Fig. 7: Modulation of the slow timescale during attention is mediated by an increase in the efficacy of network interactions. | Nature Communications

Fig. 7: Modulation of the slow timescale during attention is mediated by an increase in the efficacy of network interactions.

From: Intrinsic timescales in the visual cortex change with selective attention and reflect spatial connectivity

Fig. 7

a Effect of connectivity parameters on local timescales in the model. The fast timescale (τ1, right) mainly depends on the self-excitation probability (ps), whereas the slow timescale (τ2, left) depends on both the self-excitation (ps) and recurrent horizontal interactions (pr). The dashed rectangles indicate the range of parameters reproducing V4 timescales (mean ± s.e.m. of MAP estimates, Methods). b The slow timescale increases with the network excitability (ps + 8pr, left panel). Green and magenta dots indicate the parameters reproducing attend-away (att-away) and attend-in (att-in) timescales, respectively. Filled dots show examples of experimentally observed 20% increase in τ2 for three possible scenarios based on different changes in ps or pr (right panels). Larger changes of parameters in scenarios (2) and (3) are due to coarser grid of ps used to fit the timescales. A similar change of τ2 can also be achieved with smaller changes in ps and pr (e.g., for all 0.74 < ps < 0.745 in scenario 2). c Example autocorrelations (ACs) from the model simulations with the attend-in and attend-away parameters for the scenario (2) in b. We fitted unbiased autocorrelations from the model simulations with double exponential functions (green and pink lines) to estimate the two timescales (Methods). Source data are provided as a Source Data file.

Back to article page