Asymmetrical Dependence: Annotated Bibliography


Back to Home Asymmetrical Dependence Entry


Fodor, J.A. (1987), Psychosemantics. Cambridge, MA: MIT/Bradford.

This is the book in which Fodor first formulates his asymmetrical causal dependency theory of meaning.  The book treats many topics, ranging from Fodor’s defense of a realism of content and the propositional attitudes--his representational theory of mind (RTM)--to a defense of a view of narrow content of mental states (content that would be shared by Fodor and Twin-Fodor on Twin-Earth), an attack on meaning holism, and a defense of the language of thought hypothesis (LOT). In addition to being the first place Fodor articulates the asymmetrical causal dependency theory, it is the only place he explains one of the less articulated conditions of his theory. This is the condition that the dependencies be synchronic, not diachronic.  So if Xs cause “X”s and if Y’s cause “X”s only because Xs cause “X”s, the dependency of Y-caused-“X”s on X-caused must be simultaneous in time, not arise over time.

Dennett, D.C., (1987a), "Review of J. Fodor, Psychosemantics", Journal of Philosophy, 85, 384-389.

In this early review of Fodor's 1987 book, Dennett raises two difficulties for Fodor's asymmetrical causal dependency theory of meaning. The first is that in the Psychosemantics version of Fodor's conditions on meaning, there is nothing to block twin-earth cases. Dennett raises the issue of whether Fodor's theory violates our intuitions about meaning and twin-earth cases. Dennett says that if we suppose some stray xyz were to token Fodor's mentalese 'water' symbol, then Fodor and this twin would be broad-content twins (where as, following Putnam, if Fodor were instantaneously transported to twin-earth we would think his symbol were falsely tokened). (Fodor addresses this concern in 1990c.) The second is that Fodor's atomistic conditions on meaning raise the worry about whether a child may acquire the concept FATHER without having the concepts MALE and PARENT. Holistic theories of meaning may deny that that child has the full-blown concept of PARENT. Fodor would seem to need to acknowledge a primitive concept's existence or, at least, non-access to internal structure. Of course, Fodor can distinguish concepts from meaning, and later does so to handle just this sort of worry.

Dennett, D.C., (1987b), "Evolution, Error, and Intentionality", in The Intentional Stance (Cambridge, MA: MIT/Bradford), 287-321.

Here Dennett's treatment of Fodor's theory of meaning is brief. It is confined to a prod and a challenge. The prod is an attempt to get Fodor to reconsider his retreat from a teleological approach to naturalized meaning. The challenge is for Fodor's account of meaning to explain why a mentalese 'horse' token means horse rather than horse-or-other-quadruped-resembling-a-horse. Dennett says that Fodor's theory cannot meet the challenge. Dennett is probably right, but Fodor get's off on a technicality (of offering only sufficient conditions for meaning)

Baker, L.R., (1989), "On a Causal Theory of Content", Philosophical Perspectives, 3, 165-186.

This paper introduces Baker's well-know example of the robot-cats problem. If Jerry is shown only robot-cats to begin and then later sees a real cat chasing a robot cat and tokens 'cat', does the token mean cat, robot cat or disjunctively both? Baker argues for none of the above. She rejects the most likely choice (cat or robot cat) on the grounds that when Jerry discovers the difference between cats and robot cats and says "I was previously mistaken," Fodor's theory offers no principled way to account for the (second order) mistake. Baker concludes that Fodor has not solved the disjunction problem (or error problem).

Bogdan, Radu, (1989), "Does Semantics Run the Psyche?" Philosophy and Phenomenological Research, 49, 687-700.

This paper is largely about whether the meanings of mental state are causally relevant to psychological explanation. Bogdan also complains that Fodor is not naturalizing semantics but assuming "intentionality and naturaliz[ing] the conditions of its semantic, indeed denotational success." No specific discussion exists evaluating Fodor's asymmetrical dependency theory of meaning.

Cummins, R. (1989), "Covariance II: Fodor" In Meaning and Mental Representation. Cambridge, MA: MIT/Bradford, 56-66.

Cummins denies that Fodor's (1987) account works. He uses the example of shrews and mice. Mice cause 'mouse' tokens and shrews (let us suppose) cause 'mouse' tokens. Is the latter really asymmetrically dependent upon the former? Cummins suggests not because both mice and shrews do this by exploiting their mousey-looks ML as an intermediate causal link. Break the shrew's link between its ML and the 'mouse' and you would also break the link between the mouse's ML and 'mouse.' So there is not a causal asymmetry. In Fodor's defense, it would be possible to break the shrew-ML link without breaking the mouse-ML link. Thus, it would be possible to break the shrew-'mouse' link without breaking the mouse-'mouse' link. Also, Cummins is focusing too narrowly on look. If Fodor factors in all of the sensory projections of mice, there will surely be some ingredients of mice not shared by shrews. The fact that shrews share some of them and thereby produce tokens of 'mouse' does not show that there is no asymmetrical dependence of 'mouse' on mice in order for these other connections to hold. Cummins also calls 'ad hoc' Fodor's stipulation that the dependencies be synchronic. (But Fodor gets to stipulate sufficiency conditions!)

Godfrey-Smith, P. (1989), "Misinformation," Canadian Journal of Philosophy, 19, 533-550.

