Current events
Date: February 1st
Time: 09:30 am
Place : Gonda 2303
Title : “Experience-dependent plasticity of Network Dynamics”
Speaker: Anubhuthi Goel
Abstract: Cortical computations underlying normal and abnormal brain function are not only dependent on modifications at individual synapses but on the net interaction between many forms of plasticity at the level of the entire network. Together multiple forms of plasticity govern the complex spatio-temporal patterns of activity within local networks – that is, neural dynamics. One particular learning rule that is critical in the development of functional neural dynamics in a controlled fashion is Homeostatic plasticity. We examined plasticity of network dynamics in cortical organotypic slices in response to chronic changes in activity and found that networks rely on a balance between spontaneous and evoked activity in order to drive their average activity levels towards homeostatic set points. Importantly our data highlights the fact that at the network level homeostatic mechanisms are not restricted to simple and traditional synaptic scaling phenomena wherein all the synapses are indiscriminately scaled up or down. Rather homeostatic mechanisms involve multiple forms of plasticity operating in parallel thereby allowing circuits to independently regulate spontaneous, monosynaptic, and polysynaptic activity.
Having gained some insight as to how homeostatic plasticity influences computations in general so that recurrent cortical circuits generate functional dynamic states in a stable fashion we examined one particularly interesting type of computation, namely, how does timing emerge from the plasticity of neural dynamics. Timing is fundamental to learning and behavior, and it is increasingly clear that in many cases timing is an emergent network phenomenon; but almost nothing is known about the neural mechanisms that underlie even the simplest of temporal tasks, such as discriminating a 100 and 200 ms interval. We have recently established that when cortical organotypic slices are chronically presented with specific intervals (using electrical stimulation), the network can in a sense “learn” the trained interval: after training, presentation of a single pulse results in increased neural activity around the expected time of the second pulse—as if the network “expected” the arrival of the second pulse. We view this as an example of an emergent computation in vitro because: First, the changes in the behavior of the network seem to rely on the interaction of many neurons in a circuit rather than the simple amplification of neural responses observed in traditional synaptic plasticity studies. Second, it can be said that a simple computation is taking place because the activity patterns in the trained network provide information about elapsed time. To examine the mechanisms of temporal pattern completion, we combined electrical and optical stimulation to provide “sensory spatio-temporal experience” to cortical organotypic slices. Our data suggests that the observed timing is in part due to evoked patterns of activity—neural trajectories—in which distinct points in time can be encoded by the population of active neurons. Furthermore based on our insights from homeostatic plasticity studies we believe that homeostatic mechanisms aid in the emergence and propagation of these neural trajectories in a stable manner.
