a, Classification accuracy of natural scene categories based on compressed representations of style features extracted from natural scene images (OG, PS). The figure shows both the average accuracy across all categories and the accuracy for each individual category. b, Classification accuracy of object categories based on compressed representations of style features extracted from original object images. The format is the same as in a. Classification accuracies were significantly higher than the chance level (p < 0.05, FDR corrected) for all categories except those marked as “n.s.” (not significant) in the figure.