We applied TMS over the right AG, or a control vertex site, during an event-related, exogenously cued visuospatial attention task. Rather than comparing TMS vs. none, we made the closer comparison of high- vs. low-intensity TMS at each site. Behaviourally, we found that (high- vs. low-intensity) TMS over the right AG selectively affected perceptual discrimination of right but not left visual targets, both outside and inside the scanner, in a manner that strongly depended on cue validity. Specifically, with the present TMS protocol in the present paradigm, high- vs. low-intensity TMS over the right AG boosted accuracy specifically for invalidly cued right targets, but not for validly cued targets (see ). Thus, stimulation of the right AG evidently facilitated rightward spatial reorienting, for invalidly cued right targets. See also Hilgetag et al. (2001)
and Chambers et al. (2006)
for discussion of previous effects of parietal TMS on ipsilateral visual targets, and also for the more general point that TMS need not always lead to impairments of performance, but can also lead in some cases to specific improvements, as here.
By combining TMS with concurrent fMRI (Bohning et al., 1999
; Baudewig et al., 2001
; Ruff et al., 2006
; Sack et al., 2007
; Bestmann et al., 2008a
; Driver et al., 2009
), we were also able to examine the impact of our TMS manipulation on brain activity during task performance. This included any possible impact on areas remote from but interconnected to the right AG, and whether any such effects on brain activity might depend on the current attentional condition. With this approach, we found interhemispheric effects of the TMS (see also Bestmann et al., 2008b
; Blankenburg et al., 2008
) that depended strongly on the current attentional state. In particular, activity in the left AG (homologous to the targeted TMS site) was dependent on the interaction between TMS intensity, target hemifield and cue validity (see ), showing a similar three-way interaction as for the behavioural data. High-intensity TMS systematically increased the BOLD signal in the left AG for invalidly cued right targets (whereas the BOLD response here decreased instead for invalidly cued left targets). This pattern was specific to invalidly cued trials, did not occur when targets were validly cued, and survived cluster correction for the whole brain.
The BOLD response of the visual cortex was studied in more detail via the individual retinotopic mapping of voxels responding to our target stimuli in mapped ventral V1–V3. This allowed us to determine whether right AG TMS might exert remote, condition-specific effects on the early visual cortex, and whether any such remote effects might be analogous to the pattern shown for visual performance. As for the remote BOLD effects in the left AG, target-responsive retinotopic ROIs in the left hemisphere (contralateral to right targets) were reliably modulated by TMS of the right AG, again in a manner that depended closely on the current attentional condition. Specifically, the BOLD response of left ventral V1 and V2 to invalidly cued right targets was enhanced during high-intensity TMS (see ), in accordance with the improved performance for the same condition. Again, in line with the behavioural impact of TMS, no such effect was observed for left targets or on validly cued trials. Neither was the effect observed in dorsal visual areas (non-responsive to our upper-hemifield targets), being specific to the target-responsive retinotopic cortex instead. These remote TMS effects on the visual cortex suggest that the role of the right AG during reorienting of attention may involve modulation of stimulus responses in relatively early visual regions, possibly via the posterior cingulate cortex and left AG.
One potentially noteworthy aspect of the remote, condition-dependent TMS effects on the left ventral visual cortex (responding to right targets) was that the TMS effect appeared stronger (see , rightmost three bars) for the early visual cortex (V1) than for later areas (V3). This appears to be different from the standard findings of (TMS-unrelated) attentional modulation of visual responses that are typically stronger for later than earlier visual areas (cf. Kastner et al., 1998
). However, the present outcome appears potentially to be consistent with previous findings that some remote TMS effects can appear stronger for earlier than later areas in the visual cortex (e.g. Ruff et al., 2006
). Future extensions of the approach pioneered here could test more thoroughly how condition-specific remote TMS effects upon the visual cortex may vary for different levels of the cortical hierarchy, including use of more extensive mapping of visual areas than was possible here given the time constraints and scope of our study.
Taken together, our existing results already indicate a critical role for the right AG as one source of top-down visuospatial attention (see also Silvanto et al., 2008
), especially when reorienting of attention is required to the ipsilateral side. Our concurrent TMS–fMRI data revealed interhemispheric influences on the opposite parietal cortex, and modulation of target representations in the early visual cortex, which depended on the current event-related attentional condition. These data indicate that the behavioural consequences of parietal TMS may not depend solely on changes in local activity, but potentially also on modulation of remote interconnected brain regions (Ruff et al., 2006
; Sack et al., 2007
; Bestmann et al., 2008a
; Driver et al., 2009
). Here we showed that the remote interhemispheric consequences of right AG TMS alter in a dynamic fashion with the event-related state of attention.
In apparent contrast to the left AG, the BOLD signal in the stimulated right AG did not show a fully significant high-level three-way interaction between TMS intensity, target hemifield and cue validity, perhaps because local field distortions can potentially reduce the sensitivity of the fMRI signal immediately beneath the TMS coil (see Bestmann et al., 2008a
). Nevertheless, within a spherical ROI for the stimulated region of the right AG, we did observe an interaction of TMS intensity with cue validity, for right targets in particular, thus representing a somewhat weaker variant of the left AG pattern. Future research with increased power might be able to compare the left and right AG more directly.
It has long been suspected that interhemispheric interplay might be crucial for orienting of attention, based primarily on clinical inferences from brain-damaged patients hitherto (e.g. Kinsbourne, 1977
). The present study produced a new type of evidence for the role of interhemispheric interplay in the normal brain, by uncovering left-hemisphere effects due to right AG TMS, which showed some analogy to performance effects of the same TMS for right targets contralateral to the left hemisphere. Critically, these effects were dynamic in that they depended on the current attentional state, excluding a simpler interhemispheric inhibition account (see also Kinsbourne, 1977
). A potentially fruitful extension of our study would be to determine whether the left AG can have corresponding influences upon right-hemisphere structures, or whether instead the direction of causal influence in redirecting attention stems mainly from right-hemisphere influences upon the left hemisphere. Resolving this issue was beyond the scope of the present study, but might be addressed in future research now that we have provided a proof-of-principle for interhemisphere remote TMS effects, which depend on the current attentional condition. It may be that the right AG enjoys a privileged role in controlling reorienting of attention (see Corbetta et al., 2000
; Rushworth et al., 2001
; Macaluso et al., 2002
; Mort et al., 2003a
; Chambers et al., 2004
; Thiel et al., 2004
; Kincade et al., 2005
; Vuilleumier et al., 2008
), whereas the left AG, which we found to be influenced here, might be critical for responses to right targets in particular, if, as has been suggested, the left parietal hemisphere selectively represents contralateral space (Mesulam, 1999
Regardless of such further considerations and possibilities for future experiments, the present data provides an existence-proof that the impact of TMS on behaviour and on remote brain areas (here in the opposite hemisphere to that stimulated) can depend critically and dynamically on event-related attentional conditions [cf. Blankenburg et al. (2010)
, who could only study blocked attentional conditions in their TMS–fMRI study]. Our findings illustrate that concurrent TMS–fMRI can provide a new approach to studying how causal interplay between brain regions may vary with the current cognitive state, in this case highlighting the role of interhemispheric influences for redirecting the spatial distribution of attention.