Brain states representing dynamics of meta-control

A core assumption of this CRC is that cognitive control is not mediated by a singular central control instance, but rather emerges from the interplay of large-scale brain networks, which comprise modules such as for attention, valuation, conflict detection, inhibition, or top-down control. So far, trait-like behavioral and neural characteristics in these networks for cognitive control have been investigated within the context of the different control dilemmas. However, cognitive control and decision-making can be remarkably flexible, e.g. when contexts or task demands change. Up to now, such dynamic adaptations of cognitive control – control states – have been rarely investigated on the neural level. We postulate that the dynamics of control states are reflected by changes in activity and connectivity patterns among large scale brain networks, here called brain states. In this project, we will investigate this conception using the anticipation-discounting dilemma as an example and aim to detect dynamic changes of brain states and associated control states.
During the first funding period of this project, we studied whether real-time fMRI (rt-fMRI) neurofeedback enables participants to learn to volitionally regulate the activity of their amygdala and how this regulation alters emotional reactivity. Critically, we demonstrated that subjects can indeed learn to volitionally regulate their amygdala during a neurofeedback-training, and that they are able to down-regulate activity in their amygdala even without any feedback thereafter. In addition, with respect to emotional reactivity, voluntarily down-regulating amygdala activity abolished its reactivity to unpleasant stimuli, and, at the behavioral level, suppressed the attentional capture elicited by these stimuli. Moreover, and of importance for the second funding period, we also found that spontaneous fluctuations in amygdala activity modulate response speed in a simple reaction time (RT) task: For single trials, high amygdala activity several seconds before a reaction time trial predicted a slow RT, and low activity predicted a fast RT. The temporal precedence of a basic brain state (i.e. amygdala activity) to behavior shown in the first funding period corroborates that fluctuations of brain activity – induced as well as spontaneously occurring – can result in dynamic changes of cognitive control.
Building on these findings of the first funding period and our conceptual framework of dynamic adaptations of control states that emerge from brain states, we first will accurately describe dynamics in terms of brain states. To detect dynamic changes in large-scale brain networks, we will use brain activity in ROIs and connectivity between ROIs in combination with multivariate pattern analysis (MVPA) methods. Exemplary, we will apply this approach to the anticipation-discounting dilemma that will be assessed with intertemporal choice tasks in close collaboration with projects A6 and A8. Attractor models of this control dilemma assume two distinguishable choice states (i.e. an impulsive vs. a reflective state), the balance of which is influenced by the meta-control parameters temporal discounting rate and noise level. After having identified the respective brain states of interest using fMRI and offline classification techniques, we will independently test whether these brain states temporarily precede the assumed control states. For this purpose, we will build on the rt-fMRI technology that we implemented during the first funding period to monitor more complex brain states online. In particular, we will use so-called adaptive fMRI paradigms in combination with MVPA methods to capture brain states in real-time and then present intertemporal choice trials accordingly.
The combination of these advanced neuroimaging techniques and rigid experimental designs will enable a better understanding of the nature of brain states associated with impulsive or reflective choices. As an outlook, this approach can also be transferred to other dilemmas, such as the shielding-shifting dilemma, the dynamics of which are investigated in project B1, and to identify general brain states related to cognitive control. Furthermore, a next step would be to test whether the identified brain states in combination with real-time fMRI can be used in closed-loop training systems to foster a desired behavior.

Project Members

Principal Investigators