Abstract
In many societies, learning and experience are considered superior virtues. However, acquiring new knowledge may come with some disadvantages such as loss of open-mindedness and flexibility. Specifically, in the study of perception, it was suggested that perceptual performance is the result of an integration between prior knowledge and present sensory input. Under this Bayesian framework, prior knowledge may either hinder or enhance perception depending on its context and contents. In this pre-registered study, we examined how learning that the world is temporally systematic, affects learners’ ability to adjust to new temporal systems. Participants (N=40) performed a visual discrimination task, in which targets (slightly-oriented Gabor gratings) were preceded by visual-auditory cues. Temporal predictability at the first stage of the experiment (acquisition phase) was manipulated by setting the intervals between cue and target (foreperiods) to be either fixed (at 2700 ms) or random (sampled between 1700 and 3700 ms), between groups. Later, at the second stage (transfer phase), both groups were presented with a new fixed regularity (foreperiods= 700 ms). Findings showed that the group that was exposed to random timings of the visual targets adjusted better to the new system and its performance increased during the transfer phase. In contrast, the fixed group was less flexible to adjust to the new regularity, and its performance declined when faced with the new regularity. These behavioral effects were supported by eye-movements findings, showing deeper pre-target saccade inhibition in the transfer phase, for the random relative to the fixed group, suggesting that anticipation for the predictable target was enhanced in that group. These findings demonstrate that when learning temporal regularities sometimes more is less – learning one rule interferes with learning a new rule. These findings have implications for rehabilitation and education processes as well as for research of the mechanisms of perceptual learning.