Abstract
Background: What is the relationship between perceptual learning (PL) -- improvements in information pickup as a result of experience -- and statistical learning (SL) -- representing co-occurrences among environmental features? We investigated this question using signal detection theory methods in a paradigm that could allow both. Design: Pretest and posttest phases consisted of a coherent motion detection task: Participants observed random dot kinematograms (RDK) showing fields of dots moving mostly in random directions, but with one subset of dots moving coherently in a single direction. Participants adjusted a dial to indicate the direction of coherent motion on each trial. The training phase, between pretest and posttest, consisted of a motion discrimination task: On each trial, participants observed two sequential RDK displays, each with motion in a single coherent direction, and indicated whether the two displays showed motion in identical or different directions. The directional difference between intervals was adjusted during training using a staircase procedure to maintain 75% accuracy, with adjustment occurring every ten trials. For each participant, all discrimination trials were clustered around a constant direction; this basic direction was chosen from among 8 possible directions, separated by 45 deg, and randomized across participants. Feedback was provided after each response. Results: Preliminary results showed reliable transfer of PL effects from the motion discrimination task to improved sensitivity in the posttest to coherent motion, with these effects mostly confined to each participant’s training direction. Consistent SL effects were shown by criterion changes toward responding to the general direction of motion used in training. We will discuss correlational methods for assessing whether these are linked or independent processes. Conclusions: PL and SL can occur in the same learning context, with PL effects shown as sensitivity changes and SL effects as criterion changes. These two learning processes may proceed concurrently but separately.