Fig. 7: Summary of pooled scale invariance model and its performance relative to scale invariance.

Each of the four columns in b through e corresponds to a different tuning parameter that is predicted by scale invariant pooling, labeled on top. All functional maps shown in this figure are from the same ROI. Data yield is the same across maps, with the exception of the measured RF width at top left, which is due to it being measured with a different visual stimulus and inclusion criterion. a Bottom row shows the measured map of f_o, which is the independent variable in the models of scale invariance and pooled scale invariance. b First layer of the scale invariant pooling model consists of scale invariant simple cells. Below each map is the formula defining the given parameter. From left to right the maps are of RF width (1σ), F1/F0, SF bandwidth (1σ), and orientation bandwidth (1σ). c Output layer of the scale invariant pooling model. Under each map is the relevant parameter value(s) required to make the predicted map shown. d Measured functional maps shown above their respective model predictions. e Measured tuning and model fits of six neurons from this ROI. A number is adjacent to each tuning curve that indicates the neuron in the labeled map below in d. Black dots are data, gray line is the direct fit to the data, blue line is the scale invariance prediction, and green line is the pooled scale invariance prediction. In the case of the F1/F0 tuning curves, the dashed gray line is the baseline.