Testing assumptions of statistical learning: is it long-term and implicit?

Statistical learning has been studied as a mechanism by which people automatically and implicitly learn patterns in the environment. Here, we sought to examine general assumptions about statistical learning, including whether the learning is long-term, and whether it can occur implicitly. We exposed participants to a stream of stimuli, then tested them immediately after, or 24h after, exposure, with separate tests meant to measure implicit and explicit knowledge. To measure implicit learning, we analyzed reaction times during a rapid serial visual presentation detection task; for explicit learning, we used a matching questionnaire. Subjects' reaction time performance indicated that they did implicitly learn the exposed sequences, and furthermore, this learning was unrelated to explicit learning. These learning effects were observed both immediately after exposure and after a 24-h delay. These experiments offer concrete evidence that statistical learning is long-term and that the learning involves implicit learning mechanisms.