Abstract
According to standard teleosemantics, intentional states are selectional states. This claim is put forward not as a conceptual analysis, but as a ‘theoretical reduction’—an a posteriori hypothesis analogous to ‘water = H2O’. Critics have tried to show that this meta-theoretical conception of teleosemantics leads to unacceptable consequences. In this paper, I argue that there is indeed a fundamental problem with the water/H2O analogy, as it is usually construed, and that teleosemanticists should therefore reject it. Fortunately, there exists a viable alternative to the water/H2O model which avoids the fundamental problem, while explaining the a posteriori character of teleosemantics equally well.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
I use the term ‘intentional states’ as an umbrella term covering all states that have intentional contents, including beliefs, desires, intentions and perceptual states.
By saying that it can be shown ‘directly’, I mean that we need not appeal to considerations of ‘what matters to us’—considerations which, as indicated above, play a central role in the argument of Braddon-Mitchell and Jackson (1997).
One obvious limitation of this definition is that it only covers direct biological functions, i.e. functions that are established directly by natural selection. But it is clear that teleosemanticists must also acknowledge the existence of functions that are established indirectly—e.g. functions that items possess in virtue of being produced (under certain circumstances) by mechanisms that have direct biological functions; cf. Millikan (1984, pp. 39–49) for details.
For example, teleosemantic accounts often appeal to the functions of mechanisms that produce or ‘consume’ (i.e. respond to) representational states in order to account for the content of these states. In the case of propositional attitudes like beliefs, the functions of concept-forming mechanisms will presumably be relevant, too.
Alternatively, (S) can be interpreted as a universally quantified conditional of the form ‘∀(Fx → Gx)’ (i.e. as an abbreviation of ‘for all states x, if x is an intentional state, then x is a selectional state’). Choosing this alternative interpretation would not, however, make much difference for our discussion, since the fact that (S) is treated as an identity statement is quite inessential to its status as a genuine reductive hypothesis (cf. Section 4).
This is not to say that teleosemanticists do not recognize differences between the two cases. It is often pointed out, for instance, that the thesis that water is H2O is supported by specific experimental results, while the empirical evidence for the teleosemantic hypothesis must be of a different kind (cf. Papineau 1993, pp. 94–99). Hence, even adherents of the ‘strict water/H2O analogy’ acknowledge that the analogy is not perfectly strict.
Some authors add an actuality operator to the description on the right hand side (‘the actual odorless, transparent, potable liquid in our surroundings’). For our purposes, however, the actuality operator is not necessary, since we want to derive ‘water = H2O’ from the premises, not ‘necessarily, water = H2O’. For the irrelevance of modal status to questions of scientific reduction, cf. also fn 12.
Some theorists hold that (1) is a conceptual claim (cf. Jackson 1998, p. 59), but whether or not this is so is irrelevant for my purposes here.
Both Papineau (2001) and Braddon-Mitchell and Jackson (1997) assume that the folk role of intentional terms is wholly constituted by causal-dispositional properties. Braddon-Mitchell and Jackson (1997, p. 480) make that assumption explicit by characterizing the folk roles associated with intentional terms as “folk functional roles”, which are given by “complex interaction patterns with [the] environment”. Papineau (2001) also accepts this assumption, which is clear (i) from the fact that he endorses the model of teleosemantic reductions proposed by Braddon-Mitchell and Jackson without qualifications (Papineau 2001, p. 279, quoted in fn 8 above) and follows them in characterizing the roles associated with intentional terms as ‘folk functional roles’, as well as (ii) from his treatment of the Swampman case (cf. Section 5).
At this point, it might be tempting to argue as follows: ‘The crucial difference is that “H2O” is a rigid designator, and “the substance that covers 71% of the Earth’s surface” is not. Therefore, “water = H2O” is necessary, while “water = the substance that covers 71% of the Earth’s surface” is merely contingent, and this is why only the first statement counts as a genuine reductive identity statement.’ This argument, however, is entirely unconvincing. First, for a statement to qualify as a genuine reductive identity statement, it is clearly not sufficient that it is a necessarily true statement with rigid designators on both sides of the identity sign. Many a posteriori identity statements which are clearly not reductive fulfill this condition, cf. ‘Hesperus = Phosphorus’ or ‘water = the actual substance that covers 71% of the Earth’s surface’. Hence, the difference mentioned in the above argument is surely not the crucial difference between the two cases. Secondly, I would argue that it is not even necessary for a reductive identity statement to contain only rigid designators or to be metaphysically necessary. Suppose it turned out that ‘water’ was not a rigid designator, but a flaccid one, i.e. a term that designates H2O in our world, but other substances (e.g. XYZ) in other worlds. Would that mean that chemistry has not, after all, succeeded in reducing water to H2O? No, it would not. The semantic question of whether ‘water’ is rigid or non-rigid is completely irrelevant from the point of view of chemistry, so the success of the reductive project cannot depend on this question. Cf. Papineau (2001, pp. 285–286, 2009) for a very convincing defense of this position.
