Figure 2: QC model outperforms models without either position invariance or multi-component feature selectivity. | Nature Communications

Figure 2: QC model outperforms models without either position invariance or multi-component feature selectivity.

From: Cross-orientation suppression in visual area V2

Figure 2

Extrapolated correlation between model predictions for a novel test set and the observed responses64 is shown in all panels. The dashed lines indicate the linear regression without a constant offset. (a) The QC model outperformed the QnC; average improvement is by a factor of 1.5. (b,c) QC also outperformed an LnC and the LC by factors 3.9 and 4.4, respectively. LC, linear convolutional model; LnC, linear non-convolutional model; QnC, quadratic non-convolutional.

Back to article page