Purchase this article with an account.
Charles Liu, Takeo Watanabe; Accounting for speed-accuracy tradeoff in visual perceptual learning. Journal of Vision 2010;10(7):1111. doi: 10.1167/10.7.1111.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In the perceptual learning literature, researchers typically focus on improvements in accuracy, such as proportion correct or dprime. In contrast, researchers who investigate the learning, or practice, of cognitive skills focus on improvements in response times (RT). Here, we argue for the importance of accounting for both accuracy and RT in perceptual learning experiments, due to the phenomenon of speed-accuracy tradeoff: at a given level of discriminability, faster responses tend to produce more errors. A formal model of the decision process, such as the diffusion model (Ratcliff & McKoon, 2008), can explain the speed-accuracy tradeoff. In this model, a parameter known as the drift rate represents the perceptual strength of the stimulus: higher drift rates lead to more accurate and faster responses. We applied the diffusion model to analyze responses from a yes-no coherent motion detection task. Participants were trained for 5 days and completed 500 trials per day. On each trial, participants were shown a field of moving dots for 200 ms within a 14-degree aperture. On “signal” trials, 15% of dots moved coherently in a specific direction at a constant speed, while the remaining dots were replotted at random locations. On “noise” trials, all dots were replotted randomly. The results showed a significant range of individual differences in speed-accuracy tradeoff. When accuracy and RT measures were analyzed separately, inconsistent patterns of learning were observed across sessions. However, the diffusion model analysis indicated that drift rates improved consistently across sessions. These results suggest that part of the variability typically observed in perceptual learning experiments may be attributed to speed-accuracy tradeoff, and that drift rates offer a promising new index of perceptual learning. We discuss further advantages of diffusion modeling in perceptual learning, including the ability to dissociate decision time from non-decision time, and perceptual bias from response bias.
This PDF is available to Subscribers Only