Skip to main content
. 2017 Oct 9;8:1726. doi: 10.3389/fpsyg.2017.01726

FIGURE 4.

FIGURE 4

AlexNet performance at explaining similarity judgments: reweighting improves model performance for late layers. (A) Bars show the correlation between the similarity-judgment RDMs and each AlexNet layer RDM, with and without reweighting. A significant correlation between a layer RDM and the similarity-judgment RDMs is indicated by an asterisk (one-sided Wilcoxon signed-rank test, p < 0.05 corrected). Error bars show the standard error of the mean based on single-subject correlations, i.e., correlations between the single-subject similarity-judgment RDMs and a DNN RDM. The gray bar represents the noise ceiling, which indicates the expected performance of the true model given the noise in the data. “conv” indicates a convolutional layer and “fc” indicates a fully-connected layer. (B) Pairwise differences between model performance of AlexNet layer RDMs, with and without reweighting. Green color indicates significant pairwise differences (dark green p < 0.01, light green p < 0.05, FDR corrected across all comparisons).