Purchase this article with an account.
Johannes Burge, Wilson Geisler; Optimal speed estimation in natural image movies predicts human performance. Journal of Vision 2015;15(12):2. doi: 10.1167/15.12.2.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Accurate perception of motion depends critically on accurate estimation of retinal motion speed. Here, we first analyze natural image movies to determine the optimal space-time receptive fields for encoding local motion speed in a given direction. Next, from the receptive field 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. The space-time receptive fields (which are direction-tuned but not speed-tuned) and the speed-tuned units exhibit strong similarities to neurons in cortex. Then, in a psychophysical experiment using matched naturalistic stimuli, we show that human performance closely parallels optimal performance. Indeed, a single free parameter accurately predicts the detailed shapes of a large set of psychometric functions. For each human observer, this parameter accounts for more than 95% of the variance in the discrimination data with natural stimuli. The optimal observer also provides excellent predictions of human performance with classic artificial stimuli (e.g. drifting gabors). Importantly, the optimal observer for speed estimation was not designed to match human performance. Rather, it was constructed to maximize the accuracy of speed estimates in natural image movies given the constraints of the visual system’s front end. We conclude i) that many properties of speed selective neurons and human speed discrimination performance are predicted by the optimal computations, and ii) that natural stimulus variation affects optimal and human observers almost identically.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only