Lorenzo Peña & Manuel Liz

Review of
Philip Pettit & John McDowell (eds)
Subject, Thought and Context
Oxford: Clarendon, 1986. Pp. 300.
Philosophia 24/1-2 Bar Ilan University (Israel), december 1994 pp. 235-48. ISSN 0048-3893
The main starting point of many of the contributions collected into the book is the kind of Twin Earth considerations, along with T. Burge's and others' remarks against meaning individualism. We prefer to focus here on the former issue. The editors' Introduction (pp. 1-15) makes it abundantly clear that they accept at face value Putnam's claim about H2O in this world (call it `Houto') and XYZ (or whatever) in an alternative world (call it `Ekeino') being different stuffs. Of course the example is not by itself that important, since many other cases could be invented. Still, in the same way as that famous example has served to buttress Putnam's dictum about meaning not being in the head, the example's weakness detract plausibility from that sort of considerations. Now in fact there are such weaknesses. If the aquatic stuff in Houto is quite similar to the one in Ekeino, they are linked by the same causal chains to any other things and events in their respective worlds. Then their different composition may be held to be immaterial for individuation (in fact that point has been made by some authors). That stuff may be thought to be composed by molecules of H2O in Houto and by molecules (or parts) of XYZ in Ekeino. (We could think that water's existence supervenes upon molecules of H2O being gathered in Houto and upon molecules, or whatever, of XYZ being similarly related in Ekeino.) In fact the statement `Water is H2O can be construed in at least two ways: (1) water is composed of molecules each of which comprises two atoms of Hydrogen and one of Oxygen; (2) water is identical with H2O. In the latter construal, `H2O' may be taken to be a rigid designator of the aquatic stuff, which leaver room for there to be a possibility that H2O should not be composed by molecules each of which comprises two H atoms and one O atom. That hypothesis may be in need of further support but on the face of it it is far from obviously false. Thus one seems allowed to be skeptical about the starting point of the debate between such as emphasize broad content and those who think that the basic semantic entities are narrow contents, which, as against the broad ones, would fail to be world-dependent or world-oriented. Our foregoing remarks on individuation casts doubt on the basic argument to the effect that a subject in Houto wanting to drink water and one in Ekeino wanting to drink Ekeino-water would be indistinguishable, indiscernible, perfect replicas of one another, whereas the objects of their wanting would be different.

Admittedly, if our doubts are grounded, something not entirely unlike the radical line of response to Putnam's «intuition» seems to be in order, namely that which (according to the editors, p. 3) `takes Putnam's conclusion to call into question altogether the credentials of that conception of the psychological', i.e. the conception which posits psychological states `in the narrow sense associated with the conception which has come under pressure' -- while the less radical line of response takes meaning to be a composite state of affairs of which the [narrowly conceived] psychological state would be but one component (ibid.). Most contributors to this volume lean towards the radical line. Nevertheless, should our doubts be grounded, the challenge (or «pressure») would vanish and we could maintain that there is no such difference between narrow and broad content, or in other words that what is in the head is constituted also by relations with outside reality in such a way that, were the world different, what is in the subject's head would also be different pari passu. (Take a subject in Houto, Ram, and one in Ekeino, Psuam; is not the content or meaning of their beliefs or wantings about their respective aquatic stuff different in virtue of the mere fact that Ram is thinking about water in Houto while Psuam means water in Ekeino? Perhaps not, since, unless and until a cogent argument to the contrary is available, it seems safe to believe that the state of wanting water may be the same here and there. If you are bent on imposing very strong constraints on individuation, as D. Lewis does for individuals, we are liable in the end to espouse a rejection of a property or a state of affairs remaining the same through different possible worlds, which would spell the damnation of possible-world talk.)

