Research Projects

The primate brain has a hierarchical modular architecture, executing canonical computations across interconnected cortical areas. Communication between cortical areas is supported by long-range reciprocal connections. Feedback connections have been hypothesized to serve various functions including attentional modulation. Consequently, it is imperative to develop a multistage network model of the brain to effectively study phenomena such as attention and inter-area communication. Here, we present a stable hierarchical recurrent neural circuit model with feedback that dynamically implements divisive normalization exactly at each stage of its hierarchy. We consider a two-stage model (V1 and V4), each stage receives input from the preceding area in the hierarchy and feedback from the subsequent area, and the responses in each area are normalized by local inhibitory signals modulated by the activity of principal neurons across areas. We observe that an increase in feedback from V4 to V1, amplifies responses in both stages, with a more pronounced increase in higher cortical areas, consistent with experimental findings.

Additionally, we note that in our model’s single-stage limit cycle regime, feedback-mediated attention (i.e., endogenous attention) from a second stage stabilizes the system, yielding a stable fixed point. We further investigate the coherence of neuronal activity between cortical areas and find a peak in the gamma band whose amplitude is positively correlated with strength of feedback. Our model also admits the existence of a low-dimensional communication subspace (within and across areas) and makes predictions about how the subspace varies with feedback-mediated attention. In particular, the strength of inter-area communication is significantly increased (viz., the population activity of one area more accurately predicted from another), without affecting the subspace dimensionality, in agreement with experimental findings.

It has been shown that the brain region named Hippocampus plays an essential role in Long-term memory formation in mammals. In our study, we try to identify the possible mechanisms of long-term memory formation in mice. We tried to identify the genes most responsible for forming long-term memory.

The mice were trained with a control group with an active place avoidance task. Later, the mice were tested on their ability to remember the lesson of avoiding a shock zone. Tissues are collected from the Hippocampus region for Transcriptomic data analysis. Then, linear and non-linear correlation analyses are performed between the gene expression labels and behavior. Using Mutual Information analysis, we could identify mice's genes responsible for long-term memory formation.

Temperature Accelerated Sliced Sampling (TASS) is a hybrid enhanced sampling method that combines Metadynamics(MTD), Umbrella Sampling(US) & Temperature Accelerated Molecular Dynamics(TAMD) for efficient sampling of the high-dimensional Collective Variable(CV) space. This project proposed an efficient mean-force-based reweighting method, bypassing the tedious and cumbersome WHAM(Weighted Histogram Analysis Method).

• Mean Force-based Temperature Accelerated Sliced Sampling Method(TASS)