Fig 6. Effects of the different portions of RF interactions on correlated activity using arbitrary RF geometries.
(A) Three types of inputs to two putative neurons were considered: (1) correlated input (black; same sign; +/+); (2) anti-correlated input (white; opposite sign; -/+); and (3) uncorrelated input (gray; +/0 & 0/+). The inputs were mapped onto the neurons either in an excitatory (blue arrows) or inhibitory (red arrow) fashion. (B & C) For a fixed total input (Ntot = 1000), the ratio of correlated and anti-correlated inputs was varied in terms of both their amount (N+corr / N-corr) and in terms of their gain (gain+corr / gain-corr). This was carried out for different but fixed portions of uncorrelated inputs (shown are 0% in B and 50% in C). Roman numerals (I–IV) depict cases for which the inputs are illustrated in S10A Fig (D) Expanded model in which all possible types of RF interactions were considered: correlated inputs (+/+ & -/-), anti-correlated inputs (+/- & -/+) and uncorrelated inputs (+/0, -/0, 0/+, & 0/-). These were mapped onto the downstream neurons either excitatory (blue arrows) or inhibitory (red arrows) depending on whether a given input was part of the RF center of the RF surround. (E & F) As (B & C), but when independently varying gaincenter / gainsurround depending on whether an input was part of the RF center or surround. Roman numerals (V–VIII) depict cases for which the inputs are illustrated in S10B Fig.