It is regrettable that throughout the book no discussion is offered of the relationship between Twin Earth considerations and Quine's thesis of inscrutability of reference. Yet, as Putnam himself has remarked, there are striking similarities. With a denotation theory somehow like that proposed by Hartry Field (which Putnam deplores as positing some kind of brute, non-analyzable, metaphysical fact), there would be no need either for a duality between narrow and broad content or for the ontological relativity yielded by Quine's inscrutability thesis. What is common to both narrow/broad dichotomists and Quine is the contention that two subjects can be indistinguishable -- in relevant respects -- as regards whatever is in their heads, or nervous systems, or verbal behaviour, without its being true that they mean the same -- whether that truth failure is, as Quine thinks, ensuant upon relativity (i.e. upon its failing to be a fact of the matter that the subjects refer to this or to that) or, as the dichotomists claim, upon the fact that meaning transcends what is in the head. From a different outlook, more congenial to Field's kind of approach, if the semantical relations are different, the subjects are distinguishable, too. But that distinguishability can be conceived either as environment bound -- as, according to most contributors to the anthology, is the case for broad content -- or as constituting an additional, perhaps ultra-observational fact. If you take the former line, you can think either that water and Twin-water are the same stuff (as our remarks suggest) or that in Houto's environment the subject is wanting water while in Ekeino the subject wants Twin-water, distinguishability being then achieved by resorting to broader observations (since surely there must be some observations showing the different composition); in the latter case what is in the subjects' heads would also be different. If instead you take the alternative line of regarding the relevant distinguishability as a perhaps non-observable fact, you can think that two subjects can definitely differ in their mental states not owing to differences in their respective environments nor in virtue of empirically discoverable differences in their neuronal movements but just as an irreducible difference, which can be shown to be non-trivial through counterfactual mental experiments, and which anyway can be posited thanks to its elucidating or explicative fruitfulness upon a number of assumptions. (The assumptions can be challenged, of course, in the same way as those upon which a physical non-observable entity or situation is taken to be [more] fruitful in an explanation-oriented account.)

It is failure to entertain or debate such alternatives what constitutes the chief drawback of the book we are commenting on. In other respects the book contains a lot of interesting discussions.

In his paper «Broad-Minded Explanation and Psychology» (pp. 17-58), Philip Pettit defines the broad-narrow divide as follows. A token or particular state is narrow if the presence of that type of state is guaranteed by the context-independent character of the subject; otherwise, the state is broad. The narrow psychological states are those which supervene on how it is with the subjects independently of the nature of their environments. In the ordinary explanations of knowledge, intentional states involving a rule-following disposition, actions, and so on, such a context-dependent character of the broad psychological states is indispensable. If reference to those states is eliminated, the force of the explanation is bound to be lost. Pettit's core position is that, if for some purpose these ordinary explanations are advantageous -- and he thinks they are --, broad psychological states must be indispensable for a correct account of the mental. The standard account of action-explanation, in particular the functionalist variety, tells us that any explanation of an event is a regularizing account incorporating it into a theoretical frame trying to describe how a piece of the world can work. Narrow psychological states are sufficient for such standard accounts. But to understand and explain a mental state full of intentional content is another thing. It is a normalizing account providing explanation upon the assumption of some rationalizing rules. Within the framework of a normalizing account those rationalizing rules resort to broad psychological states in order to explain our mental life and intentional actions. Narrow psychological states are preconditions of these broad psychological phenomena. Thus objects characteristic of psychology are rationalized objects rather than pieces of the physical world. They are the result of our folkpsychological explanatory practices. And what we do in these practices we wouldn't be able to do in other ways failing to include reference to the broad context surrounding the subject. Pettit's arguments are well-taken and illuminating, but the sort of doubts set forth at the beginning of this critical notice seem to apply to his paper as much as to others.

