Fig 1.
Experimental paradigm and illustration of the TRF approach.
(A) We recorded 275-channel, whole-head magnetoencephalography signals from subjects who were viewing a 5-s circular glass pattern stimulus sequence in each trial. The global F (red) and the L (blue) of the glass pattern stimuli were continuously modulated according to 2 independent and randomly generated 5-s temporal sequences (the global F sequence and the L sequence) throughout each trial. (B) The F-TRF (red) and L-TRF (blue) responses were then calculated from the same MEG recordings based on the corresponding stimulus temporal sequences for each of the 275 MEG channels in each subject by using a linear least-squares approach. (C) Spatiotemporal dynamics of the F-TRF and L-TRF responses in source space (dSPM method for source localization) was calculated in combination with individual anatomical MRI scans. dSPM, dynamic statistical parametric mapping; F, form coherence; F-TRF, form coherence TRF; L, luminance; L-TRF, luminance TRF; MEG, magnetoencephalography; TRF, temporal response function.
Fig 2.
Experiment 1: L-TRF and F-TRF responses showed distinct neuronal spatiotemporal profiles.
(A) L-TRF responses. (B) F-TRF responses. Note that the TRF responses represent brain responses to each unit transient in luminance (L-TRF) or global form coherence (F-TRF) of the glass pattern sequences. Upper panel: Grand average (n = 20) plots for TRF waveforms (summarized as root-mean-square across all MEG channels) as a function of temporal lag (−100 to 400 ms). Gray shades indicate the confidence interval after permutation test (see details in S1 Text). Error bar indicates the standard error. Middle panel: Grand average (n = 20) plots for sensor-level topographical distribution of TRF responses at 0–100, 100–200, and 200–300 ms time range. Lower panel: Grand average (n = 20) plots for TRF source localization results in the normalized MNI template at 0–100, 100–200, and 200–300 ms time range (cluster-level permutation test across space and time, multiple comparison corrected, cluster p < 0.05). The MNI coordinates for the significant source clusters are listed in S1 Table. It is notable that F-TRF activated the IPS and V3a within the first 100 ms, followed by responses in primary visual cortices in the next 100 ms, in accordance with a reversed high-to-low activation pattern. Meanwhile, L-TRF showed a feedforward profile that started from EVAs. The data underlying Fig 2 can be found in S1 Data. EVAs, early visual areas; F-TRF, form coherence TRF; IPS, intraparietal sulcus; L-TRF, luminance TRF; MEG, magnetoencephalography; MNI, Montreal Neurological Institute; n.s., not significant.
Fig 3.
Experiment 1: TRF ROI source waveforms and granger causality analysis.
(A) Grand average (n = 20) plots for TRF source waveforms in the ROIs, defined according to source localization results, for L-TRF (left column) and F-TRF (right column) responses. Gray box indicates time ranges when the TRF responses showed significant activations compared to baseline (p < 0.0025, Bonferroni corrected). It is notable that for F-TRF responses (right column), IPS, V3a, V1 showed sequential activations (right panel), supporting a reversed hierarchical activation profile. (B) Granger causality analysis of activation time courses among ROIs for F-TRF (red) and L-TRF (blue). The solid arrows indicate significant causal effects (p < 0.05, permutation test), whereas the dashed arrows indicate nonsignificant causal effects. The data underlying Fig 3 can be found in S1 Data. EVA, early visual area; F-TRF, form coherence TRF; IPS, intraparietal sulcus; LO, lateral occipital; L-TRF, luminance TRF; ROIs, regions of interest; TRF, temporal response function.
Fig 4.
Experiment 2: L-TRF and F-TRF responses when global form property was task irrelevant.
(A) L-TRF responses. (B) F-TRF responses. Upper panel: Grand average (n = 16) plots for TRF waveforms (summarized as root-mean-square across all MEG channels) as a function of temporal lag (−100 to 400 ms). Gray shades indicate the confidence interval after permutation test (see details in S1 Text). Error bar indicates standard error. Middle panel: Grand average (n = 16) plots for sensor-level topographical distribution for TRF responses. Lower panel: Grand average (n = 16) plots for source localization results in the normalized MNI template (cluster-level permutation test across space and time, multiple comparison corrected, cluster p < 0.05). Note that L-TRF responses showed a similar feedforward profile that started from EVA as that in Experiment 1 (A). Crucially, the IPS-V3a-V1 activation sequence still emerged although was temporally delayed (B). Moreover, the VAN (orange) and DMN (blue) were also activated. The MNI coordinates for all the significant source clusters are listed in S1 Table. The data underlying Fig 4 can be found in S1 Data. DMN, default mode network; F-TRF, form coherence TRF; IPS, intraparietal sulcus; L-TRF, luminance TRF; MEG, magnetoencephalography; MNI, Montreal Neurological Institute; ROI, region of interest; TRF, temporal response function; VAN, ventral attention network.
Fig 5.
Experiment 2: TRF ROI source waveforms and granger causality analysis.
(A) Grand average (n = 16) plots for TRF source waveforms in the ROIs, defined according to source localization results, for L-TRF (left column) and F-TRF (right column) responses. Gray box indicates time ranges when the TRF responses showed significant activations compared to baseline (p < 0.0005, Bonferroni corrected). It is notable that for F-TRF responses (right column), IPS, V3a, V1, and LO showed sequential activations (right panel), which is consistent with the reversed hierarchical activation profile found in Experiment 1. Moreover, VAN and DMN showed earlier responses than IPS. (B) Granger causality analysis of activation time courses among ROIs for F-TRF (red) and L-TRF (blue). The solid arrows indicate significant causal effects (p < 0.05, permutation test); the dashed arrows indicate non-significant causal effects. The data underlying Fig 5 can be found in S1 Data. DMN, default mode network; EVA, early visual area; F-TRF, form coherence TRF; IPS, intraparietal sulcus; LO, lateral occipital; L-TRF, luminance TRF; IPS, intraparietal sulcus; ROIs, regions of interest; TRF, temporal response function; VAN, ventral attention network.
Fig 6.
Experiment 3 and rate control experiment.
(A) Experiment 3: untracked global form did not activate the IPS-to-V1 neuronal pathway. Grand average (n = 16) plots for TRF source waveforms in the ROIs (IPS, V3a, V1, and LO) for L-TRF (left column) and F-TRF (right column) responses. The gray area indicates time points when the TRF showed significant activations compared to baseline activities (p < 0.0025, Bonferroni corrected). Note the nonsignificant activations in IPS-to-V3a-to-V1 pathway for F-TRF responses (right column). (B) Results for rate control experiment. Grand averaged plots for L-TRF response under fast luminance modulation condition (solid line, Experiment 1 data, n = 20) and slow luminance modulation condition (dashed line, rate control, n = 4). The 2 L-TRF responses showed no difference, suggesting that the obtained TRF response is independent of modulation rate of the temporal sequence of the stimuli. The data underlying Fig 6 can be found in S1 Data. EVA, early visual area; Exp1, Experiment 1; F-TRF, form coherence TRF; IPS, intraparietal sulcus; LO, lateral occipital; L-TRF, luminance TRF; n.s., not significant; TRF, temporal response function.