Abstract
The effect of practice on visual short-term memory (VSTM) is unclear. While some studies found no improvement in VSTM capacity through training, it is not clear what effect practice has on the fidelity of information in VSTM. The observers in the present study trained on a continuous orientation recall task under strong or minimal memory masking interference. Method: Seventeen observers completed an orientation recall task across six one-hour sessions. After a 4.25 second retention interval, the observers entered a continuous response by directly manipulating a match Gabor to the remembered orientation of a sample. During the retention interval, the observers completed a concurrent orientation discrimination-task on a target Gabor. The discrimination boundary was identical to the mean sample axis across trials in the "congruent" group, and was orthogonal to it in the "incongruent" group. The discrimination Gabor served as a memory mask, but was otherwise unrelated to the recall task. The sample orientation, and therefore congruency, was switched during pre- and post-tests. Results: The initial recall precision (1/sd(error)) was statistically indistinguishable in the two groups. The precision improved significantly across training sessions in both groups, but the incongruent group improved faster and by a larger amount (42% improvement in the incongruent vs. 21% in the congruent group). Post-tests on switched congruency conditions revealed nearly complete transfer of learning for both groups. Discussion: These results demonstrate that training can increase VSTM precision and that congruent masks impeded learning more than incongruent masks. Critically, training on incongruent stimuli led to congruent-precision greater than the asymptotic precision obtained with congruent training. This suggests that observer's strategies may play an important role in the improvement of VSTM precision, particularly in the face of perceptual interference.
Meeting abstract presented at VSS 2014