By contrast, circuits

based upon neuronal thresholds are insensitive to loss of inputs from low-threshold inhibitory neurons but highly sensitive to loss of high-threshold inhibitory neurons. Thus, ablating high-threshold inhibitory neurons in such circuits would have a much larger effect on the drift patterns than ablating low-threshold inhibitory neurons (Figure 7B). For detailed analysis of the specific patterns of drift seen in Figure 7, we refer the reader to the simplified analytic model of the Supplemental Methods and Figure S2. A second prediction arises from analyzing the time constants of drift following inactivation. Both in the well-fit and poorly fit models, the rate of drift BI 6727 supplier following inactivation scaled approximately linearly with the inverse UMI-77 ic50 of the recurrent excitatory synaptic

time constant. To reproduce quantitatively the drift rates observed experimentally following inactivation, a recurrent excitatory synaptic time constant of ∼1 s was required. This finding predicts a role for a slow cellular component of persistence at excitatory synapses or dendrites (see Discussion). The results above show that there are multiple circuit structures, understandable by the tradeoff between two thresholding mechanisms, that could reproduce the experimental Calpain data. As shown next, however, these structural differences masked strong similarities in functional connectivity that were revealed only when the combined effects of the structural connectivity Wij, the synaptic nonlinearities s(rj), and the threshold nonlinearity of the tuning curves were considered.

To generate the functional connectivity, also known as “effective connectivity” (Sporns et al., 2004), between neurons at different eye positions, we calculated the amount of current provided by any given neuron to its postsynaptic targets at different eye positions. These currents then were normalized by the presynaptic firing rate to obtain a functional connectivity measure, current per presynaptic spike, that did not simply reflect the strength of presynaptic firing. Below-threshold neurons were assigned a functional connectivity strength of zero. The resulting functional connectivities for all circuits exhibited a striking pattern not evident in the anatomical structure: when the eyes were directed leftward, the left-side inhibitory neurons projected strong functional connections. However, the functional weights of inhibitory right-side neurons were almost zero (Figures 8D–8F). When the eyes were directed rightward, the opposite pattern emerged, with the right side inhibitory neurons dominating and those on the left side contributing little (Figures 8G–8I).