Minds: contents without vehicles

Abstract

This paper explores a new understanding of mind or mental representation by arguing that contents at the personal level are not carried by vehicles. Contentful mental states at the personal level are distinctive by virtue of their vehicle‐less nature: the subpersonal physiological or functional states that are associated with and enable personal level contents cannot be understood as their vehicles, neither can the sensations or the sensory conditions associated with perceptual contents. This result is obtained by first extending the interpretationist ideas of Davidson and Dennett to show that subpersonal physiological or functional states cannot be construed as the vehicles of personal level contents. Then the anti‐foundationalist arguments of Sellars are extended to show that sensory states cannot stand as vehicles to perceptual contents. The line of argumentation extended from Sellars also provides a critique of the current trend to posit non‐conceptual contents.

Notes

Sonia Sedivy, Department of Philosophy, 215 Huron St., University of Toronto, Toronto ON M5S 1A1 CANADA, email: [email protected]

It might be said that the vehicle of personal contents is the whole person, or the whole brain or large chunks of it functioning as a connectionist system on which multiple representations are superposed, or some wide combination of person or connectionist brain embedded in the world. However, in such claims, the notion of a vehicle is not doing any informative work since contents are no longer associated with specifiable, discrete carriers. Rather, the more specific notion of a vehicle is being used to refer to all of the conditions of whatever kind that stand in an explanatory relation to personal contents without there being anything like one vehicle per content. It seems to me that usage of this kind has the potential, at least, for being misleading.

For an extended case for the importance of retaining the notion of persons in our theoretical sights, see Hornsby (1997).

Using these notions interchangeably does not beg the question against proponents of non‐conceptual contents who hold that experiential contents are not all conceptual. The issue is not whether the personal level is the level of experience but whether that level can be partitioned into contents that have and that lack conceptual structure.

To be more precise, it is stipulated that certain differences among physical properties are not relevant so that items, states or events that differ in those respects count as tokens of one physical type and it is stipulated that all tokens of that type are associated with contents of one kind.

Though “realistic” pictures are the obvious example of representations where particular vehicles carry particular contents, I do not intend to suggest that the notion of a particular vehicle is to be understood narrowly in terms of the sort of resemblance model popular for pictorial vehicles and their subjects. I do not believe that such approaches to pictorial representations are correct (Sedivy, 1996a). More importantly, a resemblance approach is too restrictive to capture the nature of the particular vehicles on which personal contents supervene according to non‐reductive forms of physicalism—see the discussion of the physicalist model above.

I put the well‐worn expression of a “level” in scare quotes here at its first occurrence to signal that this is a metaphor that imports dangerous commitments: levels are found in vertical arrangements where some are necessarily lower than others and where there must be a lowest level that is “basic” in some sense. As Wittgenstein would say, the “conjuring trick” has already been done: what seems to be our explanatory task is set for us by the picture this bit of language creates.

For the purposes of this article, I am restricting my attention to the most recent form of neo‐empiricism that takes the form of positing non‐conceptual contents at the personal level along with conceptual contents. However, quite recent theories such as functionalism and the central state identity theory that divide perceptual states into qualia or qualitative contents on one hand, and intentional or propositional contents on the other are also neo‐empiricist in holding onto the classical dual‐capacity model of perception as an amalgam of sensory and conceptual capacities. More detailed exposition will be provided in the section dealing with empiricist and neo‐empiricist models.

See especially Davidson's (1980) “Mental Events.” I can reconstruct Davidson's work in my terms as concerning the attribution and hence individuation of experiential or personal contents because propositional attitudes, such as wishing that it were spring, are experiential or personal level attitudes to propositional contents. By mental events, Davidson means events described using mental or psychological predicates, such as the predicates used to attribute propositional attitudes. Hence, where Davidson speaks of mental events, we can understand that he means contentful, experiential events at the personal level. See Evnine (1991) for a summary discussion of the way in which Davidson's conception of the nature of events in general, and of mental events in particular, seems to have changed since the writing of his seminal “Mental Events” (Davidson, 1980). Those changes make a difference to Davidson's arguments for monism but not for his thesis that the mental is anomalous and hence they do not affect my argument.

I do not use the notion of rationalizing explanation in contrast to causal explanation. Hence, my use of the notion of rationalizing explanation does not imply that such explanation is not causal or that the contentful episodes identified thereby are not causally efficacious.

Davidson's point also applies to functional explanation (of the causal variety), which identifies phenomena that are determined by constitutive principles specifying general tendencies rather than constitutive ideals or standards.

Laws in general—and psychophysical laws in particular—would “bring together predicates that we know a priori are made for each other” (Davidson, 1980, p. 218). Hence, if we understand the nature of the regulative principles that govern mental and physical predicates, we can know whether law‐like statements that bring the two predicates together are possible. I will shortly address the concerns of those who counter that Davidson's argument turns on the impossibility of strict psychophysical laws while only non‐strict generalizations between the mental and physical descriptions are sufficient. I believe that considering the latter in terms of the associations required between contents and vehicles will illuminate the strength and import of Davidson's position.

This point also applies to functionalist cognitivism, which attributes non‐experiential contents to a person's functional parts and only thereby in turn, if at all, to a person's physical parts. The point applies because contents attributed to functional states are attributable given the way things generally tend to happen as determined by causal principles and initial conditions.

