The results presented here accord well with those previous studies that have reported a role for the RIFG in the inhibition of pre-potent responses during the GNG and SST paradigms (Menon et al., 2001, van Boxtel et al., 2001, Aron et al., 2003a, Aron et al., 2003b, Rubia et al., 2003, Aron et al., 2004, Picton et al., 2007, Verbruggen and Logan, 2008
). However, we found no evidence to support the hypothesis that the RIFG plays a unique or specialised role in inhibition and furthermore, the data run counter to this hypothesis in a number of key respects.
Firstly, if the RIFG were involved in inhibitory control specifically, then the BOLD response should have increased specifically during the inhibition of a pre-potent response. Instead, the counting of cues, the initiation of responses, and the inhibition of responses all activated the RIFG. Thus, from a functional perspective, it is more parsimonious to state that the RIFG responds whenever salient cues that have a bearing on the current task plan are detected (Hampshire et al., 2009
). Secondly, we found no evidence to support the idea that “inhibition is localized to the RIFG alone” (Aron et al., 2004
), but instead, the inhibitory manipulation in the SST recruited a network of brain regions including the IFG bilaterally, the preSMA, and the IPC bilaterally. Importantly, these additional brain regions were also co-recruited during the COUNT and RESPOND conditions. Finally, we did not find evidence that the IFG interacts with the STN to suppress initiated motor responses (Aron and Poldrack, 2006
) as no significant increase in BOLD response was observed in the INHIBIT condition. As the STN was highly active during the RESPOND condition, it seems probable that the lack of an observed effect during INHIBITION condition is due to the STN being involved to a similar extent during both motor generation and motor suppression in that block.
Whilst the results of the current study run counter to the hypothesis that the RIFG is specialised to inhibition and plays a unique role in inhibition alone, there are a number of more complex possibilities. For example, one could suggest that neurons within the RIFG were sub-divided into two distinct populations, one coding for task relevant cues, and the other for inhibitory outputs. This seems unlikely, as a disproportionate BOLD response to cues would have been predicted in the inhibition block compared with the two control conditions. Indeed, in the current task design we actively biased towards finding such a result, as the inhibitory condition was clearly a much more demanding manipulation. Instead, an increased response was seen during both response generation and response inhibition when compared with counting. Alternatively, one could argue that there were at least three distinct overlapping populations within the right IFG, one coding for cue detection, another coding for the generation of motor responses, and another coding for inhibitory control. The differential recruitment of these two additional populations could somehow conspire to mask the BOLD response of the inhibitory population. It becomes necessary, however, to complement this increasingly complex model with an increasing degree of specificity in order to explain the observed data. For example, as the BOLD response at response generation is equivalent to that at response suppression, the motor generation sub-population would have to be recruited only when responses were infrequent during RESPOND or to responses in general but to exactly half the extent as those coding for response inhibition.
Even this most complex and specified model cannot account for recent findings from the broader literature. For example, one recent study sought to examine the inhibitory control condition in the SST using functional connectivity (Duann et al., 2009
). That study demonstrated in a large cohort that there is no evidence for a direct inhibitory influence from the RIFG on components of the motor system such as the STN and the pre motor cortex whilst undertaking the SST. Instead the RIFG appeared to exert its influence over the motor system via potentiating inputs to the preSMA.
As these examples show, to account for the broad range of circumstances in which the RIFG is active, the inhibitory hypothesis must postulate hidden inhibitory components for which often there is no direct evidence. A simpler hypothesis, which we suggest can offer a more parsimonious explanation of both the data presented here and its relationship to the broader literature on executive control, is that the RIFG, along with the left IFG, the PPC and the preSMA, form a network that rapidly tunes to represent those inputs and responses that form the currently intended task schema. Thus, the response within this network is particularly strong when cues are detected that trigger effortful and task relevant behaviours.
Research into the role of right lateral prefrontal cortex–a likely analogue of the RIFG in non-human primates (Petrides and Pandya, 2002
)–offers clues regarding the probable neural mechanism by which this brain region exerts executive control. When single unit recordings are taken from this region, neurons display a highly adaptive profile, with a large proportion rapidly adapting to respond to the currently relevant stimuli, stimulus dimensions, and responses (Freedman et al., 2001, Miller and Cohen, 2001, Everling et al., 2002, Duncan, 2006
). Also, neurons in this region that respond to task specific information continue to respond when that information is being actively maintained over a delay (Fuster and Alexander, 1971, Funahashi et al., 1989, Rao et al., 1997
). Such maintenance activity may modulate stimulus processing in specialised regions of posterior cortex (Chelazzi et al., 1998, Kastner et al., 1999
). Finally, many brain regions exhibit inhibition at a local level, with inputs competing for limited capacity processing resources (Duncan, 2006
). Herein lies a potential key to the likely neural process by which inhibitory control is exerted. In terms of visual processing, inhibition of one object when attention is focused on another can be explained as a secondary effect, i.e., an emergent property of local competition when one competing item is subjected to top-down potentiating signals which have their source in the lateral prefrontal cortex (Norman and Shallice, 1980
). Thus, when considering the case of selective attention in the face of increased distraction, it may well be the case that increased inhibition of distractors is achieved simply by actively focusing more willfully on that which is attended as opposed to directly suppressing that which is distracting. It seems reasonable to suggest that at an executive level stopping and going in the SST task are represented as two alternate behaviours. If these two representations are competing for processing resources, then focusing attention on one will tend to inhibit the other.
