Abstract
Practice can induce considerable improvement in one's ability to perform a visual task. When the improvement is long lasting, it is usually referred to as perceptual learning and considered to reflect plasticity of the brain. A substantial part of this improvement occurs at the beginning of training (Ahissar and Hochstein, 1996). However such short-term improvement can also reflect other processes than brain plasticity such as adaptation and strategic shifts (Goldstone 1998). The objective of the present study was to dissociate the mechanisms that underlie long-term (LT) and short-term (ST) improvement in a pop-out detection task comprising of 8 sessions held 1 to 3 days apart from each other. Observers were required to detect a target defined as a uniquely oriented line among homogeneously oriented lines. Search displays were presented briefly and then masked. Improvement, measured as increased detection accuracy with practice, was observed both within the first session (ST improvement) and between each session and the following one (LT improvement). ST Improvement generalized to displays in which the target and distractors features on the task-relevant dimension (orientation) swapped, but not to displays in which a change occurred on an irrelevant dimension (color). By contrast, LT improvement generalized to displays in which a change occurred on color, but not to displays in which target and distractors orientations swapped. These results suggest that LT perceptual learning modulates early representations that are fine-tuned to specific features, whereas ST improvement modulates later representations in which the different stimulus features are bound.