(Presented as an invited paper for the Philosophy of Psychology session at the Twentieth World Congress of Philosophy, Boston, August 10, 1998.
Forthcoming in the Proceedings of the 20th World Congress of Philosophy, Boston, 1998.
To appear also in the journal Philosophy and Phenomenological Research.
To be published as well in the Proceedings of the International Conference "Freud at the Threshold of the 21st Century," held in Jerusalem, Israel, December 1999.)

In this paper, I criticize the claim made by Donald Davidson, among others, that Freud's psychoanalytic theory provides "a conceptual framework within which to describe and understand irrationality." Further, I defend my epistemological strictures on the explanatory and therapeutic foundations of the psychoanalytic enterprise against the efforts of Davidson, Marcia Cavell, Thomas Nagel, et al., to undermine them.

INTRODUCTION

In concert with a significant number of other philosophers, Donald Davidson (1982, p. 290) has told us that "Psychoanalytic Theory as developed by Freud claims to provide a conceptual framework within which to describe and understand irrationality." And he explains that "the sort of irrationality that makes conceptual trouble is . . . the failure, within a single person, of coherence or consistency in the pattern of beliefs, attitudes, emotions, intentions, and actions. Examples are wishful thinking, acting contrary to one's own best judgment [Aristotle's "akrasia"], self-deception, believing something that one holds to be discredited by the weight of the evidence." Wishful thinking, in turn, Davidson characterizes as "a model for the simplest kind of irrationality" (1982, p. 298). Thus, if the entire explanation of a person's belief is that it is wish-fulfilling to him/her, "then his holding the belief is irrational" (ibid.). More generally, Davidson writes:

In the cases of irrationality we have been discussing, there is a mental cause that is not a reason [in the sense of cogent evidence] for what it causes. So in wishful thinking, a desire causes a belief. But the judgment that a state of affairs is, or would be desirable is not a reason to believe that it exists (ibid.).

As for self-deception, Sebastian Gardner (1993, p. 6) elaborates on Davidson, saying: "…. in self-deception one subsystem contains the buried belief, another the promoted belief; the line between them reflects the impossibility of both beliefs being true." Incidentally, last year, Alfred Mele (1997, p. 91) challenged this depiction of self-deception as characteristically involving a contradiction: "The assumption that…. the self-deceiver believes that some proposition is true while also believing that it is false ….. produces a fundamentally mistaken view of the dynamics of self-deception. Self-deception is explicable without the assistance of mental exotica." And Erwin (1997a, p. 56) has rightly pointed out that, for any conceptual analysis of the notion of self-deception that requires the existence of a specified structure of the mind (e.g., a split self), empirical evidence is required to validate the entailed existential claim.

Proceeding from the mere description of important kinds of irrationality to their potential explanation, Davidson (1982, pp. 291-292) writes:

In attempting to explain such phenomena (along with much more, of course) Freudians have made the following claims:

First, the mind contains a number of semi-independent structures, these structures being characterized by mental attributes like thoughts, desires, and memories.

Second, parts of the mind are in important respects like people, not only in having (or consisting of) beliefs, wants and other psychological traits, but in that these factors can combine, as in intentional action, to cause further events in the mind or outside it,

Third, some of the dispositions, attitudes, and events that characterize the various substructures in the mind must be viewed on the model of physical dispositions and forces when they affect, or are affected by other substructures in the mind [e.g., Freud described the ego as exerting "a censoring force" during repression].

I shall refer to these three contentions as "Davidson'sTriplet." As he tells us (1982, p. 300, n. 6), its salient feature is "the idea of mental compartmentalization." Davidson (1982, p. 303) offers his rationale for claiming that each of these three "very general" features of psychoanalytic theory is required by any theory aiming to explain irrationality.

As to the first, which compartmentalizes the mind into two or more semi-autonomous structures, we learn that this division is "necessary to account for mental causes that are not [also] reasons for the mental states they cause" (1982, p. 303).

