Encoding Temporal Order and Visual Statistical Learning

Abstract

The literature suggests that visual statistical learning occurs from a very early age, with evidence suggesting that newborns are able to discern between familiar and novel sequences at just 2 days old. However, based on recent findings on the role of the medial temporal lobe in visual statistical learning in combination with our current understanding of this region's developmental timeline, we believe children younger than 40-months are unable to discern between the temporal regularities found between shapes in a sequence. In this particular study, we piloted two learning paradigms on adult subjects expecting to see a clear ability for the adult subjects to discriminate between our three categories of temporal order. Performance for our first paradigm, Fade-to-Reveal, revealed a significant improvement in reaction times through training, indicative of learning. For our second learning task Search-and-Find, the results of training suggested initial improvement with a regression in performance due to fatigue. Interestingly, subjects for both paradigms showed no real ability to explicitly recall the different shape-pairs at test. We interpret these opposing results to indicate that learning in these paradigms is implicit and thus the explicit recall test is not an appropriate measure of knowledge on shape-pairs.