Godfrey-Smith claims that Fodor's (1987) account does not solve the problem of misrepresentation (because it does not solve the flip-side of that problem, viz. the disjunction problem). Take the symbol 'horse'. To be falsely tokened there must be a set of things K that can cause the symbol to be tokened. K may include horses, but it will include cows on dark nights, muddied zebras, and other things of which 'horse' is to be falsely tokened. To make 'horse' mean just horse, any other element in K's causing 'horse' must asymmetrically depend upon horses causing 'horse'. But Godfrey-Smith claims that, among elements of K, since they can each cause 'horse', horses have no more claim on 'horse' (claim to be the dependent link) than any other element in K. If true, then Fodor would not have solved the problem of misrepresentation (or the disjunction problem)

Fodor, J.A. (1990a), "Psychosemantics or: Where Do Truth Conditions Come From?" In Lycan, W. (ed.) Mind and Cognition. Oxford: Blackwell, 312-337.

This is a paper that Fodor now disavows. Nonetheless, Lycan convinced him to let it see the light of day. This paper comprises Fodor’s, shall we say, flirtation with teleological semantics. This flirtation may also explain why his negative reaction to such semantic theories (in his later papers) was so ardent. This paper also flirts with something else that Fodor now strongly rejects, viz. that a part of a symbol’s meaning is its inferential role. Both of these ideas are abandoned in his asymmetrical causal dependency theory of meaning. This paper is Fodor’s attempt to show that the symbols of a representational theory of mind inherit their semantic properties from their cognitive teleological functional role.  The relevant teleological functional roles themselves are to be cashed out in terms of relevant counterfactuals about what symbols would be doing (would be causing or being caused by) in certain “normal” conditions. And what the symbols would be doing is also partly to be explained via some form of selectional advantage. The basic idea is that “X” means X if X’s would cause “X”s in one’s cognitive “belief box” under normal conditions of well functioning of one’s cognitive system. Fodor tries to cash the contents of symbols “X” in terms of their “entry conditions”—normally a state of affairs of there being an X about. (Notice the difference between this view where the content of a symbol is a proposition rather than a property, as in his later theory where it is a property.) As is well known, among other reasons, Fodor abandons this view because he maintains that there are no non-circular ways of defining “normal” conditions of cognitive well functioning.  At the end of this very paper, he is painfully aware of just these worries about a teleologically based semantics.

Fodor, J.A., (1990b), "Information and Representation" In Hanson, P. (ed.) Information, Language, and Cognition. Vancouver: University of British Columbia Press, 175-190.

In this paper, Fodor discovers a new version of the disjunction problem--one not central to the problem of error or misrepresentation (and one he seems not to have anticipated in Fodor, 1987). He divides up uses of symbols (such as "platypus") into different kinds: labelling use "that is a platypus", and representational use "a platypus is a mammal." He then suggests that information-based semantic theories must allow for representational uses in which the tokening of a symbol does not (and is not intended to) carry the information that the cause of the symbol has the property expressed by the symbol. One may token "platypus" as a result of seeing a platypus or of thinking of Melbourne. The latter would not constitute an error and yet "platypus" does not mean platypus or Melbourne. The solution suggested in this paper is that the true labelling uses of "platypus" carry the information that they are caused by platypai and the tokenings that do not carry this information are asymmetrically dependent on those that do.

Fodor, J.A. (1990c), A Theory of Content and Other Essays. Cambridge, MA: MIT/Bradford.

Except for two chapters on Fodor’s theory of content, this book is a collection of ten essays that had been previously published.  Fodor himself partitions the book into essays on intentionality and the modularity of mind.  With respect to Fodor’s theory of meaning, the essays in the first partition are especially important because they contain many keys to the development of Fodor’s “considered” view of naturalized meaning, as it has developed.  First, (in Chapter 2) we find his attack on what he calls “Wisconsin Semantics”—the early attempts to naturalize meaning by Stampe and Dretske. (This gives us Fodor’s thoughts about meaning from as early as 1984.) Here Fodor first formulates what has become know as the “disjunction problem” for certain versions of causal and/or correlational theories of semantics.  Second, (in Chapter 3) we find Fodor’s attack on teleological versions of semantics (including his own early flirtation with such views). Fodor believes all such views founder on the inability to non-circularly define a set of “normal conditions” under which a mental state would perform its normal function to indicate (thereby giving it a meaning…meaning the thing or property indicated under normal conditions).  Third (in Chapter 4) we find Fodor’s fullest development of his asymmetrical dependency theory of meaning. It takes the form of replies to objections and considers and replies to all of the significant objections he had received during the last three years (since he first formulated the view in Psychosemantics).  For our purposes, these are the crucial chapters of the book. However, it also includes very fertile work on Frege puzzles about individuation of belief, a review of Schiffer’s Remnants of Meaning, a precis of Fodor’s own Modularity of Mind, and continuation of a debate between Fodor and Paul Churchland on the observation/theory distinction and the nature of perception.

Maloney, J.C., (1990), "Mental Misrepresentation," Philosophy of Science, 57, 445-458.

Maloney denies that Fodor's (1987) account works because the required causal asymmetries do not exist. Maloney says that it is false that cows cause 'horse' tokens only because horses do. He claims that in worlds where the connection between cows and 'horse' tokens is broken, also broken will be the connection between horses and 'horse' tokens. Maloney seems to argue that cows produce a horsey manifest image and that horses do too. If it is via this manifest image that cows and horses cause 'horse' tokens, then to break the connection between cows and 'horse' tokens (via this image) would also break the connection between horses and 'horse' tokens, via this image. Maloney also takes Fodor to task for not distinguishing finely enough between meanings and concepts. Maloney says Fodor's view cannot distinguish a 'water' token's meaning H20 from a token of 'H20's meaning H20, yet these are clearly different concepts. Of this charge, Fodor's 1987 account may not be guilty, and his 1990c account is surely innocent.

