Fig. 4
From: Conscious perception of natural images is constrained by category-related visual features
DCNN representational distance and target similarity explain trials of the AB. a Depiction of analysis procedure. For each layer, DCNN representations are extracted for each image. These feature activations were then compared for all image pairs (Pearson correlation), to estimate the similarity between pairs. Due to copyright reasons, all photos except for the faces (which were photographed by one of the authors) have been replaced by representational images. b) Mean similarity between T1 and T2, based on feature activations of each layer, for lag-2 missed and hit trials separately. Separate dots represent single subjects. The mean similarity across subjects is represented by a large black diamond and black bars denote 95% confidence interval. Two-tailed dependent t-test *p < 0.05, **p < 0.01, ***p < 0.001
