Extended Data Fig. 6: Both subfields of the partial-texture VEIs are necessary and specific for evoking high in vivo responses.
From: Functional bipartite invariance in mouse primary visual cortex receptive fields
a, Examples of the MEI, MEI with either fixed or variable subfield masked out, VEIspartial, and VEIspartial with either fixed or variable subfield swapped with random content for two example neurons. The fixed subfield content was replaced with 20 different random natural patches while the variable subfield was replaced with patches cropped from 20 different random non-self neurons’ preferred textures. b,c, MEI with fixed (b) or variable (c) subfield masked out evoked weaker in vivo responses than MEI (two-sided Wilcoxon signed-rank test, W = 5347, P = 7.0 × 10−12 and W = 1570, P = 4.1 × 10−28, respectively) with 15.4% and 38.1% of all neurons showing weaker responses than their MEIs (1.9% and 27.9% after BH correction) (P < 0.05, two-sided Welch’s t-test with 33.2 and 29.9 average d.f., respectively). Data were pooled over four mice, displaying a total of 215 neurons. d,e, VEIspartial with either fixed (d) or variable (e) subfield swapped evoked weaker in vivo responses than VEIspartial (two-sided Wilcoxon signed-rank test, W = 1960, P = 6.5 × 10−35, and W = 5820, P = 7.0 × 10−20, respectively) with 30.1% and 12.7% of all neurons showing weaker responses than their partial-texture VEIs (18.9% and 0.0% after BH correction) (P < 0.05, two-sided Welch’s t-test with 30.0 and 32.5 average d.f., respectively). Data were pooled over 259 neurons from five mice.
