Optimal speed estimation in natural image movies predicts human performance.

Johannes Burge,Wilson S Geisler

doi:10.1038/ncomms8900

Johannes Burge, Wilson S Geisler

Open Access

https://doi.org/10.1038/ncomms8900

Copy DOI

Abstract

Accurate perception of motion depends critically on accurate estimation of retinal motion speed. Here we first analyse natural image movies to determine the optimal space-time receptive fields (RFs) for encoding local motion speed in a particular direction, given the constraints of the early visual system. Next, from the RF responses to natural stimuli, we determine the neural computations that are optimal for combining and decoding the responses into estimates of speed. The computations show how selective, invariant speed-tuned units might be constructed by the nervous system. Then, in a psychophysical experiment using matched stimuli, we show that human performance is nearly optimal. Indeed, a single efficiency parameter accurately predicts the detailed shapes of a large set of human psychometric functions. We conclude that many properties of speed-selective neurons and human speed discrimination performance are predicted by the optimal computations, and that natural stimulus variation affects optimal and human observers almost identically.

Highlights

Accurate perception of motion depends critically on accurate estimation of retinal motion speed
More sophisticated models are required to account for the greater speed-selectivity that is characteristic of some V1 complex cells and middle temporal (MT) neurons
The receptive fields (RFs) shapes are typically chosen for mathematical convenience, and the combination rules are based on intuitive computational principles

Summary

Introduction

Accurate perception of motion depends critically on accurate estimation of retinal motion speed. Neither the RFs nor combination rules are based on measurements of natural signals These models account for many of the response properties of cells in V1 and MT, it is not known whether neurons having these properties provide the best substrate for motion estimation with natural stimuli. When applied to natural image movies, this coding scheme produces a set of linear RFs that has some of the motion selective properties of V1 neurons (but see the study by Ringach[17]) Given this general cost function, there is no reason to expect that such RFs would be wellsuited for motion estimation or any other specific task.

Objectives

Results

Discussion

Conclusion