Sterelny, K., (1990), The Representational Theory of Mind. Oxford: Blackwell, pp. 119-121.

Sterelny suggests that Fodor needs (but lacks) a non-arbitrary way to restrict the counterfactuals that yield the asymmetrical dependencies he needs for meaning. Without restrictions, the asymmetries fail to hold. This is the same basic strategy of Godfrey-Smith and Manfredi & Summerfield, et.al. Sterelny also complains that Fodor's account fails to apply to non-natural kind concepts and their meanings and that Fodor may not have solved the "depth" problem. The "depth" problem is that of saying why "X" means distal X not proximal sensory projection of X.

Antony, L & Levine, J. (1991), "The Nomic and the Robust" In Loewer, B. & Rey, G. (eds.), Meaning in Mind: Fodor and his Critics. Oxford: Basil Blackwell, 1-16.

Antony and Levine accuse Fodor of being caught in the grips of inconsistency. On one hand he holds that only minds can detect anomic properties, such as the property of being a shirt. On the other hand he holds that terms like "shirt" get their meaning from lawful relations with shirts. Something seems to have to give. Antony and Levine suggest that the asymmetry condition in Fodor's theory of meaning can replace the anomicity condition in his theory of mindedness. Antony and Levine also raise the worry that Fodor has not solved the problem of a term like "horse" means horse rather than a disjunction of proximal sensory projections of horses.

Baker, L.R. (1991), "Has Content Been Naturalized?" In Loewer, B. & Rey, G. (eds.), Meaning in Mind: Fodor and his Critics. Oxford: Basil Blackwell, 17-32.

Baker opens with her unicorn/shunicorn example (shunicorns are small zebras with horn in the middle of their forehead). She denies that Fodor's theory can both treat 'unicorn' as an uninstantiated primitive term and meet the conditions of his theory. She claims there will be no asymmetrical dependence of shunicorn caused "unicorn" tokens on unicorn caused tokens. She also denies that Fodor could solve this problem by treating "unicorn" as a non-primitive term, claiming that to do so would conflict with naturalization of meaning. She follows with her robot-cat example from her earlier paper (Baker, 1989) and a discussion of the role of robustness in Fodor's theory (whether it can be used to rule out counterexamples to asymmetrical dependencies). Baker ends with a charge that Fodor's "robustness" condition cannot be motivated for pre-meaningful, syntactically individuated items (as it must be to be a condition on those items acquiring a meaning). Robustness, she argues, comes into play only after meaningful items exist and cannot be a basis of meaning. Finally, Baker argues (convincingly) that if "cow" is syntactically identified, it should be able to be robustly caused by anything (a probe, say)-rendering robustness trivial.

Boghossian, Paul, (1991), "Naturalizing Content" In Loewer, B. & Rey, G. (eds.), Meaning in Mind: Fodor and his Critics. Oxford: Basil Blackwell, 65-86.

Boghossian opens with worries about the project to naturalize meaning and its motivation. His challenge to the particulars of Fodor's theory begin when he tries to show that Fodor's theory reduces to a "type I" theory of the kind Fodor rejects (Fodor, 1990c). Type I theories require a set of circumstances (ideal, normal) where only the items in the extension of a term would token that term. After attempting to show that Fodor's theory reduces to a Type I theory, Boghossian attempts to show that there really are no worlds or sets of worlds which instantiate Type I theories (or at least, none can be specified in naturalistic terms). Boghossian closes by arguing that asymmetrical dependence will not explain why our term "water" means H20, not XYZ because the distance metric for dependency across possible worlds comes out wrong.

Fodor, J.A. (1991) "Replies" In Lower, B. & Rey, G. (eds.) Meaning in Mind: Fodor and His Critics. Oxford: Blackwell, 255-319.

To Antony and Levine:

Fodor admits that there is a tension and says that if forced to choose, he'd choose his theory of meaning. He suggests that we replace "anomic" with "inferential" in his account of mindedness, and it stands as before. His reply to the proximal projection problem is that large, open disjunctions of proximal stimuli are anomic and cannot satisfy the conditions of his theory of meaning.

To Baker:

Fodor is adamant that the worlds where unicorns cause "unicorn"s is closer than those were shunicorns cause "unicorn"s (though this is more by stipulation than by explanation). Failing this, Fodor acknowledges that he still has the option of allowing "unicorn" to be a non-primitive term (decompose and conquer) to get around Baker's objection. He (rightfully, I think) rejects Baker's claim that the decomposition strategy conflicts with the naturalization of meaning. To Baker's objections about robustness, Fodor replies that robustness is more than just that more than one thing can cause a syntactically identified object. It involves asymmetrical dependence, as well and this isn't trivial. True, but then Fodor seems to go into denial trying to fend off the consequences of allowing that cows cause "cow"s and probes cause "cow"s (robustly). If the causation were not asymmetrically dependent, Fodor suggests "cow" would be ambiguous. But, as we all know, "cow" is not ambiguous. So if probes would cause "cow"s (Baker seems to suggest that this would be the case) and their doing so would not be synchronically dependent upon cows causing them, this would render Fodor's theory inapplicable to our term "cow." (Unfortunately, Baker does not press this consequence and so Fodor does not have to respond to a charge of vacuity for this theory.)

To Boghossian:

Fodor defends the naturalization of semantics. He considers and rejects Boghossian's reasons for thinking that Fodor's theory is a type I theory. Fodor defends his claim that our "water" tokens mean H20 because each world in which H20 and XYZ are distinguishable and we apply "water" to H20 but not XYZ is closer to us than the world that is closest to it in which H20 and XYZ are distinguishable and we apply "water" to XYZ but not to H20.

