Tyler Burge on disjunctivism (II)

Abstract

In McDowell (2010), I responded to Burge's attack (2005) on disjunctivism. In Burge (2011) Burge rejects my response. He stands by his main claim that disjunctivism is incompatible with the science of perception, and in a supplementary spirit he argues against the detail of my attempt to defend disjunctivism. Here I explain how disjunctivism is compatible with the science, and I respond to some of Burge's supplementary arguments.

Keywords:

• Burge
• disjunctivism
• perceptual psychology
• defeasibility
• fallibility
• intellectualism

Acknowledgements

This paper has benefited from James Conant's and Charles Travis's comments, and two anonymous referees for Philosophical Explorations.

Notes

I here consider only two cases, not Burge's four. See Section 5 below for a point that would justify this, even apart from my expository policy of postponing issues about singular reference.

He says that in McDowell (1986) I hold that ‘successful perceptual states always involve representational content that refers to the object being perceived’ (Burge 2005, 43). This may reflect his not recognizing that there are two disjunctivisms in that paper. I do not talk about reference when I am spelling out my disjunctivism about the epistemological significance of experience.

‘In front of her’ marks a singular element in the knowledge enabled by the experience in my example, and perhaps in the content of the experience. But this singular element does not purport to refer to something seen. Contents expressible in the way I have indicated are de se, not de re percepta.

One of Burge's grounds for dismissing this as a possible position for a disjunctivist is that it ‘would not plausibly express the denial of a common factor among the four cases’ (2011, 47). It is mysterious how he can think this is an appropriate consideration to invoke when he is undertaking to respond to my pointing out that I do not deny a ‘common factor’.

I will follow Burge in using the phrase ‘the science of perception’ in effect as shorthand for ‘the cognitive science of perception’. I will set aside questions about whether anything besides a cognitive-scientific approach, for instance the work of J. J. Gibson, might also count as science.

I introduced the phrase, in this context, in McDowell (1998a).

Mammals are a class, not a genus. But this does not affect the analogy.

‘It is the kind of mammal it is only because it is a wolf’ would express a much less interesting thought. The parallel in the case of knowledge and belief would be the similarly trivial thought that a knowledgeable belief is the kind of belief it is (namely a knowledgeable one) only because it is knowledgeable. (This is not the implication Burge extracts from what I said, with ‘kind of’ supplied, at 2011, 73, note 20.)

See Burge (2011, 64): ‘Like McDowell, I believe that reasons are propositional.’

See Burge (2003, 528), where he implicitly equates the idea that perceptual beliefs are rationally grounded on experiences with the idea that there are pieces of reasoning that constitute normative transitions from perceptions to beliefs.

Burge writes (2011, 64): ‘McDowell has now changed his view on the point that I criticized – the point that counted perceptions propositional, and reasons for perceptual beliefs.’ I have indeed changed my view on whether perceptions are propositional. The point of the note he is commenting on was that that leaves counting perceptions as reasons untouched. I did not say, or even imply, that Burge claimed that my counting perceptions as propositional was ‘essential to the fundamental point’. What I did imply was that he failed to understand how counting perceptions as propositional relates to counting them as reasons; counting them as reasons was and remains the fundamental point. He still fails to understand that; his accusation of misdirection is point-missing.

I wrote (McDowell 2010, 247) that ‘it does not require much sophistication’ to credit oneself with the capacity for knowledge grounded in seeings. Burge expostulates (2011, 65): ‘I did not write that it does require much sophistication.’ Taken in context, the point of my sentence, obviously, is that it requires less sophistication than Burge thinks. I should not have spelled out the excess sophistication I find in his picture in terms of the idea of ‘working with sophisticated concepts like that of defeating conditions’. But what he said some ordinary human adults might lack is ‘a conceptualized “know-how” mastery’ of the territory occupied by such concepts (2003, 529; my emphasis). My point was that the idea that such a lack is relevant depends on a misconception of the conceptual surroundings that are needed for factive concepts of perceptual states. The point is not affected by Burge's insistence that he did not mean a capacity for explicit deployment of such concepts.

In the actual world, though there is no need to make this explicit. Burge's formulation – ‘in a given possible world … in that world’ – leaves no work for cross-world identification to do; the talk of possible worlds is mere decoration. (It matters in some of Burge's other candidates for things to mean by ‘indefeasible’, but they are irrelevant to the explanation I gave of what I meant by it.)

