Antony Anomalousness Mental

8/3/2019 Antony Anomalousness Mental

1/36

Philosophical Review

Anomalous Monism and the Problem of Explanatory ForceAuthor(s): Louise AntonyReviewed work(s):Source: The Philosophical Review, Vol. 98, No. 2 (Apr., 1989), pp. 153-187Published by: Duke University Press on behalf of Philosophical ReviewStable URL: http://www.jstor.org/stable/2185281 .

Accessed: 13/02/2012 12:13

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend

access to The Philosophical Review.

http://www.jstor.org
http://www.jstor.org/action/showPublisher?publisherCode=dukehttp://www.jstor.org/action/showPublisher?publisherCode=philreviewhttp://www.jstor.org/stable/2185281?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2185281?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=philreviewhttp://www.jstor.org/action/showPublisher?publisherCode=duke


2/36

ThePhilosophicalReview, Vol. XCVIII, No. 2 (April 1989)

Anomalous Monism and the Problem ofExplanatory Force*LouiseAntony

INTRODUCTION

C oncern about two problems-quite central to action theoryand the philosophy of mind-runs through the work ofDonald Davidson. They are:A) the problem of characterizing the "because" in explana-tions of human actions given in terms of the agent'sreasons for acting; that is, the problem of analyzing the

explanatory force of "rationalizations,"1andB) the problem for any materialist conception of mind posedby the apparent anomalousness of psychological states andevents.

*This paper was begun at an NEH Summer Seminar under the direc-tion of Professor Hector-Neri Castefiada at Indiana University. It wascompleted while I held a fellowship from the American Council ofLearned Societies. I would like to thank Professor Castefiada and myfellow seminarians, especially Michael Costa and James Kelly, for theirhelpful comments. I would also like to thank the National Endowment forthe Humanities and the ACLS for their support.My thanks also to members of the Triangle Language and Mind Discus-sion Group, the editors of The PhilosophicalReview, David Austin, DavidAuerbach, Randy Carter, and Hal Levin. Very special thanks to JoeLevine, who read several million earlier drafts.'This is Davidson's term in "Actions, Reasons and Causes," for the kindof explanation in question (Davidson, 1980, p. 3). I'm not altogetherhappy with it, since "rationalization," in one common use, applies to spe-cious explanations, offered by someone either insincere or self-deceptive.Rationalizations are tacitly contrastedwith the "real explanation"-an ob-viously unwanted connotation in this philosophical context. Dray used theword "rationalization" with just this connotation, and employed the term"rational explanation" to refer to explanations in terms of reasons. (SeeDray, 1957.) "Rational explanation," however, has troubles of its own, soI'll stick with "rationalization," which at least makes my terminology con-sistent with Davidson's. I simply warn the reader to guard against misun-derstanding.

153


3/36

LOUISE ANTONY

Davidson claims to have solved both problems through a unifiedtheory of action and mind expounded primarily in two articles,"Actions, Reasons, and Causes" (hereafter "ARC") and "MentalEvents" ("ME").2The theory divides into two parts: a causal theoryof action, meant to address the problem of explanatory force, andthe theory of "anomalous monism," explicitly constructed tohandle the second problem-the anomalousness of the mental (oras Davidson terms it, the "anomalism of the mental").The theory as a whole is intended to cohere with Davidson'sbackground views on semantics and psychology. Following Quine,Davidson believes that both the translation of utterances and theattribution of intentional states (enterprises which for Davidsonare simply different forms of the project of "interpretation"), issuein theories that are radically indeterminate.

In this paper, I will argue that Davidson's program contains afundamental inconsistency: that his Quinean metaphysics, whilerationalizing the doctrine of anomalous monism, makes impossiblea successful response to the problem of explanatory force in termsof a causal theory of action.

I. THE PROBLEM OF EXPLANATORY FORCEDavidson explains that his main objective in ARC is to counter(what was at the time) the increasingly popular view that explana-tions of human actions in terms of reasons differed in kind from

explanations of events in terms of antecedent causes-the view, inother words, that reasons were not causes. Not only does he be-lieve that the arguments for this view are all seriously flawed, buthe thinks that there is a strong argument for the conclusion thatreasons must be causes. Only on that assumption, he contends, canwe account for the explanatoryorce of rationalizations.

In this section, I want mainly to review and support Davidson'spositive case, but I'll begin with an overview of the anti-causalistpositions, because of their bearing on aspects of Davidson's meta-physics to be discussed below.

The arguments against reasons' being~causes are of roughly two21 will refer to articles of Davidson's by the abbreviations introduced inthe text. All articles are reprinted in Davidson (1980), and page referenceswill be to that volume.

154


4/36

ANOMALOUSMONISM

kinds: methodologicaland conceptual.Arguments of the first kindallege that differences in objectives between the natural and socialsciences generate different desiderata for explanatory adequacy.William Dray,3 for example, opposed the "methodologicalmonism"4 behind C. G. Hempel's deductive-nomological model ofexplanation.5 According to the D-N model, an explanation con-sisted in the subsumption of particular events under general''covering-laws." Dray did not challenge the adequacy of thismodel as an analysis of causal explanation in the natural sciences,but instead disputed its appropriateness for the social sciences.Dray allowed that explanations in the natural sciences seek to ac-count for particular events by placing them within a context oflawful regularity, but argued, following the Verstehen heorists be-fore him, that the goal of explanation in the social sciences is tomake individual human actions intelligible in their particularity.The aim, he said, is to show such actions to be reasonable, onceconsidered in light of "what the agent believed to be the facts of hissituation, . . . and what he wanted to accomplish."6

The primary constraint on explanations in the social sciences isthus, according to Dray, a normativeone. This makes the existenceor non-existence of lawful generalizations relating reasons to ac-tions irrelevant to the acceptability of ordinary rationalizations.Such generalizations, if they exist, would not add anything to therationalizing power of a specification of an agent's beliefs and de-sires, and the generalization alone would not serve to make theaction appear reasonable. Causal explanations and rationalizationssimply explain things in a different way.

Methodological arguments such as Dray's, if successful, showonly that reasons need not be causes in order to explain actions.The conceptualarguments, on the other hand, are meant to estab-lish something much stronger, viz., that reasons could not becauses. The "logic" of rationalizations, it's claimed, shows that ap-pealing to an agent's reasons is a way of explaining an event fun-damentally at odds with citing its cause. This claim is elaborated in

3Dray, 1957.4The term "methodological monism" was coined by G. H. von Wright.See von Wright (1971), p. 4.5Expounded in Hempel (1965).6Dray, (1963), pp. 68-69.

155


5/36

LOUISE ANTONY

a variety of ways: Anscombe7 argues that a reason is not a separateevent from, but merely a re-description of, the action being ex-plained. Others finesse the ontological issue of how many eventsare involved, but invoke the Humean requirement that causallyrelated events not bear any logical relation to each other.8 It'sclaimed that since reasons are specified in termsof the actions theyrationalize, the connection between an action and its reason is nota contingent one. Another argument on the same theme points toan inappropriate evidential connection between reasons and ac-tions: the presence of a reason cannot be ascertained indepen-dently of the occurrence of the action it rationalizes, and thus ra-tionalizations are not subject to experimental test in the way genu-inely contingent causal hypotheses are.

Anscombe and the others who argue for a conceptual connec-tion between action and their reasons cite-once again-the nor-mativityof explanations of human action as the source of this dis-tinctive feature of rationalizations. Thus: an explanation in termsof reasons only succeeds if it can show the action to be reasonablein light of the agent's beliefs and goals, and the best way to do thisis to display the existence of a rational-that is, logical-relationbetween a specification of the agent's reason and a characterizationof his projected action.

Thus it is the normativecharacter of rationalization that emerges,from all sides, as the source of the distinctive features of explana-tion by reference to reasons. Dray, Anscombe, and the others allagree that a rationalization can only be counted an adequate expla-nation of an action if it ustifies the action, in the sense of showing itto be a reasonable thing to do given the beliefs and desires theagent possessed at the time. In effect, the anti-causalists lay downthe following as necessary conditions for an acceptable rationaliza-tion:

The TruthCondition-Any attribution of mental attitudes to theagent contained in the rationalization must be true.9

7See Anscombe, (1976). Ryle takes a similar line in The Conceptof Mind.8See, for example, Melden, (1961) and Ryle, (1949).9This formulation of the requirement is meant to be neutral about the

proper logical analysis of attributions of beliefs and desires, and specifi-cally, about whether such attributions carry ontological commitments tomental entities, sentences, events, or anything else.156


6/36

ANOMALOUSMONISMTheRationalityCondition-The rationalization must display theaction as being reasonable in light of the beliefs and desiresattributed in accordance with the truth condition.

