Muzhen Ai¹; ¹UCLA

Serial dependence effect, defined as the influence of past sensory input on current perception, is a phenomenon where prior stimuli bias current perceptual judgments. This study seeks to examine the factors that modulate serial dependence magnitude by aggregating and analyzing results across 8 studies (N=211) on serial dependence. We categorized trial-wise perceptual certainty defined by noise level, spatial frequency, stimulus contrast. Through this approach, we compared serial dependence effects in high- and low-certainty conditions. Our results demonstrate that, within current trial settings, low-certainty conditions consistently exhibit a higher reliance on past sensory information than high-certainty conditions, aligning with the Bayesian account. However, previous low perceptual certainty elicits a stronger serial dependence effect. We propose that in the face of ambiguous prior sensory information, individuals may adopt a perceptual strategy that places less emphasis on current sensory input, reflecting an adaptive mechanism for managing uncertainty in perceptual environments.