In G. McCulloch's «Scientism, mind, and meaning» (pp. 59-94), the classical or Fregean notion of meaning is shown to be intimately bound up with the nature of folkpsychology. Folkpsychology and that classical notion of meaning stand or fall together. The epistemological status of folkpsychology is questioned by the «superpsychology» both of the functionalist and neurophysiological approaches. McCulloch maintains that folkpsychology offers an independent and perfectly valid descriptive network which facilitates understanding ourselves and others as rational subjects in a way beyond the scope of «superpsychology». The classical meanings of folkpsychology with which he tries to describe and understand the intentional life and actions aren't causal agents producing and explaining them. The author clarifies this point through the phrase «a pattern of engagement» (PE), which denotes a mixture of a descriptive stance and an intervention strategy with respect to some characteristic systems. A knowledgeable engager of a PE can understand its descriptive stance, perform adequate interventions on the relevant systems and give an account of the special ends the PE is aimed at. Thus, knowledgeability involves certain practical abilities. As regards PEs, folkpsychology and «superpsychology» are very dissimilar. The descriptive stance of the folkpsychological PE is not specified in physical or input-output terms, but in irreducibly intentional ones. Typical engagement in a PE is reciprocal, in the sense that such systems as confront a given engager are themselves usually engager in that very same PE. The engagers in that PE think of themselves as «we» (in a sense very near Sellars's «Philosophy and the Scientific Image of Man», in his Science, Perception and Reality, Routledge and Kegan Paul, London, 1963). Being engaged in the folkphyschological PE is not comparable to being engaged in another explanatory PE. Therefore, «superpsychology» cannot do the job folkpsychology does. «Superpsychologists» are also committed to prescientific folkpsychological PEs. Their science's value is relative to the «we» we are. The main source of distortion about all these topics is an excessively simple conception of folkpsychology taking the intentional to have a causal-explanatory role similar to the descriptive stances other PEs have. Inspite of McCulloch's strenuous endeavours, there remain some obscurities about all those technical and intricate notions. Moreover, if the sort of doubts put forward at the beginning of this notice are warranted, McCulloch's assumed dichotomy is far from obvious.

Jennifer Hornsby, in her paper «Physicalist thinking and conceptions of behaviour» (pp. 95-116), takes as her starting ground the obvious point that there are two main pictures from which we can try to understand what we are as knowing and acting creatures. One is based on what is involved when we assign mental contents to ourselves and others. Such a picture emphasizes the existence of objects of perceptions, subjects of beliefs, desires, etc., and effects of intentional actions. The other view is offered by the scientific image of ourselves as neurophysiological systems with sensory surfaces, neurophysiological states and body movements. A persistent problem in the philosophy of mind is how to articulate these two pictures into a coherent synopsis. Hornsby thinks that certain conception of «behaviour» can make some job, subsuming both the brain's outputs and the person's outputs under a common area. A purely bodily notion of behaviour (for example, Jaegwon Kim's «Psychophysical Supervenience» Philosophical Studies XLI, 1982; or Loar's Mind and Meaning, Cambridge U.P., 1981) collapses into a simple characterization of «ability-to-do-things». Thus -- Hornsby contends -- «to-do-things», to act, must be an extrapsychological topic, not the proper business for psychological theory to explain or predict. (But is such a contention not question-begging against psycho-somatic monism of any sort?) On the other hand, according to Hornsby, if we employ the ordinary and richer conception of behaviour (a conception present in some behavioral authors such as Ryle), we may become skeptical about circumscribing mental states -- as functionalism does -- to explaining proximal movements of the body, which may play no relevant role in real intentional life. How exactly we move our fingers when we turn on lights, for example, may be of no concern in a sufficient explanation of our action of turning on lights. Usually, folkpsychology doesn't assign them an explanatory role in these cases. Therefore, folkpsychology might be able to find out the relevant cognitive schemes to analyze. This leads to a rationalization work in psychological-theory explanations. Making an agent's beliefs or actions rational requires looking for the objects of the psychological explanations elsewhere than at the simple junction between the central nervous system and the world. Hornsby's arguments are very well developed, but again she is not free from all question-begging against psycho-somatic monism. A broader and suppler view of bodily motion may accommodate most of her explanation constraints.

Tyler Burge's «Cartesian Error and the objectivity of Perception» (pp. 117-36) claims that individualism is a theory originated with Descartes and nowadays re-emerged in current philosophy of mind. It is the view that the mental states of an individual can in principle be individuated in complete independence of the individual's physical and social environments. Individualism has been motivated by a wide variety of ontological and epistemological intuitions. Burge's work discusses them and proposes an argument against individualism centred on perception. According to Burge, individualism and its use of mental experiments is ensuant upon at least one of these conflations: 1) mistaking questions of counterfactually evaluating one's thoughts for questions of what thoughts one would think if one were in the counterfactual situation; and 2) mistaking our being authoritative about our actual thoughts, and about what our thoughts would be in any relevant counterfactual situation, for our being actually authoritative about certain thoughts that we would be thinking regardless of what actual or counterfactual situation we would be in.

