MAI TREBUIE DOAR SA CITESTI CARRUTHERS SI GATA Either twin-water is water, or twin-beliefs are not beliefs
This essay addresses the debate over the individuation of mental states. Putnam's "The Meaning of 'Meaning' " is a plea for externalism and for the distinction between broad and narrow content. By "externalism about mental states" I mean the doctrine that holds that properties relevant for the individuation of mental states are not intrinsic to the mental states (i.e., they are facts about the natural or social environment). "Externalism with regard to mental content says that in order to have certain types of intentional mental states (e.g. beliefs), it is necessary to be related to the environment in the right way" (Joe Lau) Following Putnam, it is usual to distinguish between two components of mental content: one intrinsic to the mental state itself, the "narrow content" and other exterior to it, the "broad content". Narrow individuation of psychological states depends only on properties of the person who has that mental state; broad individuation depends on environmental facts � social or natural facts. The main advantage of externalism is that it accomodates a causal account of reference: to mean a term is to be situated in certain ways to the environment. Its main disadvantage is that it conflicts with the powerful intuition that we have first-person authority about the content of our thoughts. While Putnam�s The Meaning of Meaning makes psychological content dependent on the hidden essence of natural kinds, Tyler Burge�s Individualism and the Mental generalizes this approach, making the content of most of our thoughts dependent on social practices in our community. Burge gives the example of someone who believes he has arthritis in the thigh, unaware of the fact that arthritis is defined as an ailment of joints. Since �arthritis in the thighs� is an empty concept (i.e., without extension), all his beliefs about arthritis in the thighs are false. But their falsity is only due to social environment, to shared linguistic pratices. If his linguistic community had accepted that there is such a thing as �arthritis in the thighs�, then his beliefs would have been true or false; they would have been indeed about arthritis. Burge�s moral, like Putnam�s, is that thoughts do not contain their truth-conditions essentially: it is social facts that ascribe to a concept its intension and to a belief its truth conditions.
Putnam's The Meaning of �Meaning� was originally intended as a plea for semantic externalism, i.e. the thesis that the meaning of a word is not a feature of the mind of the individual who "means" that word (see the "Meanings ain't in the head" section of the �Meaning of Meaning� and the distinction between words' meaning and what the words mean). It is thus relevant for linguistic content . In particular. Putnam�s mental experiment about twin earth has consequences for the Fregeean distinction between intension and extension: Putnam�s examples are the symetrical of the famous Hesperus-Phosphorus examle. As Frege proved (with the Hesperus-Phosphorus example) that different senses can correspond to one and the same reference, so Putnam�s twin earth proves that different extensions (references) can correspond to the same sense (intension). But Putnam's article has consequences that go far beyond philosophy of language: it is relevant for the debate concerning the individuation of mental states � at least if we assume that both linguistic entities (sentences) and mental entities (thoughts, beliefs) have �content� I will present Fodor's reaction to Putnam's article, as formulated in "Methodological solipsism", "Psychosemantics" and "The Elm and the Expert", Then I will attempt to devise some arguments against Putnam's externalism. I will not dwell on the mental experiments presented in in The Meaning of Meaning, which I assume to be known.
My purpose is to present and analyze some arguments offered by the two sides of the debate � broad or narrow individuation of mental states. I think that the arguments for narrow individuation are stronger � which does not mean, of course, that I agree with all the arguments against broad individuation. In particular, I will point to the weaknesses of two common arguments for narrow individuation � the argument from �causal powers� and that from �relativization to context�. I will present different arguments for narrow individuation that seem to me more convincing.
The debate between narrow versus broad individuation of mental states should not be confused with the question whether mental states are caused by factors extrinsic to the individual or not. A partisan of narrow individuation of mental states need not have any problem accepting that some mental states are caused by environmental or social factors. Rather, the debate is whether environmental or social factors essentially characterize the identity of mental states.
I. Fodor�s objections to broad individuation of mental states
I.1. Independence of psychological taxonomy
Jerry Fodor�s first reaction to Putnam�s plea for externalism appears in Methodological Solipsism . There he holds that psychology as a science needs narrow content (and thus be �internalist�), since broad individuation of mental states (�externalism�) would require that mental states� criteria of individuation be dependent on criteria of individuation of environmental facts. Thus, psychological taxonomy of intentional mental states should be dependent on the taxonomy of non-mental facts: of natural kinds, for example. If externalism were true, then we would need to wait for an adequate scientific taxonomy of natural kinds (and � via Burge, of relevant social facts) in order to provide a taxonomy of psychological states
I.2. Supervenience
Another basic objection to externalism about mental states, put forward in Psychosemantics, is that externalism violates supervenience. Mental states, broadly individuated, do not supervenec on brain states � indeed they supervene on brain states plus the environmnent. But supervenience seems to be indispensible for psychology as a science � while, if mental states were individuated broadly, criteria of individuation for mental states would depend on criteria for individuation of facts from the environment .
. In Psychosemantics, Jerry Fodor gives a more detailed account of why broad individuation violates supervenience. The reason is that mental states are broadly individuated, while brain states are narrowly individuated. It cannot be replied that supervenience could be saved by individuating brain states themselves broadly. Indeed, broad individuation is a relational individuation, and relational individuation of brain states is absurd. If brain states are broadly individuated, then any physical object could be broadly, i.e. relationally individuated. There would be no limits to the relational individuation. Anything�s individuation conditions would depend on everything else � for example every particle has the property that it is such that I raised my finger or I didn�t raise my finger, and this property would be essential to it. (This is what Fodor calls the �Mad argument�). Relational individuation of physical objects is pointless since it doesn�t account for their causal powers. Indeed, most relational properties are causally irrelevant; but criteria of identity of anything should be specified in terms of its causally relevant properties. Therefore, relational properties cannot be acceptable criteria of identity. (In Putnam�s twin earth example, it is assumed that my brain state and my twin�s state are identical � it would be pointless to insist they must be different since my brain state is the support of a thought about water and my twin�s state is the �support� of a thought about twin-water). If relational properties (according to which my brain state would be different from my twin�s brain state) cannot be acceptable criteria of identity for brain states, they cannot be acceptable criteria of identity for mental states, on pain of violating supervenience. (The suggestion that relational properties could be causally relevant is wrong: it violates the intuition that, in twin-earth examples, my mental state about water and my twin�s mental state about twin-water have the same causal powers).
