a8240cb8235e9c493a0c30607586166c-Reviews.html

This paper makes two contributions. First, this paper continues a trend of recent work by Park and Pillow (2011) and Ramirez & Paninski (2012) to provide a firm grounding of the spike-triggered methods STA and STC within forward, likelihood-based modelling. STA and STC are calculated as moments from the stimulus/response ensemble, and are popular for constructing stimulus/response models due to their speed and simplicity. Paninski 2003, Sharpee et al 2004, Samengo & Gollisch 2013), the validity and utility of these moments depends on particular restrictive conditions on the stimulus set. The prior work of Park & Pillow (2011) was able to shed light on this, by relating these moments to the parameters of a Generalised Linear Model (GLM), a popular forward model for neural responses.