Skip to main content
. 2022 Feb 7;22(2):4. doi: 10.1167/jov.22.2.4

Figure 6.

Figure 6.

Clustering of abstracted object representations in the ImageNet-trained VGG-16. Multidimensional scaling plots of the super-RDMs reflecting all pairwise dissimilarities between stimulus representations in the stimulus set. Across layers, drawing and sketch representations increasingly clustered together, whereas photo representations remained well separated.