My choice of the word “text” here echoes Hilary Putnam (1983, p. 150): “A translation scheme, however well it works on a finite amount of corpus, may always have to be modified on the basis of additional text.”

Though the present paper is not the place in which to press the point, I suggest that tracing the implications of Davidson's stress on the distinctiveness of the regulative principle of rationality with respect to the relation between contents and vehicles shows that his approach needs to relinquish the claim that background practices or conventions specifying linguistic vehicle types are inessential rather than defeasible.

I am grateful to Christopher Peacocke and Ronald de Sousa for making this objection and to William Seager for helping me see that I may maintain my position.

Let me mention a variant on Davidson's position. Davidson's arguments for monism have been criticized by philosophers such as Jennifer Hornsby (1997) and John McDowell (1985). These challenges would leave us with a position according to which the mental is anomalous but not token‐identical with the physical. Clearly, this variant position also holds that mental contents are not carried by vehicles.

Moreover, in explaining the behavior of a system in terms of the behavior of its functional parts or sub‐systems, we can, as Dennett (1978, p. 4) says, “make predictions solely from knowledge or assumptions about the system's functional design, irrespective of the physical constitution or condition of the innards of the particular object.”

Perhaps it needs to be stated explicitly that the reconstruction I am offering captures one strand in Dennett's thought but does not purport to define Dennett's position as such since I believe that Dennett's views have been changing across time.

To take the contrast one step further, Dennett holds that the distinctive identity conditions of intentional patterns preclude identification with physical pattern elements whereas Davidson argues that they do not: intentional and physical pattern elements have distinct identity conditions but are one and the same events nonetheless.

I want to urge that this … dualism of scheme and content, of organizing system and something waiting to be organized, cannot be made intelligible and defensible. It is itself a dogma of empiricism, the third dogma. The third, and perhaps the last, for if we give it up it is not clear that there is anything distinctive left to call empiricism, (Davidson, 1984, p. 189).

Sellars focuses on the phenomenalist form of empiricist foundationalism prevalent during the first half of the twentieth century, briefly considering the classical empiricism of Locke, Berkeley and Hume as well.

It might be argued that the current focus on the determinate wealth of information presented by perception is a way of recasting the problem that used to be discussed as the qualitative nature of perception.

For the most part, definitions of non‐conceptual contents and arguments for such contents neglect the distinction between descriptive and demonstrative contents. Put more precisely, the idea that there are non‐conceptual contents is the idea that there are contents that outrun our descriptive concepts such as the concepts cat, red or crimson. This level of precision is important since one argument against the proposal that there are non‐conceptual contents is that the content of complex demonstrative concepts, such as the concepts this cat, that red, or this crimson are as fully determinate as that which they represent so that such concepts can inform and structure the most determinate contents of our perceptions. Clearly, this issue lies beyond the scope of this paper. However, in the interest of accuracy I characterize the non‐conceptualist's proposal as suggesting that there are contents that outrun our descriptive concepts even though the proposal is not made with this precision.

For a detailed discussion of this proposal, see Crane (1992). I am putting forward the view that non‐conceptual contents are evidence as paradigmatic of non‐conceptual content theories even though some of the theories vary in their proposals on this point. In particular, Christopher Peacocke has advanced a different proposal (1992, 2001). However, I do not consider his proposal because its complexity lies beyond the scope of this paper and I argue against it in Sedivy (1996b). However, let me mention as what might seem to be a significant variant, Tye's (1995) approach. Tye's proposal is couched in a thoroughly non‐epistemic cognitive science vocabulary of inputs, outputs, causal relations and boundaries. Despite this terminology, it can be shown that Tye's proposal is also concerned to have nonconceptual contents play an epistemic or evidentiary function (i) in its requirement that nonconceptual contents supply “the inputs for certain cognitive processes whose job it is to produce beliefs (or desires) from the appropriate nonconceptual representations” (1995, pp. 138, 143–44) together with (ii) its concession that the content of the resultant beliefs cannot be explained in terms of causal covariance (1995, pp. 101–2).

Please recall that Crane's proposal (1992) is being taken as paradigmatic.

Consider the crucial exhibit or “piece” of evidence in a murder trial, let us say a particular gun. The only way the “piece” can support the claim that its owner is the murderer in question is if it is accompanied by a commentary establishing that and how the piece—namely, the gun—was used by its owner. But what supplies the mediating commentary in our mental lives? All there is according to non‐conceptual content theories are non‐conceptual and conceptual contents. There is no middleman in our mental lives brokering the evidential relation between perception and belief. To posit yet another mediating type of state or content would only make matters worse. Hence, perceptual content must be conceptual as well as determinate. But this is the starting point of a new substantive account rather than another maneuver in the factorizing empiricist framework we are considering.

I am not trying to suggest that the way in which Davidson's and Dennett's interpretationist theories go on to explain perception is correct. It is only their rejection of the empiricist model that I am mentioning and endorsing. To avert misunderstanding, it might be useful to state that I am also not suggesting that Sellars' substantive account of perception is correct—again, it is only his diagnosis of the empiricist model that I am endorsing (see Sedivy, 2004).

Additional information

Notes on contributors

Sonia Sedivy

Sonia Sedivy, Department of Philosophy, 215 Huron St., University of Toronto, Toronto ON M5S 1A1 CANADA, email: [email protected]