Download as RTF document (41KB)

The Theory of Ideology: Bringing the Mind Back In

Trevor Pateman

Abstract: Argues against the sociologism of much modern sociology and seeks to use contemporary cognitive science to re-work the theory of ideology. Argues that ideologies are inherently variable through time and between individuals, whatever social or cultural efforts are made to stabilise them. This is because the mind is not (or is not completely) plastic to social and cultural shaping. Ideologies are like languages, since they are internalised in individuals as generative mechanisms which are not transparent to introspection or self-monitoring. The work of Goran Therborn, Stephen Stich, Hilary Putnam and Tyler Burge is referenced.

From the standpoint of sociology, classical political theory stands accused of having sought to derive claims about society - about possible and desirable forms of association - from claims about human nature. Most sociologists think that Marx summed up the mistake of this approach when he wrote in the sixth Thesis on Feuerbach, `The essence of man is no abstraction inherent in each single individual. In its reality it is the ensemble of the social relations'. But from the standpoint of political theory and, more generally, of philosophy, sociologists necessarily assume a particular theory of human nature in the very moment of their denial that there is any such thing as human nature. For in order to sustain their preferred account of enculturation and socialization, sociologists have to assume the plasticity of human matter to cultural and social shaping - and that assumption of plasticity just is a theory of human nature.

The detailed explanation of how society gets into the individual became the task of social psychology and was exhaustingly discharged in the crushing banalities and tautologies of the social psychology textbooks of the 1950s, 1960s and probably even more recent ones. Despite attempts to 'bring men back in' (the title of an article by Homans 1964) and protests against the ' oversocialized conception of man in modern sociology ' (the title of an article by Wrong 1977), this sociologism (as I shall call attempts to exclude the idea of human nature from social theory) was still alive and well in the 1980s. Here, for example, is an influential Marxist sociologist, Goran Therborn, writing about ideology:

From what is known about the ideological plasticity of human beings and their creative capacities, we should expect the given ideologies to be almost completely reproduced in societies whose internal conditions and relationships to the natural environment and to other societies remain exactly the same from one generation to the next. (We would have to allow for only a small margin of individual `misfits' stemming from the irreducibility of psychodynamic processes to complete social control). A parental generation will always mould its children according to its own form of subjectivity; and if the ecological, demographic, socioeconomic and intersocietal relationships remain the same, the younger generation will face exactly the same affirmations and sanctions of the existing ideologies as the parental one. It follows that the explanation/investigation of the generation of ideologies will have to start from processes of change in the structure of a given society and in its relationships to its natural environment and to other societies. It is these changes that constitute the material determination of the rise of ideologies. (Therborn 1980, p. 43)

Therborn's 'argument' here is really only the ritual repetition of the positions which modern sociology has used to establish itself as what it is, in contradistinction to (bad) old political theory and (even worse) psychology. The specific challenge of psychoanalysis is evaded with a merely parenthetic concession.

Now the parenthesis might indeed tempt the classical political theorist, the Freudian or a sociologist such as Dennis Wrong, to attempt a war of attrition on Therborn's position. This would involve trying to fatten up that parenthesis so as to yield a dialectical account of `the individual' (as site of nature) and `society' (as site of culture, here in the form of ideology).

However it is my belief that recent cognitive science, as theory of mind or mental representation, can allow a move outside the boundaries set by Therborn's text and his parenthesis on to ground where a very different and much more fertile account of the life of ideology in the individual and society might be possible.

Therborn has no theory of the mind except a default account which gives it those properties that allow the dreary business of `complete reproduction' to go on. And the moment one considers what a plausible theory of the mind actually looks like, it is clear that it would not sustain such `complete reproduction', but leads to the conclusion that, even in the absence of environmental changes, ideologies always tend to change; and not because of the presence of a few `misfits', but because of the operation of ordinary, normal, mental processes. Ideologies on any plausible cognitivist account will not display Newtonian inertia but Heraclitean flux. How so?

What Therborn calls a parental generation provides what I shall call a learning generation with at least two kinds of ideological material. First, it supplies representations of actions, events, experiences, situations etc. made accountable through an ideology. Secondly, it supplies rules of an ideology through which events, etc., are supposed to be made accountable.