I'll refer to the two conditions jointly as the "justification model."But now the question arises whether these two conditions are

meant to be sufficient, as well as necessary. Do rationalizations ex-haust their explanatory function once these conditions have beenfulfilled? Is justification all there s to explaining a human action interms of reasons?The answer to this question, Davidson argues, must certainly be"no." If one maintains that justification is all there is to rationaliza-tion,

then somethingessential has certainlybeen left out, for a personcanhave a reason for an action, and perform the action, and yet thisreason not be the reason why he did it. Central to the relation be-tween a reason and an action it explains is the idea that the agentperformed the action because e had the reason. Of course we caninclude this idea too in justification;but then the notion of justifica-tion becomes as dark as the notion of reason until we can account forthe force of that 'because' ARC,p. 9).

The problem of explanatory force, then, is the problem of ex-plicating the connectionbetween actions and rationalizing reasons.Davidson thinks that the only way to do this is to treat rationaliza-tions as a species of causal explanation. Only then, he argues, canwe

turn the first 'and' to 'because' n 'He exercised and he wanted toreduce and thought exercise would do it'. . . (ARC, pp. 11-12).The inadequacy of the justification model, and the rationale for

a causal model emerge most clearly when we consider a case inwhich there are two competing rationalizations for the same ac-tion. Consider this one: Hermione is a bright, committed studentof philosophy. In any given discussion, she will likely think of sev-eral interesting points, which her general love of subject and com-mitment to intellectual progress will give her reason to expressaloud. However, Hermione is also quite insecure and overly de-pendent upon the good opinion of philosophy professors. She'saware of this neurotic tendency (I want there to be no issue about

157


7/36

LOUISE ANTONY

subconscious desires) and thus when she's afforded an opportu-nity to secure a professor's approbation by making some analyticalmasterstroke, is self-consciously possessed with a desire to make it.

Now, in February of 19-, Hermione is in the seminar of one ofher most respected teachers, and an especially brilliant commentoccurs to her. She raises her hand, is recognized, and exclaims,"But that's just verificationism!" Hermione's contribution is ap-plauded by all, and later that week she is offered a tenure-trackposition at an Ivy League university.

The question before us is, why did Hermione make her remark?On the face of it, there are three plausible answers: either she didit to foster philosophical progress, or she did it to win the approvalof her teacher, or she did it to accomplish both these things. Eachof the first two answers corresponds to a different rationalizationattributable to Hermione:

RAT 1) Hermione wanted to foster philosophical progress,and believed that making her remark would fosterphilosophical progress.

RAT 2) Hermione wanted to win the respect and approval ofher illustrious teacher, and believed that making herremark would win for her the respect and approval ofher illustrious teacher.

Both RAT 1 and RAT 2 fulfill the conditions of the justificationmodel: each attributes to Hermione a belief and desire that she infact possesses, and each shows the ensuing action to be reasonablein light of the attributed belief and desire. It is fully correct to saythat she has two distinct reasons ("primary reasons" in Davidson'sterminology)10 for performing the action she finally performs.

This might seem to favor the third answer: that Hermione actedfrom both reasons. But although Hermione clearly possessed twodifferent reasons for making the remark, it could be the case thatshe only acted from one of them. If so, then only one of the twocandidate rationalizations tells us the reason why Hermione made

'0As Davidson defines it, a "primary reason" for performing an action Aconsists in a pro attitude of an agent toward actions with a certain prop-erty, plus that agent's belief that a particular action A, conceived under adescription d, has that property (ARC, p. 5).158


8/36

ANOMALOUS MONISM

the remark. And here we reach the limits of the justificationmodel: it can tell us why each of these proposed rationalizations isa good candidate explanation, but it cannot tell us which one is thecorrectexplanation.

The anti-causalists might respond by denying that there is asingle correct explanation. They might claim that, provided boththe rationality and truth conditions are satisfied, each rationaliza-tion provides an equally correct account of why Hermione madethe remark. This would amount to insisting on the third answer-that Hermione's action was over-determined. But as I indicatedabove, such a response would be tendentious: certainly Hermionemighthave acted on the basis of both reasons, but she needn'thave.If it's even possible for a person to act from one reason when shepossesses two, then the anti-causalist is in trouble.

An appeal to ordinary language is in order here: the everydaysense of "rationalization"applies in just those cases where we rejectan offered explanation not on the grounds that we don't accept theagent's claims about what she believes and desires, but rather be-cause we don't believe that those beliefs and desires are the ones"responsible" for the action. ("Of course you want to foster philo-sophical progress, Hermione, but that's not why you spoke up.")

And on the other hand, it seems to me, human beings are notsuch pathetically Hobbesian automata that they can't once in awhile rise above banal (but omnipresent) self-concern and actfromloftier motives. ("I know that I am overly fond of the good opinionof others, but I assure you that in this case I spoke purely out oflove of the truth.") At any rate, an account of human action andrationalization ought not to rule out at the beginning possibilitiessuch as these.

But although the overdetermination response is a poor one, it isthe only sort of response open to a committed anti-causalist. Forthe problem with the justification model is not simply that itdoesn't tell us which rationalization is the correct one-it's ratherthat the justification model offers no account of what being thecorrect rationalization would consist in. What's needed is some ac-count of how a reason can be efficacious,and "efficacy" is a causalnotion.

Davidson concludes that rationalization must be regarded as aspecies of causal explanation, and that reasons-the efficaciousones-must be mental causes. This is not to deny the presence or

159


9/36

LOUISEANTONY

the importance of the normative component isolated by the anti-causalists. Davidson not only agrees that an adequate rationaliza-tion must justify the action it's meant to explain, but indeed, aswe'll see, lays heavy weight on the normative element himself. Ra-tionalizations are not simply causal explanations, on his view, butthey are at least causal explanations.

The next step, then, is to answer those who say that reasonscannot be causes. To dismiss the conceptual arguments, Davidsonmust show that the existence of a logical relation between specifica-tions of reasons and actions is consistent with their being separateevents, causally connected. This, Davidson believes, can be donewith dispatch.

Answering the methodological arguments, on the other hand,turns out to be, for Davidson, a bit more complicated. At issue iswhether or not rationalizations need to be "backed" by generallaws, as the D-N model requires for causal explanations. IfDavidson wants to say that rationalizations are a species of causalexplanation, he must either find some general laws relatingreasons to actions, or reject the covering-law model of causal ex-planation. Davidson prefers the first option, but concedes to theanti-causalists that there are no prospects for elaborating ordinaryattributions of reasons into fully general laws. Davidson's reasonsfor this concession are most interesting, and will be examinedclosely below. For now, it suffices to note that his skepticism aboutthe possibility of laws relating reasons to actions, together with hisdetermination to treat reasons as causes, present Davidson with aproblem. His solution is present in all essentials in ARC, but is fullydeveloped and "resolved" in ME, to which I now turn.

II. ANOMALOUS MONISM

The theory of anomalous monism is designed to resolve an ap-parent "paradox" Davidson thinks is generated by the followingthree principles:

(I) The Principle of CausalInteraction ("CI,"henceforth)-"atleast some mental events interact causally with physicalevents" (ME, p. 208).

(II) The Principle of the Nomological Character of Causation

160


10/36

ANOMALOUS MONISM

("NCC")-"where there is causality, there must be a law"(ME, p. 208).

(III) The Principle of the Anomalism of the Mental ("PAM")"there are no strict deterministic laws on the basis ofwhich mental events can be predicted and explained"(ME, p. 208).

The "apparent contradiction" is this: if some mental events caus-ally interact with physical events, then (presumably) some mentalevents are caused by physical events. If so, then there must be lawsunderlying the causal connections in these cases. But then, itseems, there would be laws on the basis of which mental eventscould be explained or predicted, in precise contradiction to PAM.

Clearly, this is the problem that confronted Davidson in ARC:CI is entailed by the causal connection between reasons and ac-tions,11 NCC affirms the covering-law model of causation, andPAM expresses the skepticism Davidson shares with the anti-causalists about the existence of laws governing mental events.

The solution-the theory of anomalous monism-comprisesthree theses. The first is a token-identity thesis: every mental event(where an event is taken to be an "unrepeatable, dated particular")is token-identical with some physical event. The second thesis as-serts the extensionality f singular causal claims: ".... causes "__denotes a relation that holds between particular events howeverthey are described. The third asserts the intensionalityof predicatesrelating events to laws, such as "instantiate": the truth of a claimthat an event or pair of events instantiates a law will depend uponthe description used to pick out the event or events.12

"Davidson carefully notes that one needn't embrace a causal theory ofaction to accept Principle I: perception-based belief provides as good anexample of mental-physical "interaction" as intentional action does.

