Abstract
Recent stimulus history can automatically bias the perception of a new stimulus. The present study investigated whether this stimulus history effect reflects the biased representation itself or the biased decision during the report. In the experiment, participants performed a motion direction estimation task where they reproduced the perceived motion direction from random-dot-kinematograms (RDKs). We hypothesized that response time (RT) should be faster for the trials with large direction differences from the previous trial if the stimulus history effect is driven by the decision during the report as the decision should be easier when the current motion direction differs significantly from the prior motion direction. Results showed a typical repulsive serial bias effect. More importantly, RT was faster for the trials with large direction differences, supporting the hypothesis that the serial bias is decision-driven. If this RT effect truly reflect the decision related to the serial bias, then the RT effect should be absent when the serial bias is absent. To test this prediction, we had participants perform a ‘dual-feature’ version of the RDK task where RDK was presented with colored dots and participants reported the color, instead of the direction, on the random half of the trials. We found no serial bias in the direction reports following the color report. Critically, RT effect was absent after the color report, demonstrating that the RT effect was associated with the serial bias. In subsequent analysis for the data with a large sample size (N>70), we confirm that the RT effect was mostly driven by the expedited decision for large direction differences rather than the delayed decision for small direction differences. Together, these results provide converging evidence that the prior stimulus impacts the speed of decision for a current stimulus, supporting the hypothesis that the stimulus history effect is driven by decisional processes.