Rapid association learning in the primate prefrontal cortex in the absence of behavioral reversals

Abstract

The PFC plays a central role in our ability to learn arbitrary rules, such as "green means go." Previous experiments from our laboratory have used conditional association learning to show that slow, gradual changes in PFC neural activity mirror monkeys' slow acquisition of associations. These previous experiments required monkeys to repeatedly reverse the cue-saccade associations, an ability known to be PFC-dependent. We aimed to test whether the relationship between PFC neural activity and behavior was due to the reversal requirement, so monkeys were trained to learn several new conditional cue-saccade associations without reversing them. Learning-related changes in PFC activity now appeared earlier and more suddenly in correspondence with similar changes in behavioral improvement. This suggests that learning of conditional associations is linked to PFC activity regardless of whether reversals are required. However, when previous learning does not need to be suppressed, PFC acquires associations more rapidly.

Figure 1

(A) The conditional association learning paradigm. After a cue is briefly presented at the center of gaze, monkeys maintain fixation during a memory delay before making a saccade to one of two targets presented on the right and left of the extinguished fixation spot. Only a saccade to the target associated with the cue generates a reward. (B) The block structure of the conditional association learning task without reversals. On any given block of trials, each of two novel cues is associated with a single saccade direction. Once a block of trials is completed, the old images are discarded, and two novel cues are again presented and must be learned.

Figure 2

(A) Percentage of correct performance averaged across blocks and cues for all trials (correct plus error trials). At the start of a block (Trial 1), average performance is not above chance levels (50%) but quickly jumps above chance by the second presentation of each cue and continues to gradually improve throughout the block. (B) RTs averaged across blocks and cues for correct trials to match neural data from the two monkeys. RTs are slowest at the start of new blocks but rapidly improve over the first four trials and then continue to gradually improve over the remainder of the block. Error bars show SEM.

Figure 3

(A) Change in peri-cue direction selectivity during association learning without reversals. Population percentage of variance explained by direction (PEV_dir, color scale) shown as a function of correct trials and time from cue onset, averaged across blocks and cues. Black dots indicate the half-maximal PEV_dir or rise times. (B) Rise times replotted and fit with a sigmoid curve show bistable learning, with initial late activity (Trials 1–4) followed by an increase in early trial direction selectivity starting with Trial 5. (C) Mean PEV_dir from the second half of the cue period (250–500 msec after cue onset) also shows a jump by Trial 5 but continues to increase with learning. Error bars show standard deviation of the mean.