As a model for studying motivation using monkeys, we have developed a behavioral task in which the monkey's level of performance can be con- trolled. The monkey is required to complete a series of one, two, or three successful trials to obtain a reward. In each trial the monkey must detect a color change in a small square on a TV monitor. A cue stimulus located above or below the small square changes brightness so that it is brightest when the next successfully completed trial will lead to a reward. In blocks of trials where the monkeys were rewarded with juice they performed with the highest success rate when the trial being performed would be rewarded and with the lowest success rate when the animal would have to perform other trials correctly before receiving a reward. Their reaction times for release of the manipulandum followed a similar trend, with the fastest reaction time occurring on trials that would lead directly to a reward. The monkeys thus act as if they are trying harder when the reward will be delivered after the successful completion of the next trial. We model this result as an energy surface on which the greatest energy for task performance is available when the stimulus conditions match those that will occur when a reward is deliv- ered. When intravenous cocaine was used as the reward in a block of trials, the performance level dropped and there was no difference between trials where a reward would be delivered and ones where the monkey would have to do more work to obtain a reward. However, the monkeys' reaction times were still shorter in the rewarded trials. Single neurons in the ventral striatum showed responses related to the delivery of juice rewards, to the amount of work that had been completed when the series of trials needed to obtain a reward was more than one. We suggest that these neurons would provide a good neural signal for the reinforcement of complex behavior that was rewarded. The neurons that responded when cocaine was the reward were different than those that responded to juice, and they responded equally strongly whether the trial was rewarded or not.
Showing the most recent 10 out of 25 publications