The second feature assigns to one or more subdivisions of the mind a structure similar to the one featured by Aristotle's practical syllogism, which is the pattern of explaining ordinary actions as rational: An action A is held to be carried out because the agent aims to achieve a goal G and believes that A will issue in the realization of G. Davidson calls this pattern "the principle of intentional action." And the essential point is that the ideas and affects of a person are presumed to interact so as to produce consequences in accord with the principle of intentional actions.

Yet these consequences, in turn, then serve as causes, but not reasons, for further mental events, such that "The breakdown of reason-relations defines the boundary of a subdivision….. The parts are defined in terms of function; ultimately, in terms of the concepts of reason and cause. The idea of a quasi-autonomous division is not one that demands a little agent in the division; again, the operative concepts are those of cause and reason" (1982, p. 304).

But, if so, why did Davidson (1982, p. 290) initially formulate his second feature by declaring that "parts of the mind are in important respects like people," a formulation that smacks of Freud's unfortunate homuncular reifications in his structural theory of the id, ego, and superego?

The third feature was prompted by Davidson's view (1982, p. 304) that "certain mental events take on the character of mere causes [i.e., causes that are not also reasons] relative to some other events in the same mind. This feature also we found to be required by any account of irrationality . . . in order to accommodate it we must allow a degree of autonomy to parts of the mind." He speaks of such mere causation by mental events as "blind" (1982, pp. 292, 299), since it operates "on the model of physical dispositions and forces" (1982, pp. 290-291).

Very remarkably, on the one hand, Davidson keeps insisting on the "centrality" of his triplet to Freud's edifice, but on the other hand, he writes: "my highly abstract account of the partitioning of the mind [in the triplet] deviates from Freud's. In particular, I have nothing to say about the number or nature of divisions of the mind, their permanence or aetiology. I am solely concerned to defend the idea of mental compartmentalization, and to argue that it is necessary if we are to explain a common form of irrationality" (1982, p. 300, n. 6, my italics). But this disclaimer alone already shows that Davidson's case for construing his triplet as being "central" to, or emblematic of, Freud's mental partitions dangles precariously. It will turn out that Davidson construes the centrality of a set S of hypotheses to a theory T to the effect that S is a distinctive feature of T.

II. Critique of Davidson's Freudian Plaidoyer

Importantly, Davidson (1982, p. 290) characterizes the supposed logical status of his triplet within Freudian theory as follows: They are elements in Freud's thought, "elements that consist of a few very general doctrines central to all stages of Freud's mature writings" [italics added], and he adds:"I conclude that any satisfactory [explanatory] view [of irrationality] must embrace some of Freud's most important theses" [italics added]. By way of clarification, he issues acaveat:"It perhaps needs to be emphasized that my "defense" of Freud is directed to some only of Freud's ideas, and that these ideas are at the conceptual, in contrast to the empirical, end of that vague spectrum."

Davidson is not content to point out that his triplet is "found" in Freud's theoretical edifice. The thrust of his insistence on its supposed centrality therein and on its explanatory indispensability is to confer distinctive explanatory merit on psychoanalytic theory as an account of irrationality: "I hope it will be agreed that these doctrines are all to be found in Freud, and that they are central to his theories" (1982, p. 291). Davidson thus asserts centrality, even though in the very next sentence, he reiterates correctly that the members of his triplet are logically "far less strong and detailed than Freud's views."

The notion that a particular hypothesis, or specific hypotheses, is "central" to a theoretical system is at the very core of Davidson's recurring explanatory tribute to Freud's structural theory of the mental apparatus. He apparently reasoned that it deserved this encomium, because Freud's highly specific compartmentalization of the mind entails the logically far weaker triplet, which Davidson had teased out from the topmost hallmark principles of Freud's successive bipartite and tripartite models of the mind. And, as we saw, Davidson had argued, in turn, that his triplet is indispensable to any theory aiming to explain irrationality.