John McDowell's «Singular Thought and the Extent of Inner Space» (pp. 137-68) tries to improve upon Russell's Theory of Descriptions and his epistemological notion of acquaintance (as opposed to knowledge by description). According to Russell, propositional attitudes whose contents are singular propositions entail that what exists in a mind having those attitudes is itself partly determined by which objects exist in the world. One can entertain such propositional contents only if he knows which objects the proposition is about. Knowledge by acquaintance is direct and immediate, it is a presence of the object itself to mind. The Russellian paradigm of acquaintance is perception. McDowell's paper analyses those topics on the background of Evan's account (in The Varieties of Reference, Oxford, Clarendon, 1982, ch. 6) of perceptual demonstrative modes of presentation. One can be mistaken, McDowell says, about being acquainted with a certain object. But this means no epistemological disaster about objectivity. The ultimate basis for Russell's view is a Cartesian conception of inner life, in accordance with which there is a serious and unavoidable question about how experience mirrors the world we live in. Object-dependence may fail to be an intrinsic feature of mind in certain causal descriptions, functionalist theses may be right about this, but ascription of mental contents in relation to their object-dependence takes into account not only that intrinsic nature but also external relations. This is very important because, if we exist and act in a world and think about it, our psychological self-understanding must not overlook that whole affair. Content in general, and specially singular thought, requires directedness towards reality. The Cartesian and the two-component functionalist picture forget that cognitive space incorporates, in a variety of ways, relevant portions of the «external» world.

Crispin Wright's «Does Philosophical Investigations I. 258-60 Suggest a Cogent Argument against Private Language?» (pp. 209-66) claims that the best way of defining a private language is not that according to which a private language is one such that necessarily only one person can understand it, but one no two people can have adequate reasons to believe they share. The inaccessibility might be compatible with the casual but coincidental understanding. The rejection of private languages has deep philosophical consequences against solipsism, natural pre-philosophical conceptions of mental life and the intended first/third person asymmetry. The issue really is a pivotal one in epistemology, the philosophy of mind, and the philosophy of language. A cogent argument against private language should be effective only in the moulding of reasonable people's opinions without adducing merely practical obstacles. Thus, no argument generating a «paradox» (for example, Kripke's skeptical paradox) can be cogent in this sense. Also, a cogent argument must leave communal language in its place. Any sort of epistemological skepticism about language in general must be kept apart from this kind of cogent argument. According to C. Wright, Philosophical Investigations, I. 258-60 points out a suggesting approach to the problem. The «seems right / is right» contrast is decisive. If the proponent of the possibility of a private language says that this contrast is made by reference to practical criteria, such a prospect will rest at best on very contingent circumstances. On the other hand, if he claims that the distinction can be drawn purely truth-conditionally, he commits himself to the unfortunate consequence that there may fail to be any satisfactory grounds for his belief in the possibility of private languages. In both cases, Wright says, «the ball is in the opposition's court»

Christopher Peacocke's «What Determines Truth Conditions?» (pp. 181-207) is an attempt to find a middle course between truth-conditional theories of sense and conceptual role semantics or, more specifically, acceptance-conditions semantics. He himself summarizes his position with these words: `manifestation without verificationism'. Peacocke's contribution is so outstanding that we shall devote the remainder of our review to commenting on it. Peacocke's main idea is that any content is endowed with some normative condition concerning its acceptance or rejection, although the paper tells us nothing on rejection conditions. Thus a subject's acceptance of a content (i.e. his belief that such is the case) may fail to comply with those conditions. Peacocke's account is not intended to apply to such cases at all. Now, what is the nature of the involved norms? It is strongly suggested that they are rationality (or reasonableness) requirements, a priori coherence constraints (cf. pp. 186, 187-8 -- where the notion of understanding, as opposed to that of believing, seems to be associated with rationality, the understood utterance standing by the standards bound to apply to it --, p. 188, p. 200). The suggestion is that only whatever is consistent is rational and that only whatever is irrational is a priori certainly false and so deserves rejection. Yet, two beliefs are inconsistent, according to Peacocke's approach, not only when the negation of the one can be deduced from the other but also when the former entails the negation of some acceptance conditions of the latter, and more specifically when it runs afoul of what Peacocke calls the canonical commitments of the latter. We are going to have a closer look at that pivotal notion. Anyway, what matters now is to point out that the approach encompasses laying down certain a priori nondeductive inference rules such that whoever accepts a certain content is bound to implicitly accept whatever can in accordance with those rules be inferred from the content under consideration. But it comprises more than that. The normative acceptance conditions are partly internalistic and partly externalistic. Which means that a subject cannot be said to comply with those conditions, even if he implicitly accepts the contents that can be inferred from what he accepts in accordance with the rules in question, unless those contents are in fact true -- at least some of them or on certain conditions. Such a strategy is what allows Peacocke to find a path from acceptance conditions to truth conditions, from sense to reference. There is a class of contents which are true if they are accepted with no infringement of the normative acceptance conditions. This is a consequence derivable from the first part of Peacocke's main Conjecture (p. 182). The second part is much more ambitious: `the truth conditions of contents outside this class are determined ultimately by their relations to contents inside this initial class'. Such a claim is what makes Peacocke's approach close to verificationism. The claim is not argued for in this paper, though.

