Abstract
When there are many visual items, our visual system could extract summary statistics of the items (e.g., mean, variance), facilitating efficient processing (Whitney & Yamanashi Leib, 2018). Although many researches have been conducted on mean or variance representation itself, a relationship between these two representations has not been investigated much. Thus, we investigated this relationship focusing on how perceived variance affects the estimation of mean and vice versa in two experiments using perceptual adaptation. Participants watched a sequence of orientation arrays during adaptation. In order to produce an adaptation to variance or mean, one property of the adaptor arrays (variance or mean) had a fixed value while the other property was randomly varied. After the adaptation, participants were asked to estimate the property of the test array that was varied during the adaptation. In Experiment 1, participants were adapted to two levels of variance (high, low) and then estimated the mean orientation of the test array whose physical variance remained constant. We found that the adaptation to a high variance resulted in more sensitive estimation of the mean orientation than the adaptation to a low variance. These results suggest that a perceived variance affects the estimation of mean orientation. In Experiment 2, participants were adapted to two levels of mean orientation (15°, 90° relative to the vertical) and then estimated orientation variance of test arrays whose mean orientation was vertical. Results showed that compared to the no adaptation, the adaptation to 15° mean orientation led to overestimation of the variance, but the adaptation to 90° mean orientation did not show that change. These results suggest that a perceived mean orientation affects the estimation of orientation variance. Collectively, these two experiments suggest that mean and variance representations in orientation perception are closely interrelated to each other.
Acknowledgement: This work was supported by National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (NRF-2016R1A2B4016171).