Alas, he seems to take the notion of centrality for granted, offering no articulation of his construal of it. More seriously, he leaves us completely in the dark as to how he proposes to license the cardinal inference of his Freudian plaidoyer. That crucial inference starts from the sound premise that the compartmentalization of the mind featured by Freud's successive structural theories is "central" to them in the sense of being their hallmark vis-à-vis rival theories.

Give this premise, Davidson believes he can validly infer that his logically much weaker, "very general," and "highly abstract" triplet should also be deemed "central" to Freud's theoretical edifice!

Yet, absent any statement from him of a putative license for this key inference, its conclusion is logically ill-founded.

If Davidson had in mind a cogent sophisticated licensing rationale for his vital inference, he presumably would have–and surely should have–stated it explicitly. To assume that he did have such a valid rationale, but simply left it unstated, strains charity beyond the breaking point. And to suppose, furthermore, that Davidson knowingly did himself the disservice of silence strikes me as preposterous. Thus, let me reconstruct his reasoning accordingly, while being mindful that the responsibility for the explicative and probative gaps in Davidson's argument belongs squarely on his shoulders, not on mine.

For argument's sake, let us grant Davidson (1982, pp. 303-304) that any satisfactory account of certain important species of irrationality must indeed feature his triplet. Then I claim that his characterization of it as "central" to Freud's edifice rests on a specific fallacious inference. Moreover, its conclusion is refuted by the fact that the triplet is simply not distinctively Freudian, since it is likewise instantiated by an array of diverse philosophies of mind in the history of Western philosophy.

Thus, Davidson's arrogation of explanatory merit for understanding irrationality to Freudian theory as such, merely because it affirms the triplet, will turn out to be a non sequitur with a false conclusion.

I shall argue that two sets of errors undermine Davidson's case:

(i) As I shall explain, the "central" theses of a theory are presumably, in the first instance, those of its fundamental postulates that constitute its distinctive hallmarks vis-à-vis different or rival theories pertaining to the same domain of explananda. Yet, without any supporting argument, Davidson seems to have reasoned fallaciously that the centrality of the distinctive hallmarks of a theory is preserved under logical deduction. That is to say, he infers without ado that, in the case of Freud's structural compartmentalization of the mental apparatus, centrality is deductively inherited from psychoanalytic hallmark postulates by his avowedly much less specific triplet. Yet Davidson reiterates that the members of the triplet "are . . . far less strong and detailed than Freud's views" (1982, p. 291). Furthermore he emphasizes that his triplet-based "theory [of irrationality] is acceptable" (1982, p. 304) without any assumption of unconscious components of the mind.

I contend, however, that the triplet itself is far from central to Freud's theory.

(ii) Davidson's belief in the deductive inheritance of centrality or hallmark-status seems false in the face of the venerable history of pre-Freudian philosophies of mind that differ in content from Freud's, and yet instantiate Davidson's triplet no less than psychoanalytic theory does. Major historical cases in point are the theories of the soul offered by Plato and Aristotle, as well as the faculty psychology of Leibniz's disciple Christian Wolff.

For example, Plato offers a tripartite structure of the soul akin to Freud's id, ego, and superego. And a residue of eighteenth century faculty psychology is even found in Jerry Fodor's (1983) account of brain organization, which features many "modules" of localized cells, each of which carries out a particular function, such as face-recognition.

Davidson cannot parry this historical objection by having distanced himself from the respective mental partitions avowed by the two historical doctrines he mentions. He rejects them as follows: "The partitioning I propose does not correspond in nature or function to the ancient metaphor of a battle between Virtue and Temptation or Reason and Passion. For the competing desires or values which akrasia demands do not, on my account, in themselves suggest irrationality" (1982, p. 301).

But his articulation of the flesh that he put on the mere bones of his triplet in his own explanation of irrationality clearly goes well beyond the triplet, whose centrality to Freudian theory is at issue here. Hence the details of his own partitioning cannot ward off the historical objection that the triplet is multiply instantiated in an array of otherwise diverse philosophies of mind, rather than being a badge of Freud's structural conceptions.

