Robert W. Doty

Errors by macaques performing trial-unique delayed matching-to-sample: A function of memory per se or of distinguishing between the subsequent choices?

In trial-unique delayed matching-to-sample the animal must (1) remember the item given as the sample, and (2) subsequently distinguish it from a second item to make a match. With rather short delays, even very well trained monkeys continue to make errors on this task. The question is whether these errors arise as a consequence of poor memory per se, or whether inadequate memory makes the sample difficult to distinguish from the alternative. This question was examined using pairs of items presented on multiple occasions. The role of sample was systematically interchanged between the items forming a pair. Errors were found to be well correlated between trials in which the same item served as sample, but essentially uncorrelated when those trials were compared with trials having the other item as the sample. For example, if A and B were paired items, results of trials in which A was selected as the sample were well correlated with other trials using A as the sample (run on other days); however, trials using A as the sample were uncorrelated with trials using B as the sample, even though the comparison pair (A and B presented together) was identical. These results suggest that the monkeys’ errors are not dependent on the distinguishability of the comparison pair, despite a “faded” memory on which to base the distinction.