Practical Irrationality, Reflexivity and Sartre's Regress Argument more

1 Practical Irrationality, Reflexivity and Sartre’s Regress Argument Alan Thomas Abstract This paper assesses Sartre’s critique of Freudian models of repression as inherently regressive. Sartre’s critique is itself criticized but demonstrated to raise fundamental issues about the nature of consciousness and the postulation of mental mechanisms in psychological explanation. Sartre’s argument involves the assumption that rational control over two thought contents requires a meta-representation over those contents, not simply their co-instantiation. It also seems to require, for the operation of a mental mechanism, that such a mechanism be explicitly represented not implemented. These assumptions are traced to the claim that mental states we are conscious with are those we are also conscious of. All three assumptions are rejected but are shown to involve a representative error about consciousness and rational control. Este papel valora la crítica de Sartre de modelos freudianos de la represión como intrínsecamente regresivo. La crítica de Sartre se es criticó pero demostró para levantar los asuntos fundamentales acerca de la naturaleza del conocimiento y la postulación de mecanismos mentales en la explicación psicológica. El argumento de Sartre implica la suposición ese control racional más de dos pensó que el contenido requiere una metarepresentación sobre esos contenido, no simplemente su co-instantiation. Parece también requerir, para la operación de un mecanismo mental, que tal mecanismo es representado explícitamente no es aplicado. Estas suposiciones son trazadas al reclamo de que estados mentales nosotros somos conscientes con son esos somos también conscientes. Las tres suposiciones son rechazadas pero son mostradas para implicar un error representativo acerca del conocimiento y el control racional. Cases of self-deception are a species of a wider genus, namely, the practically irrational. Understanding self-deception, therefore, depends on a satisfactory theoretical account of such irrationality, where the question is how such phenomena are even possible. Instances of practical irrationality are puzzling because, as Donald Davidson famously remarked, they constitute a “failure in the house of reason”. [Davidson, 1982] They exhibit a failure in rationality not a contrast between the rational and the irrational. A central capacity of the mind is exercised in such a way as to exhibit an internal defect. 2 The structural relationships between the usual kinds of mental state in rationally motivated action result, on a particular occasion, in a deviant outcome and the general question is how this can be so. My aim in this short paper is not, directly, to address the phenomena of practical irrationality as a whole or self-deception as a whole - both cover a very wide range of cases. It is, rather, to exhibit the dependence between how one views rationality in general and how one views those instances where rationality suffers from those internal defects that constitute ”failures in the house of reason”. My particular focus is on a famous argument of Sartre’s that Freud’s account of repression involved an infinite regress. [Sartre, 2003] Freud pioneered an explanation of how practical irrationally was possible that exploited the idea of a divided mind, understood in terms of functionally individuated discrete sub-systems within a single person. As a representative test case for this kind of explanatory strategy he suggested a distinctive treatment of repressed thoughts. Such thoughts are ascribed to an agent when an action is best explained as partly arising from a belief that an agent does not, and indeed cannot, avow. The agent’s actions are explained by the presence of such a belief, but it cannot be avowed first personally as its contents are so shameful or disgusting. The thought has been repressed in a way that Freud explained as dependent on the operation of a discrete mental mechanism. Sartre famously argued that this view was incoherent. In this short paper I want to examine Sartre’s argument both to exonerate him from the charge of simply committing the homuncular fallacy but also to suggest that the deep flaw in his argument is of representative significance. It involves a connection between the ideas of rational 3 control and models of consciousness that are surprisingly robust and yet open to serious challenge. 1 Sartre’s Critique of Freudian Repression The process of repression seems to involve, in Freud’s account, a division of the mind into discrete parts, separated in terms of what each can know and their mutual informational accessibility. [Freud, 1915/1991] In speaking of knowledge here one is describing the domain over which the discrete part of the mind operates; the class of thoughts over which it operates. Clearly, in functionally individuating a mechanism of repression, the kind of knowledge involved is knowledge of thoughts that are shameful or that the agent cannot acknowledge. That, however, is a criterion that may fail to individuate the mechanism concerned (as there are many shameful thoughts that an agent can avow, acknowledge or live with) so, following Peter Carruthers’ general style of explanation, it seems to me more helpful to individuate a mechanism of