Concepts Without Sense:
How Fodor Uses New Tools to Solve Old Problems
How Fodor Uses New Tools to Solve Old Problems
In this essay, I will explicate Fodor’s attempt to make RTM meet the requirements Frege set out for what a theory of concepts would have to be, without, as Frege does, positing Sense. I will first show why Frege thought positing Sense was necessary. I will then show how Fodor, in Concepts, responds to ‘Frege cases’ without needing to posit Sense as the Mode of Presentation (MOP). I will then illustrate how Fodor accepts the demand that concepts must be public, how Fodor criticizes the way Frege attempts to meet this demand, and show how Fodor argues a causal account of concepts could retain public concepts without needing to posit Sense. I will weigh these solutions against Fodor’s critique of the “‘classical’ RTM LOT CTM” account of the mind as found in LOT 2, and argue that Fodor’s solutions aren’t worth much if he’s right about the tripartite theory of the mind being in trouble. (Fodor 2008, 101)
The purpose of Sense
The best way to proceed would be to give the rationale for why someone would want to posit Sense; what made positing Sense necessary for Frege? Fodor’s interpretation of Frege’s motivation for positing Sense goes like this: Fodor takes ‘the typical Fregean position’ to be one in which ‘concepts are distinguished along two (...) parameters; viz. reference and (MOP).’ (Fodor 1998, 15) Fodor takes it that Frege ‘identif(ies) (MOP) with senses’. (Ibid.) Frege’s reason for adopting Sense in his ontology of concepts is that he believes Sense is necessary for a) maintaining the public nature of concepts, and b) because ‘there is more than one way to think about a referent’. (19) How Sense and Reference secures these requirements is made clear by Fodor’s exposition of the Fregean account of MOP:
‘5.1 MOPs are senses; for an expression to mean what it does is for the expression to have the MOP that it does.
5.2 Since MOPs can distinguish concepts, they explain how it is possible to entertain one but not the other, of two coreferential concepts; eg. how it is possible to have the concepts WATER but not the concept H2O, hence how it is possible to have (de dicto) beliefs about water but no (de dicto) beliefs about H2O.
5.3 MOPs are abstract objects; hence they are non-mental.’ (17-18)
5.1 and 5.3 are both necessary for Frege because of the above-mentioned requirements he has on a theory of concepts: a) 5.3 is what Frege believes secures the public aspect of concepts; concepts can only be public if one aspect of them is abstract and not-psychological because Frege thinks that mere psychologisms could not retain the public nature of concepts; and b) 5.1 is necessary because co-referring concepts can mean different things. For clarities sake: Frege thought he had to posit sense to account for the fact that concepts are public, and because co-referring concepts could mean something different. Senses are abstract objects for Frege and therefore make public concepts possible, and Sense is also what makes two co-referring concepts like ‘morning star’ and ‘evening star’ have different content. The three questions I will now address are: 1) Is Fodor sympathetic with Frege’s demands of a theory of concepts?; 2) If so, how does Fodor retain co-reference without positing sense?; and 3) If so, how does Fodor retain the public nature of concepts without positing sense?. I will now show the answer to question #1 is ‘yes’, and show how Fodor uses RTM to approach questions #2 and #3.
Co-Reference without Sense
Frege cases arise in situations where one cannot substitute concepts in propositional attitude cases, that would be identical if reference exhausted concept identity, without leading to absurdity. In LOT 2, Fodor provides the example of ‘Paderewski’. The thought experiment goes like this: John believes Paderewski was a pianist, but also believes there is a Paderewski who is a politician. Fodor argues that ‘If reference is content, then if John believes both that Paderewski was a pianist and that Paderewski was a politician, he ought to be prepared to infer that Paderewski was both a politician and a pianist. But he doesn’t. Indeed, he explicitly denies this inference. Our problem is to explain how, on referentialist assumptions, this could be true.’ (Fodor 2008,72-73) Frege’s solution was to posit Sense, and I will now show that Fodor’s solution is to show that, if one assumes RTM, Frege cases are shown to be ‘ill formed’. (73)
How are these cases ill-formed? This has something to do with the kind of philosophical atmosphere both thinkers were immersed in; insofar as Frege was interested in everyday language, he looked at the surface of expressions and worked from there. Fodor on the other hand is working within an RTM framework; one that sees language as covering up the true nature of thought; Fodor's LOT hypothesis.[1] Once this layer is removed, Sense dissipates, and fails to have a raison d’ètre in relation to co-reference. How does RTM come to the rescue in these cases? Fodor argues that ‘if RTM is among our background assumptions’ then there must be “two Mentalese names corresponding to the English word ‘Paderewski’”; 'PADEREWSKI1 and PADEREWSKI2’ would be the well-formed concepts at play. (72) Therefore, insofar as ‘(John) thinks in Mentalese, not English, (he) can’t think the thought that ‘Paderewski is tall’.’ (Ibid.) Fodor argues that ‘there is no such thought’ ‘Paderewski is tall’ if Paderweski has two tokens, there is effectively only the thoughts ‘Paderewski1 is tall’ and ‘Paderewski2 is tall’. (73). By extension, ‘John can’t think Paderewski is tall without choosing between PADEREWSKI1 and PADEREWSKI2; no more than he can think Marx was tall without choosing between Karl and Groucho(...)’. (Ibid.)