In order to avoid trivializing the first part of the conjecture, Peacocke is cautious to restrict acceptance conditions `to those for which we have an immediate account of how a thinker can manifest the fact that he is following the norms' (p. 183). The class of contents to which the paper devotes most attention is that of observational contents of a simple structure. The author specifies what he calls `the spectrum of canonical commitments' of one who judges such a content. There are many important issues about that specification. The paper does not follow what might be thought to be the normal way: to define first `commitment', then `canonical' and finally `spectrum'. In fact it is not clear what one is committed to by incurring any such commitment, the nature of the commitment itself being far from clear. Peacocke talks about commitments to believing (p. 190, p. 200), to judging (p. 199, p. 204), to accepting (p. 200). But then one's would be commitments fulfilled or met provided one, perhaps implicitly, did believe all those contents he was committed to. Still, such is far from being the case. As Peacocke makes abundantly clear, the fulfilment of the commitments does in addition comprise the contents committed-to being true (see p. 188, p. 190, p. 192, p. 194, p. 204). Yet canonical commitments can be conjoined (p. 201) and at least some of them are conditionals (p. 202). Thus they must be either contents or sentences or the like. Nevertheless Peacocke talks about their being met or fulfilled rather than their being true. It seems in fact that a commitment is fulfilled if not only the contents committed-to are in fact true but moreover the subject who incurs the commitment does somehow accept those contents.

However the paper's definition of the above-mentioned «spectrum» is expressed by a universally quantified statement. Some other things the author says, though, clearly imply that the spectrum is a set, each of its members being some «fact» or circumstance expressible through a subjunctive conditional.

There are many other problems about Peacocke's approach as sketched out in this paper (Peacocke may have dispelled them in later papers and books, but we seem entitled to remain cautious, since the difficulties are so central and so serious). For one thing, there are too many terms which are being used in a clearly technical way but for which no elucidation is provided: `stimulation' (p. 188), `experience' (p. 185 ff.), `cause' (or `being responsible for': pp. 188 ff.), `obtaining a class of descriptions' (p. 189), `concept' (p. 184), `falling under a concept' (ibid. -- notice that the intended meaning cannot be Frege's for whom concepts are functions and are not parts of thoughts), and so on. For another thing, the Fregeoid characterization of contents (pp. 184-5) seems to encounter difficulties of a Tractarian kind; e.g. combination ways are needed which in turn seem to be needed as parts of the content, an infinite regress being thus triggered, and, withal, no content being able to have several structures or parsings, with categorial unlevelments being resorted-to, which in turn entails sayability difficulties, as is widely known.

There are some more important issues, though. One of them is that the determination of truth conditions by acceptance conditions is attained by paying the price of an externalistic view of the «commitments», as has been shown above. Now, the commitments under consideration concern evidence. A subject who accepts an observational content is committed to such contents being the case as, should they fail, the content he accepts would become warrantless, or at least he would lack warrant to accept it. Thus, externalistic warrant conditions determine truth conditions. Yes, but most epistemologists harbour misgivings towards externalistic warrant, since they think that only what is in the subject's head can give warrant to his beliefs.

