Activation-flagged channels (AFCs) were defined in part by a net rise in oxyhemoglobin concentration during the stimulation interval, as in previous work.9 Specifically, the average rise in oxyhemoglobin concentration over the multiple stimulation epochs had to be significantly positive (one-sided -test). The rise was calculated using 4-s windows centered on the onset and termination of the visual stimulus. For the -test, the -value was adjusted to account for multiple comparisons19,20 by dividing the target threshold for significance by a value related to the number of independent recording channels . In this case, the number of channels analyzed was 265, but the time series were not fully independent, due both to overlaps of measurement volumes and to spatial correlations of hemodynamics over the measurement field. We chose a conservative, empirical value of and a confidence level of 95%. With those inputs, the threshold for flagging activations was set at to determine whether the average rise in was significantly greater than zero in any single channel. If a channel passed the t-test in any regression mode [VS, S, or no-regression (NR)], it passed to the second round of analysis.