Looking underneath everyday language in order to solve Frege cases shows, among other things, how philosophical approaches that acknowledge the distinction between ordinary language and the language of thought are necessary given cognitive science, and it also gives credence to the idea that a new conceptual apparatus like RTM can be fruitful insofar as it extinguishes the seeming need for philosophical ontologies of decades past that relied on notions of everyday language rather than cognitive notions; one can see how Frege cases do not warrant the adoption of Sense if one adopts a more contemporary understanding of thought. If RTM, and by extension LOT, obtain, then Frege’s notion of sense can be done away with, and doing away with sense would be nice given how much philosophical ink has been spilled on the concept. But again, its not simply about spilled ink; many analytic philosophers find Frege’s positing of sense necessary[2] given Frege cases, and thereby find it necessary to adjust their theory of concepts accordingly. I will now quickly go through some of Fodor’s points relating to this issue before moving onto public concepts.
There are a couple of questions to consider before moving on: A) how are there two mentalese tokens for Paderewski, and B) what is it about these concepts that makes them co-referential and all the while completely distinct in their causal powers i.e. why can you have one and not the other, even if they co-refer? Why is it, for instance, that PADEREWSKI1 and PADEREWSKI2 have the same referent but can’t be substituted for each other in cases with a propositional attitude?[3] Because Fodor, as I stated earlier, assumes CTM, the notion that mental states are operations on syntax, this difference in 'causal powers' cannot be determined by something like the Sense of the concept. Fodor, by his own words, is 'motivated largely by a desire to comply with the chief demand that a computational account of mental processes imposes on the theory of mental representation'; he is committed to the idea that that if mental representations are not compatible in PA cases, then they must be 'formally distinct in ways that mental processes can distinguish'. (92)
‘From CTM’s perspective, the existence of Frege problems shows at most that reference isn’t sufficient for the individuation of concepts; something further is required. But Frege’s problem doesn’t show that the ‘something’ else is a parameter of content; for example, that it is something like a sense.’ (70)
Fodor sees the problem as a by-product of ‘working on the assumption that the content of a belief exhausts its contribution to the causal consequences of having it.’ (68) If one, on the other hand, assumes that ‘the causal consequences of having on kind of belief-with-content- F(Cicero)[4] can differ from the causal consequence of having some other kind of belief-with-the content-F(Cicero)’, then, as Fodor puts it, ‘Frege cases cease to be problematic’. (Ibid.) When one assumes a framework in which operations are not sensitive of content, then ‘distinguishing the identity of beliefs from the identity of belief contents’ makes a CTM much more robust. (Ibid.) Frege cases then show that, for one, if they didn’t exist, ‘we would have to invent them’ to take into account that ‘the same belief can differ in causal powers’. (69) So in relation to question A, one can have distinct tokens of Paderewski because reference does not exhaust concept individuation, and in relation to question B, Fodor argues that ‘PADEREWSKI1 and PADEREWSKI2’ can have the same referent but different ‘causal powers’ insofar as ‘CTM distinguishes the causal powers of mental states whenever they are tokenings of type-distinct mental representations, even if the semantic contents of the representations tokened are the same’; if they have identical content, they can still play a different role causally because of 'possession conditions'. (70)
One other implication of Fodor’s solution to Frege cases, is that, interestingly enough, English has no semantics. What has semantics is Mentalese. This demonstrates how Frege’s attempt to use everyday language to look at truth-conditions and propositions was doomed from the start; Fodor argues that ‘strictly speaking, English sentences don’t have a semantics; a fortiori English words don’t have referents and mutatis mutandis English sentences don’t express propositions or have truth-conditions’. (72) Thus one can see that Fodor manages to address Frege cases without appeal to sense, or even content; if the causal powers of a token are not determined by their content, and CTM provides a way to conceive of this, then Sense is not necessary. Of course, RTM has to be cashed in for any of this to work, and Fodor is quite aware of this. One has to also wonder what the implications of Fodor's arguments against CTM are on this argument; if CTM is trouble, as Fodor argues in LOT2 as well, then why is this solution worth anyone's. attention? I'll follow up on this point at the end of the paper. I will now approach Fodor’s attempt to meet Frege’s demand that concepts be public.