Accordingly, from a quantitative perspective, the response inhibition manipulation in the GNG and SST tasks is often conceived in terms of alternate stop and go processes that compete for the earliest completion time–the horse race model (Logan and Cowan, 1984, Band et al., 2003
). If the routine go process executes before the infrequent stop process, then inhibition fails. Another perspective on this relationship is that the SST task schema is represented in two alternate action plans that compete to be executed. The stop plan is executed frequently and becomes routine and dominant, requiring minimal effort and minimal executive control from the IFG/PPC/preSMA network. Processing of the plan to stop, by comparison, is not routine and requires effortful monitoring for the stop cue and application of a top-down biasing signal in order to allow it to win the competition for execution. From the attentional tuning perspective, the contribution of the RIFG to inhibitory control can be considered akin, therefore, to that made when responding to or counting infrequent targets, i.e., the effortful maintenance and execution of a planned behaviour.
The relationship between potentiation and inhibition also becomes apparent when considering some of the other task manipulations that have previously been cited as evidence for a specific role for the RIFG in inhibitory control. For example, it has been noted that when suppressing intrusive thoughts, activity within the RIFG increases and that the RIFG appears to interact with regions of the temporal lobe that are known to be crucial to memory (Anderson et al., 2004
). On the surface, this appears to be good evidence that the RIFG is directly suppressing the representation of unwanted thoughts and memories within the temporal lobes. An alternative explanation for these findings, however, is that the RIFG is engaged in a coping strategy–for example retrieving an alternative thought or memory in order to swamp limited capacity processing resources. From a phenomenological perspective, this seems rather more likely, as when attempting not to think about something, an individual will typically try and think about something else, whilst experimentally, it is well established that trying to push a negative thought away directly can lead to increases in both the frequency and emotional impact of that thought (Wegner et al., 1987, Wegner, 1989, Purdon and Clark, 1999, Wenzlaff and Wegner, 2000
This latter take on the inhibition of thoughts ties in rather closely with the well established role of the IFG in the deliberate formation and retrieval of information in long term memory (Dove et al., 2006
). Previously, it has been suggested that the RIFG is recruited under these conditions as it becomes necessary to inhibit other memories when attempting to encode or retrieve a target memory (Aron et al., 2004
). By contrast, potentiating some sub-portion of the neurons that form the memory, and then allowing the rest of the memory to activate via a process of pattern completion is far closer to the generally accepted view of active memory retrieval via cues (Henson et al., 1999
). This interpretation is also analogous to the observation that the IFG is not just recruited during selective retrieval of semantic information, where inhibition of competing representations may be necessary, but also during the effortful retrieval of semantic information in general (Wagner et al., 2001
Finally, there is strong evidence that the RIFG plays a role in attentional switching (Dove et al., 2000, Cools et al., 2002, Hampshire and Owen, 2006
)–the process by which the focus of attention is moved from one locus to another (Monsell, 2003
). Again, on the surface, the inhibitory control hypothesis seems well able to account for the role played by the RIFG in attentional switching. The suggestion would be that the RIFG facilitates the attentional switch by inhibiting the previously attended object, location, or dimension, thereby allowing attention to shift away. It could be predicted from this account, that when switching attention away from a previously rewarded and routine response, for example during reversal learning, a much greater degree of inhibition should be necessary in order to overcome the pre-potent response. Whilst it is the case that the RIFG is recruited during reversal learning (Cools et al., 2002
) it has recently been reported that (unlike the lateral orbitofrontal cortex) activation within the RIFG is no greater at the point of a reversal than when switches are carried out between previously unrewarded objects, none of which can be considered pre-potent (Hampshire and Owen, 2006, Chaudhry et al., Under Submission
). To complicate the issue further, the extent to which the RIFG is recruited during attentional switching does appear to relate to the visual difference between the current and previous stimuli–so whilst switches of attention between similar objects recruit the RIFG more than non-switches, switches between objects drawn from different categories recruit RIFG to an even greater extent (Hampshire and Owen, 2006, Hampshire et al., 2008a
). Overall these findings accord best with a role for the RIFG in reconfiguring a representation of the currently attended input–a role that may be shared with other regions of the frontoparietal network including the IPC.
It undoubtedly remains the case that the GNG and SST paradigms are robust markers of RIFG function and in this respect they provide powerful tools for investigating the neural basis of executive dysfunction (Rubia et al., 1999, Aron et al., 2003b
) and measuring the efficacy of pharmacological interventions (Aron et al., 2003a, Chamberlain et al., 2009
). Also, one cannot entirely rule out the possibility that a sub-population of neurons within the RIFG work to exert an inhibitory influence over processes within other brain regions. However, what is clear is that the IFG plays a more general role in executive function than just the exertion of inhibitory control. Thus, the results from the tasks such as the GNG and the SST should not be over interpreted in terms of neural inhibition. For the future, a pertinent question is whether patient groups that perform poorly on GNG and SST tasks can be sub-divided according to whether the underlying impairment is an inability to maintain attention when looking for cues, or an inability to suppress a response when the cue is detected.