I.3. Causal powers
Moreover, it seems obvious that psychology needs to accommodate the intuition that mental states have causal powers, i.e. the power to cause other mental states and actions. But if mental states were individuated broadly, their causal powers would be unaccountable for by psychology � the causal powers of mental states would be realized by a sort of action at distance � the causal powers of a mental state would depend on the causal powers of facts from the environment. Causal powers depend on a microstructure: this suggests that causal powers of mental states depend on brain structure. Moreover, - since criteria of individuation need to be specified in terms of causally relevant properties, and since we have no option but to individuate brain states narrowly � we would be unable to precisely identify a mental state unless we precisely identify facts from the outer world.
The desired consequence of the �causal powers� argument is that either narrow content is the only kind of content acceptable in psychology, or � if we cannot do away with broad content - that we need to abandon the hope of the scientificity of psychology.
I.3.1. Criticism of the causal powers argument for narrow content
However, there is a problem with the �causal powers� argument for narrow individuation. There are hidden properties and phenomenal properties of a natural kind such as water. If the hidden properties (the molecular structure) is causally responsible for the phenomenal properties (the taste, smell etc. of water) then narrow content disappears. Narrow content, i.e. the content of thoughts about water that is allegedly common to me and my twin when we think about water and twin-water respectively, varies according to variations in the phenomenal properties of water or twin-water. But if phenomenal properties vary according to hidden (molecular) properties, then narrow content varies according to hidden properties. But if phenomenal properties are caused by molecular properties, it is plausible that the phenomenal varies according to variations in the hidden. Thus, narrow content is simply impossible: it is impossible to specify a level at which my thoughts about water and my twin�s thoughts about twin water are identical. An internalist account of phenomenal properties seems plausible; but an internalist account of hidden properties is purely impossible. The identity of a mental state may well be given by its causal powers, but the mental state itself is caused by phenomenal properties of the object (or natural kind) intended. But if the phenomenal properties are themselves caused.by the hidden properties, then � assuming transitivity of causation � an intentional mental state is caused by the hidden properties of what is intended.
Thus, narrow content is merely impossible, precisely because the identity of psychological states should be given in terms of their causal powers. A mental state�s causal powers would be themselves dependent on environmental factors (�The molecular structure of water� is obviously an environmental factor). Yet, the above objection cannot be, at its turn, accepted without criticism. It assumed
(a) transitivity of causation
(b) that phenomenal properties are caused by hidden properties;
(c) that this relation of causation between the hidden and the phenomenal is stronger than one of supervenience. If the phenomenal merely supervened on the hidden, then it would be impossible that the phenomenal vary without variations in the hidden, but it would be possible for variations in the, hidden level without variations in the phenomenal. On the other hand, if variations in the hidden structure are impossible without variations at the phenomenal level, then narrow content would vanish: there would vanish: there would be nothing identical in my twin and in me, for the simple reason that our mental states are essentially characterized by the hidden properties of the natural kind.
That�s why the above objection is not decisive against narrow content. But at least, it shows what proponents of narrow content assume, namely the rejection of at least one of the theses (a). (b) and (c).
I.4. Causal powers of psychological states involving empty concepts (Segal)
A different way to object from the �causal powers� direction, advanced by Segal in [Segal, 2000] points to the fact that we can have concepts without extension, such as witch, unicorn, phlogiston. Moreover, the mental states corresponding to these concepts have causal powers: they may explain other mental states and actions. According to Segal, Putnam does not consider the presence of natural kind concepts which don�t have any real, a fortiori environmental component. But we have intentional mental states about non-existent objects and these mental states may yet be causally effective. This suggests that the causal powers of intentional psychological states do not depend on the existence of natural kinds �intended�, referred to. Since it is not possible to individuate beliefs about non-existent natural kinds other than in terms of narrow content, it follows that it is the narrow content, i.e. the content which does not depend on environmental facts, that has causal powers. Externalism is thus unable to account for hallucinations and, in general, intentional mental states about non-existent entities.
1.4.1. Criticism of the �empty concepts� argument for narrow content
In my opinion, this argument is falllacious and does not undermine Putnam�s arguments for broad individuation of mental states. The crucial assumption of this argument is that beliefs about non-existent natural kinds cannot be individuated otherwise than broadly. But this assumption is mistaken. A belief about witches has a narrow component that does not contain the truth conditions of that belief; for the narrow content it is immaterial whether witches exist or not. If the explanation, as given by experts, of an action caused by mental state M is precisely the explanation in terms of broad content, and if this explanation is different from the explanation the person who is in mental state M offers for his own action, then M has a broad content � namely, the way it is characterized by experts. This is possible even if M�s content is empty (i.e., if M refers to a non-existent object or natural-kind).
