Analysing visual receptive fields through generalised additive models with interactions

Abstract

Visual receptive fields (RFs) are small areas of the visual fie ld where a stimulus induces a re- sponses of a particular neuron from the visual system. RFs ca n be mapped using reverse cross- correlation technique, which produces raw matrices contai ning both spatial and temporal informa- tion about the RF. Though this technique is frequently used i n electrophysiological experiments, it does not allow formal comparisons between RFs obtained un der different experimental condi- tions. In this paper we propose the use of Generalised Additi ve Models (GAM) including com- plex interactions, to obtain smoothed spatio-temporal ver sions of RFs. Moreover, the proposed methodology also allow for the statistical comparisons of t he RFs obtained across various experi- mental conditions. Data analysed here derive from studies o f neurons’ activity in the visual cortex of behaving monkeys. Our results suggest that the GAM-based technique proposed in this paper can be a flexible and powerful tool for assessing receptive fie ld properties