'2The claim, as stated, is admittedly obscure. However, Davidson's ownformulation of the third thesis, which I discuss at some length below, iseven worse. Nowhere does Davidson make clear exactly what he means insaying that "events can instantiate laws ... only as those events are de-scribed in one or another way" (ME, p. 215). The crucial omission is aspecification of the right ways to describe events in order to bring themunder laws. (As I'll argue below, this omission is even more importantwhen we come to deal with explanation.)On the face of it, the thesis seemsfalse-there seems to be no reason why we can't speak of an event in any

161


11/36

LOUISE ANTONY

Note that the first two theses alone-the "monistic" part of thetheory-would serve to answer those anti-causalists who base theircase on a "conceptual connection" between reasons and actions. Ifmental events are simply physical events described in a particularway, and if causation is an extensional relation that holds betweenevents however described, then the existence of a logical connec-tion between the descriptions of a pair of events is irrelevant to thematter of whether the events themselves are causally connected.

It is the third thesis that really does the work of resolving thetension generated by the full complement of Davidson's method-ological commitments, and it does this by opening a loophole inPAM. As Davidson formulates PAM ("there are no strict deter-ministic laws on the basis of which mental events can be predictedand explained"-ME, p. 208), the principle strongly suggests thatmental events are occurrences altogether outside the scope of nat-ural law-that they are, one might say, anomalous "in them-selves." If this were what Davidson meant, then the conflict be-tween PAM and the other two principles would be irresolvable,and the news that mental events are token-identical with law-abiding physical events would make matters even worse.

But the third thesis in the theory of anomalous monism is meantto warn us against interpreting PAM in this way. The third thesisasserts that all claims relating events to laws contain intensionalcontexts. Explanations and predictions, Davidson contends, incor-porate appeals to natural law, and are therefore sensitive to theway in which the explained or predicted events are described. InDavidson's own words, "events can instantiate laws, and hence beexplained or predicted in light of laws, only as those events aredescribed in one way or another" (ME, p. 215). The question ofwhether an event is explicable or predictable on the basis of lawsmust turn, then, not on the intrinsic properties of the event, butrather, on how the event is described. In order for the event to beexplained or predicted "on the basis of laws," the event must bepicked out by means of some suitable description, preferably, oneold way as being an instance of this or that law.'Why can't I say-truly andcorrectly-that my falling out of bed was an instance of the laws ofgravity? And if the third thesis doesn't rule out descriptions like "myfalling out of bed," what in the world does it rule out? The only argumentfor the third thesis is the single assertion, "laws are linguistic . . ." (ME, p.215), which serves neither to justify nor clarify the claim.

162


12/36

ANOMALOUSMONISM

presumes, a description that utilizes the vocabulary of the law in-voked by the explanation or prediction.

What PAM asserts to be "anomalous," therefore, must not be themental event itself, but rather its mentalistic description.And now itseems that PAM is badly overstated. Not only is the formulationextremely misleading in suggesting that the events denoted by psy-chological descriptions somehow fall outside the natural order, butit fails to express clearly Davidson's real point-namely, that thereare no laws expressible in the language of psychology. PAM, there-fore, ought to be replaced by, or at least understood as entailing:

The Principle of Non-Generalizability ("PNG"): There are nostrict deterministic laws formulable in the vocabulary of psy-chology.

Were there such laws, mentalistically described events would be inno sense "anomalous."

Presuming Davidson's commitment to PNG, let's look at thecomplete picture of rationalization presented by the theory ofanomalous monism: a rationalization posits a causal connection be-tween a mental event and a physical event. The truth of any sin-gular causal claim entails the existence of a natural law connectingthe two events. There are (in accordance with PNG) no laws for-mulable in psychological terms, but since every mental event isidentical with some physical event, every mental-physical interac-tion is identical with some presumably law-governed physical-physical interaction. Ordinary rationalizations can thus be countedas causal despite the non-availability of lawful generalizationsstated in termsof beliefs, desires, and the like.

Those who despaired of ever finding laws connecting reasonsand actions, per se, Davidson argues, were laboring under the mis-conception that such laws would have to use the psychological cate-gories implicit in the attribution of reasons. Instead, Davidson ex-plains:

If the causes of a class of events (actions) fall in a certain class(reasons)and there is a law to backeach singularcausal statement, tdoes not follow that there is any law connectingevents classifiedasreasonswitheventsclassifiedas actions-the classificationsmayevenbe neurological,chemical,or physical(ARC,p. 17).The general point is this: not every classification of events groupsthem into natural kinds. In making singular causal claims, our

163


13/36

LOUISE ANTONY

main concern is referential-we simply want to specify the cau-sally related events, and we are free to use any mode of designa-tion that works. Justifying the causal claim is another matter. Forthis we must show that the events picked out fall into some nomi-cally significant classes, but we're not constrained in our efforts bythe descriptions we happened on in the first step.

To illustrate the point, Davidson bids us consider a catastrophe,reported on page 13 of Wednesday's Tribune,which is the result ofa hurricane, reported on page 5 of Tuesday's Times.While it couldbe perfectly obvious that the event reported on page 5 ofTuesday's Times caused the event reported on page 13 ofWednesday's Tribune, it would be ludicrous to look for a law re-lating events of thesekinds, and, Davidson continues, "only slightlyless ridiculous to look for a law relating hurricanes and catas-trophes" (ARC, p. 17). Such are simply not the concepts of realscience.

But the example raises a concern: while anomalous monism hasso far enabled Davidson to defend his claim that reasons arecauses, insuring the view against the non-existence of laws gov-erning the class of reasons, we have yet to see how he can justifythe very different claim that rationalizations are (as he puts it inARC) "aspecies of causal explanation." To see the problem, noticethat the anomalous descriptions in the hurricane example are aspoorly suited for use in explanationsas in laws. For the same reasonthat one cannot find laws relating the class of events reported onpage 5 of Tuesday's Times to the class of events reported on page13 of Wednesday's Tribune, one cannot explainthe event reportedon page 13 of Wednesday's Tribunesimply by asserting the truththat it was caused by the event reported on page 5 of Tuesday'sTimes. Not every true causal statement counts as a causal explana-tion: for explanation, it's not enough just to pick out the cause ofan event-one must pick out the event in a certainway. In otherwords, the intensionality inherent in laws infects explanations also.

If the adequacy of a causal explanation depends not only on thetruth of the singular causal claim it entails, but also on the way inwhich the events are described, fresh questions arise about the im-plications of PNG for the status of rationalizations. Let us call adescription of an event nomic ust in case the description denotes anomologically significant class. We can then consider the followingas a possible condition of adequacy for causal explanations:

164


14/36

ANOMALOUSMONISM

TheNomic Constraint-a specification of the causal antecedentsof an event only counts as a causal explanation of that event ifthe descriptions of the event and its cause are nomic descrip-tions.Hempel may have had something like the nomic constraint inmind.13 The condition is certainly in the spirit of the D-N model,and commitment to some such constraint would explain his insis-tence on the need for general historical and psychological laws.

But what about Davidson? Despite certain remarks that suggestthat he, too, has such a constraint in mind,14 Davidson does not-and cannot-accept the nomic constraint. He cannot accept it be-cause the nomic constraint, together with PNG, immediately entailthat rationalizations are not causal explanations-indeed, that theyare not genuinely explanatory at all. And it is clearly Davidson'sview that, whatever else is true, rationalizations are perfectly legiti-mate forms of explanation. In "Hempel on Explaining Action,"Davidson ascribes to Hempel something very close to the nomicconstraint, the view that "in order to explain events we must de-scribe them in a way that reveals how laws are applicable," andvoices the following reservation about it:

We may oin in laudingas an idealexplanationa descriptionof ante-cedents and a specificationof lawssuch that the explanandumcanbededuced; but how much less still counts as explanation?It seems to'3He says in "Aspects" for example, that ". . . the law tacitly implied bythe assertion that b, as an event of kind B, was caused by a as an event ofkind A is a general statement of causal connection to the effect that, undersuitable circumstances, an instance of A is invariably accompanied by aninstance of B" (Hempel, 1964, p. 349). Davidson interprets Hempel asendorsing the nomic constraint. (See Davidson, 1980, p. 265.)'4In the original formulation of PAM, for example, Davidson speaks of

there being no "strict, deterministic laws on the basis of which mentalevents can be predicted and explained," without specifying any other pos-sible bases for the explanation and prediction of mental events.Again, in response to an objection of Attfield's (Davidson, 1980, pp.241-242), Davidson says that ".... laws (and nomologicalexplanations)donot deal directly (i.e., extensionally) with events, but with events as de-scribed in one way or another" [my emphasis]. Davidson does not say ex-actly what counts as a "nomological explanation," but presumably hemeans the term to encompass causal explanations, and thus rationaliza-tions. He does not say whichways of describing events work for nomolog-ical explanations, but the context suggests it's nomic descriptions that mustbe used.165


15/36

LOUISEANTONYme we have in actiona particularly ood specimenfor study; since weagree that one way of explaining actions is by giving the agent'sreasons, we can concentrate on the relatively clear question whatreason explanationsare like, and set aside the more diffuse problemof characterizing xplanation generally ("Hempelon ExplainingAc-tion," pp. 262-263).

