Purchase this article with an account.
Parker Banks, Allison Sekuler, Patrick Bennett; Cognitive bias and reward affect contrast and response gain. Journal of Vision 2017;17(10):510. doi: https://doi.org/10.1167/17.10.510.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Investigations of perceptual learning (PL) typically focus on stimulus and sensory factors that affect performance during training, and comparatively little is known about the roles response bias and reward structure play in determining how people learn from experience. Such questions are important because some naturalistic PL protocols (e.g., fingerprint identification) use extreme payoff structures that severely punish some responses. Therefore, we investigated how differing monetary rewards and punishments interact with PL during a texture identification task. We trained subjects on ten band-pass filtered, white noise textures in a same-different task, measuring accuracy while manipulating signal strength with the method of constant stimuli. Subjects were trained in adverse miss (AM), adverse false-alarm (AFA), and no adversity (NA) conditions over a period of five training sessions. In the NA condition, subjects received equal monetary rewards and punishments for each correct and incorrect identification. However, in the AM condition misses (identifying the same textures as different) were punished at a 100:3 ratio to rewards, and the AFA condition was subjected to similar punishment following false alarms. At the end of training, psychometric functions from subjects in the AM and AFA conditions exhibited a hard threshold: sensitivity was essentially zero to low-contrast patterns and the abruptly increased beyond a critical level of contrast. No such threshold was apparent in the NA condition. Hence, our results suggest that the reward structures in the AM and AFA conditions reduced sensitivity to weak signals and increased the slope of the psychometric function. Currently we are testing the effects of bias on PL across a wider range of payoff structures.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only