Unconscious decisional learning improves unconscious information processing

Abstract

The idea that unconscious input can result in long-term learning or task improvement has been debated for decades, yet there is still little evidence to suggest that learning outside of awareness can produce meaningful changes to decision-making. Here we trained participants using noisy motion stimuli, which require the gradual accumulation of information until a decision can be reached. These stimuli were suppressed from conscious awareness by simultaneously presenting a dynamic dichoptic mask. We show that a short period of training on either a partially or fully suppressed motion stimulus resulted in improved accuracy when tested on a partially suppressed motion stimulus traveling in the orthogonal direction. We found this improvement occurred even when performance on the training task was at chance. Performance gains generalized across motion directions, suggesting that the improvement was the result of changes to the decisional mechanisms rather than perceptual. Interestingly, unconscious learning had a stronger effect on unconscious, compared to conscious decisional accumulation. We further show that a conscious coherent percept is necessary to reap the benefits of unconscious learning. Together, these data suggest that unconscious decisional processing can be improved via training.