Well, then, given the intensionality of explanation, and the truth ofPNG, and the legitimacy of unreconstructed rationalizations, whatare "reason explanations" supposed to be like?

The key to Davidson's unique picture of rationalization is pro-vided, ironically enough, by the anti-causalists, and their emphasison the normativityof this form of explanation. As I pointed outearlier, Davidson is quite in agreement with his antagonists aboutthere being this element to rationalization; he disagrees only aboutits bearing on the ontology of the matter. Since Davidson's com-plaint against what I've called the justification model was essen-tially that it was incomplete, and not that it was incorrect, we couldappend that model to Davidson's causal theory of action to get thefollowing model of rationalization as "a species of causal explana-tion":

A Davidsonian Model of RationalizationConditions on an adequate rationalization:

(a) The Truth Condition(as before)(b) The Rationality Condition(as before)and (c) The Causal Condition-the event cited as the reason

in the explanans is the cause of the event cited asthe action in the explanandum.

Hermione's case illustrates the model: we'll give her the benefitof the doubt, and assume that it is RAT 1 ("she wanted to fosterphilosophical progress. . .") and not RAT 2 ("she wanted to securethe approbation of her professor . . .") that gives the real reasonfor her action. Obviously, RAT 1 satisifes conditions (a) and (b),because it satisfies the justification model. Additionally, condition(c) entails that the following must be true: there exists an event csuch that c is identical to the event consisting in Hermione's per-ceiving an opportunity to foster philosophical progress, and thereexists an event e such that e is identical with Hermione's makingthe remark, and c is the cause of e.

166


16/36

ANOMALOUSMONISM

In the picture I'm attributing to Davidson, then, rationalizationsexplain actions by doing two things: by citing the physical cause-despite the use of mentalistic descriptions-of a physical occur-rence, and by displaying-through the use of mentalistic descrip-tions-the action as being reasonable in light of the agent's beliefsand desires. Conditions (a-b) constrain the kinds of ways in whichthe cause of an action can be picked out in order to get a genuinerationalization, and condition (c) ensures that adequate rationaliza-tions will have genuine explanatory force.

It thus seems possible to construct a viable analysis of rationaliza-tion as causal explanation consistent with commitment to PNG.However, trouble is just around the corner.

III. THE PROBLEM: PART ONE

The first problem is one that Davidson has been primarily re-sponsible for identifying, although he certainly does not see it as aproblem for his view of rationalization. It is this: reasons can causeactions without causing them in "the right way." This is an oldchestnut by now in the action theory literature, and examplesabound, but I'll borrow one of Davidson's cases to illustrate.

A climber might want to rid himself of the weight and danger ofholdinganothermanon a rope, and he mightknow thatbylooseninghis hold on the rope he could rid himself of the weightand danger.This belief and wantmight so unnerve him as to causehim to loosenhis hold, and yet it mightbe the case thathe neverchose o loosen hishold, nor did he do it intentionally("Freedom o Act,"p. 79).

Davidson offers this case as a problem for his analysis of free ac-tion, as a problem for his analysis of intentional action, and as onereason why it's probable that there are no empirical laws governingrational behavior. What he doesn't realize is that the case providesan equally intransigent problem for his own analysis of rationaliza-tion.

Davidson notes that from the fact that there exist attitudes of anagent which both cause an action and are such as would serve torationalize the action, it does not follow that the agent performedthe action intentionally.But by the same token, we can see that arationalizationof the action in terms of those attitudes would beunacceptable as an explanation. In general, for any case involving

167


17/36

LOUISE ANTONY

a "wayward" causal chain linking reasons to actions, it will be incor-rect to offer a rationalization or the action, even though both con-ditions (a-b) and (c) of the Davidson model will be satisfied.

Davidson's model of rationalization is clearly missing something:an account of how reasons cause actions so as to rationalize themsimultaneously. The explanatory force of rationalizations is partlyexplained by the fact that reasons are things that have causal effi-cacy, but we also need to know how it is that reasons can havecausal efficacy in virtue of theirreasonableness.15How can it be thatthe causal potential of a physical event is partly determined by thelogical features of one of its descriptions?

Davidson's causal account of action is meant to underwrite theexplanatory value of an appeal to reasons; but we can see that infact, the model leaves the rational and causal aspects of rationaliza-tion radically detached. Because we are not told how a belief ordesire characteristically causes an action that it rationalizes, the en-tire appeal to the causal antecedents of the action is a red herring.If we are materialists of any stripe, we believe that the physicalgoings-on in our bodies, and especially in our brains, have some-thing to do with what kind of behavior our bodies display. None ofthe anti-causalists that I know of denied that there exist true phys-ical descriptions of the movements of our bodies during a periodof what could be called intentional action. The question wasalways, rather, what relevance any such physical description wassupposed to have to commonsense explanations of those actions.This is precisely the question Davidson does not answer.

Nor, I contend, can he. Making the needed emendations wouldimplicate him in a realism about psychology and the mental di-rectly in conflict with his deep commitment to the Quinean view ofmatters intentional.

IV. EXPLANATIONS AND LAWS

The first thing to notice is that the problem really is not specificto rationalizations: it can be illustrated by cases that don't even in-volve intentional descriptions of events..Return once more to the

'5Thiswayof puttingthe pointwassuggestedto me byJoe Levine.JerryFodormakes this point the main strut of his philosophicalargument for a"languageof thought,"about which I'll have more to say later.168


18/36

ANOMALOUSMONISM

hurricane analogy. Davidson argues that it would be silly not onlyto look for laws relating events described by reference to page ofreportage, but also to seek laws relating "hurricanes and disasters."No doubt Davidson is right about the laws-but what's interestingis that we can get perfectly good explanationsof disasters in termsof hurricanes. An astonished someone surveying the wreckage of aseaside town would be thoroughly ( atisfied to learn that there hadbeen a hurricane there the night before, and this despite the factthat "hurricane" and "disaster"are not nomic terms.

I noted above that explanation is intensional-that not just anyspecification of the causal antecedents of an event will do to ex-plain it-so that an adequate account of explanation must providesome constraints on descriptions that may be used. The nomicconstraint is one possibility, and another is the rationality con-straint. But neither would legitimate the explanation of a disasterin terms of a hurricane: "hurricane" and "disaster"are not nomicterms, and neither do hurricanes make it reasonable for disastersto have visited themselves upon coastal towns. What does, then,account for the explanatory value of reference to hurricanes?

It surely doesn't help to point out, as Davidson seems to think itdoes, that each individual hurricane and each individual disasterdoes have some nomic description, because that fact doesn't ac-count for our satisfaction with the explanation in termsof disastersand hurricanes. The same thing can be said of the identities of theevent reported on page 5 of Tuesday's Times and the event re-ported on page 13 of Wednesday's Tribune, but there are no gener-ally acceptable explanations of those events so characterized.16Thus, the token-identity of a non-nomically described event with anominally described event cannot be the whole story about the ac-ceptability of ordinary explanations.

If we can fill out the story properly for hurricanes and disasters,perhaps we'll gain insight about rationalizations. To begin with,

'61f one knew antecedently that the event reported on page 5 ofTuesday's Timeswas a hurricane, and were making inquiries of someonewho knew that it was a disaster that was reported on page 13 of theTribune, then the explanation in these terms might be OK. But notice thatthe acceptability of the explanation depends crucially on the identitiesbeing known, and not merely on their being true. And Davidson is quiteinsistent that the acceptability of rationalizations does not depend uponknowing the relevant psycho-physical identities.169


19/36

LOUISE ANTONY

notice that while "hurricane" and "disaster" are not nomic terms,they do differ significantly from descriptions like "event reportedon page 5 of Tuesday's Times" n bringing together events that arenon-accidentally elated. One can fruitfully inquire what sort of thinga hurricane is, with the intention of finding out how hurricanescause disasters. If we can understand, even roughly, how thingsdescribable in the language of physics can go to makeup a hurri-cane, we can see how the regularities describable in the languageof physics can converge to produce the regularities-apparent onthe macro level-that we describe in terms of hurricanes and dis-asters. It's the fact that we can know-even vaguely-what sort ofthing a hurricane is that makes relevant the kind of strict, micro-physical account of particular causal interactions that Davidson in-sists upon.

Thus, more important than knowing the particular token-identities involved in a given hurricane/disaster incident, isknowing, in general, how the nature of hurricanes constrains thekinds of micro-events that can constitute a hurricane (and similarlyfor disasters). For this to work, "hurricane" must enforce a princi-pled collection of events-there must be something things have incommon that makes it true that they are hurricanes, somethingnon-trivial in virtue of which they are hurricanes.