this kind in terms of what “switches on” its operation. [Carruthers, 2006] Carruthers borrows an example from Gabriel Segal to explain the general idea: you cannot help interpreting the mental life of an actor on stage even though she is only pretending to be, say, Lady Macbeth as one’s mind-reading module is switched on by the task of interpreting the mental lives of others. The operation of that module is started by inputs of the right kind. Similarly, I suggest, we can pick out a Freudian mechanism of repression by what starts its operation: the presence of shameful thoughts that cannot be acknowledged. Postulating such a mechanism solves the paradox that, in the case of a repressed thought, an agent disavows conscious awareness of a thought that the interpretative 4 process attributes to that agent as causally influencing his or her actions. The agent seems both sensitive to that reason, as exhibited in interpreted action, and not sensitive to it, in the sense of not being consciously aware of it. For example, she would not avow it and would, on being confronted with the interpretation of her actions that ascribed it, disavow it. Sartre challenged any such model by arguing that it involved the regressive postulation of an internal censor who must, in order to repress undesired thoughts, be aware of both. But he argued that that is to reproduce, within each rational agent, a functional mechanism with all the relevant features of the whole agent, hence, it involves an infinite regress that cannot be explanatory. This is an early instance of the accusation that a cognitivist model of the mind involves a homuncular fallacy: properties of people are applied, inappropriately, to sub-parts of people. [Kenny, 1871/1984] Freud has not solved the original problem: he has simply re-located it, but in doing so has achieved no more than to re-describe it. Sartre argued that Freud’s postulated inner mechanism fails as an attempt to explain the paradox of the influence on action of repressed thought as it has the same features of the initial paradoxical description. The mechanism must be both aware and not aware of one and the same thought. It must be aware of it to classify it as repressed, but not aware of it, as the thought is (ex hypothesi) repressed. This claim can be detached from some of Sartre’s distinctive reasons for asserting it, such as his belief in the coextensiveness of mentality, intentionality and consciousness and his further belief that every conscious mental state is “pre-reflectively” aware of itself in a way that is distinct from propositionally expressed self-knowledge. [Zahavi, 1999; Wider, 1997] However, it 5 cannot be detached from a corollary of these views, namely, the connection implicit in Sartre’s approach to mentality between consciousness and rational control. Sartre’s critique of Freud succeeds only if one adopts a certain view of the nature of the rational control of thought. While this view seems widely accepted it also seems to me clearly mistaken and so Sartre’s error has, as it were, representative significance. It is clear from Sartre’s argument that the functioning of the internal “censor” is understood as involving, in microcosm, the rational control of thoughts exhibited by rational agents as a whole, taken in macrocosm. But that exhibits quite clearly why it is Sartre’s argument that is regressive, not Freud’s. Whole agents have a capacity for self-conscious rational deliberation. But that capacity is developed from a rational control of thought that is neither self-conscious nor deliberative. Rational control of thought is an implementation of rationality, not a representation of it, a point made clearly in Lewis Carroll’s famous argument in ‘What the Tortoise Said to Achilles’. [Carroll/Dodgson 1895] If you assume that in order to implement a rational process a rational system must represent that process, you launch yourself on an infinite regress of representation. [Dennett, 1985] A rational agent that implements the capacity to reason in accordance with the modus ponens rule, for example, does not do so by representing within itself that rule or via explicitly representing that capacity. Of course, it may go on explicitly to represent itself as so thinking “in accordance” with modus ponens, but that is a derived and meta-level capacity that cannot, on pain of regress, be that which implements the capacity in the first place. This distinction is particularly clear in the case of those reasoned changes in view that implement logical principles, given that presumably no process of checking could dispense with those very principles themselves. However, the point generalizes to 6 rational control as a whole: a thinker can display grasp of reasoned changes in view by his or her systematic responses to changes in available evidence, without our having to ascribe explicit grasp of the principles of revision to that agent/believer. 