Event Details

Arguably, quantitative neuroscience was born when scientists started to correlate the activity of a neuron with sensory stimuli. But complex stimuli, such as natural ones, are encoded in the activity of entire populations of neurons. What is the grammar of this code? Specifically, how are the correlations among neurons and their physiological diversity involved in this code? In this talk, I will discuss analyses of the output of populations of identified and simultaneously recorded visual neurons. In these populations, and against the textbook picture of neural population coding, correlations in the spiking variability enhance the coding performance. This unexpected phenomenon relies upon a particular structure of the correlations observed in data and, surprisingly, yields a strong effect even in very small populations of neurons. I will, further, explain how the favorable structure of correlations can emerge from simple circuit features. Finally, if time allows, I will present more general theoretical extensions in which, with the use of simple models, one can illustrate the massive influence that correlations and physiological diversity can have on the precision and capacity of the neural code.