To Loar:

Fodor says that mental representations cannot be deferential-only words of a natural language may be. So he attempts to skirt some of the issues Loar raises by dividing a theory of intentionality (his) from a theory of linguistic meaning (say, Grice's). About Loar's use of intentions, Fodor reminds us that it keeps us within the intentional circle out of which his theory is designed to break. As for the others of Loar's examples that lean on a person's guiding intentions, Fodor replies with the familiar point that one may easily refer to things (stars) while having quite false guiding intentions about what one is doing (referring to holes in the fabric of the heavens). Fodor and Loar are on such different wave-lengths that there are few points of contact between them in paper and reply.

To Millikan:

Fodor addresses the difference between teleological accounts of belief and of content (apparently embracing the former and still rejecting the latter). He finds Millikan's reliance on the notion of a user of representations to be question-begging. He accuses her of describing the users in intentional terms (in terms of what they care about). This won't get us out of the intentional circle. Also, Fodor emphasizes that the users can equally be described as performing chemical processes on ambient black dots in a world where such black dots are largely food. So the disjunction problem remains: something ambient, or black or specklike or food.

Jones, T, Mulaire,E., & Stich, S. (1991), "Staving Off Catastrophe: A Critical Notice of Jerry Fodor's Psychosemantics," Mind & Language, 6, 58-82.

The authors argue that Fodor's (1987) account does not solve the disjunction problem. To show this, they introduce the notion of a horse look-a-like, viz. a "Disney donkey" (folks at Disney studios doctor up a donkey so that even on careful inspection, Jerry tokens 'horse' in response to the Disney donkey. Jones et. al. claim that Jerry's token of 'horse' means horse, but that Fodor's theory cannot account for this meaning because they deny that Disney donkey's cause 'horse' tokens only because horses do. They deny it because both horses and Disney donkey's cause 'horse' tokens via their looks (which are identical). They claim that on Fodor's account 'horse' means horse or Disney donkey. Alternatively, they might have suggested that 'horse' means horsey-look, not horse. Either would be a problem for Fodor's account which is obviously designed to explain how 'horse' means horse. In Fodor's defense, Jones et.al., may be focusing too narrowly on one sensory modality. Other sensory modalities may preserve the asymmetrical dependence, when factored in.

Loar, B., (1991), " Can We Explain Intentionality?" In Loewer, B. & Rey, G., (eds) Meaning in Mind: Fodor and His Critics. Oxford: Blackwell, 119-135.

Loar begins by suggesting that Fodor's account should be limited to terms for demonstrable properties (what Loar calls "recognitional concepts.") Loar suggests that the account will not easily fit concepts of semantic deference or theoreticals (despite Fodor's attempt to apply it to "proton"). Loar gives an example where one's term "horse" is socially to defer to the experts and which does not seem to fit Fodor's account cleanly. Also it will bring intentions and meanings into Fodor's causal conditions of meaning. (I think Fodor did not carefully address how his account might handle deference and that Loar is right to raise the issue.) Loar also presents a series of examples all designed to show the role that intentions might play in the acquisition of terms. He doubts that Fodor's theory eliminates the need to appeal to such intentions or that Fodor's theory could correctly handle his examples without appeal to referential intentions. Loar raises the problem of why terms for visual kinds don't mean projections on the retina, on Fodor's theory, despite Fodor's protests to the contrary. He suggests that "unicorn" is a vacuous term because there is no property of being a unicorn.

Millikan, R. (1991), "Speaking Up For Darwin." In Loewer, B. & Rey, G. (eds.), Meaning in Mind: Fodor and his Critics. Oxford: Basil Blackwell, 151-164.

Millikan criticizes Fodor for his attacks on teleofunctional theories (including his own earlier version). She questions what asymmetrical causal dependency could be. She suggest that Fodor's account should really be one of a teleological "because" rather than a merely causal "because" in interpreting Fodor's claim that, for example, cows can only cause "horse" tokens because horses do. Unfortunately, this is all high-level abstract criticism and it makes little contact with the nitty-gritty of Fodor's conditions. Where Millikan's comments do directly address Fodor's concerns is with respect to Fodor's rejection of teleofunctionalism on the grounds that it cannot solve the disjunction problem. In his discussion of teleological accounts of what frog snaps mean, Fodor rejects teleological accounts in part because Fodor says, on them, the meaning of frog snaps is indeterminate between flies, flies or bee-bees, and flies or small black dots. Millikan claims it is not indeterminate. Bee-bees are out, she thinks, because they are not parts of the frogs natural environments. Disjunctions are out, she thinks, because they are not parts of causal explanations. She claims snaps have the function of detecting both small black dots and flies (multiple teleofunctions yields multiple contents). Yet, to go from function to content we have factor in the user of the symbol system. When the user is factored in, the snaps come out meaning frog food (supposedly to the frog).

Adams, F. & Aizawa, K. (1992), "'X' Means X: Semantics Fodor-Style", Minds and Machines, 2, 175-183.

This paper attempts to motivate the conditions of Fodor’s asymmetrical causal dependency theory of meaning, supplementing Fodor’s own motivation. Then the paper argues that Fodor’s conditions on meaning are both too strong and too weak. They are too strong because there is no plausible reason why a purely syntactical object in one’s brain "cow" may not be caused pathologically (say by electrodes). So there is little plausibility to the claim that in a person’s brain "cow" asymmetrically depends causally on being produced by cows. Finally, the conditions of the theory are too weak. There may well be cases of causal asymmetry that do not generate meaning. The paper considers such a case.

