Purchase this article with an account.
Pan Zhang, Yukai Zhao, Barbara Dosher, Zhong-Lin Lu; Evaluating the performance of the staircase and qCD methods in measuring specificity/transfer of perceptual learning. Journal of Vision 2019;19(10):29. doi: https://doi.org/10.1167/19.10.29.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The degree of transfer/specificity is a fundamental property of perceptual learning. Many conclusions about the level and mechanisms of perceptual learning depend critically on the accuracy and precision of the estimated transfer/specificity index. Traditionally, the transfer index is derived from learning curves during initial learning and transfer measured with the staircase method. Recently, Zhao et al. (2017) developed and validated a Bayesian adaptive qCD method to measure the trial-by-trial time course of perceptual sensitivity change in dark adaptation. The method was also validated in perceptual learning in a 4-alternative forced-choice (4AFC) global motion direction task, where the qCD method characterized the time course of perceptual learning more accurately, precisely and efficiently than the 3-down/1-up staircase method. Here, we evaluated the qCD and staircase methods in assessing the transfer index in a 2AFC task. Three observers with different time constants (40, 80, and 160 trials) of an exponential learning curve were simulated, with the same time constants and 50% transfer in the transfer phase. Each simulated observer was assessed with the 3-down/1-up staircase (10% step size) and the qCD procedure with a uniform prior of the parameters of the exponential learning curve, each with five initial stimulus levels (+50%, +25%, 0, −25%, and −50% from the true threshold in the first trial). 1000 simulated runs of 1600 trials were tested for each observer. Thresholds were estimated every 80 trials in the staircase method. The estimated transfer indexes from the qCD and staircase methods, averaged over starting levels, were 0.39±0.17 (Mean±SD) and −1.21±4.91 for Observer 1, 0.43±0.14 and 0.25±2.19 for Observer 2, and 0.47±0.09 and 0.46±0.48 for Observer 3, respectively. The results indicated that the qCD method provided more accurate and precise measures on transfer of perceptual learning than the staircase method, especially when the learning is more rapid. The qCD has great potentials in studies of perceptual learning.
This PDF is available to Subscribers Only