Statistical approaches for unraveling the neural code in the visual system
MetadataShow full item record
Here we consider the neural coding problem at two levels of the macaque visual system.First, we analyze single neurons recorded in the lateral intraparietal (LIP) cortex while a monkey performed a perceptual decision-making task. We relate the single-trial responses in LIP to stochastic decision-making processes with latent dynamical models. We compare models with latent spike rates governed by either continuous diffusion-to-bound dynamics or discrete ``stepping'' dynamics. In contrast to previous findings, roughly three-quarters of the choice-selective neurons we recorded are better described by the stepping model. Second, we introduce a biophysically inspired point process model that explicitly incorporates stimulus-induced changes in synaptic conductance in a dynamical model of neuronal membrane potential. We show that our model provides a tractable model of spike responses in macaque parasol retinal ganglion cells that is both more accurate and more interpretable than the popular generalized linear model. Most importantly, we show that we can accurately infer intracellular synaptic conductances from extracellularly recorded spike trains.