Article Figures & Data
Figures
- Figure 1.
Presence of an interaction term. Significance matrices, 1-dimensional response curves, and CF-I response surface of (A-C) Neuron #4 and (D-F) Neuron #36. A, D, Significance matrices represent significant (p < 0.05, shaded dark) individual dimensions on the diagonal and interaction terms in the upper triangle. (The lower triangle is symmetric to the upper triangle and not shaded.) Red box represents the terms being considered in the 1- and 2-dimensional visualizations. B, E, One-dimensional response curve of each neuron with respect to CF (left) and I (right). Dashed lines indicate the confidence interval as ±2× SE. C, F, CF-I response surface of each neuron.
- Figure 2.
Example impact of interaction term on neuron response space. The CF-I response surface of an example neuron is visually separated into three components. For the expression, r = β0 + f1(x1) + f2(x2) + f3(x1, x2), x1 is CF, x2 is I, f1 and f2 are individual dimension terms, and f3 is an interaction term. A, Individual dimension terms (f1 + f2). B, Interaction term (f3). C, Neuron response offset by intercept (r – β0): that is (A) + (B) = (C).
- Figure 3.
A cortical neuron with multiple significant interaction terms. The significance matrix (top left) of Neuron #50 represents 5 significant interaction terms (p < 0.05). The response surfaces of these 5 terms are shown as follows: CF-I, CF-BW, CF-L, I-BW, and BW-AM.
- Figure 4.
Population analysis of A1 neurons. A, Percent of neurons driven by each significant term. Red represents highest value (100%). Green represents lowest value (10%). Lower triangle is symmetric to upper triangle and is left blank. B, Histogram of number of significant interaction terms per neuron. Median number of significant interaction terms was 2, shown by dashed vertical line.
In this issue
