Extended Data Fig. 5: Bipartite invariance quantification.
From: Functional bipartite invariance in mouse primary visual cortex receptive fields
a,b, Mean peak-normalized in silico response as a function of the variable subfield size with linear interpolation (a) and quadratic-smoothing spline (b) for 1200 random V1 neurons from six mice, 60 simulated simple cells (blue), and 60 simulated complex cells (red). c, Bipartite invariance index was defined as the ‘area under the curve” (AUC) from the quadratic-smoothing spline in b. AUC values were highly consistent regardless of whether linear interpolation or quadratic-smoothing spline was used (Pearson r = 0.995, P < 10−6, two-sided t-test). d, Bipartite invariance indices were consistent across model ensembles initialized with different random seeds (Pearson r = 0.66, P = 3.2 × 10−52, two-sided t-test). a–d, Data were pooled over 1200 neurons from six mice.
