Cognitive Neuroscience & Neurotechnology

Our lab focuses on advancing our understanding of the frontoparietal brain network mechanisms that underpin high-level cognition and adaptive behaviour.

We pursue an interdisciplinary research programme that allows studying this brain system at multiple levels of granularity. Our methodology involves subject-specific brain-computer interface technology, fMRI at 3T and ultra-high (i.e., 7T and 9.4T) magnetic field strengths (for resolving cortical layers), EEG, non-invasive brain stimulation as well as computational modelling and machine learning.

Frontoparietal brain network: key to cognition

High-level cognition is supported by large-scale brain networks, i.e., distributed regions of co-activated brain activity, and the frontoparietal network (orange) plays a central role.

High-level cognition is supported by large-scale brain networks, i.e., distributed regions of co-activated brain activity, and the frontoparietal network (orange) plays a central role.

The frontoparietal brain network plays a central role in high-level cognition and is involved in a particular diverse range of different cognitive processes (e.g., attention, working memory, reasoning, problem solving, planning, cognitive and inhibitory control). Interestingly, this brain system has the highest degree of inter-subject variability. In addition, a strikingly large range of psychiatric and neurological conditions shows disruptions in this brain system. However, despite being the focus of intensive research efforts, we are still lacking important insights into computational mechanisms underlying frontoparietal network function.

Topography of the frontoparietal network

The current understanding is that the frontoparietal system can be divided into at least two closely coupled subnetworks centred around the lateral frontoparietal and cingulo-opercular cortex. However, the unique functional role of each subnetwork remains poorly understood. Adding to the complexity, the topography of the frontoparietal network is highly variable in the individual rendering group-level approaches unfeasible. We aim to develop techniques for identifying the extent and topography of frontoparietal subnetworks reliably and efficiently in individual healthy volunteers.

Functional differentiation within the frontoparietal network

Non-invasive investigation of different cortical layers in the frontoparietal network in humans is achievable using ultrahigh field fMRI, such as 7 or 9.4T, whereas in the past, invasive electrophysiology in non-human primates was the primary means to obtain such insights.

Non-invasive investigation of different cortical layers in the frontoparietal network in humans is achievable using ultrahigh field fMRI, such as 7 or 9.4T, whereas in the past, invasive electrophysiology in non-human primates was the primary means to obtain such insights.

Frontoparietal networks, like the rest of the cerebral cortex, are organised into layers characterised by distinct cytoarchitecture (i.e., distribution of cell types) and connectivity patterns (e.g., feed-forward/feed-backward connections). Animal neurophysiology suggests that different cortical layers in the frontoparietal cortex subserve different cognitive functions. Progress in our understanding about the degree of such a functional differentiation thus depends on choosing optimal cognitive task conditions that selectively activate layers (e.g., superficial vs. deep layer dissociation). However, to date, little data on laminar functional specificity in the frontoparietal cortex exist that could guide experimental design. To resolve this problem, we aim to leverage subject-specific brain-computer interface technology (i.e. Neuroadaptive Bayesian Optimization) for functionally dissociating layers in frontoparietal subnetworks in real-time using ultrahigh-resolution fMRI at 9.4 T. This will be a crucial step for gaining more fine-grained insights into the mechanisms by which frontoparietal subnetworks support cognitive function.

Interaction within the frontoparietal network and other brain systems

Neuroadaptive Bayesian Optimization is a brain-computer interface that combines recent advances in real-time analysis of fMRI data and machine learning. The technique involves a closed-loop search across a wide variety of cognitive tasks, with fMRI data analysed in real-time and the next tasks to be chosen based on the real-time results. This approach provides a powerful strategy to efficiently explore many more experimental conditions than currently possible with standard methodology in a single scan. Figure taken from Lorenz et al. (2018) Nature Communications: https://www.nature.com/articles/s41467-018-03657-3]

Neuroadaptive Bayesian Optimization is a brain-computer interface that combines recent advances in real-time analysis of fMRI data and machine learning. The technique involves a closed-loop search across a wide variety of cognitive tasks, with fMRI data analysed in real-time and the next tasks to be chosen based on the real-time results. This approach provides a powerful strategy to efficiently explore many more experimental conditions than currently possible with standard methodology in a single scan. Figure taken from Lorenz et al. (2018) Nature Communications: https://www.nature.com/articles/s41467-018-03657-3]

Identifying causal frontoparietal network mechanisms necessitates investigating what information is computed at different cortical layers of the network, how information flows (i.e., feedforward/feedback) within the frontoparietal system (across-layers and between-subnetworks) and between the frontoparietal network and other brain systems (e.g., subcortical structures). In the long term, we aim to utilize empirical insights gained about the micro-circuitry of the frontoparietal cortex to develop more detailed and biologically constrained computational models of high-level cognition.