The lifetime of sequential memory traces

Implicit statistical learning of a sequence of information is determined by several factors, including the number of repetitions of the sequence and the distance between each repetition. In the present study, we used a Hebbian version of the lexical decision task in which participants must decide whether the items presented in the center of a computer screen are words or non-words. Unbeknownst to the participants, a sequence of three words is presented and repeated in the same order several times (30 times). For each group of participants, we manipulated the distance between each repetition of the triplet of words by inserting the same number of filler items (i.e., on average, 4, 7, 10, 20, 30 or 60). Our results show that learning of the sequence is still possible at large distances (60 filler items apart) and that the learning rate best fit a power law. Furthermore, the processing speed of the words in the repeated triplet is differentially affected by the distance between repetitions. These data thus provide novel constraints for current models of implicit statistical learning.