Does Visual Reference Depend on Sortal Classification? Reply to Clark

Abstract

Clark's main thesis is, roughly, that the visual system proper is not smart enough to sustain demonstrative reference in the way I describe. Rather, visual information processing has to be helped out by the subject's intelligent use of sortal concepts. I think there are two ways in which his line of thought underestimates the resources available to vision without appeal to the intelligent use of sortal concepts. First, the internal management of information in vision is significantly more sophisticated than Clark allows: in effect he maintains, to make his point, that amodal completion, which does seem demanded by our ordinary use of visual demonstratives, requires the intelligent use of sortal concepts, and we will see that it does not. Secondly, some of the appearance of smartness on the part of the visual system may be owed, not to its internal organization, nor to the subject's intelligent use of sortal concepts, but to the way in which the visual system is embedded in the world of concrete objects. Suppose we think of visual attention to an object as a relation in which the subject stands to a particular concrete thing. The organizing work of the visual systems in the brain may be regulated over time by the particular token object seen, so that visual attention can keep locked on to that particular thing. What does the work is the