Neural mechanisms of cognitive control
Cognitive control emerges from the coordinated activity within cortical and subcortical structures. Our ongoing research aims to delineate the sequential neuronal response profiles associated with cognitive control throughout this network. Our objective is to analyze functional connectivity, shedding light on the relationships between each node within the network. A notable observation is that lapses in focus often lead to errors. Monitoring our errors is a pivotal aspect of cognitive control. Our findings indicate that both the Error-Related Negativity (ERN) and error neuron spiking manifest initially in the pre-supplementary motor area (pre-SMA), followed by the mid-cingulate cortex (MCC). Currently, we are investigating the potential functional connectivity between the pre-SMA and MCC during error processing.
To characterize human behavior, we employ Bayesian hierarchical models in cognitive tasks. Neuronal activity is then recorded at varying levels, and this activity is aligned with behavior through two methodologies:
- Classification of individual neurons based on behavioral parameters.
- Behavior prediction from the collective activity of neuronal populations.
By mapping these parameters onto a high-dimensional space, we analyze their geometric relationship. Such approach elucidates the nuances of our error-monitoring mechanisms across diverse scenarios, offering insights into how we recalibrate focus towards our objectives.
Cognitive control in silico
The human brain represents an intricate system, and despite advanced techniques, directly measuring the electrical activity across numerous neurons remains a formidable challenge due to the scarcity of human samples and data constraints. However, with the emergence of artificial neural networks (ANN), we now possess a platform to probe computational and algorithmic principles, offering insights into the operations of the human brain.
Our objective is to train ANNs to replicate human behaviors observed in cognitive tasks. By dissecting the architecture of these networks, we aim to discern how artificial neurons address and resolve specific tasks. The computational revelations from this endeavor could potentially illuminate the mechanisms employed by the human brain in analogous scenarios.
Additionally, by introducing perturbations to the ANN, we intend to observe the resultant changes in network behavior. Such exploration presents an avenue to validate hypotheses concerning cognitive control deficits observed in psychiatric disorders, including OCD, ADHD, and schizophrenia.
Neuromodulation
Psychiatric disorders adversely impact our ability to concentrate. Is it possible to enhance focus through targeted brain stimulation? While current methods, such as deep brain stimulation employed for addressing the motor symptoms of Parkinson’s disease, have been promising, they sometimes carry unintended consequences, including deficits in cognitive control. To mitigate these risks, our primary objective is a deeper comprehension of the network mechanisms underlying cognitive control. From this foundation, we aim to engineer neuromodulation techniques that stimulate specific brain regions with well-defined parameters, improving cognitive control. Our endeavors have the potential not only to reduce the adverse effects associated with existing brain stimulation methodologies but also to introduce innovative therapeutic strategies for addressing cognitive control impairments prevalent in psychiatric conditions.
