The research reviewed in this chapter provides, in my view, compelling evidence that pigeons typically engage in active processing (rehearsal) of sample-activated codes during the retention interval in the DMTS task. One caveat to this conclusion is that some types of samples (e.g., those differing in duration) may give rise to codes less conducive to rehearsal than other types of samples (e.g., colors, line orientations). The avian information processing system is apparently more flexible and sophisticated than we had conceded in the early 1970s. Not only are pigeons capable of learning to rehearse information, they are also capable of learning to do so selectively in situations in which such rehearsal is relevant to trial outcome.

The view that avian information maintenance processes are flexible and controlled is consistent with the view that information coding processes in avians are also flexible and controlled. Recent research (e.g., Grant, 1991; Grant & Spetch, 1991, 1993; Santi, Bridson, & Ducharme, 1993; Zentall, Sherburne, & Urcuioli, 1995) has revealed that pigeons are capable of coding the same nominal event in different ways depending upon task demands (for recent reviews see, for example, Grant, 1993, and Zentall, Urcuioli, Jackson-Smith, & Steirn, 1991). Thus, the cognitive capacity to actively rehearse and code information, and the ability to exert control over those processes, may be phylogenetically far more primitive (and hence more extensive) than we had once thought.