Now let me justify basically my statement that Davidson's peremptory assumption of the deductive heritability of centrality is fallacious.

What does he mean, in the first instance, when he says that a hypothesis H is "central" to a theory T, or that H counts as one of T's "core" hypotheses, or yet that H is one of T's "most important theses"? This question requires a precise answer. My answer will serve to reject Davidson's claim of the centrality of his triplet to Freudian theory. And thereby I shall undermine his special explanatory tribute to Freud's theory as an account of irrationality.

Following Tarski, I shall use the term "the consequence class of a theory T" to designate the set of all deductive consequences of T. Now, surely a statement S's meremembership of the consequence class of T does not make S"central" to T. It does not do so, if only because that class has some members which have watered down much of the content of T's axiomatic base, as in Davidson's triplet. And Davidson (1982, p. 300) himself characterizes the triplet as "highly abstract."

To render the intent of his statements containing the locution "central," or its equivalent, centrality apparently needs to be construed, in the first instance, as a property of those hypotheses that are characteristic hallmarks of T and identifyit by distinguishing it from its known rivals or from different theories. In this sense of "central," a hypothesis H can bepeculiar to T by being incompatible with T's rivals, or because one or more such rivals do not enunciate H.

Thus construed, Euclid's parallel postulate, for example, would be "central" to Euclidean geometry, because it is incompatible with both hyperbolic and spherical non-Euclidean geometry. On the other hand, Euclid's first four postulates do not distinguish his geometry from hyperbolic non-Euclidean geometry, although these postulates do collectively distinguish both of these geometries from spherical non-Euclidean geometry. Clearly, the notion of centrality requires refined statement, but no such nuanced formulation is found in Davidson's paper.

As we saw, he acknowledges (1982, p. 291) that his triplet of logical consequences of Freud's own formulations is "far less strong and detailed than Freud's views." How then does he reason that, despite this much weaker logical status, the triplet is nonetheless "central" to Freud's psychoanalytic edifice, presumably in the sense of still being distinctive to it vis-à-vis other theories?

Absent any statement by him as to a licensing rationale, I cannot escape the conclusion that Davidson relied on the following inferential principle: If H is a distinctive or "central" hypothesis of a theory T, then so is any of its deductive consequences, however much weaker than H. Note at once that this principle is not licensed by the following elementary logical fact: If any logical consequence of H, however weak, were false, then–by modus tollens–H itself and indeed T would also be false. Clearly this obvious fact does not vouchsafe that the centrality of a hypothesis to a theory is preserved under logical deduction. But this principle is untenable, if only because it would just trivialize the relevant notion of centrality to regard even some of the weakest, most general, nondistinctive abstract logical consequences of the hallmark postulates of T as central to it!

In short, Davidson reasoned fallaciously that Freud'sdistinctive compartmentalization of the mind into the ego, id, and superego licences the conclusion that the very abstract triplet is likewise distinctive to Freud's theory. This conclusion is false.

I have argued that Davidson's insistence on Freud à propos of the triplet is, alas, a case of special pleading: As I see it, Davidson should have been content to let his own theory of irrationality stand on its own feet, along with noting that the triplet on which it relies is likewise a logically very weak deductive consequence of Freud's psychoanalytic partition of the mind, whereby Freud also has the resources to explain irrationality in Davidson's sense. Yet paying tribute to the ill-begotten centrality of the triplet to Freud's theory is a recurring item on the agenda of Davidson's essay "The Paradoxes of Irrationality," which evolved from his 1978 Ernest Jones Lecture to the British Psycho-analytical Association.

