2016/02/23 by Alison I. Weber, Jonathan W. Pillow, Weber, Alison I. +1 · 2 citations
Engineering · Neuroscience · #Advanced Memory and Neural Computing #FOS: Biological sciences #Neural dynamics and brain function #Neurons and Cognition (q-bio.NC) #Neuroscience and Neural Engineering
paper · pdf · doi:10.48550/arxiv.1602.07389
openalex publication_date 2016/02/23 · openalex created_date 2022/10/04 · openalex updated_date 2026/07/28
A key problem in computational neuroscience is to find simple, tractable\nmodels that are nevertheless flexible enough to capture the response properties\nof real neurons. Here we examine the capabilities of recurrent point process\nmodels known as Poisson generalized linear models (GLMs). These models are\ndefined by a set of linear filters, a point nonlinearity, and conditionally\nPoisson spiking. They have desirable statistical properties for fitting and\nhave been widely used to analyze spike trains from electrophysiological\nrecordings. However, the dynamical repertoire of GLMs has not been\nsystematically compared to that of real neurons. Here we show that GLMs can\nreproduce a comprehensive suite of canonical neural response behaviors,\nincluding tonic and phasic spiking, bursting, spike rate adaptation, type I and\ntype II excitation, and two forms of bistability. GLMs can also capture\nstimulus-dependent changes in spike timing precision and reliability that mimic\nthose observed in real neurons, and can exhibit varying degrees of\nstochasticity, from virtually deterministic responses to greater-than-Poisson\nvariability. These results show that Poisson GLMs can exhibit a wide range of\ndynamic spiking behaviors found in real neurons, making them well suited for\nqualitative dynamical as well as quantitative statistical studies of\nsingle-neuron and population response properties.\n