The broad content of the belief about witches is the concent ascribed to the believer by an �expert�, by a person who knows there are no witches. (Perhaps a different way to make this point would be to say that witches, phlogiston and unicorns have only phenomenal properties, but no hidden properties). But the explanation of the actions and beliefs of someone who believes in the existence of witches is different, according to whether the person who offers the explanation believes there are witches or not. Thus, we may distinguish between the broad and the narrow content of thoughts about non-existent objects. It is plainly false that such thoughts have only narrow content: the best proof is that the thoughts about witches entertained by someone who believes there are witches are different and have different �causal powers� from the thought about witches of someone who knows (or at least believes) there are no witches. (Someone may have thoughts about non-existent objects even if s/he knows their non-existence: for example I may desire the existence of witches, or I may think that science would have developed in a certain way if phlogiston really existed). Therefore, externalism can accommodate thoughts about non-existent objects (or non-existent natural kinds). The broad content of these thoughts is the content that is accessible to a �person who is aware of the emptiness of the concepts involved. The narrow content is the content accessible to a person who doesn�t know their emptiness.
Note that it can be tempting for externalism to defend itself against the �empty concepts� argument is to say that someone who thinks he believes something about unicorns does actually believe something about the environmental phenomena that actually cause his unicorn-thoughts. For example, someone may ingest a substance S that makes him entertain thoughts about unicorns. It could be said that it is that substance that is the reference of the unicorn-thoughts. But this is not a promising strategy, in my opinion, since it blurs the distinction between genuinely meaning substance S and meaning unicorns.
My argument that a thought involving empty concepts can still be broadly individuated makes room for a broad individuation of content that does not consider its extension as being essential. �Broad individuation�, on this account, is only that kind of individuation of mental states that pays attention to facts (or properties) extrinsic to the person who has that mental states.
Segal�s �empty concepts� argument for narrow content misses the fact that someone can have beliefs (or other intentional mental states) involving an empty concept whoile knowing that the concept in cause is empty. But since there is obviously a difference between the causal powers of the states �believing there is a ghost out-there� and �desiring that my interlocutor believes that there is a ghost out there�, my thesis is that mental states involving empty con cepts can well be individuated by facts outside the person�s mind.
In order to realize that the �empty concepts� argument for narrow content is wrong, we can also point to the fact that a concept about a non-existent object (or natural kind) may yet have a possible extension. It could be the case that entities like witches and phlogiston could exist, i.e. they are logically possible altough nomologically impossible. Even if these entities are logically impossible, it would still remain true that some mental states about non-actual, but possible entities occur and that they can be individuated broadly, according to the strategy outlined above.
I.5. Relativization to context By insisting that intentional mental states should be individuated broadly, externalism derives from the win-earth examples the conclusion that my mental state about water and my twin�s mental state about twin-water are essentially different. Internalism, faithful to narrow individuation, should therefore prove that my mental state and my twin�s are essentially identical.
Fodor argues in Psychosemantics that insisting that my mental states and my twin�s are different in an important way means to ignore relativization to context . It is the difference of context that accounts for apparent difference of the menta states in twin-earth examples, and there is no reason why this difference of context should be incorporated into the mental states themselves. Narrow content can be preserved, as long as we accept that the relation between my mental state and water is the same as the relation between my twin�s mental state and twin-water. And speaking this way about mental states amounts to their narrow individuation. Narrow content thus becomes (somehow similar to Fregean senses) a function from contexts to truth-conditions. But truth conditions depend on the broad content. Thie desired consequence of this argument is that mental states do not have their truth conditions essentially, i.e. that they are narrowly individuated.
The definition of narrow content as a function from context to truth-conditions cannot be accepted without objections. First: in some cases there are obvious differences between two mental states A and B, no matter what contexts we adopt. It is obvious that �relativization to context� cannot be a solution for obtaining narrow contexts from broad contexts in any situation. But it is not obvious how the difference between the situation where, on the one hand, two different mental states can become identical if we focus on contexts and, on the other hand, the situation where the difference between two mental states should be preserved no matter the cotnexts we choose, can be accounted for. If two mental states are genuinely different, i.e. they have no common narrow content, then � according to Fodor � they cannot be made identical no matter the contexts we choose. But it seesms that, by appropriate gerrymandering, we can find a narrow context common between any two apparently different mental states. My objection to Fodor�s treatment as narrow context as a function from context to truth conditions is that it is not clear what �context� is. It is not clear, in other words, how large the context should be.
In order to realize this, we can consider the case of thought about ducks and the thought about rabbits. They seem different in a more radical way than the thought about water and the thought about twin-water. But it is possible to devise mental experiment of two people who are really in the same mental state, when one of them refers to (�means�) �rabbit� and the other one means �duck�. Such a mental experiment would be coherent if the phenomenal properties of ducks for one of these persons were identical with the phenomenal properties of rabbits for the other person. Since phenomenal properties, the way things look like, are essentially subjective, there is no reason in principle why this mental experiment should not be coherent. (A sufficient but not necessary condion for such a situation to obtain is that one and and the same person is unable to distinguish between ducks and rabbits; Putnam�s twin-earth experiment actually assumes that neither I, nor my twin, are able to distinguish between water and twin-water).
Another example can be be devised (following Quine) using rabbits and undetachable rabbit-parts (or temporal stages of rabbits). Since � according to Quine � language could as well function by understanding an alien term as standing for rabbit or for part of a rabbit, it is plausible to suppose that someone can be in one and the same mental state when meaning �rabbit� and when meaning �(spatial or temporal) part of a rabbit�. It follows that �rabbit� and �part of a rabbit� may stand in the same relation as H2O and XYZ, given an appropriate context. In a radical translation situation they have the same causal powers � indeed they are stimulus equivalent. However, it seems pointless to postulate a narrow context that is identical to a thought about rabbits and a thought about parts of rabbits, and even more pointless to claim that these two thoughts are essentially identical since they share there is a context in which they have the same causal powers and they have the same narrow content. My point is that, for any two intentional mental states, it is possible to devise a context where they have the same narrow content; but this is not a reason to consider the two thoughts as identical. Therefore, the �relativization to context� argument is wrong.