III. Critique of Cavell and Nagel

I now turn to Marcia Cavell's and Thomas Nagel's epistemological assimilation of psychoanalytic explanations of irrationality to the practical syllogism, qua purported "Extensions of Desire-cum-Belief Explanations of Actions." In her book, The Psychoanalytic Mind (1993, pp. 79-80), Cavell contests my thesis (Grünbaum 1984, 1993) that Freud failed to provided cogent evidence for his cardinal hypothesis that repression is causally necessary for neurosis, and that undoing it is therapeutic. Along with such other philosophers as Donald Davidson and Thomas Nagel (1994a, 1994b), she claims that my epistemological challenge is met by realizing that "the causal connection between repression and symptoms is fundamentally the same–though more complex and less obvious–as that between desire and action, and action and belief " (1993, p. 80). In short, Cavell asserts that psychoanalytic causal explanations of neurotic symptoms, in Freud's broad sense of including slips and manifest dream content, areextensions of commonsense folk psychology, because they too conform to the pattern of the practical syllogism (hereafter "PS").

That pattern features a desire-cum-belief set, which is held to cause an action. Thus, Cavell speaks of an action as "fully intentional" and of the PS as "the reason-explanation model" (1993, p. 179). Yet later she allows that "No one model will fit all the cases of repression, any more than it will self-deception," explaining that she has focused on the large species of cases that pertain to psychoanalytic treatment and do conform to the PS model (1993, p. 186). As we know, the psychoanalysis of patients draws on each of the three major branches of Freud's cornerstone theory of repression: psychopathology, dreaming, and slips.

Now, as a basis for contesting my epistemological challenge, Cavell enlists the alleged PS-pattern of Freud's repression-etiologies. Speaking of the putative applicability of that explanatory pattern, she maintains: "If this were so, then no inductive evidence of some special kind would be needed to establish a causal connection in any special case" (1993, p. 80). But she does not tell us at this point how this schema is to validate the universal generalizations that are the hallmark of the three pillars of Freud's theory of repression. Presumably, she intends the kind of induction by enumeration on which Thomas Nagel (1994b, p. 56) relied, when, in his retort to me, he invoked the intuitive credibility ofsingular psychoanalytic causal attributions:

intuitive plausibility . . . necessarily applies in the first instance to specific explanations, rather than general principles . . . confirmation goes from the particular to the general; the general theory of repression and psychosexual development has to be supported by its individual instances, rather than the reverse.

Yet in a forceful and cogent counter-argument, Erwin (1996, pp. 13-19, and further pp. 106-107, 26-40, and 122-123) concludes: "Nagel's principle [of confirmation] is generally insufficient to justify singular Freudian causal inferences. The reason is that one or more of the required empirical presuppositions are generally unmet" (1996, p. 18).

Below I shall deny the general assimilability of typical psychoanalytic explanations of neurotic symptoms, dreaming or bungled actions to the PS. But, before I do so, I need to stress that, even if there were such assimilability, these explanations are nipped in the bud epistemically, because Freud's imputations of his hypothesized repressed desires have no cogent independent support (Grünbaum 1984, 1986, 1993). Among other difficulties, the method of "free association," I maintain, is not causally probative.

Moreover, in ordinary applications of the PS to explain many of our actions, such motives as wanting to achieve a good family life, economic security, or a good reputation are independently and multiply attested. But, for example, as I have argued in my writings, Freud's imputation of strongly repressed homosexual desires to paranoiacs (1911, 1915, 1922) is sorely devoid of cogent evidence. For these reasons alone, the characterization of psychoanalytic explanations as "extensions" of reason-explanations of actions à la Thomas Nagel, Marcia Cavell, et al., simply begs the epistemological question.

So much for a preparatory caveat.

Ironically, Richard Wollheim, Nagel's and Cavell's stoutly pro-Freudian confrere, explicitly rejects the purely enumerative inductive stance championed by Nagel, and presumably also by Cavell, without mention of these authors. Wollheim (1993, Chap. VI) presents his rival account of psychoanalytic confirmation in a chapter entitled "Desire, Belief, and Professor Grünbaum's Freud." Departing from Nagel's supposedly intuitive confirmation of individual psychoanalytic explanations as credible extensions of commonsense understanding, Wollheim (1993, pp. 104-110) deems clinical testing in the psychoanalytic treatment sessions fitting in cases such as the pathogenesis of the Rat Man's obsessions.

