Murali Lab

Strategies of Control

In everyday life, we are faced with situations where we have to control what to do, what to say, and what we think. How are we able to stop unwanted actions, intrusive thoughts that bother us, and negative emotions which hinder our ability to perform daily tasks?

Impairments in inhibitory control manifests as disorders such as Tourette's Syndrome, OCD, and anxiety disorders. Using neurofeedback, behavioral, and brain stimulation approaches, we explore ways in which the underlying brain networks can be modulated to control actions and thoughts.

Functional role of Neural Oscillations

Neural oscillations have been implicated in the cognitive control of our movements, actions, memory, and thoughts. Often occurring as brief "bursts" of activity, these oscillations are transient in nature. What functional role do they serve in supporting cognition?

Understanding these can pave ways to show how brain dynamics change in normal and disease conditions. For example, in Parkinson's disease, bursty oscillations in the beta band (13-30 Hz) are enhanced in both cortical and sub-cortical regions, relating to symptoms such as slowness in movements and tremor.

Computational principles underlying cognitive control

Theory building requires a good understanding of the mechanistic and computational aspects of the underlying cognitive construct. We focus on building computational models that bridge the gap between neural processes and behavior.

Our work ranges from modeling how one can stop or control their actions, to how cognitive control networks achieve this via interaction with sub-cortical networks.

BCI & Motor Control

We are working on advancing the precision of Brain-Computer Interfaces (BCI). We are moving beyond the limitations of traditional binary control commands to fine-grained control. Our work involves characterizing neural markers that robustly decode intensity of imagined movements. This can enable users to exert fine-tuned, continuous control over external devices. Our research mostly leverages non-invasive EEG to record data.

We also rely on immersive Virtual Reality (VR) in combination with EEG to study and enhance Motor Imagery and immersive and embodied feedback, allowing for nuanced control systems where users can modulate effort and intensity in real-time.

Ecologically-valid neuroscience

We are striving to understand whether neural circuits involved in motor and cognitive control and their signatures generalise to naturalistic settings. This will allow us to develop applications that can operate in the real-world where there is less control over influencing variables.

We use Augmented and Virtual Reality as a tool to support this transition from lab-based tasks to realistic and immersive settings and finding a middle ground for manipulating task-relevant factors.

Neuro-inspired AI

We work at the intersection of computational neuroscience and machine learning and AI which can inspire new architectures capable of learning, adapting, and behaving more like the brain. We draw from principles of reinforcement learning, neural computations, cognitive control to develop models capable of both generating neural dynamics and behavior. This work advances both AI and generates testable predictions to understand brain function, thereby supporting neurotechnology, human-AI interaction and adaptive intelligent systems.