Cram, Hans-Robert, (1992), "Fodor's Causal Theory of Representation", The Philosophical Quarterly, 42, 56-70.

Cram laments that Fodor's account gives no explanation of why the required causal asymmetries would have the features they are supposed to have in order to yield meaning. Cram then compares Fodor's account of meaning with Dretske's (1986) account of misrepresentation and borrows ideas from that account to interpret Fodor's theory. Cram's basic idea is to find a causal route from, say, horse to "horse" that nothing else shares even though there is a route from horse to "horse" that cows may share (their look). The unique route from horses to "horses" would preserve the asymmetrical dependence, according to Cram.

Manfredi, P., & Summerfield, D. (1992), "Robustness Without Asymmetry: A Flaw in Fodor's Theory of Content," Philosophical Studies, 66, 261-283.

Manfredi and Summerfield suggest that there is not a single problem that goes by the name "disjunction problem," but many, including "error" and "robustness." They give an attempted counterexample. Let there be a cow-'cow' connection and a horse-'cow' connection and then allow cows to change most of their perceptible properties. They claim that the first connection may be broken but not the second, yet 'cow' would still mean cow, contrary to Fodor's theory. Fodor might well respond that cows would still cause 'cow's (e.g., stuffed cows would, pictures of cows would, and so on) so that the first connection had not in fact been broken (though a new connection between altered looking cows and 'cow's might have been established). Manfredi and Summerfield contrast their purported counterexample with those of Cummins and Maloney, claiming that the latter ignore the heterogeneity of mechanisms that mediate semantic referent and mental token. They close their paper with objections and replies to their counterexample.

Millar, Alan, (1992), "Critical Notice of 'A Theory of Content and Other Essays.' " The Philosophical Quarterly, 42, 367-372.

Millar notes that the fact of asymmetric dependencies seems to cry out for further explanation (assuming that Fodor is right about their existence). Millar also challenges Fodor's rejection of meaning holism and meaning normativity.

Myin, E., (1992), "Some Problems for Fodor's Theory of Content." Philosophica, 50, 101-121.

Largely expository, this piece highlights the important insight that meaning and mental states might come apart. Things other than mental states might have genuine, semantic, meaning (at least, according to Fodor). It also offers the criticism that Fodor's theory of content cannot satisfy some of Fodor's own desiderata. Specifically, it is argued that his theory cannot make all meaning determinate. It cannot handle things like the meaning of frog snaps or handle the meanings of "virtuous" or "justified" or other predicates about properties over which people disagree. The claim is that Fodor's theory cannot be applied in practice unless these disputes were settled. But nor could any theory, so this seems like no objection at all. Myin's second objection is that there are too many things that will satisfy Fodor's conditions for meaning and not be belief states. He notes that Fodor explicitly allows these two things to come apart, but argues that this split is inconsistent with other of Fodor's desiderata concerning content and psychological laws. Overall, the paper's exposition receives higher marks than its critical appraisal.

Wallis, C., (1992), "Asymmetric Dependence and Mental Representation." Psycoloquy, 3 (70) Fodor Representation (1). avalaible online

This paper is a shortened but otherwise identical paper to Wallis, 1995.

Adams, F. & Aizawa, K. (1993), "Fodorian Semantics, Pathologies, and Block's Problem", Mind and Machines, 3, 97-104.

This paper compares and contrasts the "pathologies" problem (from Adams & Aizawa, 1992) with an objection that Block raises for Fodor's account of meaning. Block suggests that if a term "cow" is taken as an orthographic item, then there is no asymmetric dependency of non-cows causing "cow"s upon cows causing "cow"s. He suggests that in nearby worlds "cow" may mean tree and trees may cause "cow"s. Block's objection bears a strong resemblance to an objection of Adams & Aizawa (1992), but this paper shows that Fodor has a way out of Block's problem that does not count as a way out of the pathologies problem.

DeWitt, R., (1993), "Representations and the Foundations of Cognitive Science: Commentary on Wallis on Fodor-Representation", Psycoloquy: 4 (11) Fodor Representation (3). avalaible online

DeWitt largely focuses on the problem of whether representations can be causally explanatory in cognitive science. He suggests that they cannot, but gives no reasons to support this. He acknowledges that Fodor's intentional realism commits him to the view that representations are causally explanatory.

Livingston, K. (1993), "What Fodor Means: Some Thoughts on Reading Jerry Fodor's A Theory of Content and Other Essays," Philosophical Psychology, 6, 289-301.

The title of the essay is descriptively correct. Livingston fits Fodor's theory of meaning into the context of a host of other familiar Fodorian themes: intentional realism, anti-relativism, anti-holism, and modularity of mind, theory-neutrality of perception, among others. Specifically with respect to the asymmetrical theory of meaning, Livingston suggests that cognitive science should turn to the search for "mechanisms" that could produce the asymmetrical dependencies needed for Fodorian semantics. Livingston seems to suggest that empirical search for such the possibility of such mechanisms might determine whether Fodor's theory is correct.

Mortensen, C., & O'Brien, G., (1993), "Representation and Causal Asymmetry. Commentary on Wallis on Fodor-Representation" Psycoloquy, 4 (19) Fodor Representation (5). avalaible online

The authors attempt to defend Fodor against Wallis' (1992) objections. The most significant defense they offer is that Fodor's (1990c) account has an historical component. They suggest that the historical component can be used to generate asymmetries where Wallis claims there are none. Interestingly, this flies in the face of Fodor's own rejection of the (1990c) historical component in his later work (1994).