2 Consciousness and Rational Control The seductive line of reasoning in Sartre that leads to his misguided critique of Freud is one that is detectable in the work of several philosophers writing about consciousness and rationality. [Siewert, 2000, ch. 6] The argument typically proceeds as follows. Inference is primarily conscious inference, in that a person may only be committed to inconsistent beliefs if he or she fails to be aware of an inconsistency in belief. Conscious awareness cannot tolerate such inconsistency and when a person becomes aware of holding inconsistent beliefs he or she revises his or her belief set to eliminate it. If one is consciously aware with a mental state, that state is one that the subject is consciously aware of being in. So the rational control of thought involves a conscious mental subject being aware of two mental states and of the rational relations between them, such that a process of inference can be applied to the embedded intentional content of the two states if that content is truth evaluable. Summarily: rational control involves metarepresentation. However, if you are committed to the thesis that if you are consciously aware with a mental state then you are conscious of it, then consciousness, too, involves metarepresentation. Two lines of argument seem to converge on a single best explanation of core features of our mental life: both a constitutive account of consciousness and a plausible model of rational control can be explained by the mind’s capacity for metarepresentation. 7 However, both of these arguments are mistaken. The mis-step here is in the claim that “if one is consciously aware with a mental state, that state is one that the subject is consciously aware of being in”. Fred Dretske has convincingly argued that it is a non sequitur. [Dretske, 1995] Conscious mental states are states with which a person makes herself conscious of other things and they can discharge this role without themselves being the object of some further mental state. Sartre, as is well known, argues that while a certain kind of reflective self-knowledge of conscious mental states is possible, that higher order awareness depends on the fact that the “first order” state which is its object is already conscious. It is already conscious in two ways: it consciously and intentionally represents an object and contains an implicit self-awareness. But Sartre’s argument for that last claim, that such a state is implicitly self-aware, is woefully thin. He argues that if a first order conscious state were not implicitly self-aware, he argues, then it would be an “unconscious” state, which is a nonsense. But Jay Rosenberg levels against this argument precisely Dretske’s point: the first order conscious state is still conscious in the sense that it consciously represents its intentional object. [Rosenberg, 1981] Sartre’s argument only goes through on the assumption that those states we are conscious with are states that we are conscious of, and that is precisely the point at issue. This claim about consciousness, then, that conscious mental states are states of which one is conscious, combined with the claim that the rational control of thought involves its explicit representation to the agent, leads to a conception of rational control as involving reflexive self-scanning: a model of “rationality as reflexivity”. The connection between a constitutive account of consciousness and an account of the rational endorsement of thought is so close that Sartre can treat them as mutually re- 8 inforcing and on that basis develop his critique of Freud. But that critique is a reductio ad absurdum of Sartre’s assumptions. Both of the assumptions underlying his argument against Freud’s postulated mechanism of repression are equally open to question. A corollary of Sartre’s rationality as reflexivity argument is that, if it is correct, it is going to be very difficult to distinguish rationality, mentality and consciousness for reasons independent of his own bifuraction of reality into that which is “in itself” (reality) and that which is “for itself” (conscious and subjective reality). The idea that the mind’s contents are fully transparent to itself, or that all mental states are conscious, is usually rejected on the ground that such a view exemplifies a discredited Cartesianism about the mental. Is it not clear that we can distinguish rational animals from those with mental states, and mental subjects from conscious subjects? Indeed, such distinctions are clear and important: rationality and mentality are more widespread phenomena throughout the natural world than consciousness. But those philosophers who are committed to the rationality as reflexivity view undermine this attractive view from within: if rational control involves reflexive self-scanning, then it requires an analogue of selfconsciousness. But if self-consciousness is clearly anything, it is clearly a sub-part of consciousness. One could hardly deny that a self-conscious mental state is at least conscious.1 [Thomas, 2003] Thus any agent with rational control over its mental states is going to have to be conscious, surely a very implausible and unwelcome result given that our pre-theoretical intuitions are that rational mentality is found far more widely in nature than consciousness. We can agree with Bennett that simple honeybees are not rational agents who act in accord with rules even if they exhibit regularity. [Bennett, 1989] However, it does not take a substantial further extension of the capacities of this 9 uncomplicated species to interpret it as implementing rationality. [Collins, 1968; Brooks, 1991] However, no one defends the very implausible view that insects slightly more complicated