In the first situation, where he or she is confronted with representations, the learner's task is to arrive at a competence in the ideology. The favoured cognitivist construal of such competence is that the learner's competence takes the form of a mentally represented generative theory which could, for example, be modelled in a computer program. Now if there are no very rigid constraints on the class of learnable ideologies (such that one should really think of ideologies as archetypally represented in the (collective) unconscious with incoming information merely providing their remembrance) then it is no more than the application of a general theoretical principle to say that the representation of an ideology which any individual learner will arrive at will be underdetermined by the data (the accounts of the parental generation) on which it is based. In other words, there can be no principle of induction which could lead the learner from a set of data to exactly the theory or ideology which generated it. Furthermore, at least some of the cognitive processes involved in the derivation of a productive mental representation of an ideology will be inaccessible to introspection and, hence, in principle, not controllable in their operation by reflexive self - monitoring (in the sense of Giddens 1979) of the processes of ideological growth. This means that if the learner's accounts generated from his or her mentally represented ideology are in any way unsatisfactory, attracting sanctions (the negative reinforcement of the parent or teacher saying, That is NOT what we believe!) then the learner may only be capable of responding by making ad hoc adaptations which modify the output of the mentally represented ideology without altering its fundamental or core structure. This is because the learner does not have (computational) access to all that is going on in the development of a mentally represented ideology (cf. the brilliant work of Henning Andersen 1973; and also Fodor 1983). Underdetermination and inaccessibility identify sources of ideological transformation generated independently of the environmental changes Therborn lists as necessary and sufficient conditions of ideological change.

What Therborn has `overlooked' is that the mind is itself an environment for ideology and an environment which is intrinsically creative because necessarily abductive in the sense of C S Peirce: abduction is that kind of thinking, theorising and hypothesising (not necessarily conscious) which takes us beyond the information given. And note that in locating at least part of this creative power in regions inaccessible to introspection and reflexive self - monitoring, the cognitivist approach I am sketching does not equate with a voluntarist theory of subjective agency or individual will. On my account, ideologies are as much things to which individuals are liable as things of which they are capable. It is just that liability is no longer equated with a tendency to reproduce unchanged (Therborn's Newtonian doctrine) but rather with a tendency to reproduce in changed form (the Heraclitean doctrine). Ideology is inherently variable from generation to generation, and individual to individual. This is what so horrifies defenders of all ideologies, whether of religious doctrines or the grammar prescribed by the Academie Francaise

Turning now to the second kind of input which a learning generation receives we can ask, What happens on this cognitivist account if we consider not the growth of competence in an ideology from the evidence of accounts and representations, but the inculcation of an ideology as itself a set of rules? (This is what one might call the catechism model of ideology) To be sure, individuals are quite capable of learning and repeating lists, but the moment they try to use their lists (their rules) as a basis for judgement (for giving accounts and producing representations themselves) underdetermination intervenes to make it the case that the rules do not determine the judgement, even though the rules can be perfectly determinate in the mind. How else could casuistry be possible? But since cases are no more determined by principles than theories by data, the tendency to ideological variation cannot be thwarted by an emphasis on catechising learners. They will always and necessarily go their own (ideological) way. Teaching a catechism is as useless as telling the waves to retreat.

The argument set out in the previous paragraphs generalizes to any form of representation. An apprentice to a system of visual representation (a visual ideology for a writer such as Hadjinicolaou 1978) is presented with both instances or exemplars of the system of representation in use (individual pictures) and with explicitly formulated rules of the system (in drawing manuals, for instance). Yet even in the training routines of the most hidebound academies (of the sort described in Grosz 1982), the apprentice necessarily `internalizes' the system of representation idiosyncratically as a visual idiolect. Because we are at home with `artistic creativity', even in this modest form, this fact does not strike us as at all remarkable, yet it is strictly unaccountable within Therborn's framework.