This "something" is not going to be a simple function of thehurricane's physical or chemical composition-otherwise, "hurri-cane"17 would be a nomic term, and we're supposing along withDavidson that there are no strict laws relating the class of hurri-canes to the class of disasters. Rather, the features shared aregoing to be abstract characteristics-functional or structural prop-erties that any complex of events must display in order to becounted a hurricane.

This is, of course, simply the familiar functionalist point thatthere are regularities that only become apparent when we employrelatively high-level descriptions of objects and events.18 But therecognition that there can be regularities that are neither nomicnor accidental requires us to complicate significantly the picture oflaws, causation, and explanation that we've been assuming.

170r some extensionally equivalent predicate, which could be readilytranslated to "hurricane" through a simple bridge law.'8See Putnam, 1960, for the classic statement.

170


20/36

ANOMALOUSMONISM

In his recent work, The Nature of Psychological Explanation,'9Robert Cummins points out that there's been a general failure torecognize the difference between two distinct explanatory strate-gies: subsumptionand analysis.This distinction is orthogonal to anyof the distinctions in terms of discipline or subject matter urged bythe anti-causalists, accommodating my observation that we have asmuch difficulty accounting for the explanatory value of appeal tohurricanes as for the value of appeal to reasons.

The subsumption strategy is the style of explanation we've beendiscussing: explaining an event pair by showing the pair to be aninstance of some nomic generalization. This strategy is well-suitedto the explanatory demands placed upon what Cummins calls"transition theories": theories that are meant to specify the condi-tions under which particular systems will undergo various changesof state. But the explanation of state transitions by subsumptionunder causal generalizations is, while important, by no means theonly form of theoretical explanation. Even transition theories de-part from the D-N paradigm, both by employing laws other thancausal generalizations, and by using them for purposes other thanthe subsumption of individual causal pairs.

But more important for current purposes than the complexityof the explanatory structure of transition theories is the fact thatthere exists a completely different kind of theory, which has asso-ciated with it a completely different kind of explanatory strategy.A "property theory" is meant to explain, not how state changesoccur, but what it is for an object or a system to have a certainproperty or to be in a certain state. The strategy here is analytical

the theory must say what it is about the composition or organiza-tion of the system that makes the system an instantiation or realiza-tion of the property.

Cummins emphasizes that the kind of laws that do thiswork-"instantiation laws"- do not license type-type reductions,precisely because a single property may be instantiated in manydifferent ways. And we can see that it is because of the possibility

'9Cummins, 1984. In the succeeding paragraphs, I draw from ChapterI. I cannot begin to do justice to the richly detailed account of the forms ofexplanation and the variety of laws that is contained in Cummins's book,and so anything in the discussion below that seems crude or overly simpleshould be assumed to be my fault and not his.171


21/36

LOUISE ANTONY

of multiple instantiation that the having of higher-level functionalor architectural properties admits of explanation. Since "being acell" is not simply a matter of having a certain physical or evenchemical composition, there is something to learn about how "cell"constrains the range of chemical structures: it constrains them ineffect by setting a very specific functional problem, which a struc-ture must "solve" in order to be properly called a cell. A propertytheory tells us what the task is, and how this or that structure hasmanaged to accomplish it.

Obviously transition theories and property theories can and dohelp each other out. If we want to understand a state transition ina complex system, we may need to learn first through a propertytheory how the antecedent and consequent states are realized inthat system, and only then look for transition theories to subsumethe lower-level events under causal generalizations. This is ob-viously how it's going to work for hurricanes and disasters. Ananalysis of hurricanes in terms of temperatures and velocities ofair and water molecules, plus some transition theories predictingthe effects of such molecules in such states on various kinds ofsolids, topped off with some functional analyses of things likepeople, livestock, houses, trees, and beachfront properties, and wehave the complete story, not only about what caused the disaster,but about why talk of hurricanes explains the disaster.

Well. Now that we've got all that settled, it's pretty obvious howDavidson's model of the explanatory value of rationalizationsought to be filled out. Davidson is right that we don't need tobother about finding empirically substantive transitiontheories re-lating reasons to actions, and he's right that the causal element inappeals to reasons is underwritten by the existence of true causalgeneralizations relating the state transitions of the realizing physi-ological system. All that's missing is the account of whythose statetransitions add up to a reason's producing an action-that is, aproperty theory for psychological properties.20

If this account of the causal production of actions by reasons iseven roughly correct, then there is a way to deal with the examplesthat seem to threaten causal theories of intentional action. What

20This is precisely what Cummins develops in Cummins, 1983. SeeChapter III, "Understanding Cognitive Capacities," especially Sections 6and 7.172


22/36

ANOMALOUSMONISM

we now have is the possibility of an account of what "the right way"is for reasons to cause actions in order truly to rationalize them.Recall the case of the discomposed climber, cited above. Theclimber's desire to be free of his partner, in conjunction with hisbelief that letting go of the rope would free him, caused him to letgo of the rope, although not in the way that would warrant oursaying that he let go of the rope "because he wanted to be free ofhis partner." For comparison's sake, let's imagine another climber,a vicious one this time, who, having the same desire to be free ofher partner, deliberatelydoes what she believes will fulfill that de-sire, viz., lets go of the rope. The problem for Davidson here is tosay why a rationalization is a proper explanation in the second casebut not in the first.

Once we grant property theories their proper place in the on-tology of explanations, we can see a principled difference betweenthe two cases. In the case of the vicious climber, an analysis of theclimber's psychology will reveal a complex system of intercon-nected physiological sub-systems organized in such a way that,when the system comes to realize the psychological state of wantingto be free of one's partner, relevant sub-systems will be engagedand (as predicted by relevant transition theories) will shift into astate intentionally characterized as "letting go the rope." Thingsare only superficially similar with the discomposed climber. Here,we may suppose, the physiological properties of the belief and de-sire states that were causally implicated in the motor movementswe call "letting go of the rope" were not the physiological proper-ties in virtue of which those states instantiate the psychologicalproperties of wanting to be free and believing letting go the ropewill do it. The situation is rather like that of a badly housed com-puter, which, whenever it begins to compute the square root of 16,begins to vibrate so heartily that itjogs the printer, which, being sojogged, prints out the numeral "4."

In the case of the vicious climber, the reference to beliefs anddesires is explanatory because it points us toward a functional anal-ysis of the climber as a psychological system, and because in thiscase, the event in question is properly viewed as the proper resultof the proper operation of this psychological system. In the case ofthe discomposed climber, however, it is close to an accident thatthe physiological antecedents of the explanandum had psycholog-ical descriptions that appeared to rationalize the "action."(Indeed,

173


23/36

LOUISE ANTONY

the fact that everyone who discusses these cases finds it odd to callthe consequent events "actions" lends plausibility to the hypothesisthat in these cases, the psychological mechanisms adverted to in arationalization are not actually engaged.)

All physiological states have an enormous number of properties,each property representing a variety of causal potentials. What thefunctional analysis of the climber's psychology will give us is anaccount of how particular intensional properties-like "being adesire to be free of one's partner"-can be co-instantiated and co-ordinated with just the right causal potentials. In other words, it isa property theory of psychological states that will tell us how ac-tions can be caused in virtueof their reasonableness.

This is what Davidson needs. Unfortunately for him, he can'thave it. The difficulty stems from Davidson's commitment to PAMand PNG-not so much from the principles themselves, forperhaps they could be charitably recast in light of the complexitiesintroduced by Cummins's discussion-but from the basis of thesecommitments. I will argue that Davidson's belief in the "anoma-lism" of the mental is part of a strong skepticism about the psycho-logical, derived from Davidson's understanding of centralQuinean doctrines, and radically incompatible with Davidson's so-lution to the problem of explanatory force.

V. DAVIDSON'S PSYCHOLOGICAL SKEPTICISMLet's survey these commitments of Davidson's. There is, first of

all, PAM (there are no strict deterministic laws on the basis ofwhich mental events can be predicted and explained). PAM, I'veargued, is obscure, but at least must be taken to entail PNG (thereare no strict deterministic laws formulable in the vocabulary ofpsychology). Cummins's discussion of the role of laws in explana-tion raises questions about the significance of these two principles:if Davidson has in mind causal generalizations when he speaks of"strict deterministic laws" (and there's every reason to think hedoes), then the two principles, even if true, are irrelevant to issuesabout the status of psychological explanations, for both could betrue even if there are excellent property theories explaining therealization of psychological kinds in physical systems.

It's tempting to see Davidson's emphasis on the "anomalism" ofthe mental as fundamentally an attack on strong reductionism,174


24/36

ANOMALOUSMONISM

and to see the theory of anomalous monism as primarily an at-tempt to demonstrate the possibility of materialism in the face ofthe non-reducibility of psychological states. But to interpretDavidson in this way-as a closet functionalist-is a mistake. He isrejecting far more than strong reductionism, and his reasons forrejecting even that are not the familiar functionalist ones.

