Attention in humans and animals: is there a capacity limitation at the time of encoding?

Investigators have suggested that pigeons have a limited capacity input channel and that as a result performance suffers when information load is great. This information overload hypothesis is based on converging findings from experiments using the matching-to-sample paradigm. Fundamental among these findings are that (a) performance improves as sample duration increases, (b) performance is better when the sample contains one relevant feature (elements) than when it contains two (compounds), and (c) this element superiority effect can be reduced by making the relevant feature of compound samples predictable. Experiments 1 and 2 show that these effects occur for humans as well. However, Experiments 3, 4, and 5 show that in humans at least some of these effects do not result from information overload at the time of encoding. Thus, the assertion that these effects do reflect such a limitation in pigeons must be reevaluated.