Public concepts without Sense
For Frege and Fodor, the thought that concepts turn out to be public is indispensable. In respect to Fodor, his rejection of Quine’s IRS (inferential role semantics), a position which he himself supported in his early work[5], is predicated on how IRS fails to account, or make room for, public concepts.[6] Fodor argues that ‘since practically everybody has some eccentric beliefs about practically everything, holism has it that nobody shares any concepts with anybody else’. (Fodor 2004, 35) The idea is that concepts should prove not to be epistemically relative, and if they are, the notion that we could communicate with them would be implausible; if every concept is linked with another concept in a kind of ‘web’, then what constitutes a concept are these very relations; having one concept C would require having every other concept C is related to. Frege himself feared positions that would fail to make concepts public, and in fact, Fodor argues that this fear is what guided Frege into making concepts have a Sense. I mentioned earlier that, according to Fodor, Frege saw concepts as distinguished by virtue of their reference and their MOP. Fodor argues that Sense is what Frege argues is a concept’s MOP, and furthermore, that ‘(Frege) thinks, quite wrongly, that if MOPs are mental then concepts won’t turn out to be public.’ (Fodor 1998, 20) I will now briefly discuss what public concepts should look like, explain why Sense will not work for Fodor, and then go over what Fodor hopes will retain public concepts; there doesn’t seem to be a clear-cut answer on how exactly RTM retains concepts as public, but the theory is interestingly build on the assumption that they have to be.
How does Fodor retain public concepts without positing sense? For one, it would be good to state what Fodor means by public: ‘Concepts are public; they’re the sorts of things that lots of people can, and do, share’. (28) What this means is that there is a type, like TRIANGLE for instance, that cognisors have a token of. With a concept like FOOD for instance, ‘it should turn out that people who live in very different cultures and/or at very different times (me and Aristotle, for example) have the same concept FOOD’. (29) To put this into context, other theories of mind like Theory-theory, which is influence by the Kuhnian notion of ‘Incommensurability’, would argue that insofar as people could have a different mental theory of TRIANGLE, that their concept of TRIANGLE would in fact not not be the same. (113) Fodor argues on the other hand, that it is perfectly reasonable to assume that he has the same concept of TRIANGLE as Einstein did, in the same way that it is perfectly reasonable to assume that Rousseau and Hume had the same concept of DOG. Fodor rejects what he calls 'variet(ies) of kinds of conceptual relativism; the idea that concepts are incommensurable:
‘If everybody else’s concept WATER is different from mine, then it is literally true that only I have ever wanted a drink of water, and that the intentional generalization ‘Thirsty people seek water’ applies only to me. (And, of course, only I can state that generalization; words express concepts, so if your WATER concept is different from mine, ‘Thirsty people seek water’ means something different when you say it and when I do.’ (29)
As stated earlier, Frege thought what made concepts public couldn’t be mental, and a fortiori, Sense had to be an abstract object. This is obviously not going to work for someone who is interested in mental states. So Fodor has to show how a) MOP could not be senses, and b) How one could conceive of public concepts within RTM. Fodor has already given a hint as to how concepts could be public within RTM; concepts are acquired by virtue of some sort of ‘latching’ on to properties in the world. Thus, WATER and REDNESS are by definition public because WATER is always caused by ‘water’, and REDNESS is always caused by something red. Because concepts are sub-doxastic, they do not depend on one’s thoughts about them, there isn’t really a problem with concepts being public; concepts simply have to be public in some sense insofar as they are all caused by tokenings of the properties they are concepts of. Therefore, what is left is to show why this is incompatible with MOPs as Sense.
