Realidades y Riesgos en medio del conflicto armado en el proceso de

The main question of this study was whether the change in percept dominance durations due to parietal cTBS was accompanied by a modulation in BOLD signal. We analysed activity related to perceptual switches (i.e. beta-estimates obtained for the button-response regressor during viewing of SFM stimuli), for pre- and post-TMS runs for the IPS and the vertex site. First, I tested if the ROI activity was significantly lateralised between hemispheres by modelling a repeated measures ANOVA on normalised beta estimates using the factors TMS site, whether the data is pre or post TMS, and hemisphere. This model did not include M1-l, since this region was only extracted unilaterally. We found a significant interaction between TMS site and pre-post (F(1,19) = 7.18, p < 0.002), but no main effect of hemisphere or interaction term including hemisphere (F(1,19) < 1, ns). We therefore averaged normalised beta estimates across hemispheres in bilateral ROIs in every participant and henceforth analysed ROIs irrespective of hemisphere.

Figure 6 — Percent signal change per ROI and TMS condition during percept switches

Percent signal change for each previously defined ROI across the different TMS sites as well as pre or post TMS. Consistent with whole-brain results, the strongest activity was observed in primary motor cortex. Consistently, post IPS signal was weaker than pre IPS, while this effect was reversed for vertex. Error bars are ± 1 s.e.m.

Next, I performed the main repeated measures ANOVA on normalised beta estimates using the factors TMS site, pre-post, and ROI. Mauchly´s test indicated that the assumption of sphericity had been violated, hence degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity for the main effect of ROI (χ2_{(5) = 55.60, p > 0.0001, ε = 0.52) as well as the three-way}

−0.005 0.000 0.005 0.010 0.015 0.020

Thal V1 MT SPL IPS M1−l IFG

percent signal change

vertex pre vertex post IPS pre IPS post

interaction term (χ2_{(5) = 37.28, p = 0.002, ε = 0.60). We found a significant main effect of ROI}

(F(6,114) = 57.32, p < 0.0001), as well as a significant three-way interaction between all factors (F(6,114) = , p = 0.019) (figure 6). Interestingly, IPS cTBS in all ROIs except M1 tended to decrease activity, while vertex cTBS led to the opposite.

Figure 7 — TMS modulation of percent signal change across ROIs

Modulation of percent signal chance for each ROI calculated as directed interaction contrast between vertex-pre, vertex-post, IPS- pre, IPS-post [1 -1 -1 1]. * t-test significance p < 0.05 Bonferroni corrected. Error bars are ± 1 s.e.m.

Table 2 — t-tests on TMS modulation across ROIs

One-sample t-tests for each ROI on the modulation of percent signal change, i.e. the difference pre-post vertex TMS subtracted from the difference pre-post IPS TMS. * t-test significance p < 0.05 (Bonferroni corrected). Error bars are ± 1 s.e.m.

In what followed, I examined not the normalised beta-estimates, but rather their TMS modulation, i.e. the difference pre-post vertex cTBS subtracted from the difference pre-post IPS cTBS. We entered this modulated estimate into a one-way ANOVA using ROI as factor. Mauchly´s test indicated that the assumption of sphericity had been violated, hence degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (χ2_{(20) = 35.20, p > 0.02, ε = 0.60). We}

p = 0.038 * p = 0.027 * −0.0100 −0.0075 −0.0050 −0.0025 0.0000 0.0025

Thal V1 MT SPL IPS M1−l IFG

modulation of percent signal change

ROI one sample t-tests (df = 19)

ROI t-statistic p-value (uncorrected) p-value (Bonferroni) significance

Thalamus -3.13 0.0054 0.038 * V1 -1.89 0.073 0.52 V5/MT-r -1.49 0.15 1 SPL -3.29 0.0038 0.027 * IPS -0.95 0.35 1 M1-l -0.17 0.87 1 IFG -1.97 0.06 0.44

found a significant main effect of ROI (F(6,114) = 3.31, p = 0.019). We hence proceeded with post hoc Bonferroni corrected one-sample t-tests to find ROIs for which the TMS modulation was significantly different from zero (figure 7, table 2).

Figure 8 — Correlation of TMS modulation of percent signal change and SFM percepts

Modulation of percent signal chance for SPL (left) and thalamus (right), given as directed interaction contrast between vertex-pre, vertex-post, IPS-pre, IPS-post [1 -1 -1 1], are correlated against modulation of SFM bistable dominance durations. Neither was statistically significant (|r| < 0.25, ns).

At last, I attempted to correlate the cTBS modulation of percent signal change with the cTBS modulation of SFM percept durations using Pearson’s r. The presence of such a correlation would substantially strengthen confidence in the result, since it would entail that participants who respond favourably to TMS in the behavioural domain also show a larger TMS modulation of their BOLD signal. Alas, I found no such correlation for the two significant ROIs (SPL: r = -0.06, p = 0.81; Thalamus: r = -0.21, p = 0.37) (figure 8).

2.3.2 Whole-brain

In order to discern any brain areas whose BOLD signal may have been modulated by cTBS, but that did not appear in the ROI analysis, I also performed a whole-brain survey. As a first step, I charted how the signal associated with SFM button presses was affected by vertex cTBS. We found no suprathreshold voxels deactivated by cTBS, but several small clusters that showed increased activity post TMS (figure 9, I., table 3, I.). Given their spread, the liberal alpha threshold of 0.001 uncorrected, as well as the result that none of the clusters are statistically significant following cluster-correction, I interpret these as noise. In contrast, when examining IPS stimulation, I found

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.03 −0.02 −0.01 0.00 −50 0 50 100

percent change in dominance

modulation of percent signal change

SPL ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.02 −0.01 0.00 −50 0 50 100

percent change in dominance

modulation of percent signal change

no activation, but instead inhibition of BOLD activity in the bilateral thalamus, left V1 as well as right posterior medial frontal cortex (figure 9, II., table 3, II.). Finally, I turned to the main contrast, namely BOLD modulation following IPS stimulation, with the vertex (TMS site unspecific) effect subtracted (I. > II.). This revealed that IPS cTBS inhibited BOLD signal in the thalamus, left V1, left middle frontal gyrus and right posterior medial frontal cortex (figure 9, III., table 3, III.).

Figure 9 — Whole-brain fMRI results

Brain areas showing differential activation post vs pre TMS when vertex was stimulated (top left) and when IPS was stimulated (button left). Decrease in BOLD activity when IPS is stimulated with the vertex effect subtracted, when on sagittal slices of the MNI T1-template (right), directed interaction contrast between vertex-pre, vertex-post, IPS-pre, IPS-post [1 -1 -1 1]. Threshold p < 0.001 uncorrected.