task with variable list lengths of one to six items.
1. Procedure. The number of list items in each trial could be 1, 2, 3, 4, or 6 items. The experiment consisted of 216 unique trials (108 Same trials and 108 Different trials) which were constructed from 864 unique items. Because the Kodak Carousel projectors hold only 140 items these trials were distributed into six different carousel projector trays. They were actually pairs of trays but are referred to here as tray 1, 2, 3, etc. The list and probe items were in separate trays and projected onto separate screens. List items were projected onto an upper screen and probe items onto a lower screen. Image size was 18.4 cm horizontal by 12.4 cm vertical. The screens were spaced 16 cm center to center and were at a distance of approximately 50 cm from the monkeys and 62 cm from the pigeons. The pigeons viewed the stimuli through a large (8.75 cm square) and thin Plexiglass center pecking key. Two choice keys were located on either side of the center pecking "window." A peck on the red right one was a Same response and a peck on the green left one was a Different response. Monkeys moved a lever to the right (Same) or left (Different) to indicate their choice responses. Pigeons initiated trials by a center key peck and monkeys by pressing down on a 3-position lever. Both were cued by a ready "clicker" signal. The six different trays of items were run in blocks with different orders within each block. The items were presented for 1-sec for the monkeys and 2-sec for the pigeons with a 1-sec delay between list items. Following the lost list item a probe item was presented following a delay of 1-sec for the monkeys and a 0.5-sec for the pigeons. A 3-sec intertrial interval separated successive trials. All correct responses were rewarded (grain for pigeons, orange juice or banana pellets for monkeys). All incorrect responses were punished with a time out (with house light turned on) of 10-sec for monkeys and 6-sec for pigeons. The entire set of 216 trials were usually conducted daily with each of the monkeys and pigeons. Occasionally, only a portion of the set was conducted, and training resumed the next day at the position in the set where it had been terminated the day before.
2. Results.Figure 20.3 shows the average performance for the three monkeys (Joe, Max, and Linus) and the three pigeons which were run in this experiment. The monkeys began at better than chance performance and showed a gradual and steady increase in performance eventually reaching better than 90% correct after 7,200 trials. Their initial performance of 63% correct is not terribly surprising, considering that just previous to this experiment they were performing very accurately in a Same/Different task and had revealed nearly perfect transfer performance in a test with novel items. The pigeons, on the other hand, showed no signs of acquiring 1-6 item variable list length serial probe recognition task even after more than 2,300 trials. Their performance stayed near chance performance (52% correct). The pigeons too had performed accurately in the Same/Different task and had been given training identical to the monkeys. The pigeons, however, did not transfer well to novel