Skip to main content
. Author manuscript; available in PMC: 2011 Mar 17.
Published in final edited form as: Neuroscience. 2009 Dec 29;166(2):698–711. doi: 10.1016/j.neuroscience.2009.12.048

Figure 4. Partial coherence analysis reveals cortex and GCL as significant predictors in CTX-GCL-DCN recordings.

Figure 4

For each pairwise combination of signals in a recording, the maximum coherence between the two signals was determined for the frequency range of 0.5 to 4 Hz and compared with the partial coherence calculated by subtracting the cross spectrum of a third simultaneously recorded signal. Recordings in which the maximum coherence in this range did not exceed the upper confidence limit were excluded from analysis. A: The contribution of each cortical area to the network was evaluated using partial coherence analysis. The coherence between each pair of cortical areas was compared after removing the third cortical signal. This resulted in a significant reduction in coherence for all cortical areas (SMA:MI with SI removed, p = 0.004, n = 9, paired t-test; SMA:SI with MI removed, p < 0.001, n = 9, paired t-test; MI:SI with SMA removed, p = 0.004, n = 9, paired t-test) B: The frequency at which the peak coherence was measured between cortical pairs occurred in the 1.5 – 2.5 Hz range (SMA:MI, 2.21 ± 0.28 Hz; SMA:SI, 1.87 ± 0.21 Hz; MI:SI, 1.87 ± 0.21 Hz). These frequencies were not different between cortical pairs (p = 0.550, n = 9, Kruskal-Wallis test) C: There was a significant (p<0.05) phase shift in the oscillation between SMA and SI and SMA and MI at the frequency of peak coherence, but not between MI and SI (SMA:MI, 9.43 ± 2.04 deg; SMA:SI, 17.90 ± 4.48 deg; MI:SI, 6.91 ± 4.03 deg, n = 9). D: Removing the cortical signal significantly reduced the coherence shared between GCL and DCN for all cortical areas (SMA, p < 0.001, n = 25; MI, p = 0.004, n = 12; SI, p = 0.011, n = 11; paired t-test used for all comparisons). E: Similarly, removing the GCL signal significantly reduced the coherence shared between CTX and DCN for all cortical areas (SMA, p < 0.001, n = 26; MI, p = 0.007, n = 12; SI, p = 0.023, n = 11; paired t-test used in all comparisons). F: Removing the DCN signal significantly reduced the coherence shared between GCL and SMA and MI but not SI (SMA, p < 0.001, n = 27; MI, p = 0.034, n = 12; SI, p = 0.105, n = 11; paired t-test used in all comparisons). Standard error bars plotted in panels A-F.