Yet, Wollheim unwittingly concedes one of my principal epistemological animadversions against Freudian theory (Grünbaum 1984, p. 278) when he declares (1993, p. 109):

In the first place, if psychoanalytic hypotheses are to be clinically tested, then we must assume, at least for the duration of the test, certain more general psychological principles for which the hope must be that eventually tests will be devised, presumably of an extra-clinical kind, that will establish them. These principles will include such things as symptom formation, the mechanisms of defence, primary process or concrete thinking, regression under anxiety, types of character-structure such as the narcissistic or the paranoid-schizoid, and the stages of libidinal organization. How credible we find these principles will decide for us how rational it is to assume them in advance of well-designed, extra-clinical tests [italics added].

Erwin (1996, pp. 134-135) aptly explains that this Wollheim passage boomerangs, and is ammunition for my view:

This concession on Wollheim's part creates a puzzle. Let us assume that he is right: clinical confirmation of particular psychoanalytic hypotheses requires that we assume the truth of other parts of Freudian theory, but these in turn have no empirical support (indeed, there is only a hope that the means for testing them will eventually be devised). If Wollheim is right, we have an alternative argument for Grünbaum's (1984) conclusion that the support provided by the clinical evidence is, at best, remarkably weak.

Relatedly, in his systematization of psychoanalytic explanations, Wollheim (1993, pp. 95- 102) pulls the rug from under Nagel, Cavell, and others who, as we recall, reason as follows: Psychoanalytic explanations invoking repression typically conform to the folk-psychological PS, and therefore can dispense with "inductive evidence of some special kind" (Cavell 1993, p. 80). As her basis for just this inference, Cavell (1993, p. 180) had averred that Freud's "last revisionary work on repression and related matters–Inhibitions, Symptoms and Anxiety (1926 [1925])–builds squarely on a reason-explanation model."

To the contrary, Wollheim is concerned to emphasize that there are three variants of explanations of human conduct–even including some, though not all, slips–for which "It becomes progressively less plausible to see the behavior at issue as the outcome of a practical syllogism, and correspondingly less and less convincing to think of it as an action" (Wollheim 1993, p. 96). Regrettably, Wollheim does not acknowledge that in my 1984 book, I had adduced major violations of the PS by some of Freud's best known explanations: To wit, the explanations furnished by his example of the supposedly paradigmatic aliquis slip, by the early and late versions of his etiology of paranoia, and–following Michael Moore–by Freud's entire theory of dreams. None of these, I argued, conform to the intentional structure of the PS (Grünbaum 1984, pp. 77, 76-77, and 82, respectively).

As for Freud's celebrated case of a young man's aliquis slip, it features that man's forgetting this Latin word in a quotation from Virgil's Aenead. I had used just this famous slip to illustrate the breakdown of the PS due to the absence of the requisite belief in a means-end connection (Grünbaum 1984, p. 77):

Thus, take Freud's aforementioned report of the memory-slip of forgetting the Latin word aliquis, which he attributed to the subject's repressed pregnancy fear . . . [concerning his paramour]. This well-known case fails to conform to the practical syllogism. For there is not a shred of evidence that the male subject underwent his memory lapse in the unconscious belief–however foolish–of thereby realizing [fulfilling] his desire (hope) that his paramour is not pregnant. Therefore, . . . [there is] no basis for claiming that the subject unconsciously intended his own mnemonic failure as a means of achieving the much desired freedom from pregnancy, even if his desire itself is deemed to qualify as an "intention." As Morris Eagle (1980, pp. 368-369) has stressed, there is a major gap between a mere wish and a plan to realize it.

Both Cavell and Nagel ignored this passage in my book. Yet the absence of the means-end belief required by the PS undermines Nagel's view (1994a, p. 35) that Freud's explanation of the memory-slip here might be an "intuitively credible extension" of the PS "general structure of explanation" to cover this case.

