@santiruiznavas @mkpsyx @ScienceAdvances @WilmaBainbridge @Chris_I_Baker We defined *semantic* or *conceptual* as non-visual dimensions (i.e. not red, not textured, etc.) from the set of 49 dimensions identified in the study below, which underlie similarit
5,160 followers
5,160 followers
As a bit of background, we previously used a smaller dataset of 1.46 million odd-one out judgments for the same images to demonstrate that similarity can be explained with a surprisingly low-dimensional embedding: we found 49 interpretable dimensions. 2/n
5,160 followers
@jcrwhittington @IbanDlank @fpereira Looking forward to reading your work in detail! We’re coming more from the empirical side :-) My working hypothesis has been that disentangled representations of frequent concepts aid readout by reduce wiring costs. Her
75 followers
Revealing the multidimensional mental representations of natural objects underlying human similarity judgements. https://t.co/TPfX3i1zYv