In the current study, we disrupted function in a PFC control region, the right IFJ, with rTMS and assessed the consequences on top-down modulation in posterior cortical regions and working memory performance. Results indicated that top-down modulation during very early visual processing stages of the memoranda and subsequent working memory performance were causally related. There are four results supporting this conclusion. First, color processing showed an rTMS-related decline in both P1 modulation and working memory accuracy. Second, as the P1 modulation recovered with time (i.e., in the second half of the block), so did working memory performance. Third, on an individual participant basis, the rTMS-induced effect on the P1 modulation during color processing predicted changes in working memory accuracy. Finally, motion processing, which exhibited bilateral connectivity, did not show rTMS-related effects on P1 modulation or working memory accuracy. Thus, we conclude that early top-down activity modulation during stimulus processing driven by attentional demands is causally related to subsequent working memory performance (see Supplementary Discussion
for further justification). Notably, the data revealed the IFJ to be a PFC control region that mediates this causal connection between top-down modulation in the service of attentional goals and subsequent working memory performance. Moreover, top-down modulation may be subserved by alpha phase coherence for long-distance communication.
Although functional connectivity analysis is informative in characterizing neural networks that may be involved in top-down modulation, it is a correlational measure and cannot be used to make strong statements of causality. In recent years, rTMS has been coupled with physiological recordings to provide causal evidence in humans that fronto-parietal regions are a true source of top-down modulation of activity in sensory cortical areas. However, only a few of the fronto-parietal regions proposed to be cognitive control regions underlying top-down modulation have been evaluated with this approach. TMS coupled with neuroimaging has been used to show that visual ERPs are under top-down influences from frontal eye fields (FEF)20, 35
and posterior parietal cortex (PPC)36
. Top-down influences operate on areas V1 through V5 and have differential roles based on whether frontal (FEF) or parietal (interparietal sulcus) cortex is the source of modulation22
. Research within the last two years has indicated that neural modulation occurs early during visual attention to features (~100 ms post stimulus onset)12, 26
. Here, we contribute to this literature by showing that the right IFJ modulates early visual cortical processing. Specifically, both enhancement and suppression of the P1 amplitude for relevant and irrelevant color stimuli, respectively, were diminished post-rTMS. This suggests that suppression is not simply the lack of enhancement. Indeed, recent research has shown a selective decline in older adults' ability to suppress irrelevant information, suggesting that enhancement and suppression rely on distinct mechanisms37
. An interesting question remains as to how one frontal region may be involved in both enhancing and suppressing posterior neural activity based solely on the goals of the task. One possibility is that functional segregation of the IFJ may exist at a scale smaller than the area impacted by rTMS, consistent with the notion of a topographical organization and functional specialization of this region38
Despite the largely distinct fronto-parietal networks associated with attention to color and motion features, the right IFJ was common to both networks (albeit neighboring subregions within the IFJ). Recent research has shown the IFJ to be involved in many different tasks, including task switching, interference control and working memory39
. As such, it has been hypothesized that the IFJ plays a role in updating relevant task representations40
. The current data supports this hypothesis and extends it by suggesting that updating relevant task representations may occur via goal-directed biasing of neural activity in distal cortical regions. Furthermore, we show here that this neural biasing optimizes working memory performance. Interestingly, the IFJ and other frontal cortical regions display topographic organization and functional specialization across multiple spatial tasks, including memory-guided saccades, spatial working memory, and finger pointing38
. Thus, spatial encoding cannot be discounted as an alternative hypothesis for the proposed role of the IFJ. However, given the many different tasks that have now been shown to utilize the IFJ that do not rely on spatial information39, 40
, it seems more plausible that the IFJ has a generalized role in updating task representations. This is not to imply a lack of topographical organization within the IFJ. Color and motion processing engage ventral and dorsal visual processing streams, respectively, and this domain specificity is thought to be maintained in frontal regions41
. The current data replicated our previous report that IFJ activity during color processing is localized ventrally within the IFJ relative to regions engaged for motion processing24
, thereby supporting selective topographical organization and functional specialization of this region.
While long-distance cortical communication seems likely to occur via phase information42
, evidence that top-down modulation utilizes such a mechanism has only recently been appreciated. Alpha band activity appears to be a prime candidate for such processes, as it is modulated by attention43
and linked to fronto-parietal networks subserving attention21
and working memory44
. Here, we show pre-stimulus alpha-band phase coherence between frontal and posterior regions is modulated by attention and that perturbing function in the IFJ disrupts this modulation. These results support recent findings that top-down modulation mediated by PFC networks bias activity in sensory cortical regions prior to stimulus onset24, 45
and extends it to phase coherence being an operational mechanism. Given that all stimuli within each trial were presented randomly, anticipatory top-down modulation must have occurred as a cognitive set, as opposed to “switching” between enhancement and suppressive modes. A recent review has suggested alpha phase synchronization may be interpreted in terms of processes that coordinate top-down control and access to memory traces27
. Here, we provide direct evidence to support this hypothesis by showing that alpha phase coherence is modulated by attention and that IFJ rTMS disrupts alpha phase coherence modulation.
Based on the current results and previous research, we hypothesize that when participants are instructed to remember color, the right IFJ biases visual cortical regions via alpha phase coherence prior to stimulus onset. Anticipatory biasing allows visual information (i.e. color features) to be enhanced or suppressed early
in the visual processing stream, depending on their relationship to task goals. This contrast between enhanced and suppressed activity yields increased fidelity of the representation of the memoranda, which leads to improved working memory performance. The current study identified P1 modulation as a bridge between selective attention and working memory performance, and lends support to the supposition that selective attention and working memory encoding may not be dissociable at the neural level46
. However, this may not be true in all cases. Attention operates during both perceptual and post-perceptual stages of stimulus processing, and its influence on working memory may depend on the timing of activity modulation, as well as the type of attention and working memory task47
. Although the current results were identified during the encoding period, this does not preclude that top-down modulation during other stages of the task (e.g., the delay period) influence working memory performance. Furthermore, it should be noted that the IFJ is only one region identified in widespread fronto-parietal neural networks utilized by selective color and motion processing. Future research will assess the necessity of other putative control regions within the frontal and parietal cortex.