II. Fodor�s arguments against the distinction between broad and narrow content In The Elm and the Expert, Fodor continues to think that twins are a counterexample to psychological generalization, that's why his strategy is to play down their importance. Fodor�s attack against Putnam no longer arises from his commitment to narrow individuation of mental states; on the contrary, Fodor now accepts that broad content is definitory. Fodor�s conversion to externalism does not imply that narrow individuation is wrong ; it only implies narrow content is superfluous, i.e. it has no explanatory virtues that broad content hasn�t. But Putnam�s The Meaning of Meaning is still a threat to the methodology of psychology as Fodor wants it since Putnam believes in the explanatory virtues of the distinction between narrow and broad content and that broad content is incompatible with supervenience. Fodor�s positon in The Elm and the Expert can more be adequately rephrased by saying that it states how broad individuation can be reconciled with supervenience.
Fodor�s argument is now that twins are nomologically implausible, and psychology is a nomological science. Even if twins are logically possible, their existence would be such a far-fetched possibility that psychology should not take care of it.
Apart from twins, a possible argument against supervenience consists in the deferential concepts, like elm and becch, whose extensions cannot be discriminated by a non-expert. Fodor�s strategy is to treat all concepts as being more or less deferential: indeed, a non-expert can distinguish elms from beeches with the help of an expert; the expert plays here the role of an instrument of measure, like a diagnostic substance that can help a non-chemist distinguish between acids and bases. He explicitly compares �I cannot tell elms from beeches, so I defer to experts� to �I cannot tell Sundays from Mondays, so I defer to the calendar� . In Fodor�s account, deferential concepts would prove that broad content violates supervenience only if there would be creatures who
- could not tell apart natural kind A from natural kind B;
- have the concepts of A and B;
- one of them means only A by both concepts, the other means only B .
If these conditions held, psychology should generalize over these creatures. But broad individuation cannot generalize, since it cannot account for what is identical in the two creatures. These conditions are satisfied by the twin cases (who are accidental and nomologically impossible) and they are not satisfied by the deferential concepts � since non-experts can tell apart between the extensions of any two concepts they have, even if aided by experts.
II.1. Deferential concepts and meta-experts
But Putnam�s article contains more than the twin-earth story. In order to prove his claim that thoughts don�t have their extensions (and thus their truth conditions and, perhaps, their causal powers, essentially, Putnam devised the example of deferential concepts, i.e. concepts which may be �had� by a person who is unable to tell apart between their extensions.. For an average person it is impossible to distinguish between elms and beeches. It follows � according to Putnam � that, since an average person still has different concepts for �elm� and �beech�, that the extensions of these concepts is not essential to their individuation. Difference in extension between the concept of elm and the concept of beech is ascribable to experts. (This example serves as an argument for Putnam�s hypothesis of the division of linguistic labour: i.e., there are words whose meaning is only known to experts). DETALIAZA DESPRE HDLL!!!
Fodor�s reply is that the hypothesis of division of linguistic labour is not a decisive argument for the distinction between broad and narrow content. FII FOARTE ATENT, PENTRU C� �N E&E FODOR DEVINE BROAD!! We can distinguish elms from beeches, via experts. (Fodor assumes twin-earthians can't distinguish water from twater, even via experts; otherwise he should give a unified rejection of the Twin-water and elm/beech example, which he doesn't).. The need for experts proves indeed that individuation of concepts is broad, i.e. it depends on facts (or people) different from the person who has those concepts. But this example leaves no room for a narrow content that would be different from broad content. Thus, we can simply do away with narrow content: If there were any place for narrow content, then two people could be in the same mental state yet use words with different meanings. This would happen if DA CITAT DIN EE! UNDE FODOR SPECIFICA CDLE IN CARE BROCO AR FI UN CONTRAEX
It should be noted that Putnam's point with the elm/beech example was precisely to strengthen the case for "experts" and broad content. The elm/beech is just a "our-worldly" analogy of the water/twater situation. It makes more plausible the intuition suggested by the water/twater example, that we can be in the same mental state while relating to two different concepts.
I will make a few comments on the elm/beech example. This is only one of the many cases of "deferential concepts". It is a rather simple case, where I have two concepts and I am unable to tell apart between their extensions. There can be cases in which I have both concepts A and B, and wrongly think I can tell apart As from Bs (I actually cannot); or I can tell apart among them only to a limited extent. More (philosophically) relevant cases: I may wrongly trust an expert; I can be as knowledgeable as the expert but mistrust my own knowledge. For example, it is traditionally held that we have first person authority about our psychological states, at least about non-intentional psychological states or the non-intentional component of our psychologicl states: pain is the most often cited example. There may yet be people who mistrust the descriptions (or �avowals�) they would be prepared to offer to their psychological states and instead ask for experts to tell them what psychological state they are experiencing. (Behaviorism can be an extreme illustration of this situation: a genuine behaviorist doesn�t believe in first person authority and would not be sure he feels pain until an expert (in this case a doctor) confirms his belief that he is in pain. If behaviorism is false and we have indeed first person authority about our non-intentional psychological states, then we have here a case where the resort to experts is simply illegitimate. If we have first person authority about our (non-intentional) mental states, then everyone is an expert about the psychological states s/he is experiencing. Experts, may be sometimes illegitimate.
It is beyond the scope of this paper to evaluate the implications of these cases for the internalism/externalism debate, but I think nevertheless they should be explored. I just want to point to the fact that the very notion of �deferential concepts� is more problematic than it may seem. It is not clear which concepts are deferential and which aren't.. It could be said that deferential concepts are just those concepts which someone claims to have, yet be unable to recognize their extension. But this definition of deferential concepts is wrong, since it might be the case that someone only thinks he is able to distinguish As from Bs (elms from beeches), although he actually isn�t. He does not resort to experts, although experts subsequently prove him wrong..