As noted, skepticism about reductive psychological laws is quitecompatible with optimism about property theories for psychology.But Davidson categorically denies the possibility of any systematicempirical treatment of psychological states and events. He doesnot make the easily remedied mistake of inferring the impossibilityof psychology from the falsity of reductionism-Davidson's infer-ence goes precisely the other way around. His rejection of reduc-tive laws depends upon a prior, radical rejection of psychology toutcourt.

Functionalists reject strong reductionism because of the possi-bility of (or likelihood of, or demonstrated fact of) multiple real-izations of psychological states. Davidson's arguments, on theother hand, stem from general arguments against the possibility ofany kind of psychological laws. These arguments comprise varia-tions on two themes: (a) beliefs, desires, and other psychologicalstates do not form a "closed system," and (b) that the methodologyof psychological ascription involves non-empirical elements. Noneof these arguments is successful, and I'll indicate briefly where Ithink they go wrong. But my main objective at this point is not tocriticize Davidson's anti-realist arguments,2' but rather to docu-ment his metaphysical commitments, and to expose their source.I'll begin with a brief look at set (a), and then concentrate on set(b).

What does Davidson mean by saying that mental events do notform a "closed system"? At points, Davidson seems to be sayingsimply that psychological events interact with events that are notpsychological.22 But that fact alone would hardly impugn thescience of psychology, any more than the fact that people trip over

21A comprehensive and detailed critique of Davidson's argumentsagainst psychological laws is to be found in Lycan, 1981.22For example, in "Philosophy as Psychology," Davidson says that ". . .psychological events clearly cannot constitute a closed system; muchhappens that is not psychological, and affects the psychological" (Da-vidson, 1980, p. 230).

175


25/36

LOUISE ANTONY

rocks impugns the science of biology. What Davidson is really get-ting at is revealed more clearly in this passage:

Physical theory promises to provide a comprehensive closed systemguaranteedto yield a standardized,unique descriptionof every phys-ical event couched in a vocabularyamenableto law.It is not plausible that mental concepts alone can provide such aframework,simply because the mental does not, by [the principle ofmental-physical nteraction]constitute a closed system (ME, p. 224).Here we see that Davidson's objection to psychology stems notfrom the fact that psychological events interact with non-psycho-logical events, but from the more fundamental fact that non-psychological events even exist-or, to be more precise, from thefact that there exist events which the language of psychology isinadequate to describe. Physics (Davidson presumes) does notsuffer this limitation. It is comprehensiven the sense that every pro-cess in nature, whatever else is true of it, admits of some character-ization in physical terms. And from the fact that physics compre-hends (in this sense) all of nature, Davidson argues, it follows thatphysics is fundamental.

To the extent that Davidson's basis for rejecting psychology isthis invidious comparison between the prospective scopesof psy-chology and physics, his argument is essentially a reiteration of oneof Quine's arguments for indeterminacy.23 But this argument-whether we consider Davidson's version or Quine's-shows eithertoo much or too little. It either demonstrates that physics, beingthe only truly fundamental science, is the only genuine science; or,it shows that since non-fundamental systems like biology can con-stitute genuine sciences, that fundamental-ness should not be ac-cepted as a criterion of scientific legitimacy.

In either case, psychology remains no worse off than biology.

23An illustrative passage from "Facts of the Matter": "What [physicalismsays] about the life of the mind is that there is no mental differencewithout a physical difference.... It is a way of saying that the funda-mental objects are the physical objects. It accords physics its rightful placeas the basic natural science without venturing any dubious hopes of re-duction of other disciplines" (Quine, 1979, p. 163). Significantly, Quineapprovingly cites "Mental Events" in support of his claim that "thegroupings of events in mentalistic terms need not stand in any systematicrelation to biological groupings" (ibid.).176


26/36

ANOMALOUS MONISM

Yet Davidson wrote no companion piece, "Biological Events," sup-porting a thesis about the anomalism of the biological with the ob-servation that the biological does not constitute a closed system.Why not?

Quine's defense of the legitimacy of the biological sciences (andindeed, psychology as long as it's behaviorist) can be inferred fromanother familiar argument for indeterminacy, the so-called "argu-ment from above."24 According to that argument, an arbitrary''setting" of all physical parameters is sufficient for fixing thetruth-conditions of claims in physics, and (presumably) in higher-level sciences like biology, but not for the claims of semantics andmentalistic psychology. Hence, psychology is not simply underde-termined with respect to the physical facts, but underdeterminedwith respect to physical theory. It is this "second level" of under-determination that constitutes genuine indeterminacy. Theorychoice in such a domain can therefore not be a matter of empiri-cally discovering antecedentally existing states of affairs, but mustrather be, ultimately, a matter of convention, convenience, or aes-thetic preference. When it comes to psychology, there is "no fact ofthe matter."

Davidson, I contend, shares this view of psychology with Quine,but holds it for interestingly different reasons. For Davidson, asfor Quine, psychology is not fully empirical because its domain isnot "objective," in the sense that psychological attributions arehuman constructions, and answer as much to human concerns andperspectives, as to a pre-existing, mind-independent reality. Butthis alleged non-objectivity is not, in Davidson's mind, due to asecond level of underdetermination,25 but rather to what he re-gards as an essential methodological feature of psychologicaltheorizing. This brings us then, to the second of Davidson's two

24See Quine, 1970, "On the Reasons for Indeterminacy of Translation."An argument against applying the argument from above to the problemat hand is that in this article Quine makes the puzzling admission thatmuch of theoretical physics, and all of our ordinary material-object talk isunderdetermined by all possible evidence. Still he does say that mentalisticattributions suffer from an "additional" layer of underdetermination, thatis, underdetermination with respect to physical theory and material-objecttalk. Whether Quine's position can be made coherent and plausible is anissue, thank heavens, beyond the scope of this paper.25Davidson seems actually to reject the argument from above at ME, p.221.

177


27/36

LOUISE ANTONY

themes: the methodological peculiarity of psychology. Davidsonbelieves, I'm about to argue, that in psychology, non-empirical "in-terpretive" concerns trump all others.

The source of the non-objectivity of psychology, according toDavidson, and the feature that finally sets psychology (and any"intentional" science) apart from physics, biology and the others,turns out to be, ironically, just what the methodological anti-causalists always said it was-the normativity of rationalization.Let's look at the arguments in some detail.

Davidson first notes that the ascription of mental contents is ho-listic: there can be no evidence for a particular psychological as-cription unless a host of other ascriptions are assumed to be true.For example, nothing counts as evidence for or against the hy-pothesis that Hermione believes it's raining unless assumptions aremade about what Hermione desires (for example, Hermionewants to keep dry).

There is no assigningbeliefs to a person one by one on the basisofhis verbal behavior, his choices, or other local signs no matter howplainand evident, for we make sense of particularbeliefs only as theycohere with other beliefs, with preferences,with intentions, hopes,fears, expectations,and the rest (ME, p. 221).From the first feature, the holistic character of belief/desire ascrip-tion, Davidson rightly infers that our "theoretical" conclusion thatHermione believes it's raining is vulnerable to a host of data otherthan the behavioral data which may have immediately promptedthis ascription. Moreover, he continues, since only a finite numberof these data are available at any time, our theorizing about mentalstates is always defeasible.

But, as Davidson recognizes, the holistic nature of informal psy-chological theorizing would not, on its own, be enough to distin-guish psychology from any of the natural sciences. Rather, he says,it's the holistic nature of psychological ascription taken togetherwith its normativity that makes the difference. Put these two fea-tures together, and the result will be the impossibility of any sys-tematic relation between psychological kinds and physical kinds:

The nomological irreducibility of the mental does not derivemerelyfromthe seamlessnatureof the world of thought, preference,and intention, for such interdependence is common to physical

178


28/36

ANOMALOUSMONISM

theory,and is compatiblewith there being a single right wayof inter-preting a man's [sic] attitudeswithout relativization o a scheme oftranslation.... The point is ratherthat when we use the concepts ofbelief, desire, and the rest, we must stand prepared,as the evidenceaccumulates, to adjust our theory in the light of considerationsofoverall cogency: the constitutiveideal of rationalitypartly controlseach phase in the evolution of what must be an evolving theory....We must conclude ... thatnomologicalslackbetweenthe mentalandthe physical s essential s long as we conceiveof man [sic]as a rationalanimal (ME, pp. 222-223). [Myemphases.]