So far in this sketch I have, like Therborn, assumed environmental constancy. But the cognitivist model I have been sketching also holds the prospect of providing a theory of how environmental change gets into the mind, and hence into the process of ideological reproduction, without it being necessary to postulate bad faith on the part of individuals. For example, suppose that in the abduction or growth of mentally represented ideologies from accounts and representations, or in the derivation of case judgements from principles, there are preference rules involved (for one account of preference rules see Lerdahl and Jackendoff 1983) so that all other things being equal certain ways of representing situations are favoured over others. Then one could argue that a changing environment acts on individuals below the level of conscious awareness or of the introspective mind, causing alterations in their preference rule orderings and weightings and yielding ideological change as output. Things which were once seen in that way are now seen in this way. On similar lines, one could reintroduce affective and instinctual components into what may seem an overly intellectualist account and, of course, simulations of hot cognition were among the earliest achievements of work in Artificial Intelligence (I am thinking of Colby's and Abelson's work, reviewed in Boden 1977, chs. 2 - 4).

In sum I have tried to suggest how, using cognitive theory, we can bring the mind back into the theory of ideology and with it incorporate an autonomous learning- or cognitive development- theoretic element into theories of ideological and, more generally, cultural change. But the viability of the Heraclitean model I have sketched depends crucially on sustaining a viable cognitivist concept of mind in the face of considerable philosophical and sociological hostility to the very idea of mental representation. I have tried to confront that hostility, and show it to be misguided, in another essay on this website. "Wittgensteinians and Chomskyans: Mentalism in Linguistics" and the reader should go there for the arguments.

Technical Stuff: The Personal and The Sub-Personal: Beliefs and Sub-Doxastic States:

Ideology has a public and social side when it is deployed in normative social practices - making judgements, evaluations, criticisms, and so on. But on the cognitivist account it also has a non - introspectible existence as a generative set or system of mental representations. This two-sided existence provides for a division of labour between the social scientist and the cognitive scientist. The former engages in such activities as ascribing beliefs to persons (subjects) and the latter attributes states and processes to organisms.

Sociologists and anthropologists do not need to be told that it is often a very tricky business indeed to decide what belief to ascribe to some person, especially if that person is mentally ill or belongs to an exotic culture. Stich 1983 argues that this is, in part, because our everyday (folk psychological) concept of belief attributes beliefs to others on the basis of presumed similarities to ourselves, rather than on the basis of strict identities. Thus Stich argues that when we say `S believes that p' what we mean is that S is in a belief state similar to the one which would play the typical causal role if my utterances of `p' had had a typical causal history (Stich 1983, p. 88). Furthermore, and more importantly in the present context, the content of someone's belief state is not specifiable independently of facts about the context in which the believer is situated. If this is so, the putative claim of a mentalist and individualist cognitive science to study beliefs would be obliterated and the claims of those who have championed the irreducibility of the social would be vindicated. What, then, are the arguments which lead Stich to accept the case for the context dependence of belief ascription?

Essentially, they are the arguments of Putnam 1975 and Burge 1978: arguments which are continuous with the Wittgensteinian private language arguments discussed elsewhere on this website in my essay "Wittgensteinians and Chomskyans". Their upshot is to place persons, beliefs and meaning in the public and social sphere. Here I will illustrate the general form of the argument with an example from Burge's influential paper, `Individualism and the Mental' (Burge 1978). Stich tidies up Burge's argument, but for present purposes I think we can work from it.

Burge invites us to contemplate a man who has and believes he has arthritis in the ankle, wrist, etc., waking up one day with a pain in his thigh and going off to the doctor to report that he believes his arthritis has spread to his thigh. No it hasn't, says the doctor, Arthritis is a disease of the joints. Compare this, says Burge, with a community where everything is the same except that doctors, etc., believe that you can have arthritis of the thigh. In other words, it's a community which has a different concept of arthritis from our one.

Burge's central idea is that, although the second community has a different concept of arthritis from us, the individual in the first community does not, despite his erroneous belief. It would be wrong, Burge thinks, to reconstruct the content of his thought in such a way that he comes out believing something other than that he has arthritis in his thigh. It would be wrong to ascribe to him a variant belief that he has tharthritis. If this is so, it follows that the meaning of `arthritis' is not in his head, since what is in his head is inconsistent with the meaning of `arthritis'. Further, the content of what is in his head is partly constituted by something outside his head in the community: we say he believes he has arthritis in his thigh, not tharthritis, and we do so in virtue of facts about the community to which he belongs, not about him. Burge says this is a common situation, and what he says relates to familiar theories in the literature on understanding other cultures: Gellner's defence of the right of natives to get it wrong, for instance (Gellner 1970); and `surplus of meaning' arguments in literary theory according to which we are obliged to mean more than we can possibly intend. Do Burge's arguments knock out all forms of cognitive theory which resolutely place meanings "in the head"?

