Neural Circuit Mechanisms for Neural Computation
Spatio-temporally dynamic sequences of spikes are believed to encode information that are critical in neural computation such as sensory and spatial information processing as well as memory encoding/recall. Generally, two different types of neural coding schemes exist: temporal code that encodes information in the precisely timed spikes and rate code that encodes information in the number of spikes in a give time-window. Our lab is interested in how these two different types of neural codes are generated and used for neural computation, which is the core basis of our behavior. Through optogenetic manipulations of distinct subtypes of inhibitory interneurons such as parvalbumin(PV)- and somatostatin(SST)-expressing interneurons, and performing in vivo electrophysiology (network oscillations, single-unit recordings) in vivo calcium imaging during behavior in mice,
we experimentally investigate the neural circuit mechanisms underlying neural computation.
Currently, some of the projects of the lab include:
-
The role of interneurons in neural code generation, synchronization and propagation
-
The role of distinct subtypes of interneurons in memory encoding and its application to Alzheimer’s disease therapeutics
Neuroscience-inspired computational models of the brain for AI and neuromorphic robot applications
The neural circuits of the brain are extremely complex and dynamic thus it is difficult to capture the full details using the current experimental tools. To complement this, using the experimentally-derived data, we build physiologically and anatomically realistic computational models of synapses, neurons and neural circuits. Using these models, we can extrapolate our experimental data as well as come up with new predictions on neural computational rules that can again be experimentally tested.
In addition, the computational models of the neural circuits and the new neural computational rules can provide new insights to developing “neuroscience-inspired AI algorithms” and neuromorphic systems.
Current projects of the lab include:
-
Development of “neuroscience-inspired” feedforward neural networks models for AI applications
-
Development of neuromorphic robot for spatial navigation