Fodor argues that ‘there are good reasons to believe that 5.2 excludes both 5.1 and 5.3’.[7] (16) Why is it that MOPs cannot be abstract objects (5.3) or individuated by their Senses (5.1) if one wants to hold on to the thesis that insofar as ‘MOPs can distinguish concepts, they explain how it is possible to explain one and not the other, of two co-referential concepts(...)’? (Ibid.) In terms of 5.1, Fodor argues that ‘if MOPs are senses, and distinct but co-extensive concepts are distinguished (solely) by their MOPs, then synonymous concepts must be identical, and it must not be possible to think one without thinking the other’. (Ibid.) However, when looking at the logical syntax of concepts, Fodor argues that ‘(i)f ‘a’ and ‘b’ are different names, then the inference from ‘Fa’ to ‘Fb’ is never conceptually necessary’.[8] (17) For instance, if someone told you that ‘Jackson was a painter’ and that ‘Pollock was a painter’, it would seem like ‘that fixes the sense of both names’. (16) However, Fodor claims that it would be perfectly reasonable for one to ‘wonder whether Jackson and Pollock were the same painter’. (Ibid) This leads Fodor to claim that Sense is a fortiori ‘patently’ not what MOPs are; ‘if concepts with the same sense can be different MOPs then (..) MOPs can’t be senses.’ (Ibid.) So in response to 5.1, Fodor is arguing that the content of a concept cannot be what makes it individuated in the way that an MOP would individuate concepts; concepts with the same content can have different MOPs, and therefore ‘Frege’s substitution test doesn’t identify senses’. (17) Fodor concludes that ‘individuating MOPs is more like individuating forms of words than it is like individuating meanings’. (Ibid.)
In terms of 5.3, why can’t Sense be an abstract object if one holds on to 5.2? As stated in an earlier section, Frege needs to invoke Sense as an MOP to individuate concepts because ‘different concepts can have the same referent.’ (19) However, Fodor argues that insofar as MOPs are what ‘Frege holds (...) can individuate concepts’, he cannot then ‘allow that a MOP can correspond to a concept in more than one way.’ (Ibid.) While Frege has no way to stop this beyond ‘sheer stipulation’, if this does manage to enter his theory of concepts, then ‘each way of entertaining the MOP would (presumably) correspond to a different way of thinking the referent, and hence (presumably) to a different concept of the referent’; MOPs however should stand in a bijective relationship to the concept it is a part of for Frege. As Fodor argues: ‘MOPs are supposed to correspond to concepts one-to-one.’ (Ibid) While 5.3 was supposed to retain concepts as public, Fodor argues that it leads to a fatal flaw in the Fregean architecture.This leads quite nicely into Fodor’s attempt to make sense of MOPs coherent in RTM.
Insofar as Fodor is assuming RTM, the discussion of the ‘Fregean Architecture’ is in fact included in the description of what he takes to be the fifth thesis of RTM[9], it is clear that MOPs are going to have to be naturalised in some sense. Not only are MOPs going to have to be naturalised, and I say this because I take it that Sense is non-physical for Frege, but they will also have to fit within RTM, and by extension CTM. In the previous section, I showed how distinctions among concepts must be defined purely syntactically within an RTM framework; in such a way that is negligent of theircontent. Concepts must be syntactically individuated; they will be different in their causal relations. This leads Fodor to simply state that ‘MOPs are mental representations’. (21) Frege’s notion of Sense is thus ill-suited within RTM because, again, Sense is concerned with the content of a representation, while RTM is only concerned with the syntax of representations. Furthermore, per thesis five of RTM, its unclear how MOPs could be causally defined if its a) non-physical, and b) not in the head.
‘If, as seems likely, the identity of a mental state turns on its causal role, then if MOPs are to individuate mental states they will have to be the sorts of things that the causal role of a mental state can turn on. But its a mystery how a MOP could be that sort of thing if MOPs aren’t in the head.’ (Ibid.)
If MOPs are non-mental, then it seems like they could not be the ‘proximal determinants of mental processes (as per thesis five)’. (Ibid) Thus, it looks like: a) Fregean architecture cannot hold if one assumes 5.1, 5.2, and 5.3; and b) 5.1 and 5.3 make it difficult to conceive how MOPs can have causal powers. Furthermore, concepts can retain their public nature because they are sub-doxastic and rely on outside properties to be instantiated; my concept RED is the same as your concept RED because they are both caused by ‘redness’. Again, this is all fine and good, but RTM has to be cashed in for any of this to be meaningful; do Fodor's own arguments against CTM make this solution, as well as the solution to Frege cases, worth mentioning?
What are the problems for CTM? Fodor sketches his general argument as such:
“(1) Computation, as our current cognitive science understands it, is an intrinsically local process; when a computation 'looks at' a representation in its domain, what it is able to 'see', or to operate upon, is the identity and arrangements of its constituents. Nothing else.
