Jenny W.S. Chiu¹ (), Alan L.F. Lee¹; ¹Lingnan University, Hong Kong

Our perception of a visual stimulus tends to depend on previously presented ones, a phenomenon known as serial dependence. Previous studies have shown that confidence judgments on the preceding stimulus can influence the magnitude of serial-dependent bias (e.g., Samaha et al., 2019), but it remains unclear whether we are metacognitively sensitive to perceptual effects arising from serial dependence. In the present study, we investigated the metacognitive sensitivity of serial dependence effect in a visual working memory task. In each trial, one or six Gabor patches (as a between-subjects factor) were briefly presented at locations around 5-6 degrees away from fixation. After a delay with a randomly-sampled duration (normally-distributed with mean = 2,000ms, SD = 500ms, limited within 250 - 3,750ms), a cue would appear at one of the Gabor’s locations. Participants first reproduced the orientation of the Gabor patch presented at the cued location by adjusting the orientation of an on-screen line, and then rated their confidence on the orientation response using a slider bar. We manipulated the relative difference in orientation between every two consecutive trials such that there were 11 possible relative differences (inclusively from -75 to +75 degrees, 15 degrees apart). When there was only one Gabor presented (i.e., a set size of 1), we found that precision in the reported orientations (measured as the inverse of across-trials response variability) was significantly higher when the target orientation was the same as the one in the preceding trial. Interestingly, confidence ratings remained constant across values of relative differences. In a separate experiment with a set size of 6, we did not find any serial dependence effects on either first-order or second-order responses. Our findings suggest that, while serial stability could improve the precision of stimulus representation, we are metacognitively insensitive to such improvement.

Acknowledgements: This work was supported by the Research Matching Grant Scheme (LU Fund Code: 185218) from the University Grants Committee of the Hong Kong SAR