Moreover, if a deferential concept is just any concept had by a person who sincerely claims to be unable to identify its extension, then the extension of the concept �deferential concept� would depend on ordinary linguistic practices. In order to recognize a concept as deferential, we would only need to follow our belief that it is deferential. (Externalism would not be true about the very concept �deferential concept�), since any concept would be a deferential concept if it just seemed to us to be a deferential concept. �Deferential concept� would be a concept individuated in terms of its phenomenal properties. But in order to know the extension of deferential concepts, we first need to know what the deferential concepts are. If � as I tried to suggest � the term �deferential concept� itself is narrowly individuated, then narrow content is, surprisingly, prior to broad content.
Of course, the argument above holds only if a deferential concept is really anything that seems to be a deferential concept. But if we could be mistaken about whether a concept is deferential or not, then we would need experts (a sort of �meta-experts�) to teach us the extension of the concept �deferential concept�. This is compatible with the possibility that some experts are illegitimate (i.e., they claim to be experts for concepts whose extension is actually known to the person who has the concept). I am not sure such a complicated account of the division of the linguistic labour is acceptable. If we had to rely on meta-experts to tell us what are the deferential concepts and what are the non-deferential concepts, then it is possible that, pushed to the extreme, all concepts are deferential � we should refrain from claiming that we may distinguish between the extensions of any two concepts because it would be possible that experts prove us wrong. The same would hold true also for identifying the extension of any concept C we have: it could be possible for future experts to discover that there were actually two different extensions of C, and thus that the meaning of our words is ambiguous when we refer to C. Putnam�s example is the situation before 1750, when the chemical composition of water was unknown and when it would have been impossible to tell apart between water (H2O) and twin-water (XYZ). Yet, since water and twin-water had different chemical compositions even then, the word �water� was used illegitimately by earthians and twin-earthians. Indeed, it is a consequence of Putnam�s article that the concept of water was a deferential concept even before it was known to be deferential. .�Water� was discovered to be a deferential concept. As scientists discover essences (according to Putnam), so meta-experts discover deferrential concepts: meta-experts� job is not that of typical scientists, i.e. they do not discover the essences of deferential concepts. Their job is rather to classify those concepts which are deferential, whose extensions have a hidden essence that needs to be discovered by scientists..But there is nothing that guarantees that , with respect to any of the concepts we know have, we are in the better situation than earthians (and twin-earthians) were around 1750. Perhaps scientists will discover the hidden essence corresponding to them, and meta-experts will confirm that this hidden essence belongs to the meaning of the terms denoting them. It is also possible that meta-experts first discover that a concept is a deferential concept, i.e. its extension is determined by facts known to the scientists, although nobody yet knows what this hidden essence is. As it happens to be, most psychological terms are now in this situation: we believe there must be a scientifically discoverable meaning of terms like �attention�, �memory�, sensation� and the like, and thus do not trust our common-sense intuition about these terms, although it is far from being established what is the hidden essence of these psychological processes and thus their scientific meaning.
The resort to experts thus leads to skepticism about grasping a concept. Every concept is a potentially deferential concept, therefore we can never be sure that we can appropriately identify the extension of any concept. History contains many relevant examples: for example, it is plausible that several diseases were named �plague� before the discovery of the bacillus of plague. Perhaps similar discoveries will be made in the future. any concept is a potentially deferential concept. If any concept can be a deferential concept, then it is possible that we are largely ignorant about the identity of our psychological states, broadly individuated. Not only we are ignorant about them, but everyone else shares this ignorance � except for the future experts who discover the extension(s) corresponding to our concept. Indeed, if we wrongly think that A-s are a natural kind, and it is proven in turn that there are two different natural kinds, Aici and A2, corresponding to our A, then we actually have a disjunctive belief whenever we think we have a belief about A. If all concepts are potentially deferential concepts, and if individuation of intentional mental states is broad, then we are potentially ignorant about all of our mental states.
The hypothesis of the division of linguistic labour and the resort to experts has another paradoxical consequence for the debate over the individuation of mental states. If resort to experts is indeed an argument that individuation of mental states should be broad (�meanings ain�t in the head�), its assumption is that it is experts� mental states that distinguish between the extensions of the two concepts. Broad content becomes just another name for the narrow content of experts� mental states. Indeed, a chemist who discovered that water is different from twin-water would no longer be in one and the same psychological state when thinking about water and thinking about twin water. In the case of the expert, meanings are indeed in the head. Thus, the hypothesis of the division of linguistic labor only defers the authority about the identity of mental states to the mind of experts. A different way to understand this argument: the distinction between broad content and narrow content parallels the distinction between hidden and phenomenal properties of facts. Experts have access to the hidden properties of natural kinds (and, in general, of facts) and it is these hidden properties that are essential for their identity. .If intentiononal mental states are individuated broadly, then their identity depends on the hidden properties of objects in the world. But since experts have access to the hidden (essential, identity-conferring) properties, their intentiononal mental states have are different from those of the non-experts. Non-experts have access only to the phenomenal, non-essential properties of objects. An expert�s belief narrowly individuated is the same as a non-experts� belief broadly individuated. This happens because a non expert�s mental state, broadly individuated, is the mental state the expert ascribes to him. Broad content is a description of a psychological state, a rephrasal of it as given by an expert. If this is true, then broad content may be defined as the narrow content of an expert. Meanings are in experts� heads.
III. Rationality and externalism The claim was made by Segal in the same book (Segal 2000) that narrow individuation is required since it better accomodates the explanations of people�s actions that preserve their rationality. The intuition which I think lies at the basis of this claim is that narrow individuation of mental states (like beliefs) can make a person�s thoughts appear more coherent than would appear if broadly individuated. After all, if we had to rely on experts � as externalism wants us to � in order to individuate our mental states, then it could turn out that we only believe that we have many beliefs (or other mental states) without actually having them; we could be mistaken about the inferences which we are the surest about, since we can ignore the meaning of the terms involved.
