1990 年 8 巻 2 号 p. 61-68
This experiment using a straight alley with rats tested a prediction from Hulse's rule-encoding hypothesis concerning serial pattern learning when reward of elements was changed from food pellets (original training) to saccharin solution (transfer training). Initially, three basic groups learned the monotonic (M) pattern 14-7-3-1-0, the nonmonotonic (NM) pattern 14-1-3-7-0, or the random (R) pattern under food pellets. These patterns differ as to whether there is any single rule or not. Then half of each basic group learned either the three-element (10-1-0) or the six-element (15-10-6-3-1-0) monotonic patterns using saccharin solution. During transfer training, pattern tracking-anticipated running corresponded to differential amounts or reward-was better in the rats originally trained under the M pattern than those under NM or R patterns. However, positive transfer across training phases was weaker than that predicted from the hypothesis. The results indicate that the rule learning established under extended training could be biased by peripheral hedonic stimulation, such as the quality of reward.