ABSTRACT
Even for the theories discussed above that implicate visual similarity as a key element in explaining visual recognition impairments specific to biological objects, some additional factor seems to be required for a complete account of this type of disorder. In the context of visual object recognition, the concept of similarity requires a reference group that specifies what a particular target object is similar to. The relevant reference group for all of the visual similarity theories of CSVA discussed above is always made of objects that belong to the same semantic category as the target. The reasons why this is so are not always transparent, however. One clear and explicit statement that has been made in this regard is by Riddoch, Humphreys, and their collaborators (Forde et al., 1997; Humphreys et al., 1988, 1995; Riddoch & Humphreys, 1987b). For these authors, the reason why the reference group against which visual similarity is assessed is made of objects of the same category is because the visual object recognition system operates in cascade. This implies the occurrence of interactive effects of factors that tap separate processing stages, notably those concerned with the visual shape of an object (i.e. perceptual encoding and structural descriptions) and those concerned with its meaning (i.e. semantic memory). Thus, what their theory predicts is that increased visual similarity between objects will cause some uncertainty on the representation of the target at the stages processing its shape. This uncertainty will then be transmitted to semantic memory, where it will be magnified if the items that are visually similar to the target are also semantically related to it. In other words, the prediction is for an overadditive interaction of visual similarity and semantic proximity, such that the effect of visual similarity will be exclusive to, or much greater for, semantically related objects than for items that belong to separate semantic categories. A complementary hypothesis suggested by Arguin et al. (1996b) that could contribute to the interaction of visual similarity with semantic proximity is that of feedback from semantic memory to structural descriptions, which could help separate the shape representations of visually similar items if they are semantically very distinct from one another.
