Abstract

The increasing effect of reinforcement in task learning is variable across subjects. We studied the role of 2 types of reinforcement on conceptual learning. Sixty (29 female) medical students (18-20 years), participated on a feedback-based learning task (eight blocks, 40 trials each). The subjects had to predict a binary outcome using 8 different patterns, deterministically associated with the outcome, and an immediate correct/incorrect feedback was presented (Kumaran et al. 2007). The expected performance was ?70% in the first 3 blocks, ?80% in blocks 4-6 and ?90% in blocks 7-8. Subjects were randomly assigned to two types of monetary reward protocols: (1) a progressive increase (R+) if the expected block performance was obtained and (2) a progressive decrease(R-) if the expected block performance was not obtained. Average performance significantly increased with training, from 52.3±9.0 in the first block to 87.5±15.0 at the 8th block (F=101, p< 0.001). Performance was normally distributed up to the 4th block (mean ±SD=68.5± 15.5) and exhibited a progressive ceiling effect in the subsequent blocks, thus we did the analysis for the first 4 blocks. Overall, men (70.5± 15.5) performed better than women (66.0± 15.0) in the test (f=4.5, p=0.04) without sex/ block interaction. In addition, performance was increased with R- (73.0± 16.0) compared to the R+ (64.0± 13.0), with significant interaction between reinforcement type and blocks (F=6.7, p=0.012). A separate analysis for sex and type of reinforcement indicated that women and men with R- had similar performances, while women trained with R+ performed poorer than men and than women with R-. We conclude that in this conceptual task women are more sensitive to the reinforcement type than men, exhibiting an increase in their performance and learning speed with decreasing reinforcement.