At (2011, 76, note 42), Burge uses this claim as a ground for refusing to recognize my gloss as an explanation of ‘indefeasible warrant’.

Burge formulates the later items in his catalogue in terms of properties of beliefs also. He does not seem to think it follows that none of them is a type of indefeasibility of warrant.

As in Burge (2005), he talks mostly about ‘referential infallibility’, which implicates the issues about singular reference that I am keeping in abeyance for now. But he does say something that is closer to what matters for present purposes; he says I hold ‘that successful perceptual experiences … are infallible as regards veridicality’ (2011, 55). Veridicality is not quite what is needed, as I explain at McDowell (2010, 245). But it is as close as Burge gets to the idea that is needed for understanding my epistemological disjunctivism.

See, for instance, 2011, 54: ‘I think that intuitive considerations suggest that a given perceptual event that is referentially successful could have failed to be. Successful perceptual events are not referentially infallible.’ The restriction to referential success or failure is inessential; see note 16 above.

For instance (2003, 536): ‘No perceptual state type can be reliably veridical under all conditions. Perceptual states are components of perceptual competencies. Our perceptual competencies are fallible. All perceptual states (type or token) and all perceptual competencies are subject to possible error.’ I am not sure what to make of Burge's denial. He thinks I accused him of ‘a simple fallacy about entailment’ (2011, 74, note 26). But the target of my analogy is a mistake not about entailment, but about what it is for a capacity to be fallible.

Compare Burge (2011, 56), where he says I intend my uses of expressions like ‘there for the subject, perceptually present to her’ in a ‘philosophically loaded sense’. I think my use of these expressions would strike anyone not corrupted by philosophy as obviously right; the loading with philosophy is all on the side of resistance to the ideas they express.

In the arguments from McDowell (1998b) and McDowell (2009) that Burge dismisses at (2011, 61–3), I was trying to bring out how a question to this effect ought to embarrass people who have persuaded themselves that non-conclusive warrants can suffice for knowledge. Burge's responses do not address the question. The point does not turn on the fact that in those papers I assume a Sellarsian conception of the knowledge of rational subjects, in terms of standings in the space of reasons. (See Burge's dismissive remarks about this at 2011, 77, note 47.) The question arises in connection with knowledge possessed by subjects whose warrants are conceptually accessible to them, whether or not we take a Sellarsian view of such knowledge.

For an illuminating discussion of the distinction between kinds of scepticism that I am exploiting here, see Conant (2012).

Burge gives some fine examples in Burge (2005).

That was Daniel Dennett's thesis in the paper that was my target, Dennett (1978). In Burge (2005, 74, note 50), Burge says: ‘McDowell inveighs against “equating” the animal's perception with “computationally described goings on in their interiors”.’ And he says that is a straw man. But my target was something Dennett certainly did: equating the presence in consciousness of the content of perceptual states of individuals with access by the individuals to the content of states described (explained) as upshots of computational processes in their interiors. (Dennett took it, and I followed suit, that states so described would be sub-individual. This will matter as I go along.)

Many, not all. Rational individuals are computationally describable in so far as they are capable of computation, and they are not merely syntactic engines.

In Burge (2005, 50), Burge says: ‘I see no sense in which the perceptual system is merely a “syntactic engine”.’ That reflects the fact that what he means by ‘the perceptual system’ is something that embraces individual as well as sub-individual states. That is his right; he can use the idea of ‘the perceptual system’ however he chooses. But as I have noted, he does not engage with the idea that within what he calls ‘the perceptual system’, we can isolate a sub-system within which the generation of states is computationally explained; and that the isolated sub-system, a perceptual system in the sense of my dispute with Dennett, is sub-individual, and aptly described by Dennett's image.

In Burge (2005, 14), Burge speaks of ‘the visual system’ as computing ratios between light intensities along edges. (He does not mean that that is something perceiving individuals do.) That is a perfectly proper use of the notion of computing. But in the primary sense computing is something human beings learn to do, something that can be done only comprehendingly. The sense Burge employs (which is the sense in which the machines called ‘computers’ compute) is an extended sense, to be understood in terms of how illuminating it is to model the activities in question on computing in the primary sense. When a use of a concept is modelled on another use, some of the content of the concept is carried over but some is left behind. See Sellars on models and commentary in scientific concept-formation: Sellars (1997, Section 51).

Where I have spoken here of extended senses, I spoke there of metaphorical uses. That was perhaps unduly provocative. Metaphor is just one way of generating uses of terms that are extended from the primary uses.