Figure 5.

Removing of noise correlations

(A) We are testing the effect of removing correlations between all neuronal pairs (case 1, left), and across the groups of neurons with the same sign of the weight (case 2, right). Note that for the case 2, correlations within each group remain intact (right). Figure reprinted with permission from Koren et al., 2020a.

(B) Noise correlations are correlations that arise in the same trial. We remove them by randomly permuting the trial order. In the case 1 (remove all correlations), we permute the order of trials for every neuron independently. In the case 2 (remove correlations across groups), we assign a different trial order to each group of neurons, while neurons from the same group maintain the same order of trials. In all cases, the permutation of trial order is limited to trials within a given condition.