Pietroski, P., (1993), "Fodor Unscathed: Commentary on Wallis on Fodor-Representation." Psycoloquy, 4 (10) Fodor Representation (2). avalaible online

Pietroski defends Fodor from Wallis' charges. Correctly, Pietroski points out that much of Wallis' criticism is misdirected at natural meaning as opposed to "non-natural" meaning that Fodor is trying to capture. Pietroski also correctly suggests that Wallis focuses too narrowly on cases where there is but one type of sensory modality from external object (barn) to symbol ("barn"). These features make Wallis' objections wide of the mark.

Seager, W., (1993), "Fodor's Theory of Content: Problems and Objections." Philosophy of Science, 60, 262-277.

Seager raises a large number of issues for Fodor's theory. According to Seager, Fodor sometimes treats his conditions for meaning as necessary, as well as, sufficient conditions. Also, Fodor does not define syntactically primitive predicates. Fodor's theory entails that all terms must be capable of misapplication (which Seager finds "bizzare"). Fodor's theory assigns the wrong meaning to uninstantiated terms such as 'hobbit;' says Seager. And, if disjunctive properties and laws are allowed, neither version of Fodor's theory (the one with the condition that the laws between properties and symbols be instantiated or the one without) solves the disjunction problem. Seager also give examples to indicate that asymmetric dependencies of the kind Fodor's theory requires simply do not exist (malapropisms, spoonerisms). Also, Seager believes that meaning comes first and asymmetrical dependence follows (not the other way around).

Wallis, C., (1993a), "Mental Representation and Cognitive Science: Reply to DeWitt and to Pietroski on Wallis on Fodor-Representation." Psycoloquy: 4 (45) Fodor Representation (4). avalaible online

Wallis tries to repel Pietroski's defense of Fodor. Attention turns to whether Pietroski's rescue of Fodor depends on a holism. I don't think it does, but Wallis interprets it in such a way that he can use this against Pietroski. The idea is that Wallis charges Pietroski with saying that the barn-"barn" law is the asymmetrical dependent law over façade-"barn" because one has beliefs about the depths of barns. But surely Fodor could not exploit this because one has no beliefs about barns at all until one acquires the term "barn". The asymmetry must be in place long before there are beliefs about barns (or facades). I don't think Pietroski (or Fodor) would make this mistake. All Pietroski (or Fodor) need to repel Wallis' case is that when "barn" means barn, facades (suitably inspected) simply would not cause "barns"…period. Otherwise Fodor's theory may say "barn" means barn or façade. Wallis has to say that "barn" means barn when there is no asymmetry at all of any kind. That seems enormously unlikely. Wallis correctly defends Fodor against some of DeWitt's charges and points out that Fodor's theory of representation is not unique in having the problem of explaining how mental representations can have causal power.

Wallis, C., (1993b) "Counterfactuals, Asymmetry, and Representaiton: Reply to Mortensen & O'Brien on Fodor-Representation." Psycoloquy: 4 (45) Fodor Representation (6)

Pietroski correctly noted that most of Wallis' examples of symmetries in cognitive science are relevant only to natural meaning and not to semantic content of symbols that can be falsely tokened. Wallis' replies are to Mortensen and O'Brien on the particulars of taking Wallis' symmetries from science to show more. Wallis does a good job of defending against Mortensen and O'Brien's suggestions, which take Wallis' examples to present obstacles to Fodor's account of meaning (at the level of false tokening). Neither Wallis nor Mortensen & O'Brien seem to notice that their exchanges about an organism's history and its affect on counterfactuals likely violate Fodor's stipulation that the asymmetrical dependencies of his theory be synchronic, not diachronic.

Adams, F. & Aizawa, K. (1994a), "'X' Means X: Fodor/Warfield Semantics", Minds and Machines, 4, 215-231.

This paper argues that the Warfield's attempt to rescue Fodorian semantics from difficulty does not succeed. Warfield's emendation requiring that the laws of the asymmetrical dependence condition be instantiated only requires a change of example. Twin-Earth cases are given to show that Fodor's conditions will not be met and yet Ted's "water" tokens will mean water. Other cases are given to show that when Fodor's conditions are met, they give the wrong content assignments. For example, Ted in a world of both H20 and XYZ, and where by coincidence he instantiates only the H20-"water" law life long, should mean H20 by "water" (on Warfield's modified theory). Adams & Aizawa claim (plausibly) that in such an environment, Ted's "water" tokens actually would mean H20 or XYZ, nonetheless. Hence, Fodor's theory threatens to be vacuous-giving incorrect content assignments. Finally, an example is given to show that a perfectly meaningless item in the language of thought "Framus" could satisfy Warfields conditions for being meaningful, including that symbols be syntactic items in a language of thought, and yet be meaningless (despite satisfying Fodor/Warfield's conditions for having a meaning).

Adams, F. & Aizawa, K. (1994b), "Fodorian Semantics" In Stich, S. and Warfield, T. (eds.) Mental Representations. Oxford: Basil Blackwell, 223-242.

This paper considers several different interpretations of Fodor's conditions for meaning. It shows difficulties for his theory on each known interpretation of the conditions. Fodor's theory is shown to be unable to block attributions of meaning to items that are meaningless. The theory is also shown to founder on assignments of meaning in typical Twin-Earth cases. Additionally, it is shown that Fodor's theory very likely is vacuous-not applying to human brains or any actual items that have meaning. Lastly, it is suggested that the sorts of asymmetrical causal dependencies Fodor discovers are a consequence of meaning, not a source of meaning.