This was discovered by Henry Cavendish in 1781. However, Cavendish, being still very much in the grip of phlogiston theory, did not yet draw the right conclusion from his experiments (cf. Partington 1962, pp. 325–338).
The first experiment of this kind was performed by William Nicholson and Anthony Carlisle in 1800.
The qualification ‘almost’ is necessary for two reasons. First because samples of water found in nature always contain impurities (trace elements like sodium, potassium and chloride), and secondly because some H2O molecules dissociate into H + and OH- ions (cf. Hendry 2006, pp. 869–873). I largely ignore these complications in the following.
For a compelling case that emergentism is best understood along these lines, cf. Beckermann (2009, pp. 156–157).
According to the standard definition in chemistry, “a liquid is matter in a fluid state [i.e. with no fixed shape; PS] that is relatively incompressible” (Considine 2005, p. 937).
A brief caveat: this description should not be taken to imply that the surface properties of water/H2O are identical to its microphysical properties. In fact, as will become clear in the following, I think that the reductive explanation of the macro-properties F1, …, Fn by the micro-properties P1, …, Pn only requires that F1, …, Fn are realized by P1, …, Pn. (Note, however, that ‘the stuff that has F1, …, Fn = the stuff that has P1, …, Pn’ can still be a perfectly true identity statement!) .
Walter (2010, pp. 211–213) helpfully distinguishes between two notions of realization, a functionalist and an explanatory notion (my terms). The notion of realization I am appealing to here is the explanatory one (see e.g. Lepore and Loewer 1989, pp. 179, Clapp 2001, pp. 112–113), which I take to be equivalent to the notion of grounding.
This interpretation of (so-called) theoretical identifications is favored e.g. by Soames (2002, pp. 255–257).
A partial reductive explanation would be enough since teleosemanticists only claim that selectional properties play a crucial role in a reductive account of intentional states. (This is also a difference between the case of teleosemantics and the water/H2O case, but not the decisive one).
For the sake of simplicity, I follow Braddon-Mitchell and Jackson and Papineau in ignoring the compositional structure of expressions like ‘the belief that it is raining’, and treat them as if they were unstructured natural kind terms analogous to ‘water’.
This claim is reminiscent of Dretske’s (1988) thesis that instantiations of selectional properties by neural states can act as the “structuring causes” of behavior.
I ignore at this point any complications arising from the conventional nature of sentence meanings. To avoid these complications altogether, one can simply consider other propensities, e.g. the propensity to use an umbrella (instead of the propensity to utter ‘it’s raining’) and the propensity to wear sunglasses (instead of the propensity to utter ‘the sun is shining’).
It is interesting to analyze Papineau’s (2001) argument from this perspective. As mentioned before, Papineau accepts that intentional terms are associated with merely causal folk roles. He then argues that these roles are realized in the actual world by selectional states, but he does not try to show that the roles are realized by selectional states qua selectional states, i.e. in virtue of their selectional properties. Indeed, his remarks on Swampman strongly suggest that selectional properties are not essential to the realization of causal folk roles (cf. Section 5). While the latter claim is plausible in itself (as I’ve argued above), it generates a fundamental problem for Papineau, for in combination with his assumption that intentional terms are associated with those very causal roles, it appears to lead to a theory that treats (S), in effect, as a merely accidental empirical truth about content. In my view, this is what enables Braddon-Mitchell and Jackson (2002, p. 376) to argue that Papineau’s teleosemantics collapses into “a species of analytical functionalism – the species that insists on the theoretical importance of selectional classifications”.
When confronted with this conception of teleosemantic reductions, some teleosemanticists might want to argue that these weakly normative elements are more central to the folk roles of intentional states than mere causal role properties, or maybe even that causal role properties are largely irrelevant. For the purposes of my argument, however, the weaker claim – the claim that folk roles are partly normative – suffices.