Stich argues that the Putnam-Burge critique of individualism in semantics does indeed undermine cognitive theory if this theory takes the form of what he calls a representational theory of mind (RTM). An RTM can take two forms, either strong or weak. The weak theory assigns content to non - introspectible subpersonal mental representations. The strong theory additionally assigns a causal role to the content of such representations in the explanation of mental processes. Either kind of theory is undermined by the Putnam - Burge anti - individualist arguments. But cognitive theory need not take the form of an RTM. It can also take the form of a syntactic theory of mind (STM) which assigns to non - introspectible subpersonal representations only formal, syntactic properties: the organism has subdoxastic states, not beliefs (Such 1978; Evans 1978 develops similar ideas.)

An STM, like any cognitive theory, posits an intermediate level of reality between neurology and what one might call social phenomenology. It is still the sort of theory to which, for instance, Dreyfus would object: `conceptual confusion . . . results from trying to define a level of discourse between the physiological and the phenomenological' (Dreyfus 1979, p. 182). For, as Stich puts it,

The core idea of the STM - the idea that makes it syntactic - is that generalizations detailing causal relations among the hypothesized neurological states are to be specified indirectly via the formal relations among the syntactic objects to which the neurological state types are mapped. Similarly, generalizations specifying causal relations between stimuli and neurological states will identify the neurological states not by adverting to their essential neurological types but, rather, by adverting to the syntactic objects to which the neurological types are mapped. (Stich 1983, p. 151)

The virtue of an STM theory, for Stich, is that it makes no use of the vague and inherently social concepts of folk psychology. It is strictly a theory of the individual organism, which is what Stich thinks psychology must be (as does Fodor 1981, ch. 9, who also advocates an STM in response to Putnam - Burge type criticism). As Stich says `STM theories are able to characterize the cognitive states of a subject in terms appropriate to the subject rather than in terms that force a comparison between the subject and ourselves' (Stich 1983, p. 158).

All this adds up to the outline of a possible division of labour between the social scientist and the cognitive scientist. But if there is a division, there is also a connection. Just as syntactically specifiable states run on neurological states, as urged by both type and token physicalism, so beliefs run on syntactically specifiable states, though certainly not in a way which allows type-type correlation, rather in a way which allows only token-token correlation. Stich 1983 writes of the states postulated by cognitive theory as subserving the beliefs and desires ascribed to persons by folk psychology .

Now the claim I would wish to make is this: that though both the sub-personal and personal levels enjoy a relative autonomy of operation, the autonomous operation of sub - personal processes which I characterized at the outset as underdetermined and inaccessible to introspection has consequences for the kinds of beliefs which can plausibly be ascribed to subjects at the personal level and is, specifically, a cause of variation in plausibly ascribable beliefs. Expressed differently, we could say that though the content of a belief `ends up' being ascribed only at the personal and interpersonal level, the processes which fix belief are not all and not at all personal level processes, as is indicated by the fact that we cannot decide what to believe (cf. Williams 1973).

I do not know what Stich would make of this idea, but something like it needs to be true if the project sketched in this essay is to be defended against the serious threat posed to all cognitive theory by the anti - individualist arguments of Putnam, Burge, and others. Where the line between the personal and the subpersonal should be drawn, however, is far from clear. It is equally unclear whether the personal can be equated entirely with the social or cultural, and the subpersonal with the mental or natural.

For example, in cases of hysterical paralysis, the paralysed part of the body will correspond to a body part as defined in folk terms (arm,leg, etc.) not to a neurophysiologically distinguished part of the body (see Boden 1981, ch. 2). In this case, whether the description and explanation of the paralysis should be pursued in personal or subpersonal terms or both is unclear: the distinction does not solve any theoretical problems, though Stich 1983 claims that one of the virtues of an STM is that it allows for a unified explanatory theory which treats the mentally ill, along with children and members of exotic cultures, in the same way. The problem posed for Stich by the case of hysterical paralysis seems to me to be this: that the explanation of hysterical paralysis seems to involve necessary reference to a folk level concept (contentful, semantic, etc.) operating in a case (mental illness) where an STM approach (i.e. formal, syntactic) is supposed to be adequate.