than honeybees are candidates for the ascriptions of consciousness (as opposed to rational mentality). Sartre’s views come as a package: accept his critique of Freud and you are driven to accept his wider views about the relation between rationality, consciousness and mentality. But, once again, this ought to be viewed as a reductio of his critique of Feud. 3 Representation and Homuncularity We are the most sophisticated and cognitively developed form of rational creature; our higher capacities of thought disguise from us their more mundane origins, origins more continuous with the phenomena of rationality found more widely in the natural world. Sartre takes our highest capacities for fully self-conscious rational deliberation and applies that model to the postulated inner censor of Freud’s mechanism of repression, in order to generate his purported regress argument. But if all thought involved the explicit representation to itself of thought contents and their relations, then thinking, itself, would be launched on an infinite regress. At some point a distinction has to be introduced between the contents of thoughts and the cognitive architecture of thoughts, described in functional terms. Cross classifying any account of the contents of a mind we also need an itemization of its capacities, understood as psychologically real powers. [Geach, 1964] Freud’s mechanism of repression is an implementation of a cognitive mechanism, not a representation of it. It does not consciously represent to itself the repressed thought, in order to repress it. When we interpret its function, we represent it as both aware and 10 not aware of one and the same thought but that is, as it were, an artefact of our interpretative stance. The mechanism of repression functions to sort thoughts into one of two different “locations”, placing some thoughts in a functional category where they are not available to conscious awareness without interpretative redescription, and where their relation to other thoughts is associative, rather than rational. But these metaphors of place simply are metaphors for functional categories. Like a set of points switching train trucks into one of two sidings, we have one functional device that sorts thoughts into two functional categories: the access conscious and the non-access conscious. [Block, 1997] In Block’s terminology, a mental state is access conscious if it is poised to be used for the direct rational control of thought and action. It is important to the defensibility of this explanation that some of the theoretical baggage attaching to such terms as “functional mechanism” is disavowed. The postulated functional mechanism is psychologically real, causally efficacious and rational in this sense: it operates over rational contents. The beliefs that it sorts are equally real and causally efficacious in Davidson’s original sense that reasons simply are causes. But its rationality is, precisely, borrowed from ascriptions of rationality to the whole agent: its task is simply, as it were, to operate on contents. But that contrast is precisely how Freud dissolves the apparent paradox attaching to the motivated irrationality of repression. To appreciate that point one must ask the obvious question: must not such a mechanism be sensitive to some feature of the thoughts that it sorts, in order to explain why some are repressed and some are not? Is that not all we need to generate Sartre’s paradox? I think that question betrays a misunderstanding: the sorting process itself is the reason why some thoughts are repressed and some are not. The mechanism of repression 11 is interpretable as a rational, content sensitive, process. That is why its operations are within the house of reason. But its process of sorting is that which repression consists in, and has no further explanation – that is why it is both functional and the operation of a mechanism. That is why it is a failure in the house of reason. (This is not for the “blanket” and very unhelpful reason that it is a causal mechanism, but because it supplies the wrong kind of cause: the operation of repressed thought in generating actions is associational, not rational.) It is for those reasons that, far from being paradoxical, the postulation of such a mechanism meets precisely the criteria required to instantiate a paradox of irrationality. The answer to the sceptic’s question is that it is no more problematic to explain a mechanism such as repression as switched on by the presence of shameful thoughts that the agent cannot tolerate within conscious awareness than it is problematic to explain one’s mind reading module as switched on by the actions of intentional agents, primarily people. 