In this context, Kripke�s famous puzzle about belief was invoked in support of narrow content. To explain the rationality of the person who has two seemingly contradictory beliefs (that London is pretty and London is not pretty, or that Paderewski has musical talent and Paderewski lacks musical talent). The internalist (narrow content) account of Kripke�s puzzle will say this: if we narrowly individuate the beliefs, then it will turn out that someone has the belief that London-1 is pretty and London-2 is not pretty, i.e. the contradictory predicates apply to two different concepts . Segal�s point is that it is only narrow content that allows the person in Kripke�s example to be rational since narrow individuation ascribes to him two different concepts. If we stick to the externalist view and claim that the meaning of the terms (proper names, in Kripke�s examples) depends on non-psychological facts and that it is fixed by experts, it will follow whenever a person believes both that London is pretty and London is not pretty he is having contradictory beliefs (is making a logical mistake). But Kripke�s point was precisely to highlight cases in which entertaining such beliefs does not amount to logical inconsistency.
In my opinion, Kripke�s puzzle cannot receive an internalist solution, at least if it is followed in detail. Kripke�s thesis is that the puzzle cannot be eliminated if we are sensitive to the difference between the definite descriptions under which the object referred to by the proper name falls. Even if an object is identified through the same definite description, the puzzle may still appear, provided the descriptions are formulated in different languages, and/or that the person ignores that they denote the same object. Now, the �narrow content� solution to Kripke�s puzzle about belief parallels its (alleged) descriptionalist solution. Indeed, if we identify two concepts associated with the proper ame about which contradictory beliefs are hold, it is yet possible for the puzzle to reappear, thus making the believer continue to appear irrational. For example, an internalist would claim that in the Paderewski example there are two different concepts involved, �Paderewski the musician� and �Paderewski the politician�, and that these two concepts can only be individuated narrowly (since they have the same extension, although unbeknownst to the believer). But the description �Paderewski the politician� can be understood as to refer to two different persons � if, for example, someone believes that there are two politicians with the same name. In such a situation the narrowly individuated concepts would multiply accordingly. Segal would want us to attribute two different concepts to the believer in Kripke�s example and � since these concepts have the same extension � concludes they should be narrowly individuated. But their narrow individuation amounts to their individuation by description. In order to preclude the possibility of such a multiplication of the narrowly individuated beliefs, we would need to forbid rephrasing the descriptions that express the (narrow) content of these concepts. Thus, the internalist solution to Kripke�s puzzle may be rejected with the same arguments Kripke presents against the descriptionalist solution to the puzzle: concepts� narrow content cannot be specified otherwise than by descriptions; but the puzzle can arise at the level of descriptions, therefore it can arise at the level of narrow content.
The issue of �externalism and rationality� was addressed by Fodor in �The Elm and the Expert�, where (contrary to the views he expressed in �Psychosemantics� and �Methodological Solipsism�, he claims that narrow content is superfluous . The thesis that � contrary to Segal � it is precisely broad individuation that can account for people�s rational behaviour is formulated with respect to Frege�s intensional context (two proper names with the same extension � like Hesperus and Phosphorus). Frege�s cases appear when an individual believes that Fa and ~Fb, although (unbeknownst to him) a=b . Fodor� thesis is that Frege cases are indeed a good argument against broad individuation, but they are an argument for believers� irrationality. But we need to assume that people are generally rational � therefore we can exclude Frege cases from psychology. It is apparent from this argument that it is precisely narrow content that makes room for accounting for people�s irrational beliefs . If this argument is sound, it proves that Segal�s contrary claim (that narrow content accounts for rationality of beliefs which appear irrational if broadly individuated) is wrong. Therefore we can do away with intensionality and with narrow content.
IV.
My second objection against externalism arises from the distinction between content and psychological type. To believe that water is wet and to hope or fear that water is wet is to entertain several intentional mental states of the different psychological types but with the same content. To believe that water is wet and to believe that the sun will raise tomorrow are mental states of the same psychological type but with different contents. Now, arguments for externalism point only to the lack of transparency of mental content. According to externalism, mental states should be broadly individuated, since their content depends on properties extrinsic to the individual who has these mental states: experts can discover that we don�t really know the extension of the concepts intended by our psychological states. I argued above (in section ?) that this leads to skepticism about content and, more generally, to skepticism about the meaning of our words. By a shift of emphasis from content to psychological type, we can ask: If mental content is not transparent, why should the psychological type of a mental state be transparent? In other words, how can we be sure that we really believe something, and not desire or fear it? When someone on twin-earth thinks he wants to drink water, how can he (or we, for that matter) be sure that he really wants to drink water and is not actually afraid of water? Perhaps there are experts who can not only discover the hidden essences of natural kinds, but also discover the hidden essences of psychological types. If we (for good reasons) do not trust common-sense taxonomy of natural kinds, why should we trust common-sense taxonomy of psychological types? Environmental facts like natural kinds and diseases are not classified according to their phenomenal properties, but to their hidden essense, discovered by scientists. But then why preserve commonsense authority about psychological types?
Broad individuation is incompattible with first person authority about mental content, although commonsense generally accepts it. (We are usually sure what we are thinking about). But there is no reason, besides commonsense, to classify psychological types according to phenomenal properties. Water is not what has the phenomenal properties of water; but why should a belief be what has the phenomenal properties of a belief? Pushing this issue to the extreme, we can imagine that a twin-earthian who is in a mental state with the phenomenal properties of �wanting water� actually doesn�t want water: he is afraid of water. But since there is no water on twin-earth, his mental state should be redefined as �being afraid of twin-water�. If twin-water is not genuine water, why should twin-beliefs be genuine beliefs?
