Fig. 2: Mapping invariance properties of perceptual priors.
From: Neural and computational mechanisms underlying one-shot perceptual learning in humans
Top row: Experiment 1 (n = 30 subjects). Bottom row: Experiment 2 (n = 12 subjects). a and f Learning effect in response to the original grayscale images and catch images in Experiments 1 and 2, respectively. Data underlying (a) are identical to those plotted in Fig. 1d. b–e Learning effects in response to left-right inverted grayscale images (b), 90° rotated grayscale images (c), size-manipulated grayscale images (d), and left/right visual-field shifted grayscale images (e). g–j Learning effects in response to a different grayscale image from the same category (g), high-contrast line drawings (h), magnocellular pathway-biasing low-contrast images (i), and parvocellular pathway-biasing red-green iso-luminant images (j). In (i), image contrast is artificially increased for visualization purposes. Asterisks denote statistically significant interaction effects in a two-way repeated measures ANOVA compared to original (green), high-contrast line drawing (blue), or catch (orange) trials. The central white dot of each violin plot represents the median, the gray vertical bar represents the interquartile range (25th to 75th percentiles), the violin plot bounds represent the minima and maxima, and the plot curvature represents the density estimate of the data distribution. Source data are provided as a Source Data file. All images adapted from the Caltech 10170 and Pascal VOC71 databases.