At one point, Papineau himself hints in this direction: “I think that everyday psychology is already very sensitive to issues of cognitive design. It types states by what they are designed to do, rather than in purely causal terms, and this typing plays a significant part in everyday psychological understanding” (Papineau 2001, p. 288). But, first, I do not think that this is quite the right way to supplement the causal folk role (moreover, it may be seen as question-begging by some opponents of teleosemantics, since ‘cognitive design’ is clearly a historical notion), and secondly, I would argue that Papineau fails to draw the right conclusions from the fact that the folk role associated with intentional terms is not purely causal.
Note that I am still using the standard model of reduction; the crucial difference lies in the role described in (1). Another option for teleosemanticists would be to attack the standard model, but one of the advantages of my answer is that it does not rely on such radical measures.
Cf. also Millikan (2004, p. 63), who insists that teleosemantics is “a theory only of how representations can be false or mistaken” (though I would argue that this is an overstatement).
The theses that Peters and Sebastián argue for differ in strength. Peters confidently claims that “there were (and will be) plenty of Swampman-like creatures in the actual world” (Peters 2014, p. 280), while Sebastián argues for the more cautious thesis that “Randoman”—a Swampman-like creature—is nomologically possible and not all too improbable, so that we cannot definitely rule out that such a being will exist in the future (Sebastián 2017, pp. 323–324, p. 332).
Both Peters (2014) and Sebastián (2017) build on their Swampman-like cases to offer further arguments against teleosemantics. Sebastián, for instance, argues that his Randoman case makes it vivid that teleosemanticists cannot account for the explanatory relevance of representational properties (Sebastián 2017, p. 328–330), an issue that has usually been discussed under the heading of ‘the problem of explanatory irrelevance’, ‘the epiphenomenalism problem’ or ‘the Soprano problem’ (after Dretske 1988, p. 79). While I think that there are promising answers to this type of worry (see e.g. Neander 2017, pp. 40–46, pp. 58–60, pp. 73–89), it would take us too far afield to examine them here. In any case, the metatheoretical conception of teleosemantic reductions defended in this paper is not supposed to provide an answer to this worry all by itself (see also fn 34 below).
The reason, briefly put, is this: if the belief-that-it-is-raining role is purely causal-dispositional, then one of Swampman's states surely plays that role, and that means that (S) must be false if Swampman is actual.
If this is true, how can teleosemanticists explain why many people have the intuition that Swampman does believe that it is raining? First, they can point out that Swampman’s state partly realizes the belief-that-it-is-raining-role, since the state has the relevant causal-dispositional properties, and this may account for our inclination to describe Swampman as having the belief that it is raining. Secondly, teleosemanticists can question the reliability of our intuitive judgments about such contrived and unrealistic (or, at the very least, extremely rare and unusual) cases.
Starting from the Swampman case, several critics have raised challenges that concern (a) the connection between intentional states and phenomenal consciousness, (b) the connection between intentional states and moral status, and (c) the question of whether teleosemantic theories can account for the explanatory relevance of intentional properties (see e.g. Antony 1996, Peters 2014, Sebastián 2017). All these worries deserve a full discussion, and some of them have already been discussed to a greater or lesser extent in the teleosemantics literature (see e.g. Dretske 1988; Neander 1996, 2017; Papineau 2001). They are not the topic of this paper, however, so I will not try to address them here.
To restate my previous point about Swampman in terms of realization, teleosemanticists are forced to deny that weakly normative properties can be realized by the intrinsic properties of Swampman’s brain states. This is due to fact that, according to teleosemantics, weakly normative properties are not realized by the intrinsic properties of Davidson’s brain states, and since Swampman’s states share all their intrinsic properties with those of Davidson, it follows that the intrinsic properties of Swampman’s states cannot realize weakly normative properties. This does not entail, however, that teleosemanticists have to deny the possibility of alternative realizers in general.
References
Antony, L. (1996). Equal rights for swamp-persons. Mind and Language,11, 70–75.
Beckermann, A. (2009). What is property physicalism? In B. McLaughlin, A. Beckermann, & S. Walter (Eds.), The Oxford Handbook of Philosophy of Mind. Oxford: Oxford University Press.
Braddon-Mitchell, D., & Jackson, F. (1997). The teleological theory of content. Australasian Journal of Philosophy,75, 474–489.