Putnam's Meaning of �Meaning� is, inter alia, a plea for scientific realism. One of its purposes is to accommodate psychology within the framework of scientific realism: words� meanings and reference do not vary with age or �paradigm�, �gold��s meaning has always depended on the atomic structure of gold and �water��s meaning has always depended on the molecular structure of water � even unbeknown to people. But his realism actually amounts to skepticism about the meaning of our words. We can never know what we are thinking about, until an expert discovers the essence of our intentional object. Moreover, we can never know whehther we think of anything, or hope for it, or desire it and so on.
This generalized skepticism is the logical consequence of psychological externalism. But this generalized skepticism is inconsistent: indeed, in order to specify criteria of individuation of a psychological state we need to hold something fixed in order to know what we refer to. If we want to discover what someone thinks about, we need to assume he at least thinks of it; one the other hand, if we want to discover whether someone thinks about something, we need to assume that he really means that something. If we doubt about the content of someone�s mental states (as externalism does), we need to assume he correctly identified the psychological type of his mental states; if we wanted to doubt about the psychological type of someone�s mental states, we need to assume he correctly identified the content of his states. If we wanted to doubt both, we would not know what mental state we are referring to; indeed, we would have no reason to claim we are referring to a mental state that both the person who �has� that mental state and we identify as that mental state.
Although doubting both about the content and the psychological type of a mental state is impossible, this is the logical consequence of broad individuation of mental states. This, together with the remark that resort to experts in order to discover the content of one�s mental state leads to a more moderate skepticism (or, if experts are indeed needed, then �broad content� is nothing more than the content of a mental state had by an expert), are in my opinion sufficient reasons to believe that narrow content is more basic than broad content. However, the arguments for narrow content based on individuation of mental states according to causal powers and �relativization to context� are wrong.
The distinction between broad content and narrow content is similar in some respects to that between the referential and attributive usage of definite descriptions, as formulated by Donnellan in Reference and definite descriptions. In the referential usage, a description refers to anything that the utterer intents to pick out; it is a tool used to draw the hearer�s attention to a particular object. On the contrary, in the atrributive usage, a description refers (even unbwknonst to the speaker) to any entity that has the property that the description ascribes to it. Kripke, in Speaker�s Reference and Semantic Reference generalized this distinction to proper names, interpreting it in the terms of Grice�s distinction between what speakers mean by their words and words� meaning. This distinction is similar to the one drawn by Putnam in The Meaning of �Meaning�. This remark may have consequences for the individeuation of menta states. It should be said that the referential usage of a definite description (or of a name) is its usage for which the mental state of the speaker is essential.(narrow content). The attributive usage of a term corresponds to the usage for which facts unknown to the speaker may be essential (broad content). It may be tempting to equate the distinction between narrow versus broad content with the distinction between de dicto and de re entry of a term (description, natural-kind term or proper name). However, as Kripke points in the same article (Speakr�s Reference and Semantic Reference), a term with a de dicto entry does not refer to anything. On the contrary, a term used attributively refers to something (in Donnellan�s terminology); by it the speakers means something (in Putnam�s terminology) � even if this �something� is not the term�s meaning. Therefore, the distinction between broad and narrow content is not similar to that between de dicto and de re entry of a term � unless we want to say, following Frege, that a a term in a de dicto context refers to its sense. The term �methodological solipsism� was coined by Putnam in The Meaning of Meaning: �[Traditional philosophers] made an assumption about what we may call methodological solipsism. This assumption is the assumption that no psychological state presupposes the existence of any individual other than the subject to whom that state is ascribed.� (p. 220, in Mind, Language and Reality) ��if you re interested in causal explanation, it would be mad to distinguish between Oscar�s brain states and Oscar-2�s brain states; their brain states have identical causal powers. That�s why we individuate brain states individualistically. And f you are interested in causal explanation, it would be mad to distinguish between Oscar�s mental states and Oscar-2�s mental states; their mental states have identical causal powers. (Psychosemantics, p. 34). In order to strengthen the case for the identity of the twin�s mental states (and of their causal powers), Fodor develops the argument from �relativization to context�. I have tried to prove (in section I.5.) that the argument from relativization to context is wrong. Segal�s argumentruns like this: �Take two planets TE1 And TE2. Take twin subjects or groups on the two planets. Let TE1 subjects use the word W1 and TE2 subjects use the word W2 as if they were kind words. Let W2 apply to some real sample on TE2. Let W1 be empty on TE1 the externalist nowfaces a dilemma. He must either deny that W1 expresses a concept or concede that it does. But the former option is obviously not acceptable. And the latter option is not acceptable to the externalist. For if W1 epresses and empty concept, that concept will also be expressed by W2, and there would be a shared world-independent content�. This argument states that if a word expresses an empty concept (for an externalist all concepts have extensions) , its extension will coincide with the extension of the same word used in an environment where it is not empty. But, as I have argued, externalism can be reconciled with the view that a concept may lack an extension. This s obvious if the concept has a possible extension, but the possession of a possible extension is not a necessary condition (although it is a sufficient one) for its broad individuation. That�s why Fodor�s account of the identity of causal powers of mental states is counterfactual: �What is relevant to the question of identity of causal powers is the following pair of counterfactuals: (a) if his utterance (/thought) had occurred in my context, it would have had the effects that my utterance (/thought) did have; and (b) if my utterance (/thought) had occurred in his context, it would have had the effects that his utterance (/thought) did have. (Psychosemantics This example is obviously adapted from Wittgenstein�s discussion about �seeing aspects� in Philosophical Investigations, IIxi. Wittgenstein�s