Fodor, J.A., (1994), The Elm and the Expert. Cambridge, MA: MIT/Bradford.

These are Fodor’s 1993 Nicod Lectures. Perhaps the most interesting thing that happens in this book is that Fodor gives up narrow content. He now denies that it is necessary to save a unified psychology (one that applies to both Fodor and Twin-Fodor). The book begins trying to reconcile Frege Puzzles and Twin Earth Puzzles with computational implementation of broad content intentional psychological laws. Fodor’s way out is to abandon the goal. He now claims that there will be mechanisms that prevent Twin cases and Frege cases from arising in creatures that satisfy broad content intentional laws. (A neat trick if it worked, but it probably doesn’t. Fodor seems to think psychology is still safe if Twin and Frege cases don’t arise very often. But they surely seem to arise all the time. There need not be a Twin-Earth to have twin-like cases. Jade/Jadite, gold/fool’s gold serve equally well to water/twin-water.) There is also a lecture on Fodor’s answer to Quine on the inscrutability of reference, and there is a very curious lecture on knowing the content of one’s thoughts. For our purposes, the two most important sections of the book are the two appendices. In the first appendix, Fodor applies his theory of content to the meanings of (mental) names.  This should be a problem for him because his theory was built to apply to laws ranging over properties (Xs) and predicates (“X”s). Content laws for names would range over individuals (Aristotle) and names “Aristotle”). A problem is that individuals seem not to feature in laws. Fodor tries to solve the problem by having laws connect the property of being Aristotle with “Aristotle.” (This probably won’t work, for it would have names meaning properties, not individuals.)  In the second appendix, Fodor completely abandons what may be called his flirtation with an historical condition in his theory of meaning. He at one time included but now strongly rejects that the law “Xs cause ‘X’s” must be instantiated. He now thinks the counterfactuals (“Xs would cause ‘X’s”) alone can establish meaning. This move entails that Davidson’s Swampman has thoughts the instant he assembles! (This surely seems false to historicalists like me. It also takes away Fodor’s ability to solve the water/twin-water problem. He now solves the latter by saying it won’t happen…or won’t happen much).

Wallis, C., (1994), "Representation and the Imperfect Ideal" Philosophy of Science, 61,407-428.

This paper considers attempts to define meaning in terms of co-variation of representation and represented under idealized conditions. Reminiscent of Fodor's own attacks, it claims that co-variationists cannot solve the disjunction problem by saying that when fake Xs cause "X"s the conditions are always non-ideal (without circularity). Wallace contrasts idealization in physics with that in naturalized semantics. He further argues that idealizing assumptions are contrary to current cognitive science where real systems exploit numerous heuristics to make up for the lack of ideal conditions. Wallace then denies what Fodor's theory requires for "barn" to mean barn, viz. that barns cause "barn"s and fake barns cause "barns" and there are conditions under which the fake barns cause "barn"s only because barns cause "barn"s. There are no latter conditions because, Wallace claims, they would require "ideal" conditions of the type Wallace claims do not exist in cognitive science (and may not exist in science).

Warfield, T., (1994), "Fodorian Semantics: A Reply to Adams and Aizawa," Minds & Machines, 4, 205-214.

Warfield argues that Fodor can get around the "pathologies problem" of Adams & Aizawa (1992), where they claim Fodor's conditions are too strong, because Fodor's asymmetry condition applies only to instantiated laws and the pathologies cited by Adams & Aizawa are not instantiated. Warfield claims that Fodor can get around the claim that Fodor's conditions are too weak, by strengthening what counts as a symbol. This would block the example Adams & Aizawa offer as something that meets Fodor's conditions, but is not a symbol with a meaning. Warfield suggests restricting Fodor's conditions to syntactic items in a language of thought (not unreasonably, given Fodor's other views).

Neander, K., (1995), "Misrepresenting & Malfunctioning." Philosophical Studies, 79, 109-141.

Among other things going on in this paper, Neander takes on Fodor's challenge that teleology cannot help naturalize semantics because natural selection comprises an extensional context. Neander argues that any apparent indeterminacy of function attribution is generated not by extensionality of context but by natural causal hierarchies and "by" relations (x does y by doing z). She argues that specific functions can be determined. Suppose x does y and z but that the normal environment that allows z changes. Now x does y but y does not lead to z. X still does not necessarily malfunction. Its doing z may still be its specific function. This all plays out when we turn to Dretske's bacterium that seek magnetic north in order to reach oxygen poor environments and frogs that zap small dark moving dots in order to get food. Neander wants to say we can give determinate functions (and contents) in theses cases by resorting to her account of "most specific function". Neander also points out (correctly, I think) that it is because Fodor rejects accounts of meaning (and function) that appeal to causal history, rather than only counterfactuals, that he thinks teleology is not going to help with naturalized semantics. Neander argues that since Fodor is wrong about causal history not being relevant to selection, this contributes to his being wrong about causal history not being relevant to semantics.

Voltolini, A., (1995), "Is Meaning Without Actually Existing Reference Naturalizable?" Grazer Philosophische Studien, 50, 397-414.

According to Fodor (1990c), meaningful expressions denoting no actual entity, like "unicorn", do not constitute an exception to his project of semantic naturalization based on the notion of asymmetrical dependence between causal relations. But Fodor does not give any principled reason in order to show that, say, a non-unicorn caused "unicorn"-token means UNICORN, as he on the contrary does regarding a non-X caused "X"-token for any existing X. Nevertheless, his claim that one such expression has a mere denotational meaning can be accounted for, though in a non-naturalistic way. Suffice it that one appeals to the weak Meinongianism contained in the thesis that one can directly refer to possible entities by means of suitable fixing reference description.