Turning next to Freud's explanations of dreaming and of manifest dream-contents, I had written (Grünbaum 1984, p. 82):

.quite apart from Freud's own explicit views, Michael Moore (1980, 1983) has perceptively argued in detail that, on Freud's evidence, the dreaming of manifest dream contents simply does not pass muster as an intended action. For it turns out that, despite the linguistic regalia of "intention" in which he clothed his explanations of various dreams, conceptual examination of these explanations reveals that "dreams are not productions [actions] we stage for [unconscious] reasons, but are events caused by wishes" (Moore 1980, p. 538; 1983, p. 64).

Moreover, as Owen Flanagan (1995, p. 5) has noted, Augustine likewise contended that dreams are not intended actions we carry out for reasons; they are experiences thathappen to us, even if they are caused by our desires. Hence he denied that he was violating God's command against fornication, when he had a sexual wet dream: Dreams can have sinful content without being committed sins (cf. Augustine 1961, pp. 233-234).

Relatedly, Mele (1997, p. 101) defends "an explanatory model for self-deception that diverges from models for the explanation of intentional conduct." In short, he points out (idem.) that "motivated behavior is not coextensive with intended behavior."

Finally, concerning Freud's etiology of paranoia, which attributes this disorder to repressed homosexuality, I had asked rhetorically (Grünbaum 1984, pp. 76-77):

In the psychoanalytic explanation of a paranoiac's delusional conduct, can the afflicted agent be warrantedly held to have "reasons" for his/her behavior such that he/she unconsciously believes it to be a means of attaining the fulfillment of his/her homosexual longings? Can the paranoiac be warrantedly said to have unconsciouslyintended his delusional persecutory thoughts and comportment to accomplish his erotic objectives?

Just for argument's sake, postulate the existence of clinical evidence such that the paranoid agent could be held to regard his persecutory thoughts unconsciously as a means of coping with his homosexually engendered anxiety. And suppose further that this belief was unencumbered by the realization that this anxiety-reduction is purchased at the cost of generating other anxieties by the negative responses of those who are victimized by his ill-founded suspicions. Even then, such a putative unconscious belief would hardly be tantamount to the quite different belief that these suspicions conduce to the realization of his homosexual goals.

The upshot of my argument in this passage (1984, p. 77) was that there is just no empirical evidence for the existence of the latter belief in the achievement of erotic fulfillment, which is required by the PS in this context. Furthermore, in a 1986 review-symposium on my Foundations, I had observed (Grünbaum 1986, p. 269) that on Freud's perennial view of paranoia, the delusional symptoms represent compromise gratifications of two conflicting wishes: The wish for homosexual erotic fulfillment and the anxiety-laden desire to banish this wish from conscious awareness.

But Cavell (1993, p. 80) took issue with my inclusion of the erotic wish in the etiologic scenario, objecting that it is "a misunderstanding" on my part based on an exegetical anachronism. As she would have it (idem):

First, the wish that repression and symptom-formation satisfy on Freud's later [1926] instinctual model is not itself erotic but rather the wish not to acknowledge an erotic wish, or to avoid something that makes one anxious. Second, if repression did display the full intentional structure, the repressed aims would be rational, from their own point of view: If I thought (consciously or unconsciously) that I might be less anxious if I had a certain belief, I would have good reason for acquiring it if I could. I would have a reason for being a believer in x, though I would not have evidence in the light of which it was reasonable to think x true . . . . In that [latter] sense [only], it is irrational.

As to Cavell's charge that I am guilty of a misconstrual of the post-1926 Freud, Erwin has documented by reference to Freud's 1933 clarification (S.E. 1933, 20:94) of his 1926 view that Cavell's own exegesis, not mine, is defective on precisely the etiologic role that the late Freud attributes to the erotic wish. Citing Freud's 1926 paper and, more importantly, his 1933 elaboration, Erwin explains (1996, pp. 108-109):

. . . Freud speaks of repression and the avoidance of anxiety, but he does not disavow his earlier claim that the etiology of psychoneuroses involves erotic desires. Rather, he holds that there is an initial repression and repressions that occur later. The later repressions involve anxiety avoidance ["anxiety is awakened as a signal of an earlier situation of danger"] but the initial repression is in response to libidinal demands [i.e., the tabooed intense wish for homosexual fulfillment] (S.E. 1933, 22:94). When Grünbaum refers to the paranoid's erotic desires (1984, p. 79), he is talking about the desires that according to Freud, are initially repressed, and that later lead through reaction formation and projection to paranoia.

