"Thinking ahead" in rat discrimination learning

Abstract

If rats chose S+ in a brightness discrimination in a T-maze, they experienced, on that run and over four forced runs to S+ which followed, a pattern of reinforcement in which quantity of reward in the goalbox increased from 0 to 14 food pellets, decreased from 14 to 0 food pellets, or varied randomly. If the rats erred and chose S-, reinforcement was withheld, and they were forced a second time to 0 reward in the S- goalbox. The results indicate that rats readily learn the brightness discrimination under these conditions, the animals exposed to the sequentially increasing pattern learning somewhat slower than the others. This was true in spite of substantial delay of reward. Theoretical accounts based on perseverative inhibitory or facilitatory tendencies, or upon frustration, fail to describe the data accurately. Instead, analysis of the data shows that the animals were accurately anticipating the quantity of reward to be obtained on each run, running fast for large quantities and slowly for small. © 1975 Psychonomic Society, Inc.