point is that two different mental states can correspond to identifying the same object (aspects being somehow similar to Fregean senses): one and the same visual object can be identified via different aspects. The symmetrical of this experiment is the situation where two different visual objects are identified via the same aspect � which parallels Putnam�s example with water(H2O) and twin-water (XYZ) The essential properties of the two objects are the same, but their phenomenal properties are identical. I believe myself that a consequence of the externalism (at least in Putnam�s version, i.e. assuming that we need experts to fix the reference of the terms which we refer to) has the consequence that all concepts may be (or become) deferential and that this leads to skepticism. That is, unlike Fodor, I think that the thesis �All concepts are potentially deferential concepts� is not a good argument for externalism, that�s why I think that some concepts are not deferential. A way to understand this is the following: �I can�t tell Mondays from Sundays so I defer to experts� is different from �I can�t tell elms from beeches so I defer to experts�, since in the second case, but not in the first, the narrow content of another person�s mental state (the expert�s) is used in fixing the extension of my concepts. I am sensitive to the difference between �I can�t tell Mondays from Sundays so I defer to the calendar� and �I can�t tell Mondays from Sundays so I defer to an expert in calendars�. I believe that the second, not the first, is the more exact counterpart of Putnam�s example with elms and beeches. There is a difference between (a) doubting about the extension of my concepts and using extrinsic facts in order to fix it; and (b) doubting about the extension of my concepts and using (and trusting) another person to fix it. This is a rephrasal of what Fodor says in The Elm and the Expert, p. 38: �To construct a twin case that would really embarrass referential construals of mental content, you would have to show that it is nomologically possible that there are creatures for which it is nomologically impossible to distinguish between As and Bs, but of which an externalist theory of content is required to say that they have the concept A but don�t have the concept B�. Kripke in A Puzzle about belief devises several mental experiments that apparently rejects the direct reference theory of proper names and points to the fact that it can be interpreted equally as evidence against a description-theory of proper names. One example involves a Frenchman, Pierre, who believes that London is pretty and expresses this belief in French (�Londres est jolie�), then moves to London and learns English by �direct exposure�, coming to the belief that London is not pretty (he ignores the fact that London and Londres are two names of the same city). Another example doesn�t involve translation: Paderewski was a pianist who became a Polish statesman. Someone may believe that Paderewski had musical talent (knowing Paderewski was a pianist) and that Paderewski lacks musical talent (knowing that Paderewski was a politician, mistakenly believing that Paderewski-the-politican is a different person from Paderewski-the-musician). The puzzle consists in the fact that two contradictory beliefs may be hold by a person who is perfectly rational. It may be tempting to see these examples as pleading for a desciptionalist view of proper names (which would claim that if we specify the descriptions according to which the proper names refer to the object, the contradiction dissipates). Kripke however remarks that the puzzle can reappear at the level of the definite descriptions associated with the noun (for example, someone may understand �London� as �the capital of England� and �Londres� as �la capitale d�Angletterre�, without realizing England and Angleterre refer to the same country). A similar example can be coined by elaborating on Wittgenstein�s remarks about seeing aspects. Someone can see the picture of the duck-rabbit alternatively as a duck or as a rabbit. When he sees the picture as a rabbit, he may think �this is a very nice picture� (since it niely depicts a rabbit). When he sees it under the duck-aspect, he may think �this is a very poor picture� (since it poorely depicts a duck). If he doesn�t realize that it is the same picture that is seen under the two aspects, he will come to think that �the rabbit-picture is a nice picture� and �the duck-picture is a poor picture�. As the argument from rationality for narrow content goes, it I hard to avoid attributing to the person in cause contradictory beliefs if we individuate them broadly. It is important to notice that Fodor�s conversion to externalism and broad content does not signify rejection of supervenience of the mental states upon the (narrowly individuated) brain states. Fodor addresses the objection (previously made by him) that broad individuation violates supervenience, since broad psychological laws cannot generalize over twins (whose mental states and brain are �identical�, at least at some level). The general form of this claim is that a twin cannot distinguish X from Y but his concept applies to X only, whereas the other twin cannot distinguish between X and Y but his concept applies to Y only. Fodor treats these cases as accidental ones, and uses their implausibility as a reason to reject the need for psychology to account for them. (Narrow content is supposed by the claim that psychological laws should generalize over twins � since both twins have the same narrow content. Therefore rejecting the need for generalizing over twins is a rejection of the need for narrow content). �Putnam�s [Meaning of �Meaning�] shows something about our concept of content: it shows that supervenience of the broadly individuated intentional upon the computational isn�t conceptually necessary. But it does not argue against the nomological supervenience of broad content on computation since, as far as anybody knows, XYZ is nomologically impossible. � (p. 28). The argument I presented in section I.3.1. (that the hidden essence of natural kinds can be conceived as causing their phenomenal properties) makes Fodor�s argument in The Elm and the Expert more plausible. These cases are, ceteris paribus, similar with Kripke�s mental experiments in A Puzzle about Belief. Pierre in Kripke�s example is rational since he thinks that F(London) and ~F(Londres) but does not think that (F&~F)(London) since he doesn�t know that London=Londres. CAND PIZDA MASII E RATIONAL? CAND STIE MAI MULT!!!!! LA SEGAL IESE CA TC CAND STIE MAI PUTIN!!!! �Rational agents reliably make a point of knowing the facts that the success of their behaviour depending upon� (The Elm and the Expert, p. 46)� �If intentional laws are broad and Frege cases are allowed to proliferate, then the fact that rational behavior is generally successful is unintelligible�. (ibidem, p. 48). Scientific psychology can well be broad, even if it doesn�t account for intensional contexts like Frege�s. Of course, if we accept that rational behavior is one that does not generate intensional contexts � and this is the consequence of accepting that rational behavior is one that presupposes knowing as many (relevant) facts as possible about one�s actions or beliefs.