Purchase this article with an account.
Alan A. Stocker, Najib Majaj, Chris Tailby, J. Anthony Movshon, Eero P. Simoncelli; Decoding velocity from population responses in area MT of the macaque. Journal of Vision 2009;9(8):741. doi: https://doi.org/10.1167/9.8.741.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The responses of neurons in area MT are thought to underlie the perception of visual motion in primates. However, recent studies indicate that the speed tuning of these neurons changes substantially as contrast is reduced (Pack et al., 2003; Krekelberg et al., 2006), in a way that seems inconsistent with the reduction in perceived velocity seen psychophysically.
To understand this apparent discrepancy, we measured the responses of MT neurons in anaesthetized macaque to a broad-band compound grating stimulus for a broad range of velocities and contrasts. We presented the same stimuli to all neurons, adjusted only for receptive field location and preferred direction. As in previous studies, contrast profoundly affected the responses of the neurons, producing shifts of their preferred velocities toward slower speeds, as well as changes in response amplitude and tuning bandwidth. We constructed a “labeled-line” velocity decoder that operates on a neural population that includes the measured set of 59 neurons, along with a “mirror” set tuned for the opposite direction. We find that operating on this population that represents both positive and negative velocities allows the decoder to capture the key characteristics of human velocity estimation and discrimination, including speed biases at low stimulus contrast.
We also examined optimal linear and (non-linear) Bayesian decoders, and found that they produce nearly veridical percepts when operating on the full neural population. Restricting these decoders to operate on a small set of model neurons whose response properties are obtained by averaging over subsets of neurons with similar tuning leads to qualitatively good matches to the perceptual data, but only when the decoder is optimized for stimuli drawn from naturalistic prior distributions over speed and contrast. We conclude that MT response characteristics are well matched to the statistics of the natural world, such that linear decoding is close to optimal.
This PDF is available to Subscribers Only