(2) But constituent is ipso facto local property of representations.
(3) So, according to CTM, mental processes are themselves ipso facto local, and their locality imposes substantial constraints on what models of the mind CTM can allow.
(4) But broadly empirical considerations suggest that these constraints can't be me.”(2008, 107)
While Fodor’s critique of CTM would need to be considered in detail, something that goes beyond the scope of this paper, it is fair to say at this juncture that, if there is a possibility that CTM as it stands isn't true, it is about the worst thing that's ever happened to cognitive science, as well as Fodor's solution to Frege cases, as well as his attempt to replace Sense. If Fodor is right in doubting that 'the 'classical' RTM LOT CTM model is anything like a general account of the mind', as he claims to have warned in LOT 1, then Fodor's solutions are in serious trouble. (101) Some have doubted that Fodor's arguments against CTM do not succeed, and he should hope they're right.
Fodor’s attempt to meet the requirements Frege set out for what a theory of concepts gets off the ground without positing Sense, as long as one assumes RTM. Fodor accepts the demand that concepts must be public, yet shows the tools Frege uses to meet this demand are ill-suited for the problem, and shows how a causal account of concepts that posits MOPs could retain concepts being public without needing to posit Sinn, or in other words, assuming that content is relevant to computations. However, as Fodor himself point out, in rather paradoxical fashion, there are good reasons to think that RTM, as it is currently, is in serious trouble. Fodor’s seeming rejection of RTM isn’t so surprising though, insofar as he seems to have only assumed RTM in the past because he felt every other option failed to account for something a theory of concepts would have to be. If RTM turns out to be untenable, and other views of concepts have to be assumed, then perhaps concepts won’t turn out to be public and have Senses.
“It’s really the basic idea of RTM that Turing’s story about the nature of mental processes provides the very candidates for MOP-hood that Frege’s story about the individuation of mental states independently requires.” (Fodor 1998, 22)
There’s a sense in which Fodor rejects the tools of the past, while assuming the problem is the same; concepts need to be public and have something to do with the external and the internal, while being rational. If Fodor is right about RTM in LOT 2, this prospect is also in serious trouble. Fodor seems to have a lot more on his plate than a mere debunking of Frege’s ontology; he has to make what might turn out to be a failed theory meet Frege’s ideals for what concepts must be. One should that hope Fodor thinks a theory of concepts is worth having even if these ideals can’t be met.
Work Cited:
Fodor, Jerry. Having Concepts: A Brief Refutation Of The Twentieth Century.Mind and Language 19.1 (2004): 29-47. Print.
Fodor, Jerry A., and Ernest LePore. Holism: a shopper's guide. Oxford: Blackwell, 1992. Print.
Fodor, Jerry A.. Concepts: where cognitive science went wrong. Oxford: Clarendon Press ;, 1998. Print.
Fodor, Jerry A.. LOT 2: the language of thought revisited. Oxford: Clarendon Press;, 2008. Print.
Wittgenstein, Ludwig, David Pears, and Brian McGuinness. Tractatus logico-philosophicus. London: Routledge, 2001. Print.
[1] Wittgenstein hinted at the independence of thought from everyday language in §4.002 of the Tractatus.
[2] Fodor claims Frege’s conclusions are highly influenced by appeals to intuitions. While I will look at Fodor’s assessment of Frege’s work, taking the time to go back to Sinn und Bedeutung itself and seeing the philosophical phenomena at play there would definitely strengthen this paper. However, based on my understanding of Sinn und Bedeutung, Fodor’s analysis does justice to Frege’s work.
[3] The most intuitive of examples I’ve read is: Lois Lane sees Clark Kent going to work. While Clarke Kent and Superman are the same referent, It’s clear that Lois doesn’t believe Superman is going to work. Thus, substituting ‘Lane thinks Kent is going to work’ with ‘Lane thinks Superman is going to work’ doesn’t work- Lane just does not believe that!
[4] Cicero has the property ‘fat’.
[6] This is made explicit in Fodor and Lepore’s: Holism: A Shopper’s Guide (1992), as well as Fodor’s essay: Having Concepts: A Brief Refutation of the Twentieth Century (2004).
[7] The numbers stated here refer to the Fregean account of MOPs on page 15-16 of concepts, and listed on page one of this essay.
[9] ‘Whatever distinguishes coextensive concepts is ipso facto ‘in the head’; this means that it’s available to be a proximal cause (/effect) of mental processes.’ (15)
No comments:
Post a Comment