Stephen Stich commented on the previous paragraph as follows: `I think there is an ambiguity in the interesting hysterical paralysis example sketched ... "The paralysed part of the body", you write, "will correspond to a body part as defined in folk terms." But now consider an example parallel to Burge's. Suppose that in one linguistic community the word "torso" includes the neck in its extension, while in another linguistic community it does not. Suppose further that in the second community there is a subject who (mistakenly) believes that the neck is included in the extension of "torso". Now, when this subject suffers hysterical paralysis because he believes his torso is evil (or whatever), what happens to his neck? Is the neck paralysed or not? Your reference to "a body part as defined in folk terms" leaves this unclear. It would be a real difficulty for a syntactic theory of the mind if this subject's neck was not paralysed, though his Burgean replica in the other society did have a paralysed neck. But as far as I know, there is no evidence at all that this would happen' (Personal communication 1985).

Again, to equate the personal with the social and cultural, so that the person becomes the interface which society creates in human organisms in order to control and exploit the workings of those organisms. seems to rule out a priori that a nativist account of the origins of personhood (on the lines, say, of Trevarthen 1979) should be true. In relation to Stich 1983, the problem is simply that though he discusses the nature of `believes' and `p' in the locution `S believes that p', he nowhere discusses the nature of S. To open up that issue is beyond the scope of this essay sufficient to say that while defenders of AI and cognitive theory such as Dennett claim that it has solved the homunculus problem (see Dennett 1979, ch. 7), it has had very little to say about the nature of the person, the subject which emerges from the hommelette. I pursue the topic elsewhere on this website, in the essay "The Subject and the Speaking Subject".


Andersen, H (1973) "Abductive and Deductive Change" in Language, vol 40, pp 765 - 793

Boden, M (1977) Artificial Intelligence and Natural Man. Hassocks: Harvester Press

Boden, M (1981) Minds and Mechanisms. Brighton: Harvester Press

Burge, T (1978) "Individualism and the Mental" in P French et al, eds, Midwest Studies in Philosophy vol IV, pp 73 - 121. University of Minnesota Press

Dennett, D (1979) Brainstorms. Hassocks: Harvester Press

Dreyfus, H (1979) What Computers Can't Do. New York: Harper and Row

Evans, G (1978) "Semantic Theory and Tacit Knowledge" in S Holtzman and C Leich, eds, Wittgenstein: To Follow a Rule, pp 118 - 137 London: Routledge and Kegan Paul Fodor, J (1981) Representations. Brighton: Harvester Press

Fodor, J (1983) The Modularity of Mind. Massachusetts Institute of Technology Press

Gellner, E (1970) "Concepts and Society" in D Emmett and A MacIntyre, eds, Sociological Theory and Philosophical Analysis pp 115 - 149. London: Macmillan

Giddens, A (1979) Central Problems in Social Theory. London: Macmillan

Grosz, G (1982) A Small Yes and a Big No. London: Allison and Busby

Hadjinicolau, N (1978) Art History and Class Struggle. London: Pluto Press

Homans, G (1964) "Bringing Men Back In" in American Sociological review, vol 29, pp 809 - 818

Lerdahl, F and Jackendoff, R (1983) A Generatve Theory of Tonal Music. Massachsuetts Institute of Technology Press

Putnam, H (1975) "The Meaning of 'Meaning'" in his Mind, Language and reality, pp 215 - 271. Cambridge University Press

Stich, S (1983) Beyond Belief: From Folk Psychology to Cognitive Science. Massachusetts Institute of Technology Press

Therborn, G (1980) The Ideology of Poweere and the Power of Ideology. London: New Left Books

Trevarrthen, C (1979) "Communication and Co operation in Early Infancy" in J Bullowa, ed, Before Speech pp 321 - 347. Cambridge University Press

Williams, B (1973) "Deciding to Believe" in his Problems of the Self, pp 136 - 151. Cambridge University Press

Wrong, D (1977) "The Oversocialized Conception of Man in Modern Sociology" in his Skeptical Sociology pp 31 - 54. London: Heinemann

First publication in this form on this website in 2005. Based on material in chapter 5 of my book, "Language in Mind and Language in Society" (Oxford University Press 1987), where even more Technical Stuff can be found.