Another difficulty is that the thus laid down acceptance conditions hardly resemble anything the average subject would be prepared to volunteer or even acquiesce to, since they essentially involve technical terms elucidation of which calls for a hard and lengthy theoretical development. Peacocke gives us assurance that `[t]he thinker need not, of course, have these specifications of canonical commitments consciously in mind, or be capable of formulating them' (p. 186). Yet he incurs the commitment. But, since such commitments are normative acceptance conditions, the subject is surely unwarranted in accepting a certain content when he fails to incur those commitments, i.e. to at least implicitly believe they are fulfilled. Yet the present writers find those commitments hard to understand, and doubt whether they «incur» them upon their acceptance of observational contents.

The third main difficulty concerns the notion of an observational concept. Upon a more austere view of the matter observations and observation sentences seem more likely to be dealt with holophrastically, for reasons Quine has dwelt on time and again. Now, if deprived of the structural conception of observational content set up on p. 185, the road from acceptance conditions to truth conditions argued for on pp. 190-1 seems to be blocked. (By the way the argument on p. 190 obviously presupposes a number of premises kept unuttered: they may be assumed in virtue of a suitableness constraint referred to on p. 189. In fact it is not entirely dear that the paper proves its claim that observational contents are individuated by their canonical commitments, and so that no two different contents can share all those commitments; notice that if the claim is true the content expressible by `This table is quadrangular' seems bound to be the same which is expressible by `This table is four-sided'.)

There are some philosophical presuppositions underlying Peacocke's approach which, widely accepted though they are, deserve being discussed all the same. For instance, that `[i]n any plausible theory, there will be something which is the object of propositional attitudes' (p. 182), which, unless it is a (definitional?) constraint on plausibility, can be shown not to be the case since de dicto attitudes can be plausibly thought to reduce to de re attitudes (see for instance D. Dennett's «Beyond Belief», in Thought and Object, ed. by A. Woodfield, Clarendon, 1982, pp. 1-96, esp. p. 87); in accordance with some sort or reductive approaches, some regenerated Russellian theory of belief could turn out to be, after all, the best account. (And a weakened and subtler notion of acquaintance, or presence-to-mind, may allow us to dispose of Fregean or skew-Fregean presentation modes and the like, which Peacocke cherishes; only such immediacy as is required for acquaintance can be attained through other things at several removes, which means some kind of «mediated immediacy», as Hegelians and Hermeneutic philosophers are fond or saying.) Another assumption which Peacocke takes for granted is that there are no degrees of observationality (a contention which Quine has endeavoured to refute), or of acceptance, or of experiencing-as, or of being-committed-to, or of a commitment being met, or of suitableness, rationality and so on. All or nothing, no in-betweenness is allowed for. Likewise, truth is assumed to lack any degrees at all. (In connection with that, coherence seems to be equated with negational consistency, inspite of the existence of paraconsistent logics.) Then, it is also assumed (p. 188) that a causal link is required for the fulfilment of acceptance conditions, which again seems to be dubious from an internalistic conception of warrant. (In fact it can even be argued that there is no need for the subject to believe that the object's being, say, red is what causes his viewing it as red. A Malebranchian and a Leibnizian may be warranted in believing that that object yonder is red.) Finally, such as reject the cleavage between a priori and a posteriori truths are likely to be unsympathetic towards Peacocke's approach.

All those problems do not detract from the paper's great interest. Peacocke's later writings (esp. his Thoughts: An Essay on Content, Blackwell, 1986) do elaborate and improve on this paper. But the problems just discussed show that the kind of motivations prompting the essays collected into the book under consideration are likely to be regarded as in need of a deeper elucidation by such as have been more or less influenced by Quinean philosophizing.

The volume also comprises two other papers: J. Lear's «Transcendental Anthropology» and D. Wiggins's «On singling out an object determinately», the latter concerning the issue of indeterminate identity, to which we shall devote a later paper elsewhere.

Inspite of a few minor weaknesses, Pettit & McDowell's collection of essays is one of the books most scholars interested in the confines of philosophy of language and philosophy of mind will find worth reading. A subject index would have been welcome, and in fact its absence is quite annoying. Some misprints are irksome (e.g. on p. 196, line 12 up, `objectional' for `objectual').