This passage is difficult to unpack: first of all, let me note onceagain that Davidson's line represents an innovation in the defenseof the indeterminacy thesis. For whereas Quine rests his case on ade facto underdetermination of psychology by physics, Davidsonmeans to show that psychological claims oughtnot bytheir nature tobe determined by the physical evidence. Within this same passage,Davidson says that the irreducibility of the mental is not duesimply to the availability of more than one acceptable translationscheme, because determinacy could be secured easily enough bythe arbitrary selection of a scheme.26 The real problem with such astrategy, Davidson declares, is not the arbitrariness in the selectionof schemes, but rather that the fact that any fixing of psycho-physical relations would preclude the "opportunistic tempering oftheory," prompted and required by "the constitutive ideal of ratio-nality" (ME, p. 223).Davidson clearly believes not only that certain normative con-straints are essential to psychological theorizing, but that theirpresence gives psychological claims a different kind of groundingfrom that of fully empirical claims. He feels that a proper sensi-tivity to "considerations of overall cogency" requires not simply theusual epistemological willingness to revise belief/desire ascriptionsin light of further developments, but rather, a radically relativisticattitude toward the whole enterprise. He believes that one shouldlook upon psychological hypotheses not-as one views hypothesesin the physical sciences-as our best guesses about the nature ofan objective phenomenon, but rather as artifacts reflective of ourcurrent stage of self-interpretation.

26This is, after all, a strategy Quine endorses for resolving residual in-determinacy in physics if any remains when all possible data is in. (SeeQuine, 1970.)179


29/36

LOUISE ANTONY

Davidson provides no argument for this view, nor, indeed, anyexplicit avowal of it. But if we attribute to him this radically con-structivist position on the nature of psychological "facts," it be-comes possible to explain (though not to justify) some otherwisepuzzling aspects of his view-notably, his insistence on the in-prin-ciple impossibility of any systematic, nomic connection betweenpsychological and physiological states. For notice that even instan-tiation laws would (or could) provide physical criteria (or at leastsufficient conditions) for psychological ascriptions. That, in turn,seems to allow the possibility of our obtaining the means to make aparticular psychological ascription, independently of any moreglobal information about the subject's other psychological states,and regardless of any normative considerations about the subject'srationality. The existence of physical criteria of psychologicalstates would thus threaten to ossify the process of psychologicalinterpretation, not by making it insensitive to new data, but bymaking it sensitive to the wrong kind of data. Davidson's fear, Ibelieve, is that if psychological hypotheses could be decisivelybacked by physical evidence, then they could not play the expres-sive, interpretive role he has assigned them.

In short, he believes that the only way to ensure the satisfaction ofthe normative constraints distinctive of psychological theorizing isto insulate psychological ascriptions from any bodies of evidencethat could potentiallycompete with "considerations of overall co-gency." Because Davidson is a radical constructivist about psycho-logical states, potential conflicts between the normative constraintsoperative in our "folk" psychologizing and the physical facts arenot to be passively borne and devoutly hoped-against, as is the casewith potential conflicts between other pre-theoretic taxonomiesand their scientific refinements.

To get the feel for the sort of situation I believe Davidson wantsto guard against, consider the following hypothetical case. Sup-pose that from extensive observation of Hermione we have devel-oped an extraordinarily well-confirmed instantiation theory. Thismeans, among other things, that we have good reason to believethat in Hermione, neurophysiological state NP6 realizes the mentalstate of intending to write a book, and neurophysiological stateNP7 realizes the mental state of believing that she will never write abook. Now suppose that we discover, by neurophysiological exami-nation, that Hermione is in bothNP6 and NP7. Our well-confirmed

180


30/36

ANOMALOUS MONISM

instantiation theory tells us that Hermione both intends to write abook and believes that she will never write a book. But our appreci-ation of the normative aspects of psychological ascription tells usthat this particular compound ascription borders on the inco-herent: in the ordinary case, good evidence for the second ascrip-tion is strong evidence against the first.27 What are we to do? Thephysical data militate in one direction, "considerations of overallcogency" in precisely the other.28

Now for any materialistic functionalist, the possibility of such a

27Davidson does not make the claim that belief and intention are relatedin the way I've suggested. I offer this particular case simply as an illustra-tion of the sort of thing I think Davidson has in mind. I should note thatsome philosophers have argued against the particular connection I'm as-suming. See, for example, Brand, 1984.280ne referee has suggested that the problem I sketch would not arise ifwe attended properly to Shoemaker's distinction between core and total

realizations of psychological states (see Shoemaker, 1981). NP6 and NP7must be considered, at best, only core realizations, it was argued. The totalrealizations of the intention and belief in question would surely embodythe relevant normative constraint by making it a nomological impossibilityfor the two core states to be co-instantiated. Thus the full functional speci-fication of these states would entail that either Hermione cannot be in NP6and NP7 simultaneously, or else that, if she is in both states at the sametime, that we are wrong in regarding one or both physiological states asthe core instantiation of the relevant psychological state.While it's true that one could rule out the co-instantiation of core real-izations of "intending to write a book" and "believing I will never write abook" in this way, I don't think this strategy can really dissolve theproblem, nor can it be of much help to Davidson. Suppose that we doattempt to embody our pre-theoretic normative constraints within func-tional definitions in this way (and I am very sympathetic to this approachthough see Section VII below for my own somewhat different sugges-tions about how to handle the normative constraints). The problem I en-vision can still arise if-and I take this situation to be a simple empiricalpossibility-it turns out that there is in fact no way to obtain total realiza-tions of functional psychologies containing such strict constraints. Wehave two options, and neither will be of comfort to Davidson. Either weallow a modification of the normative constraints, in light of the totality ofevidence, behavioral and neuro-physiological-anathema to Davidson,I've argued; or we insist on preserving the normative constraints, with theresult that our best theory turns out not to be literally true of anyone.Thisin itself needn't bother Davidson, especially if, as I've argued, he's a rad-ical anti-realist about psychology. We get no embodiment of our functionalpsychology at all. But it will be an embarassment when it comes time tosolve the explanatory force problem, since it means that there will be noembodiment f our best psychological theory. It's awfully hard to constructan identity theory on a basis like that.

181


31/36

LOUISE ANTONY

situation is disconcerting. Too manycases of this sort would surelybe evidence against the truth of the instantiation theory, for thesimple reason that when pre-theoretic constraints are violated tooblatantly and too frequently, the theory loses its claim to being atheory of its particular domain. If a psychological theory entailedthat human beings normallyand self-consciously ossessed contradic-tory beliefs, which were in turn never implicated in the productionof behavior, but did reliably cause a swelling of the left ankle, onecould legitimately protest that this was hardly a theory of beliefatall.The real issue is not whether an empirical theory of the realiza-tion of belief can give any attention to normative, pre-theoreticconceptual connections among mental states-it seems clearly thatit must. (I'll sketch my own view of how the normative constraintsare to be accommodated in the last section.) Rather, the issue iswhether or not these considerations are to be assigned the status ofa priori truth. Davidson treats them in precisely this way, and hasthe metaphysics that goes along with such a position. But a natu-ralistic functionalist, whatever she decides in a case like the oneabove, must have the optionof revision. What to do in such a casemust be an open question.

Davidson's theory gives and is designed to give an a priori andunequivocal answer in this and similar cases of apparent floutingof conceptual dependencies among psychological notions. Ascrip-tions of intention and belief, like all psychological ascriptions, arepart of a complex interpretive project, the point of which is torender as complete and coherent a construal of the agent's be-havior as possible. As a consequence, the applicability of psycho-logical concepts is as much constrained by those concepts' relation-ships with each other as by the behavioral evidence they are in-voked to explain. To link individual psychological ascriptions withevidentiary conditions outside the intentional realm would be, onDavidson's view, to destroy the flexibility of application necessaryfor the normative goal to be realized, and to introduce the possi-bility of conflicts like the one described above.

This interpretation, with its echoes of the conceptualists' em-phasis on logical connections, illuminates Davidson's remarksabout "the different natures of the evidence" for physical and psy-chological claims. He says in ME that if denying the possibility ofpsycho-physical laws is not to be judged a prioristic,

182


32/36

ANOMALOUSMONISM

it can only be because to allow the possibilityof such laws wouldamountto changingthe subject.By changingthe subjectI meanhere:deciding not to acceptthe criterionof the mentalin termsof the vo-cabularyof the propositionalattitudes(ME,p. 219).

I conclude that Davidson is not only arguing against strong reduc-tionism, but is opposed in principle to there being any systematicnomic relation between psychological and physical kinds. I'll con-clude by rehearsing the consequences of this stance for his positionon the explanatory force of rationalizations.

VI. THE PROBLEM: PART Two

Davidson's model of rationalization has it that the following istrue: if RAT 1, rather than RAT 2, gives the reason why Her-mione made the remark, that's true because he primary reason em-bodied in RAT 1 is identicalwith some physical event c, and c is thecause of Hermione's making the remark. The rationalization,"Hermione made the remark because she wanted to foster philo-sophical progress" is metaphysically underwritten by a physical,causal explanation in the same way ordinary explanations of dis-asters in terms of hurricanes are underwritten by more precise sci-entific accounts of the destructive force of water and wind.

