compound (column 2), performance is superior relative to when the cue comes after the compound (column 3). The comparable data in the Unified columns on the right do not contain significant differences. These data are consistent with the assumption that the birds are processing the two components of the Separated condition separately. That is, the effectiveness of cueing depends on the nature of the stimulus compound. In our displays, if the two relevant attributes are combined in a single "object," as in the Unified condition, selective processing of one attribute is difficult. If the two attributes are properties of different "objects," as in the Separated condition, then the cueing procedure is effective.
The relevance of Lamb's research to the issue of whether or not pigeons decompose compound stimuli into their components should be clear. The answer is: not necessarily, but if the stimulus arrangements are favorable then decomposition will occur. The cueing task allows these effects to be seen in the Separated arrangements. Under other circumstances, such as in the Unified condition, stimuli are treated more holistically. Finally, the difference between the pre-cued and post-cued results suggests that the decomposition occurs during information uptake, a conclusion inconsistent with the previously described failure to find element-compound convergence with long sample durations ( Cook, Riley, & Brown, 1982; Lamb & Riley, 1981). Obviously, our understanding of these effects is not perfect. (Since this material was written, the three other birds in the experiment have finished. Data from all five birds are consistent, with one interesting exception. One of the birds shows the selective attention effect in both the Separated and Unified conditions. Apparently, a pigeon can decompose the Unified stimulus, suggesting that decomposition need not depend on where the pigeon is looking.)
There appears to be clear evidence of compound stimulus decomposition in both Cook, Riley, and Brown's experiments, which varied spatial separation of elements, and in Lamb's work, which used a cueing procedure. There is also, in Lamb's work, now evidence supporting the limited capacity hypothesis, but the status of this hypothesis is less certain because of the apparent negative evidence. Also, on the issue of decomposition, it seems apparent that the question is not whether it occurs, but what are the conditions that facilitate or inhibit its occurrence? In the search for these factors, useful insights may be gained from the study of research on perceptual organization and memory in humans (cf. Riley, Cook, & Lamb, 1981). Such work suggests the possibility that decomposition may not be a high level cognitive achievement, as suggested by Cox and D'Amato, but a fundamental fact of perception. In that event, its manifestation in pigeons should not be surprising. But one should then also expect a similar outcome with monkeys. If Cox and D'Ameto's monkeys did not decompose the stimulus compounds in their displays, why not? One possible answer might be that viewing distance was much greater in their experiments and, because their stimuli were the same size as ours, visual angle would have been much smaller. Spatial proximity