Effects of list length and first-trial reward upon concurrent discrimination performance

Rhesus monkeys were given 2304 two-trial object discrimination problems that were presented as members of lists consisting of two, four, or eight problems. The problems within each list were presented concurrently, Trial 1 for all problems in the list being presented before initiation of Trial 2 for any problem. Each problem involved exposure of a single object on Trial 1, which was either rewarded or not rewarded, followed by presentation of the Trial 1 object along with a second object on Trial 2. Performance varied inversely with list length, and this effect became more prominent as a function of training. Performance after Trial 1 nonreward was substantially better than that after Trial 1 reward, and this effect became enhanced both with continued training and with increased list length. The findings were discussed in terms of Bessemer's short-term memory analysis of learning set.