The first part of the problem concerned the difficulty Davidsonmust have in explaining why reference to the causally efficaciousphysical events in psychological terms sometimes does and some-times does not serve to explain the ensuing event, even if the psy-chological attributions were true, and even if they serve to rational-ize the event qua action. I argued that what Davidson needed wasan account of how reasons can cause in virtueof theirrationalproper-ties, and that doing so would require a property theory of mentalstates. Now we can see why Davidson cannot accept such a friendlyamendment. But additionally, we can now see that his "solution"tothe problem of explanatory force in terms of a causal theory ofaction has utterly dissolved.

On Davidson's view it cannot be a genuine fact that any physicalevent is identical with any mental event, for the application ofmental predicates to events is always radically indeterminate.Whether the neurophysiological event c is identical with this pri-mary reason or that one, must depend upon which psychological

183


33/36

LOUISE ANTONY

ascription we're inclined to make at a particular point in an "essen-tially evolving interpretation" of Hermione's behavior. Today we'llidentify c with Hermione's wanting to foster philosophical prog-ress, but tomorrow we'll find that we get a more cogent view of heroverall if we identify the cause of her making the remark with herwanting to impress her professor.

The point is not merely epistemological: the pertinent identitiesare non-existent, not merely unknowable. Since there are no gen-uine facts about the contents of Hermione's mental states, therecan be no genuine facts about their relations with physical states. Ifit is a fact that c is identical with one rather than another mentalevent, it is a fact in virtue of our decision to accept one rather thananother proposed rationalization of Hermione's action. The inter-pretation-relativity of psychological attribution infects psycho-physical identities, even token ones.

In short, it is the acceptability of particular rationalizations thatmetaphysically ground psycho-physical identities, and not theother way around. Davidson cannot explain the explanatory forceof the rational "because" by appeal to underlying causal connec-tions because there is no objective attachment between the inter-pretive psychological story we decide to tell and the physiologicalgoings-on in a person's body. A reason cannot rationalize an actionin virtue of some causal link between some physical event and itsphysical effect, simply because there is no fact of the matter whichreason the cause is identical with. Davidson turns out to be com-mitted to just what the anti-causalists have been trying to tell himall along: the goings-on in your body have nothing to do with whyyou do what you do.

VII. CONCLUSIONSDavidson's commitment to the indeterminacy thesis makes his

anomalous monism irreconcilable with his solution to the problemof explanatory force. But since anomalous monism is internallyconsistent, one option for Davidson would be simply to give up hiscausal account of rationalization. Indeed, it might seem reasonablefor Davidson to give up altogether on the problem of explanatoryforce-why would someone who believes in indeterminacy thinkthere's a fact of the matter about why Hermione made the remark,anyway?

184


34/36

ANOMALOUSMONISM

But those of us without Davidson's Quinean commitments will, Ithink, want to retain Davidson's solution to the problem of explan-atory force, for the problem is compelling, and the solution is agood one. What we need, then, is a good functionalist account ofpsychological states.

Right. And what the Middle East needs is a good regional peaceplan. Mindful of the improbability of my saying anything eithersubstantive or original on the topic, let me conclude with a briefadvertisement for my personal favorite functionalist theory.Jerry Fodor has long argued that beliefs and other cognitivestates are to be analyzed as relations between an agent and physi-cally realized tokens of an internal medium of representation andcomputation, affectionately known as "mentalese." Because thesetokens are physically realized, they have causal properties; becausethey realize linguistic items, they have syntacticand semanticproper-ties. One begins to see, on this view, how beliefs and desires can

both cause and rationalize human behavior, and-because thesetokens are functionally individuated-how one could get a theoryof theright way for reasons to cause actions.29There are other, non-representational versions of functiona-lism, to be sure, but it is the strong representationalism of Fodor'sview that particularly suits it for use in an account of the explana-tory force of rationalizations. I should, however, own up to the factthat this same feature makes it a very real possibility that there willbe just the kinds of evidentiary conflicts-between physical evi-dence and considerations of cogency-that Davidson warns usof.30 Fortunately, I don't think that's a problem.

Davidson's admonitions about retaining "allegiance" to "propersources of evidence" are really echoes of the old anti-materialistarguments against the claim that mental states are physical states.Where they appealed to forms of privileged access ("no amount of

29The canonical arguments can be found in Fodor, 1975, and Fodor,1978. As I noted earlier, the later work is of particular relevance here.30There are brands of functionalism that don't carry this risk. Dennett,for example, is a functionalist; but because he denies that the realizingphysical states map in any interesting way onto the contents of ordinarypsychological ascriptions, he doesn't have to contend with Davidson'sproblem. But neither, I think, does he have a satisfactory answer to theproblem of explanatory force. Dennett's, incidentally, is the theory ofmind Davidson should really have.185


35/36

LOUISE ANTONY

physical evidence could persuade me I wasn't in pain if I knew Iwas!"), Davidson appeals to "considerations of cogency." The gen-eral response is the same in both cases. Scientific theories must payproper respect to the pre-theoretic judgments that, after all, give aworking definition of their domains; but, as a famous philosopheronce said, nothing is immune from revision.

A theory of pain that overrides informants' judgments abouttheir own feelings too often is a bad theory of pain. We can makethe required degree of match as high as we like. But at some point,we may well allow that, strange as it may seem, people are some-times wrong about whether or not they're in pain.

If representationalism is correct, it could well turn out that phys-ical evidence sometimes competes with considerations of cogencyin the ascription of psychological states. But the unsatisfactory feelof this result is mitigated by the consideration that we can ensure,as with our imagined theory of pain, that the number of such con-flicts is acceptably small. Specifically, quasi-conceptual connectionsbetween propositional attitudes, like the one between intentionand belief suggested in Section II, can be preserved as constraintson the interpretation of mentalese tokens. Thus, while it remains apossibility for the normal basis for ascription of a psychologicalstate (Hermione believes that P), to conflict with some piece ofphysical evidence (there's no token of P in Hermione's belief box),conditions of theoretical adequacy on the interpretation of a real-izing physical state can ensure that this happens only rarely.

Obviously, functionalism in general, and representationalism inparticular, need a more thorough defense than I can possibly offerhere. Perhaps there can be no psychological theory of the sort I'vedescribed-there are many who think there can't, and for manydifferent reasons.31 My aim has been simply to show what's neededto make Davidson's original project work. If you want to accountfor the explanatory force of rationalizations, you can't be aQuinean about psychology. And that's the truth.North Carolina State University

3'See, for example, Stich, 1983; Burge, 1979; and Dennett, 1980.186


36/36

ANOMALOUS MONISM

REFERENCESAnscombe, Elizabeth (1976). Intention. Ithaca, N.Y., Cornell UniversityPress.Brand, Myles (1984). Intending and Acting: Toward a Naturalized ActionTheory.Cambridge, Mass., The MIT Press.Burge, Tyler (1979). "Individualism and the Mental," Midwest Studies inPhilosophy:Studiesin Metaphysics,Vol. IV. Edited by P. A. French, T. E.Uehling, and H. K. Wettstein. Minneapolis, Minn., University of Min-nesota Press.Cummins, Robert (1983). The Nature of PsychologicalExplanation. Cam-bridge, Mass., Bradford Books/The MIT Press.Davidson, Donald (1980). Essays on Actions and Events. New York, N.Y.,Oxford University Press.Dennett, Daniel (1980). "Beyond Belief," Thoughtand Object:Essayson In-tentionality.Edited by Andrew Woodfield. Oxford, England, ClarendonPress of Oxford University Press.Dray, W. H. (1957). Lawsand Explanationin History.Oxford, England, Ox-ford University Press.,(1963). "Historical Explanation of Actions Reconsidered," ThePhi-losophyof History.Edited by Patrick Gardiner. Oxford, England, Oxford

University Press (1974).Fodor, J. (1975). The Language of Thought.New York, N.Y., Thomas Cro-well., (1978). "Propositional Attitudes," TheMonist 61.Hempel, Carl G. (1965). Aspectsof ScientificExplanation. New York, N.Y.,Free Press.Lycan, William G. (1981). "Psychological Laws," Mind, Brain, Function:Essays in the Philosophy of Mind. Edited by J. J. Biro and Robert W.Shahan. Norman, Okla., University of Oklahoma Press.Melden, A. I. (1961). Free Action. Boston, Mass., Routledge and KeganPaul.Putnam, Hilary (1960). "Minds and Machines," reprinted in HilaryPutnam, Mind, Language, and Reality:PhilosophicalPapers, Vol. 2. Cam-bridge, England, Cambridge University Press, 1975.Quine, W. V. 0. (1970). "On the Reasons for Indeterminacy of Transla-tion,"Journal of Philosophy67, pp. 178-183., (1978). "Factsof the Matter,"TheSouthwesternJournalf Philosophy .Ryle, Gilbert (1940). The Conceptof Mind. New York, N.Y., Barnes andNoble.Shoemaker, Sydney (1981). "Some Varieties of Functionalism," reprintedin Shoemaker (1984)

Antony Anomalousness Mental

Documents