Braddon-Mitchell, D., & Jackson, F. (2002). A phyrric victory for teleonomy. Australasian Journal of Philosophy,80, 372–377.
Broad, C. D. (1925). The Mind and its Place in Nature. London: Kegan Paul.
Clapp, L. (2001). Disjunctive properties and multiple realizations. The Journal of Philosophy,98, 111–136.
Considine, G. D. (Ed.). (2005). Van Nostrand’s Encyclopedia of Chemistry. Hoboken, NJ: Wiley.
Davidson, D. (1987). Knowing One’s Own Mind. Proceedings and addresses of the American philosophical association,60, 441–458.
Dretske, F. (1988). Explaining behavior. Cambridge, MA: MIT Press.
Friedman, H. L. (1993). Water. In S. P. Parker (Ed.), McGraw-Hill encyclopedia of chemistry (2nd ed., pp. 1155–1157). New York: McGraw-Hill.
Hendry, R. F. (2006). Elements, compounds, and other chemical kinds. Philosophy of Science,73, 864–875.
Jackson, F. (1998). From metaphysics to ethics. Oxford: Oxford University Press.
Lepore, E., & Loewer, B. (1989). Making mind matter more. Philosophical Topics,18, 175–191.
Mill, J. S. (1843/1973). System of Logic, Ratiocinative and Inductive. In J. M. Robson (ed.): Collected works of John Stuart mill (Vols. 7–8), Toronto: University of Toronto Press.
Millikan, R. (1984). Language, thought, and other biological categories. Cambridge, MA: MIT Press.
Millikan, R. (1989). In defense of proper functions. Philosophy of Science,56, 288–302.
Millikan, R. (2000). On clear and confused ideas. Cambridge: Cambridge University Press.
Millikan, R. (2004). Varieties of meaning. Cambridge, MA: MIT Press.
Millikan, R. (2010). On knowing the meaning; with a coda on Swampman. Mind,119, 43–81.
Neander, K. (1991). The teleological notion of ‘function’. Australasian Journal of Philosophy,69, 454–468.
Neander, K. (1996). Swampman meets Swampcow. Mind and Language,11, 118–129.
Neander, K. (2012). Teleological Theories of Mental Content. In E. N. Zalta (Ed.), The stanford encyclopedia of philosophy. Berlin: Springer.
Neander, K. (2017). A mark of the mental. Cambridge, MA: MIT Press.
Papineau, D. (1993). Philosophical naturalism. Oxford: Blackwell.
Papineau, D. (1998). Teleosemantics and indeterminacy. Australasian Journal of Philosophy,76, 1–14.
Papineau, D. (2001). The status of teleosemantics, or how to stop worrying about Swampman. Australasian Journal of Philosophy,79, 279–289.
Papineau, D. (2009). The poverty of analysis. Proceedings of the Aristotelian Society Suppl.,83, 1–30.
Papineau, D. (2016). Teleosemantics. In D. L. Smith (Ed.), How biology shapes philosophy (pp. 95–120). Cambridge: Cambridge University Press.
Partington, J. R. (1962). A history of chemistry (Vol. III). London: Macmillan.
Peters, Uwe. (2014). Teleosemantics, Swampman, and strong representationalism. Grazer Philosophische Studien,90, 273–288.
Sebastián, M. Á. (2017). Functions and mental representation: The theoretical role of representations and its real nature. Phenomenology and the Cognitive Sciences,16, 317–336.
Soames, S. (2002). Beyond rigidity. Oxford: Oxford University Press.
Walter, S. (2010). Taking realization seriously: no cure for epiphobia. Philosophical Studies,151, 207–226.
Acknowledgements
This paper has greatly benefited from the constructive criticisms of five anonymous referees for this journal. For helpful feedback on (much) earlier versions of this paper, I would also like to thank Ansgar Beckermann, Frank Hofmann, Joachim Horvath, Fabian Hundertmark, Brian Leahy, Manolo Martínez, Bence Nanay, Christian Nimtz, David Papineau, Eva Schmidt and other participants of the ‘Minds without Magic’ workshop 2013 and the DGPhil conference 2014. This research was supported by the project ‘Advancing Teleosemantics’ (SCHU 2860/2-1, NI 1320/2-1), funded by the German Research Foundation (DFG).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Schulte, P. Why mental content is not like water: reconsidering the reductive claims of teleosemantics. Synthese 197, 2271–2290 (2020). https://doi.org/10.1007/s11229-018-1808-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11229-018-1808-6