Figure 3: (A) The left column shows the significant cluster in
cortical source space of the comparison between the “attend to both
rings” and “attend to the central fixation cross” pre-cue conditions.
The upper panel of the right columns depicts the mean relative power
across the sources pertaining to the significant cluster. The lower
panel shows the corresponding paired observation plot for mean
normalized values. Red lines represent participants with a difference
exceeding the lower bound of the 95% confidence interval of paired
differences. Grey lines represent differences that did not meet this
criterion. (B) The upper panel shows the mean power across the
significant source cluster of (A) for the right and homologous left
hemisphere regions. The lower panel depicts the corresponding paired
observation plot for mean normalized values, so that individual
differences can be observed at the same scale. Red lines for the right
hemisphere represent participants with a difference exceeding the lower
bound of the 95% confidence interval of differences. Grey lines
represent data that do not meet this criterion. Red lines for the left
hemisphere represent participants with a difference within the 95%
confidence interval including zero (indicating the absence of an
effect). Grey lines represent differences outside these boundaries.
Error bars in all subplots indicate the s. e. m. ri = rings, cr = cross.
However, the main objective of this study was to characterize how
stimulus driven ssVEF responses are modulated by shifts of spatial
attention when spatial attention is already directed towards the
peripheral or central visual field (rings vs. cross). Therefore, we
assessed the interaction of relative ssVEF power changes across
experimental phases (pre-cue baseline, left or right visual hemifield
not attended or attended) and pre-cue baseline condition (rings vs.
cross). The nonparametric cluster-based permutation test based on
previously formed F clusters (see methods) for the left visual hemifield
(lvf) stimulation indicated an interaction between the pre-cue baseline
condition (attend rings vs. central fixation cross) and experimental
phase (pre-cue baseline, lvf not attended, lvf attended). The
significant source cluster encompassed V1, V2, hMT+, the
occipito-parietal and inferior-temporal visual cortex of the right
hemisphere (summed F value = 1044, p < 0.001, maximum
parametric F(2, 38) = 9.71, p < 0.001, ε = 0.84,
η2 = 0.34; Figure 4A). However, right visual hemifield
stimulation (rvf) resulted in an interaction between pre-cue baseline
condition and experimental phase only at trend level localized in left
V1, V2, and a small part of the occipital-parietal region (summed F =
137, p = 0.084; maximum parametric F (2, 38) = 8.09, p = 0.004, ε =
0.73, η2 = 0.30, Figure 4A).
Given the height of the left hemispheric cluster (F(2, 38) = 8.09) the
trend level cluster based significance was probably due to the smaller
extension of the source cluster. Therefore, in order not to rule out a
significant interaction for the rvf stimulation in the left hemisphere,
the mean ssVEF power values across the significant source clusters of
the left and right hemispheres were entered in a repeated measurements
ANOVA with the within-subject factors pre-cue baseline, experimental
phase and hemisphere. An absent interaction between the pre-cue baseline
condition and experimental phase in the left hemisphere should result in
a pre-cue baseline by experimental phase by hemisphere interaction.
However, no such interaction was observed (F(2, 38) = 1.66, p = 0.20, ε
= 0.93, η2 = 0.08). Across hemispheres, the
interaction pre-cue baseline by experimental phase indicated that ssVEF
power was differently modulated depending on the pre-cue baseline
condition (F(2, 38) = 8.57, p < 0.001, ε = 0.71,
η2 = 0.31). Figure 4B depicts the mean relative ssVEF
power changes across all significant cortical sources pertaining to the
previously identified source clusters in the left and right cortical
hemisphere for all experimental phases (pre-cue baseline, post-cue not
attended, and post-cue attended) and the two different pre-cue baseline
conditions (attend rings vs. fixation cross). Whereas the participants
attended the peripheral rings during the pre-cue baseline task, the
ssVEF responses across both cortical source clusters were enhanced for
the pre-cue baseline and post-cue attend experimental phases compared to
the unattended visual hemifield (pre-cue vs. unattended: t(19) = 3.80, p
= 0.001, Cohen’s d = 0.85; attended vs. unattended: t(19) = 4.24, p
< 0.001, Cohen’s d = 0.95). However, when the participants
attended to small variations of fixation cross sizes during the pre-cue
experimental phase, relative ssVEF power was similar to the ssVEF
response elicited by the unattended hemifield stimulation during the
post-cue phase (t(19) = 1.68, p = 0.110, Cohen’s d = 0.37). In contrast,
attending the cued hemifield during the post-cue phase provoked enhanced
ssVEF responses in comparison when the hemifield was not attended (t(19)
= 4.43, p < 0.001, Cohen’s d = 0.99). Furthermore, when we
tested the activation level for the attended side between the conditions
cross vs. rings we found a trend for higher activation when subjects
attended to central fixation first (t(19) = 1.93, p = 0.070, Cohen’s d =
0.43).