Abstract
A recent report (Tarr et al., 1998) documented that a change in the direction of illumination interfered with performance on “same” trials when a sequentially-presented pair of objects were being matched on the basis of their shape. The implication of that study was that direction of illumination, which participants were instructed to ignore, was coded as part of the object representation. However, on identical trials (i.e., same direction, same shape), the absence of any change in the images of S1 and S2 could serve as a perfectly reliable cue that the correct response should be “same.” We replicated the Tarr et al. study (their Exp. 1) but included a condition in which S2 was translated with respect to S1. This position shift completely eliminated the cost of changes in illumination, consistent with recent evidence for invariance in IT cell tuning to illumination direction. The transient-like cue potentially plays a role in all same-different matching tasks in which S1 and S2 are presented sequentially, even with an intervening mask. Tarr et al.'s (Exp. 3) documents that the absence of illumination effects in our shifted condition should not be interpreted as indicating that surface information, in general, cannot be incorporated into longer-term object representations.