Natural versus Synthetic Stimuli for Estimating Receptive Field Models: A Comparison of Predictive Robustness

Abstract

An ultimate goal of visual neuroscience is to understand the neural encoding of complex, everyday scenes. Yet most of our knowledge of neuronal receptive fields has come from studies using simple artificial stimuli (e.g., bars, gratings) that may fail to reveal the full nature of a neuron's actual response properties. Our goal was to compare the utility of artificial and natural stimuli for estimating receptive field (RF) models. Using extracellular recordings from simple type cells in cat A18, we acquired responses to three types of broadband stimulus ensembles: two widely used artificial patterns (white noise and short bars), and natural images. We used a primary dataset to estimate the spatiotemporal receptive field (STRF) with two hold-back datasets for regularization and validation. STRFs were estimated using an iterative regression algorithm with regularization and subsequently fit with a zero-memory nonlinearity. Each RF model (STRF and zero-memory nonlinearity) was then used in simulations to predict responses to the same stimulus type used to estimate it, as well as to other broadband stimuli and sinewave gratings. White noise stimuli often elicited poor responses leading to noisy RF estimates, while short bars and natural image stimuli were more successful in driving A18 neurons and producing clear RF estimates with strong predictive ability. Natural image-derived RF models were the most robust at predicting responses to other broadband stimulus ensembles that were not used in their estimation and also provided good predictions of tuning curves for sinewave gratings.

Making Sense of Real-World Scenes

“…to demonstrate spatial frequency biases in higher-order visual regions with filtered stimuli that contain exclusively high or low spatial frequencies [e.g., 143]. However, brain responses obtained using manipulated images do not necessarily generalize to intact natural scenes [144,145], and it is important not to attribute the operation performed to achieve an image manipulation to a neural computation without considering its biological plausibility. Instead, it might be more useful to think of image manipulation as emphasizing specific aspects that are more or less diagnostic for a given task [146,147].…”

Stimulus-Driven Population Activity Patterns in Macaque Primary Visual Cortex

“…Previous studies have found that the ability of a RF model to predict a neuron’s response can depend on the stimulus class on which the model was trained [20–22], suggesting that the population activity might use somewhat distinct basis patterns for different stimulus classes. On the other hand, if basis patterns are influenced by the shared underlying network structure [4, 10], then we would expect them to be shared across responses to different stimuli.…”

“…Although the RF models do not capture every aspect of V1 neuronal activity [20–22, 24], we can apply the same dimensionality reduction methods to the activity generated by an RF model to help interpret the outputs of dimensionality reduction. We consider it a strength that, in many cases (described below), the outputs of dimensionality reduction applied to population activity show the same trends as when applied to activity from an RF model.…”

“…Many previous studies have compared visual cortical responses to natural and artificial stimuli on a single-neuron level (e.g., [20–22, 31, 32]). Their predominant approach was to define a parameterized RF model to relate a neuron’s activity to the visual stimulus.…”

Visual pattern discrimination by population retinal ganglion cells’ activities during natural movie stimulation

“…However in natural vision, the visual system is facing to complex and dynamic natural stimuli, which are very different from the simple stimuli (Reinagel 2001;Carandini et al 2005;Felsen and Dan 2005;Geisler 2008). In recent years, naturalistic visual stimuli have been increasingly used in the visual system (Lesica et al 2008;Mante et al 2008;Endeman and Kamermans 2010;Talebi and Baker 2012), and it has been demonstrated that the neuronal responses to natural input are not results of superposition of simple constituents (Einhauser and Konig 2010). Thus it is desirable to investigate the information processing of the visual system in response to natural stimuli.…”