0, p = 0.04) and higher order areas (t10 = 2.6, p = 0.01). Although we lacked coverage of early visual areas in the medial and posterior cortex, we observed a trend from midlevel visual areas in the ventral and dorsal stream toward larger TRWs in higher order visual areas. The TRW values from frontal cortical electrodes were higher than in all other ROIs buy FG-4592 (Figure 5B). Having found TRW patterns in ECoG that substantially match prior neuroimaging results (Hasson et al.,

2008; Lerner et al., 2011), we next tested the hypothesis that regions with longer TRWs should exhibit a shift toward a slower timescale of dynamics. We assessed the timescales of neuronal population dynamics using two metrics: first, a measure of low-frequency variance in the power time courses, and second, a measure of temporal autocorrelation in the power time courses. To measure the low-frequency variance in the power fluctuations, we first

calculated the “modulation spectrum” of each electrode: this is the power spectrum of the 64–200 Hz power fluctuations at each site. After dividing the electrodes via a median split on TRW values (median TRW value = 0.11), we averaged the modulation spectra within the “long TRW” and “short TRW” groups. The group of long TRW electrodes showed relatively more slow fluctuations than the group of short TRW electrodes (Figure 6A). The increase was most apparent below 0.1 Hz, and was seen in both the intact and fine-scrambled conditions. To quantify the strength of the slow fluctuations, we computed the fraction of the modulation spectrum that was below 0.1 Hz at each BLZ945 clinical trial site. We refer to this normalized about amplitude of slow fluctuations as “LowFq” (see Experimental Procedures; and also Zuo et al. [2010]). LowFq values range from 0 (indicating faster dynamics) to 1 (indicating slower dynamics). LowFq values were higher in the long TRW group

than in the group of short TRW electrodes (Figure 6B). This was evident for both the intact and fine-scrambled movie conditions. These observations were confirmed in a 2-way ANOVA with factors of stimulus (intact/fine-scrambled) and TRW (long/short): both factors significantly modulated LowFq (p < 0.01) but the interaction was not significant (p = 0.24). The fraction of slow fluctuations in power was also associated with TRWs on an electrode-by-electrode basis. LowFq values measured during the intact movie were robustly correlated across electrodes with TRW values (r = 0.46, p = 3e-5; Figure 6C). The same effect was observed when measuring LowFq in the fine-scrambled movie (r = 0.37, p = 0.001; Figure 6D). Partial correlations between LowFq and TRW values, with repeat reliability (rINTACT or rFINE) included as a covariate, were also highly significant (p < 0.01 all comparisons). This indicates that the relationship between LowFq and TRW was not due to a link between LowFq and electrode responsiveness within a single condition.