The amplification effect is not driven by perceptual sampling bias

Abstract When observers are confronted with a set of stimuli of varying sizes, the relatively salient large stimuli among them tend to influence their judgments, leading them to overestimate the average size of the set of stimuli—a phenomenon known as the "amplification effect." This effect is a prevalent phenomenon in the research on ensemble perception, yet its mechanism remains unclear. This study aims to address this issue by manipulating stimulus presentation and introducing adaptation to distinguish between early perceptual encoding and later responding as potential loci of the effect. Firstly, comparing multiple stimuli of varying sizes with a single stimulus produces an amplification effect, resulting in the overestimation of the average size of multiple stimuli, which is in line with the former theory. However, this bias disappears when comparing the sizes of multiple stimuli of varying sizes and multiple stimuli of uniform size. Secondly, the adaptation paradigm provides converging evidence in support of this hypothesis. Adaptation to both multiple large stimuli and a single large stimulus induce adaptation aftereffects, whereas adaptation to multiple stimuli of varying sizes does not, suggesting that stimulus salience does not bias the perception of average size during the adaptation period. Experiment 1 was conducted in the central visual field, and Experiment 2 replicated these findings in the left and right hemifield fields. In conclusion, the amplification effect does not constitute a stable bias arising at the perceptual sampling stage, which is driven by stimulus size salience. Instead, it likely emerges during intermediate processing or the subsequent response selection stage after initial perception. These findings suggest a reassessment of the perceptual saliency-based weighting theory.