Context-dependent modulation of cognitive control involves different temporal profiles of fronto-parietal activity

Highlights

• When cognitive control is rarely needed, it is reactively implemented.

• Reactive control involves on-trial transient activation of fronto-parietal areas.

• In contexts that require control frequently, proactive control is implemented.

• Proactive control entails intertrial activity in areas also active in reactive control.

Abstract

To efficiently deal with quickly changing task demands, we often need to organize our behaviour on different time scales. For example, to ignore irrelevant and select relevant information, cognitive control might be applied in reactive (short time scale) or proactive (long time scale) mode. These two control modes play a pivotal role in cognitive-neuroscientific theorizing but the temporal dissociation of the underlying neural mechanisms is not well established empirically. In this fMRI study, a cognitive control task was administered in contexts with mainly congruent (MC) and mainly incongruent (MI) trials to induce reactive and proactive control, respectively. Based on behavioural profiles, we expected cognitive control in the MC context to be characterized by transient activity (measured on-trial) in task-relevant areas. In the MI context, cognitive control was expected to be reflected in sustained activity (measured in the intertrial interval) in similar or different areas. Results show that in the MC context, on-trial transient activity (incongruent – congruent trials) was increased in fronto-parietal areas, compared to the MI context. These areas included dorsolateral prefrontal cortex (dlPFC) and intraparietal sulcus (IPS). In the MI context, sustained activity in similar fronto-parietal areas during the intertrial interval was increased, compared to the MC context. These results illuminate how context-dependent reactive and proactive control subtend the same brain areas but operate on different time scales.

Introduction

When our GPS and the road signs point us in opposite directions, we need to overcome automatic tendencies (e.g., adhering to an outdated GPS) in favour of more appropriate responses (e.g., following the road signs). Exerting such cognitive control is crucial in everyday life but also cognitively demanding (Kool et al., 2010; Shenhav et al., 2017; Vassena et al., 2014; Verguts et al., 2015). Adaptive allocation of cognitive control therefore requires a balance between exerting a sufficient amount (to solve hard tasks), but not too much (to spare costly cognitive effort). As a result, cognitive control is best applied sparsely and should only be sustained when it is frequently needed (Botvinick et al., 2001; Ridderinkhof, 2002; Braver et al., 2007; Braver, 2012; Jiang et al., 2014).

One way to minimize cognitive expenses while maintaining acceptable performance levels is by applying cognitive control on different time scales. This is mirrored in the way incongruent (i.e., difficult) trials are handled in typical cognitive control tasks. Examples of incongruency can be found in the Stroop task (Stroop, 1935), where word colour must be ignored in favour of word meaning, and in the flanker task (Eriksen and Eriksen, 1974), where one should respond to the central target and ignore the flankers. When incongruent trials are rare, reactive control is thought to be active, meaning that control operates in a just-in-time regime (short time scale). Reactive control is thought to be implemented by transient reactivation of task-relevant brain areas (Braver, 2012; Bugg and Crump, 2012; Logan and Zbrodoff, 1979). On tasks with mainly incongruent trials, this transient strategy could lead to frequent errors and delays. Here, a proactive control mode is optimal, operating on a long time scale.

It has been proposed that the brain regions activated during proactive control show anatomical or functional overlap with those activated during reactive control. For example, similar or closely related regions in lateral PFC have been suggested to be involved in reactive and proactive control depending on the specifics of task demands (Braver, 2012; De Pisapia and Braver, 2006). First preliminary support for this context-dependent two-mode (i.e., reactive and proactive) theory of cognitive control came from studies that compared cognitive control between mainly congruent (MC) and mainly incongruent (MI) contexts. In MC contexts, transient (e.g., on-trial) activity in anterior cingulate cortex (ACC) and other fronto-parietal areas is typically higher on incongruent than on congruent trials (Carter et al., 2000; De Pisapia and Braver, 2006; Grandjean et al., 2012; Jaspar et al., 2016). This is taken as an instance of reactive control and results in slow response times (RTs) and low accuracy on incongruent compared to congruent trials. In MI contexts, the same areas are often found to be activated but the difference between incongruent and congruent trials is less strong or even absent (Carter et al., 2000; De Pisapia and Braver, 2006; Grandjean et al., 2012; Marini et al., 2016). This is interpreted as an indicator of sustained control across all trials (congruent and incongruent) in proactive control mode, reducing or eliminating neural and behavioural differences between trial types.

Another way of indexing time scale differences in cognitive control is through hybrid fMRI designs that combine block- and event-related responses (Petersen and Dubis, 2012; Visscher et al., 2003). Here, reactive control is indexed by transient activity on the trial level and proactive control by sustained activity on the block level. Using these designs, the same fronto-parietal areas have been found active in transient and sustained manners, depending on whether the context was MC or MI (Braver et al., 2003; Marini et al., 2016). However, similar designs have also led to claims that two control modes indeed exist but that they not only differ in temporal activation profile but also comprise different brain areas (e.g., Olsen et al., 2010; Seeley et al., 2007). For example, Dosenbach et al. (2008) suggested that transient adjustments in control are initiated by a fronto-parietal network including lateral prefrontal and superior parietal cortices. Sustained control, on the other hand, was supported by a cingulo-opercular network comprising dorsal ACC and the anterior insula. Consistent with this dual-network perspective on cognitive control, Wilk et al. (2012) also showed that transient and sustained activity can arise from different brain areas. However, in this case, the ACC, anterior insula, and inferior parietal cortex showed transient activity, while sustained activity was found in medial superior frontal gyrus.

In the current study, we adapted the event-related fMRI paradigm to investigate time scale differences in cognitive control. Typically, event-related paradigms measure on-trial activation and are therefore informative about transient, but not sustained activation. To address this issue, we used activity measured in intertrial intervals as a proxy for sustained activation to identify how cognitive control operates in contexts with mainly congruent or mainly incongruent trials. Thus, “active” flanker trials were interleaved with “blank” trials that consisted of a prolonged fixation cross, which allowed measuring context-dependent intertrial activation (Horga et al., 2011). This method offers a novel perspective on two-mode theories about cognitive control and tests whether cognitive control can indeed be allocated with different temporal profiles (reactively and proactively) depending on the context (MC or MI), and whether this involves similar or different brain areas. The blank trials have the additional advantage that they allow measuring activation independent of differences in stimulus-response contingencies, trial difficulty, motor response, accuracy, or response time (Grinband et al., 2011; Schmidt and Besner, 2008).

If context-dependent cognitive control is indeed differently implemented through transient and sustained neural activation, then two neural patterns can be predicted. First, increased on-trial (i.e., transient) activity in typical cognitive control areas, such as dorsolateral prefrontal cortex (dlPFC), dorsal anterior cingulate cortex (dACC), and parietal cortices (Corbetta and Shulman, 2002; Dosenbach et al., 2008; Niendam et al., 2012) is expected on incongruent compared to congruent trials in the MC block but not in the MI block. Second, increased blank-trial activation is expected in the MI block compared to the MC block, reflecting the sustained recruitment of cognitive control in the MI block. Transient and sustained control activation might be located in similar (Braver, 2012; Kerns et al., 2004; Marini et al., 2016) or different (Dosenbach et al., 2008; Olsen et al., 2010; Wilk et al., 2012) brain areas.