“The conscious relationship of the subject to the world surrounding him...can be likened to what the portrait artist does, and how he does it. The painter, as we know, places in front of him the model, along with a canvas on an easel and then begins to purposefully bring the representation on the canvas into correspondence with the model. The portrait or the landscape taking shape on the canvas is a reflection, an image of the model. But this reflection, like the model itself, is located outside the artist, as the object and product of his activity. It is the artist himself, as the subject of activity, who compares the representation with the model from the side, from a third position. Both the object represented and the representation of the object are counterposed to the   artist as two objects located outside of him and which can be compared to one another” (Ilyenkov, 1997: 46). 

In the case of the governor we begin instead with the representation of the object and then proceed to create the object represented. The coming into being of the real device is preceded and mediated at all stages by the “ideal” device which portrays its mode of functioning in all essential details. As the real device takes shape in accordance with the design, the one can be compared with the other and the two objects can be “purposefully brought into correspondence” from the third position of the designer. Finally, the functioning device stands on its own, operating independently in accordance with physical laws which are blind not only to the problem Watt started with but to theory in general. 

The case of the governor, therefore, is a good illustration of the form taken by the cognitive process in human social activity: ”Man, and only man, ceases to be ‘merged’ with the form of his life activity; he separates it from himself and, giving it his attention, transforms it into an idea” whereupon “the form itself of the activity corresponding to the form of the external object is transformed for man into a special object with which he can operate specially without touching and without changing the real object up to a certain point” (Ilyenkov, 1977: 278). Watt’s activity here is “activity on the plane of representation, altering the ideal image of an object” (Ilyenkov, 1977: 280). Such activity, Ilyenkov stresses, is also “sensuous objective activity” in which Watt interacts bodily with a “sensuously perceived object” (Ilyenkov, 1977: 280)  - eg a diagram of the governor - and physically alters that object in some way. However, 

‘the thing altered here is special; it is only the objectified idea or form of the             person’s activity taken as a thing. That circumstance also makes it possible to slur over the fundamental, philosophical, epistemological difference between material activity and the activity of the theoretician and ideologist who directly alters only the verbal, token objectification of the ideal image’ (Ilyenkov, 1977:   280).

Thelen and Smith’s treatment of cognition involves precisely the kind of philosophical slurring over referred to by Ilyenkov. While they commendably attempt to examine the real interconnections between thinking and doing as aspects of practical activity they are reluctant to treat thinking as ideal activity and thereby miss the significance and role of the ideal, including theoretical and scientific representations, in social practice. The problem of the “ideal” is one of the most difficult in Materialist philosophy, but its difficulty is proportional to its importance for an understanding of specifically human cognition, and of social practice in general: 

 “Without an ideal image man cannot in general exchange matter with nature, and the individual cannot operate with things involved in the process of social production” (Ilyenkov, 1977: 274).

While acknowledging, therefore, the power and explanatory potential of Thelen and Smith’s “embodied” account of cognition in action, one must recognize that its inability to theorize the role of the ideal leaves a rather large hole at its center.

On a final note, it is perhaps worth thinking briefly about how the experiential realism of Lakoff and Johnson would cope with Watt’s centrifugal governor. We can assume that as Watt works out his design he categorizes and makes use of a variety of imaginative processes such as metaphor. The embodied view would insist that the resulting ideas could not be objectively true in the sense of corresponding to the inherent properties of the real world, although they may be true “relative to understanding”. The problem the becomes how to explain why the real piece of machinery works in the way it was designed to do. How can the systematic correspondence between idea and real object be accounted for? Clearly, the actual behaviour of the governor cannot be explained in terms of human bodily and social experience. By the same token, we cannot say that Watt’s ideas are true only “relative to understanding” because the governor, as an unthinking lump of metal, is completely indifferent to any understanding. If human ideas cannot be separated from the body which makes them possible, then why do they apply so accurately to the non-human “body” of the governor? If we know only “interactional properties” then how can we know the “inherent properties” of a mechanical system? How is it possible, using cognitive mechanisms that “do not mirror nature”, to represent so faithfully the properties and interactions of natural phenomena? 