Wallis, C., (1995), "Asymmetric Dependence, Representation, and Cognitive Science." The Southern Journal of Philosophy, 33, 373-401.

Wallis charges that any adequate semantic theory must fit what an accurate cognitive science discovers about the human brain. Wallis charges that Fodor's theory does not fit with what we know from cognitive science about signal detection or feature detection (looking at color perception and Biederman's theory of "geons"). Wallis further claims that Fodor's theory cannot handle uninstantiated predicates (such as "unicorn" or "gape" …giant ape). Wallis also claims that Fodor's theory of meaning is inconsistent with his theory of mind (that minds must be able to detect non-nomic properties…properties that cannot enter into nomic relations qua those kinds of properties, such as "shirt"). Lastly, Wallis notes that Fodor is committed to the following interconnection of epistemics with semantics: the view if 'X' means X, then there are some circumstances in which one would be able to discriminate X-facsimilies from Xs (so that 'X' has univocal meaning). Wallis thinks his cited findings of cognitive science contradict this. However, if one's symbol 'Robin' means Robin (the bird), nothing Wallis cites shows that one's symbol can mean Robin without there being circumstances under which one could discriminate Robins from Robin look-a-likes.

Gates, Gary, (1996), "The Price of Information." Synthese, 107, 325-347.

This paper could be called "naturalized meaning theories encounter Quinean indeterminacy examples." Gates suggests that neither Fodor's nor Dretske's theories of meaning solve Quine's indeterminacy of meaning problems. Does "horse" mean horse or horse-stage or undetached horse parts? Gates runs through the various tools available to naturalizers of meaning and suggests that none of them eliminate Quinean indeterminacy.

Gibson, M. (1996) "Asymmetric Dependencies, Ideal Conditions, and Meaning." Philosophical Psychology, 9, 235-259.

Gibson gives a lengthy defense of Stampe and Dretske against Fodor's early attack on Wisconsin Semantics. She claims that the ceteris paribus conditions on the laws in Fodor's theory (' Xs cause "Xs"', etc.) exploit the same sorts of idealization conditions exploited in Wisconsin Semantics. These are the sorts of "normal conditions" that Fodor pinpoints as the problem with this approach to naturalization of meaning. She also (rightly) gives Fodor his usual C- in exegesis of the text of Wisconsin Semantics and straightens out quite a number of discrepancies. Gibson claims that there is no disjunction problem in Wisconsin Semantics. (This interpretation may require a somewhat implausibly restrictive reading of what the disjunction problem comes to.) Gibson also detects a sleight of hand on Fodor's part where meaning generates asymmetries, not the other way around (as Fodor's theory requires).

Adams, F. & Aizawa, K. (1997a), "Rock Beats Scissors: Historicalism Fights Back," Analysis, 57, 273-281.

This paper examines Fodor's rejection of his own earlier version of his theory that incorporated an historical condition of meaning (that some Xs actually cause "X"s). Now Fodor thinks it is "just the counterfactuals" that fix meaning. We argue that Fodor has not demonstrated that meaning is a-historical. Such a theory founders on problems of the meaning of names, typical Twin-Earth cases, and Swampman cases. Despite Fodor's claim that Swampan has meaningful thoughts, we provide reasons to deny this.

Adams, F. & Aizawa, K. (1997b), "Fodor's Asymmetrical Causal Dependency and Proximal Projections", Southern Journal of Philosophy, 35, 433-437.

Fodor requires that meaningful items be capable of being robustly caused. This becomes worrisome when one looks at proximal projections onto the retina or other sensory receptors. The worry is over what keeps one's representation "cow" from meaning retinal projection of a cow (or some such). Fodor argues that such meaning attributions are blocked because they don't satisfy his "robustness" condition. In this paper we show that there may well be an unlimited supply of robust causes of "cow" that satisfy all of Fodor's conditions and which, thereby, imply that on his theory "cow" means cow projection. We close by considering objections and replies to our examples.

Papineau, D., (1998), "Teleosemantics and Indeterminacy." Australasian Journal of Philosophy, 76, 1-14.

Papineau considers Fodor's rejection of teleological theories of content, in virtue of the frog snaps and dark moving spots. Fodor says teleological theories are indeterminate about whether the content is fly or dark moving spot. Fodor diagnosis the problem as resulting from natural selection's generating an extensional context. Papineau denies that natural selection is extensional, but says that the threat of indeterminacy in the teleological contexts does not arise out of mere extensionality. One problem is to choose among competing causally relevant descriptions. After considering and rejecting Millikan's and Neander's replies to Fodor, Papineau claims that content can only be ascribed to systems with a belief-desire psychology, and claims that frogs lack that psychology. So he thinks frogs lack a definite content and that to answer Fodor's objection to teleosemantics, one must not consider the content of frogs. Papineau thinks that one can only ascribe determinate belief content to things that have determinate desire content. But where does the determinate desire content come from? And how do we get out of the intentional circle? Papineau doesn't think desire content can be fixed independently of belief content. To determine contents of desires one must look for specific results that it is the function of those desires to produce. A desire for food has the effects of getting the organism to put food in the mouth, allow digestion, and foster health. It is when desire for food causes this whole sequence that selection preserved that type of desire. What makes the desire specific to food is that the desire could continue to perform its specific function even if other support systems ceased to foster health or allow digestion, and so on. Papineau embraces Neander's account of specific function.

Top of page

e-mail Adams' Homepage Adams' Homepage