4 Original and Derived Intentionality A crosscutting issue that obscures the argument I have put forward and makes it more difficult to appreciate is between original and derived intentionality and rationality. For example, one can describe a logic gate in a computer as implementing the logical connective, disjunction. Clearly, the computer implements disjunction without representing it to itself. (Some of its operations may be represented within itself in the sense that the system as a whole may carry reflexive information about its own information processing but we can take if for the simple case of the implementation of a logic gate that is not the case.) But when we discuss the operations of this computer, 12 given that we know that rationality and intentionality of the machine is ascribed by us and that it does not possess these features intrinsically, we represent that logic gate to ourselves using the term “or”. We both ascribe intentionality and rationality and we name those features of the computer that implement those features. People, however, are different (as are those mental subjects relevantly like them.) We implement intrinsic or original or underived intentional mental states that stand in rational relations. In the case of mental subjects like us we both implement intentional and rational arrangements, without ascribing them (we do not need to ascribe them as they are already there) and describe them. That gives us a set of homonymic expressions: “deciding on the next move”, for example, uses a term, “deciding”, that we might apply to chess computer or a human playing chess. But these are homonyms, referring to ascribed (derived) intentional states and rational relations in one case and original (underived) states and relations in the other. But it would be a mistake to infer from these facts that implementation in any way depends on representation. Implementation depends on representation in the case of the chess computer in the trivial sense that its electro-mechanical operations are what they are, but that they only become “deciding on the next move” by virtue of our ascribing that state to it and also representing it in that way. That makes it look, in this case, as thought the implementation is a shadow of the grammar of mental ascriptions. But that case is disanalogous to that of people who possess original intentionality: in our case, we implement rationality and intentional mental states without having to represent them to ourselves or to anyone else. To argue that implementation depends on representation is to run together three issues, one about rational control, one about how ascriptions operate and one about original versus derived intentionality, all of which need to kept apart. 13 Conclusion The interaction between the macro-level of the whole agent and the micro-level of discrete functional mechanisms within the cognitive architecture of the agent explains the distinctive strength of Freud’s explanation and is not, as Sartre claims, its central defect. The model of rationality as involving reflexive control analogous to self-conscious deliberation, which fundamentally motivates Sartre’s critique, is the assumption that needs to be questioned. It would be a further problem for such a view of rationality if it thereby cuts itself off from the kind of explanation of the paradoxes of irrationality that Freud pioneered.2 List of Works Cited BENNETT, Jonathan [1989] Rationality, Hackett Publishing. BLOCK, Ned [1997] ‘On a Confusion about a Function of Consciousness’, in BLOCK. N. & FLANAGAN. O. & GUZELDERE. G., [1997] BLOCK, N & FLANAGAN O & GUZELDERE G. (eds) [1997] The Nature of Consciousness, M. I. T. Press. BROOKS, Rodney [191] ‘Intelligence Without Representation’, Artificial Intelligence, 47, pp. 139-159. CARRUTHERS, Peter [2006] The Architecture of the Mind, Oxford University Press. CARROLL, Lewis (Charles Lutwidge Dodgson) [1895] ‘What the Tortoise Said to Achilles’, Mind, (NS), vol. IV, No. 14, April, pp 278-280. 14 COLLINS, Arthur [1968] ‘How One Could Tell Were a Bee to Guide his Behaviour by a Rule’, Mind, vol. 77 no. 308, pp. 556-560. DAVIDSON, Donald [1982] ‘Paradoxes of Irrationality’, reprinted in WOLLHEIM, R & HOPKINS, J (eds) [1982] pp. 289-305. DENNETT, Daniel [1985] Brainstorms, M. I. T. Press. FREUD, Sigmund [1915] ‘On Repression’, reprinted in Freud [1991] FREUD, Sigmund [1991] On Metapsychology, Penguin Freud Library. GEACH, Peter [1964] Mental Acts, Routledge. KENNY, Anthony [1971/1984] ‘The Homuncular Fallacy’, reprinted in The Legacy of Wittgenstein, Basil Blackwell. ROSENBERG, Jay [1981] ‘Apperception and Sartre’s “Pre-Reflective Cogito”’, American Philosophical Quarterly. SARTRE, Jean-Paul [2003] Being and Nothingness, trans Hazel Barnes, Routledge Classics, Routledge. SIEWERT, Charles [2000] The Significance of Consciousness, Princeton University Press. THOMAS, Alan [2003] ‘An Adverbial Theory of Consciousness’, Phenomenology and the Cognitive Sciences, vol 2, no 3, pp. 161 – 185. THOMAS, Alan [2004] ‘Reconciling Conscious Absorption and the Ubiquity of SelfAwareness’, seminar presentation, CUNY. WIDER, Kathleen [1997] The Bodily Nature of Consciousness: Sartre and Contemporary Philosophy of Mind, Cornell University Press. WOLLHEIM, Richard & HOPKINS, James [1982] Philosophical Essays on Freud, Cambridge University Press. 15 ZAHAVI, Dan [1999] Self-Awareness and Alterity: A Phenomenological Investigation, Northwestern University Press. 1 This is separate from the argument that self-consciousness must be more than consciousness, perhaps, consciousness plus a capacity to represent thoughts to oneself in a distinctively de se way. That argument would make it even harder to argue that when, for example, one ascribes rationality to honey bees one thereby ascribes a capacity on the part of honeybees to represent their mental states to themselves. 2 Thanks for discussion of this paper to Edward Harcourt and to Kathryn Brown.