ラットにおける継時的に提示された物体刺激の数的弁別学習

抄録

The present study examined numerical discrimination learning of sequentially presented objects in rats. Rats were trained to respond to a third object among six same ones arranged in a row. The objects were put on one of ten goal wells filled with sand in a rectangular open field. A food reward was given at the bottom of the correct well (under the third object). Since the assignment of objects to the wells was changed from trial to trial, rats could not identify the correct stimulus based on either location of specific goal box or distance from start point. Inaccessible food rewards put in wire mesh tea strainers were set in the incorrect wells to control possible odor cues. After rats learned the numerical discrimination task reliably, partitions with openings were inserted between the goal wells and thus rats could encounter the object stimuli only sequentially. Although rats’ performances deteriorated initially, they recovered rapidly and reached reliable levels. However, when novel test objects were introduced to the training, rats’ performance deteriorated to chance level. To eliminate possible trial-and-error learning to the test stimuli, food rewards were presented not only for the third but also second and fourth test stimuli. Because performance deteriorated on the training trials as well as the test trials, it was suggested that rats learned to respond to the second or fourth stimulus through the nondifferential reinforcement for the second, third, and fourth test stimuli. Although these results suggest that rats can discriminate numerical aspects of sequentially encountered objects, it also suggests that another method for testing transfer of the numerical discrimination to novel test stimuli is needed.