As for Cavell's second complaint of misunderstanding against me, it is ill-founded, being solely a matter of her very odd misreading of my p. 79. The point I made there is clearly introduced on the preceding p. 78, where I wrote: "It would be an error of moment to invoke Freud's likening of repressed mental states to conscious ones in an endeavor to assimilate psychoanalytic explanations to the practical syllogism after all" (italics added).

Thus, replying tellingly to Cavell's second complaint, Erwin (1996, pp. 108-109) explains that her point is irrelevant to Grünbaum's criticism. His point [here Erwin, relying on Cavell's incorrect page reference, again incorrectly lists my p. 79 in lieu of pp. 76-77] is not that the agent's symptom formation would be irrational but that there is no evidence that the agent has the belief that would relate the repressed wish to the symptom. Without such evidence, we lack grounds for assimilating the case to the practical syllogism (italics added).

Paranoiacs are "not usually amenable to analytic [clinical] investigation" (S.E. 1922, 18: 225), if only because they suspiciously resist forming a therapeutic alliance with the analyst. Therefore, it is very unlikely that there could be trustworthy clinical evidence for the theoretical imputation of the requisite beliefs to the paranoiac. Similarly for the other "narcissistic" neuroses.

As we saw, Cavell and Nagel adduced the extended psychological credentials of the pattern of intentional actions as evidential grounds for Freud's causal explanations in his theory of repression. But it now emerges that their attempted assimilation of Freud's etiologic explanations to the structure of intentional actions cannot bear the epistemic and explanatory burden that they have placed on it.

As a corollary of Part III, it emerges that Marcia Cavell, Thomas Nagel, et al., have failed to undermine my epistemological strictures on the explanatory and therapeutic foundations of the psychoanalytic enterprise.

ENDNOTES

1. But see S. Gardner's (1993, Chap. 3, p. 50) for a discussion of whether Davidsons version of mental subsystems provides a successfi.il explanation of irrationality.

2. I am greatly indebted to the psychoanalytic historian Rosemarie Sand for providing me with much information concerning these historical examples.

3. This chapter uniformly disparages my views, and alas is replete with contrived fault-finding and red herrings; see the rebuttals in Erwin (1996, pp. 130–136) and in Grtinbaum (1994).

4. Cavell is referring to Freud's 1926 signal theory of anxiety. Eric Gillett (1990, Section II, pp. 558-568) has urged both logical and empirical difficulties against this theory.

5. All citations of Freud's writings in English are from the Standard Edition of the Complete P~sychological Works of SigmundFreud, translated by J. Strachey et al. London: Llogarth Press, 1953–1974. 24 vols. Each reference will use the abbreviation "S.E." followed by the year of first appearance, volume number, and page(s).

6. The reference to p. 79 is incorrect and should have been to pp. 76–77. It matters, because p.79 is the one from which Cavell generated her erroneous statement that my objection to the applicability of the PS here is that paranoid behavior is irrational. I had neither said nor implied that!

7. Besides Rosemarie Sand (see note 2 above), I am greatly indebted to Edward Erwin for ongoing helpful discussions, and for his further defenses of my views (see Erwin 1993, 1997b). Besides, I thank Frederick Crews for having suggested a number of expository improvements in the text above. Finally, I am greatly indebted to Jerome Wakefield, a quondam student and philosophical partisan of Davidson's, for having offered me challenging criticisms of the original text of this paper. His challenge has enabled me to articulate my strictures further, but has not persuaded me to retract any of them.