Fig. 2: Evidence accumulation signals.
From: Cortical evidence accumulation for visual perception occurs irrespective of reports
A Averaged HGA traces from single channels in the VVC (top) and anterior insula (bottom). HGA in miss trials (red) and in hit trials separated in three terciles based on reaction times (blue). The vertical dashed lines indicate the onset and offset times of the face stimuli. The black bars indicate significant effects of detection (cluster-based permutation tests). Note that the channel in the insula was localized by the atlas as part of the IFC. B Averaged HGA across all responsive channels in the VVC (top) and IFC (bottom). The black bars indicate main effects of detection as assessed with sample-by-sample hierarchical mixed-effects regressions (corrected for false discovery rate across time). Response-locked plots can be found in Supplementary Fig. 5. C Representation of the decoding algorithm, with example hit (blue) and miss (red) trials for 3 channels (“chan”). A trial-by-trial detection signal was obtained by applying the weights of the multivariate classifier at each time point (b = bias; w1, w2, …, wn = weights). The weights were chosen to optimize the distance between the maximum values of the latent variables in hit vs. miss trials. Trials were classified as hit if the maximum values of the latent variable crossed a threshold (purple line), and as miss otherwise. D Latent variables decoded from the VVC (dark blue) of one participant (G5; left panel) and the IFC (green) of another participant (G21; right panel), averaged over all miss trials and separated in three terciles based on reaction times for hit trials. Insets: Trial-by-trial depiction of the relationship between reaction times and the predicted time of threshold crossing of the decoded latent variable, with Spearman correlations. E Average Spearman correlation between observed and decoded response times across participants (bar plot: average; dots: individual participants). Red lines indicate group-level chance-level performance assessed by permuting labels across trials (VVC: n = 4, dark blue; SPC: n = 5, light blue; IFC: n = 13, green; DLPFC: n = 4, yellow). Red dot surrounds indicate participants who displayed above chance-level performance after false discovery rate correction. All shaded areas represent 95% confidence intervals.