To a Materialist, the case of Watt’s governor highlights serious contradictions in the philosophy of the “embodied mind” and provides powerful evidence of the objectivity of human thinking.


Bakhurst, D (1991) Consciousness and Revolution in Soviet Philosophy: From the Bolsheviks to Evald Ilyenkov, Cambridge: Cambridge University Press

Bakhurst, D (1995) ‘Lessons from Ilyenkov’, in The Communication Review, Vol1 (2): 155-178

Cole, M (1996) Cultural Psychology: A Once and Future Discipline, Cambridge, Mass: Harvard University Press

Ilyenkov, E V (1974) ‘Hegel and hermeneutics’ [in Russian], Voprosii Filosofii 8: 66-78

Ilyenkov, E V (1977) Dialectical Logic, Moscow: Progress

Ilyenkov, E V (1982) The Dialectics of the Abstract and the Concrete in Marx’s ‘Capital’, Moscow: Progress

Ilyenkov, E V (1997) The Dialectics of the Abstract and the Concrete in Scientific-theoretical Thinking [in  Russian], Moscow: Rosspen

Johnson, M (1987) The Body in the Mind, Chicago: University of Chicago Press

Jones, P E (1994) ‘Evald Ilyenkov and the history of Marxism in the USSR’, Review of Bakhurst (1991), in History of the Human Sciences, Vol 7 (4): 105-118

Jones, P E (1997) ‘Language and cognition in goal-directed, practical activity’, paper presented at the 5th International Cognitive Linguistics Conference, Amsterdam, July 14-19

Jones, P E (1998) ‘Symbols, tools, and ideality in Ilyenkov’, paper presented at the 4th Congress of the International Society for Cultural Research and Activity Theory, Aarhus, June 7-11

Keller, C & Keller, J (1996) ‘Imaging in iron, or thought is not inner speech’, in J Gumperz & S             Levinson (eds) Rethinking Linguistic Relativity, Cambridge University Press

Lakoff, G (1987) Women, Fire, and Dangerous Things, Chicago: University of Chicago Press

Lakoff, G & Johnson, M (1980) Metaphors We Live By, Chicago: University of Chicago Press

Lakoff, G & Johnson, M (1999) Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought, New York: Basic Books

Lemke, J (1996) ‘Self-organization and psychological theory’, Review of Smith & Thelen (eds) (1993), Theory and Psychology, 6 (2): 352-356

Rips, L J (1975) ‘Inductive Judgements about Natural Categories’, Journal of Verbal Learning and Verbal Behaviour, 14: 665-681

Sinha, C (1997) ‘Signifying Subjects’, International Journal of Communication, Vol VII (1-2)

Smith, L B & Thelen, E (eds) (1993) A Dynamic Systems Approach to Development:Applications, Cambridge, Mass: MIT Press/Bradford Books

Thelen, E & Smith, L B (1994) A Dynamic Systems Approach to the Development of Cognition and Action, Cambridge, Mass: MIT Press/Bradford Books.

Tolman, C (1987) ‘Dialectical Materialism as Psychological Metatheory’, in H J Stam et al (eds) The Analysis of Psychological Theory: Metapsychological Perspectives,Washington: Hemisphere Publishing Corporation van Gelder, T (1992) ‘What might cognition be if not computation?’, Indiana University Cognitive Science Research Report 75.

[i] A version of this paper was presented at a meeting of Sheffield Hallam University Language and Linguistics Group on 4th March, 1999. I would like to thank all those who attended for their comments and encouragement. Passages from Ilyenkov (1997) were translated with help from Marianna Ivanova.

[ii] A critical review of Lakoff and Johnson (1999) is currently in preparation.

[iii] The pronoun ‘he’ has been used throughout the translated passages to conform with Ilyenkov’s use of the masculine singular pronoun to refer to ‘the child’, ‘the adult’, etc.

[iv] Cf Jones (1998) for discussion.