FLEXIBLE CODING STRATEGIES IN PIGEONS: RETROSPECTIVE AND PROSPECTIVE CODING USING A RADIAL MAZE ANALOG TASK

Zentall, Steirn, and Jackson-Smith (1990) found evidence for dual coding in pigeons in a radial maze analog task. Specifically, they found that pigeons used retrospective coding in which previously chosen keys were remembered when a delay was interpolated early in a trial and prospective coding in which to-be-visited keys were remembered when a delay was interpolated late in a trial. An alternative explanation, the criterion shift hypothesis proposed by Brown, Wheeler, and Riley (1989), suggests that these data are consistent with dual coding because of an artifact of the correction procedures used by Zentall et al. The criterion hypothesis suggests that retrospective coding is used and that pigeons make choices more carefully after many choices have been made on delay trials as compared to control trials, which creates the appearance of prospective coding later in a trial. The present experiments tested this hypothesis using a new testing trial procedure and new, more conservative control trials. In experiment 1, the results of Zentall et al. were replicated using a fixed delay procedure instead of their original progressive delay procedure. Experiment 2 used a forced choice procedure after the delay to make the probability of making an error 50% on each trial type. Control trials also included a forced choice procedure to eliminate the assumptions required by the corrections procedure used by Zentall et al. The results were inconsistent with the retrospective coding account predicted by the criterion shift hypothesis and with the dual coding hypothesis. Instead, the results were consistent with a prospective coding account in which to-be-visited keys were remembered. These results were replicated in Experiment 3 using the pigeons from Experiment 1. The present findings have important implications for the field of comparative cognition