Exploring and explaining properties of motion processing in biological brains using a neural network.

Reuben Rideaux,Andrew E Welchman

doi:10.1167/jov.21.2.11

Abstract

Visual motion perception underpins behaviors ranging from navigation to depth perception and grasping. Our limited access to biological systems constrains our understanding of how motion is processed within the brain. Here we explore properties of motion perception in biological systems by training a neural network to estimate the velocity of image sequences. The network recapitulates key characteristics of motion processing in biological brains, and we use our access to its structure to explore and understand motion (mis)perception. We find that the network captures the biological response to reverse-phi motion in terms of direction. We further find that it overestimates and underestimates the speed of slow and fast reverse-phi motion, respectively, because of the correlation between reverse-phi motion and the spatiotemporal receptive fields tuned to motion in opposite directions. Second, we find that the distribution of spatiotemporal tuning properties in the V1 and middle temporal (MT) layers of the network are similar to those observed in biological systems. We then show that, in comparison to MT units tuned to fast speeds, those tuned to slow speeds primarily receive input from V1 units tuned to high spatial frequency and low temporal frequency. Next, we find that there is a positive correlation between the pattern-motion and speed selectivity of MT units. Finally, we show that the network captures human underestimation of low coherence motion stimuli, and that this is due to pooling of noise and signal motion. These findings provide biologically plausible explanations for well-known phenomena and produce concrete predictions for future psychophysical and neurophysiological experiments.

Highlights

The transduction of changing patterns of light into the perception of motion underpins adaptive behaviors ranging from depth estimation to navigation and grasping
We find that the network exhibits the same biases, and use our access to the system to show that this is due to the similarity between reverse-phi motion and the receptive fields of spatiotemporal neurons tuned to opposite directions
Similar diversity across middle temporal (MT) neurons is observed for direction selectivity, that is, whether a neuron responds to the individual components or combined pattern of a moving object (Movshon, Adelson, Gizzi, & Newsome, 1986). These two properties index the complexity of the information that is encoded by MT neurons in terms of speed and direction, and we find that they are positively correlated in the network, that is, MT units tuned to speed are more likely to be tuned to pattern motion

Summary

Introduction

The transduction of changing patterns of light into the perception of motion underpins adaptive behaviors ranging from depth estimation to navigation and grasping. For motion perception to guide these behaviors effectively, changes in visual input must be translated into accurate estimation of both direction and speed. Many biological systems appear to be highly proficient at this task; for example, humans can reliably discriminate differences in speeds between 5% to 7% (de Bruyn & Orban, 1988; McKee, 1981) and over a century of research on motion processing has expanded our understanding of the neural computations that underlie this ability. We can measure the output of the system in response to different inputs (i.e., psychophysics), gross population activity (e.g., fMRI or EEG), or point measurements (i.e., cell recordings), but combining this information to extract the underlying neural computations and principles remains a challenge

Methods

Results

Conclusion