Smith's Argument Against Naturalistic Accounts of Intentionality

(Very rough draft)

So far, we've looked at Smith's arguments against naturalistic accounts of (i) perceptual knowledge, (ii) concept formation, and (ii) concept correction. In this post, I aim to finish (or at least nearly finish) discussing Smith's book by focusing on Smith's last main type of argument against naturalism: (iv) arguments against the compatibility of perceptual knowledge and naturalistic accounts of intentionality.

As before, Smith's focus is (broadly speaking) Dretske-style accounts of intentionality. According to such accounts, a concept or perceptual state is of or about its referent just in case the former reliably covaries with the latter when functioning properly. So, for example, when functioning properly under normal conditions, thermometer readings co-vary with temperature, and scale readings co-vary with the weights of objects on the scale. Because of this, thermometer readings reliably represent temperatures, and scales reliably represent the weights of objects. In a similar way, sensations reliably represent the external environment when functioning properly under normal conditions. 

The heart of Smith's criticism is that if naturalistic accounts of intentionality are correct, then we can't know if our perceptions are veridical. Smith uses two main arguments to support this conclusion. Smith's first main argument is that if naturalistic accounts of intentionality are correct (again, think Dretske et al.), then intentionality is essentially a matter of a law-like correlation between external object and internal brain state. On such accounts, the object of immediate awareness is the internal brain state, and there is no way to "see" beyond it to see if such states are caused by, and reliably represent, their external referents. And finally, if that's right, then there is know way to know if the former reliably represents the latter, since (argues Smith) knowledge of some x requires the potential knower to have immediate, direct acquaintance with x. Therefore, naturalistic accounts of intentionality entail that we can't have perceptual knowledge of the external world.

What to make of this argument? As we mentioned in the last post in this series, Smith is aware that Dretske and other naturalists can appeal to an externalist account of knowledge, according to which knowledge doesn't require "getting outside one's own skin", as it were, to check if one's beliefs and/or internal representations reliably represent their external referents. Rather, all that's required for knowledge, on such accounts, is that the connection between beliefs and/or referents are, in fact, reliable. However, Smith replies that this sort of response is inconclusive at best, since he thinks knowledge requires conceptualization of its object, and he thinks he has shown that the forming the requisite concepts is impossible given naturalism. Therefore (argues Smith), epistemic externalism is at best necessary, but not sufficient, for knowledge. Unfortunately, as we saw in the previous post in this series, it's not at all clear that Smith is right about that.

Smith second main argument is that if naturalistic accounts of intentionality are correct (again, think Dretske et al.), then we shouldn't think our concepts are veridical, on the grounds that intentionality is at root a causal process. But the problem is that causes always modify what they act upon. But if so, then intentionality must modify the relevant brain states involved in perception. And if that's right, then (claims Smith) perceptual experience must be (or at least probably is?) distorted. And if it's distorted, then it's not veridical. Smith then goes on to use the point as a reply to Dretske's epistemic externalist rejoinder to Smith's "causal chain" objection to naturalistic perceptual knowledge. For if intentionality distorts perception, then we have grounds for thinking that perception is distorted.

What to make of this argument? Perhaps the most obvious worry is the inference from "causes modify" to "causes distort". For while a cause can modify its object for the worse, it can also modify it for the better. And this is no less true when the object of modification is representational. To take an obvious example, consider perceptual registrations in the eye. In this sort of process, the pupil of the eye receives light from the external environment, which in turn registers an accurate two-dimensional image of its referent on the retina at the back of the eye. So here we have a case of a cause that modifies its object for the better from a representational point of view. And given the prima facie plausibility of the view that reliably tracking one's environment is conducive to survival and reproduction, there are strong reasons to think that evolution would select for reliable representational processes in organisms. (Smith claims that his argument doesn't depend on Plantinga's evolutionary argument against naturalism, so let's leave discussion of the latter for another day).  I therefore find this argument unpersuasive.