Abstract
Our visual system must solve the difficult problem of stabilizing noisy and ever-changing visual input. Previous studies have demonstrated that serial dependence in visual perception is a potential mechanism to promote perceptual stability over time: the percepts of current stimuli are biased toward previously seen stimuli (Fischer & Whitney, 2014). Here, using a simple detection task and classification image technique, we show that such serial dependency alters the perceptual templates of orientation representations. In the experiment, a low-contrast near-vertical Gabor was embedded in a static white noise on half the trials, and subjects were asked to judge whether the target Gabor was present or absent in the noisy image. The target contrast was adjusted for each individual so that d’ was around 1.5. Every 5 to 11 trials, a high-contrast Gabor was presented as an inducer with its orientation tilted 10 degrees clockwise or counter-clockwise relative to vertical. We drew classification images separately for trials following clockwise or counter-clockwise inducers and found that the classification images were significantly biased toward the most recent inducer orientation. Given the task was a simple detection task, subjects were not required to encode, maintain, or recall stimulus orientation from memory. Nevertheless our results indicate that the perceptual templates of orientation were biased toward stimuli presented up to 10 trials before the current trial, suggesting the presence of a long-lasting mechanism to stabilize orientation perception. Our results also pose a methodological question for the classification image technique itself. This widely used psychophysical method typically assumes that a subject’s response is determined only based on the current stimulus input. However, this assumption may be violated since our perception is serially dependent. We will discuss the spatiotemporal continuity of serial dependence in classification images and the fruitfulness of investigating higher-order reverse correlation.