Mechanisms of cognitive control
At first glance, two great hallmarks of cognitive control appear to be in opposition. First, cognitive control must be selective: in a capacity limited system, processing of task-relevant information must be prioritised above the rest. Second, control must be flexible. After selectively attending to one set of information in one moment, we must be able to shift to a new set of information in the next, as we move through our task and mental focus changes. Our research advances the proposal that these features are two sides of the same neural coin, arising from a single neural system - the “multiple demand” or MD network - that drives selective, yet flexible, processing of task relevant information.
We study the temporal dynamics and causal interactions of information coding in and between the MD and other brain networks that might give rise to this ability. Current lines of research focus on whether cognitive control is achieved through selection vs. inhibition of information, developing new methods to track information exchange between networks and the flexible mechanisms underpinning our ability to solve complex task by scheduling series of simpler "attentional episodes“.
Selection or inhibition?
Selective attention is necessary to prioritise processing, in our capacity-limited brain, in favour of what is currently important. There is a long-standing debate regarding whether attention affects neural activity by enhancing the representation of information that is relevant to us, and/or by inhibiting the representation of distracting information. Current projects use transcranial magnetic stimulation (TMS) - a technique that temporarily changes neural activity - in combination with fMRI to examine the causal role of the multiple demand system in attentional selection. We examine the consequence of disruptive TMS on information coding elsewhere in the brain: does it affect processing of attended information, or the information we are trying to ignore? We are also implementing classic selective attention-based tasks, such as the Stroop, to understand how neural codes change in specialised cortices when participants attend to or ignore the current input.
Information Flow Analysis
Beyond knowing what information is encoded in which brain region and when, we seek to understand how information is exchanged between networks as information flows from input to output in behaviour. Using the logic of granger causality applied to multivariate MEG data, we developed a method to track how much information coding in one brain region is influenced by information coding in another. We have found that the timecourse with which information flow changes from predominantly feedforward (occipital to frontal) to feedback (frontal to occipital) matches the timecourse with which attention affects coding in visual cortices. Ongoing projects seek to develop the mathematical approach, garner detail on the spatial sources of feedback, and understand what type of information is exchanged under different task demands.
Every day, we flexibly perform an extraordinary range of tasks. The theory of ‘attentional episodes’ proposes that we do this by breaking each task down into simple steps, and addressing those steps in sequence. Within an episode, we need to selective focus on the currently relevant information. Between episodes, we need to rapidly reconfigure our attentional system and re-orient to what is now relevant. Current projects use time-resolved neuroimaging methods (MEG/EEG), encoding and decoding models, and behaviour, to probe the temporal limits of this reconfiguration in the brain. We ask how effectively the multiple demand system can reconfigure within a task, under what circumstances it does so, and whether there are individual differences in the tendency to break a task into episodes or try to solve it as a whole.