On Ch. 1 of Smith's Naturalism and Our Knowledge of Reality

In chapter 1, Smith begins his case against naturalism's ability to account for our capacity for knowledge of the external world. Toward that end, he offers a brief explication and critique of D.M. Armstrong's naturalistic account of direct realism with respect to perception. Below is a quick sketch of some particulars of Armstrong's account, and some of his motivations for it, that are relevant to Smith's evaluation.

Armstrong is an empiricist. However, unlike traditional empiricists (e.g., Hume, Locke, et al.), Armstrong rejects indirect realist accounts of perception, according to which perceptual awareness of external objects is mediated through sense-data. Armstrong worries that sense-data theories fall prey to radical skepticism with respect to knowledge of the external world. For if sense-data stand as intermediaries between ourselves and the outside world, then there is no way to know if they are caused by or accurately represent it. 

In the place of sense-data theories, Armstrong offers a direct realist account of perception, according to which the immediate and direct object of perception is the external world itself. In its barest essentials, Armstrong's account is that perceptual knowledge is just reliably caused belief about the world by means of the senses.[1] Armstrong is thus a reliabilist and externalist about perceptual knowledge. 

Smith raises three main criticisms against Armstrong's account. First, he argues that it suffers from the so-called "causal chain" problem. For if our beliefs are the result of a long causal chain, and we can only be aware of the last member of that chain, then we can never know if our belief is caused by, and accurately represents, an object in the external world.

Smith acknowledges that Armstrong has a reply to the above criticism, viz., that this objection rests on a mistake, conflating the causal conditions for perception (which involve intermediaries between perceiver and object) with the epistemic state of perception itself (which is immediate). However, Smith is not persuaded:

. . .the central issue posed by the argument . . .is not identifying perception with the causal conditions themselves. Rather, it is our inability to traverse, or transcend, the causal chain. . .and have epistemic access to the originating object itself. . .without that ability, Armstrong seems left with no way for us to know that we match up with the real, physical objects in the world.  (p. 17)

I'm not sure what to make of this criticism, however. For while Smith is aware that Armstrong's theory is a version of epistemic externalism, Smith seems not to appreciate that externalist theories of knowledge (including Armstrong's) lack such stringent epistemic requirements. For according to externalist theories, it's enough for the process of perception to be reliable in order for it to produce knowledge; one need not also be privy to internally accessible evidence that the process of perception is reliable in order to have knowledge via perception.[2] Perhaps Smith wants to argue that externalist theories of knowledge are inadequate, but no such argument is made here.

Smith's second criticism is that Armstrong's theory of perceptual knowledge can't account for the possibility of forming accurate concepts of objects in the external world. For (argues Smith) accurate concept formation of an object requires the ability to distinguish between accurate and inaccurate conceptions of thing, which in turn requires direct awareness of it.

I find the second criticism puzzling: why think such an ability requires direct awareness of the referent? Smith thinks it's because "otherwise, we will be forming concepts of something else [besides the external referent]". But why think that? It seems to me that even indirect realist/sense-data theories of perceptual knowledge seem capable of accounting for accurately forming concepts of objects, so long as the relevant sense-data are caused by, and accurately represent, their referents. Smith might reply that one could never "see beyond" the veil of sense-data to the external world to verify that they are veridical, but we've already seen that such a worry seems answered by an externalist account of perceptual knowledge. For as long as the sense data reliably represent external objects, the basic conditions seem to be in place for accurate concept formation.

Perhaps there is a further worry about distinguishing between veridical and illusory perceptual states. Smith seems to have such a further concern when he raises the familiar example of the straight stick that appears bent when submerged in water (p. 18). But surely even an epistemic internalist and indirect realist has the resources to handle cases of accurate concept formation in such cases. For they could give an explanatory justification (in terms of, say, an account of light refraction) for thinking that the straight-stick experiences are veridical, while bent-stick experiences are not. Such accounts have of course been around since at least Descartes.

Finally, Smith argues that Armstrong's theory of perceptual knowledge can't account for the intentionality or "aboutness" of perceptual beliefs. For beliefs are about things -- in the cases at issue here, they are about objects in the external, mind-independent world. But this "aboutness" is not one of the basic properties described in physics and chemistry textbooks, and it is hard to see how intentionality could be built up from such properties. There should thus be an initial presumption against the ability of Armstrong's theory of perception to account for intentionality.

One worry I have with Smith's third criticism is the one I mentioned in the inagural post in this series, viz., that Smith equates naturalism with physicalism.  But as I mentioned last time,  there are other versions of naturalism that accept non-physical entities into their ontology (e.g., abstract objects). But if so, then even if we assume arguendo that intentionality can't be reduced to the physical, there are other versions of naturalism that are perfectly compatible with that.

That's it for now. My next post will take a look at ch. 2.

-------

[1] Notice the implication that on Armstrong's account, perceptual experiences -- i.e., sensations -- that seem to be of the world, are not essential to perceptual knowledge (cf. blindsight cases, etc.).
[2] Alternatively, an externalist could appeal to Sosa's distinction between animal knowledge and reflective knowledge, and then say that basic externalist perceptual knowledge is of the former sort, but not the latter sort.