Unformatted text preview:

Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11The simultaneous coding of orientation and contrast in the responses of V1 complex cellsTimothy J. GawneUniversity of Alabama at Birminghampresented by Drew AltschulBackground●Major questions in this field–How much can a single cell tell us?–How can we reduce the ambiguity of a single cell?●Previous research–V1 cells can encode information in temporal codes.–A single neuron's output is ambiguous.–Ambiguity reduced by temporal coding.●Recently found correlations–response strength to stimulus orientation–response latency to stimulus contrast●Temporal coding–Has been seen in these cells–Exact meaning is unclear in this contextFocus of this Study●Purpose: to determine how effective responses of individual and groups of V1 complex cells are at representing stimulus orientation and contrast when both factors are allowed to vary.●Main findings–Contrast is related to latency and temporal modulation–Power of a single neuron–Multiple cells, multivariate–Pooling improves spike code responsesMethodsData Collection●Data was taken from individual complex cells in V1 of two rhesus monkeys.●The monkeys viewed bars with varying (uniformly distributed) orientations and contrasts to the background.–The first monkey viewed sets of 12 differently oriented bars.–The second monkey viewed sets of 8 bars.–Each monkey viewed 8 sets: 4 black and 4 white sets.●The individual spikes of individual cells were measured via extracellular methods, and further analyzed.One of the stimulus sets presented to the monkey subject.MethodsData Analysis●Quantification Methods–Spike count after stimulus–Replace each spike with Gaussians and summed–Principle Component Analysis●Simplifies data into a few coefficients of waves●Neural Network–Combining multiple neuron signals–Synthetic data set of group responsesResults ●Latency –reflected stimulus contrast better than stimulus orientation–not an explicit function of the strength of the response●Contrast increases, latency decreases●If contrast remained constant, spike counts could predict orientation.Plot of the average responses of the neurons to different stimuli.Results●Orientation –held constant, spike counts were not very predictive contrast very well.●Both factors varying –orientation prediction was still effective–almost no relationship could be seen with contrast prediction●Coefficients of the principle components were used to evaluate temporal coding.Regression plots of actual contrast/orientation vs. predicted from single complex cells.Results●Temporal modulation –stimuli were allowed to vary–orientation prediction was maintained–contrast prediction improved to the best results yet●Improvements using temporal modulation were partially due to actual changes in response latency.●Also the result of some changes in the strength of the response.●Overall improvement over spike count performance.Regression plots of actual contrast/orientation vs. predicted from all 11 complex cells.ResultsRegression plots of actual stimulus contrast and orientation - Left: individual cells- Right: multiple cellsConclusions●The use of temporal coding in complex V1 neurons could theoretically be utilized to send signals for both orientation and contrast.●However, just because this is possible does not mean this is what the complex cells are doing. ●Using temporal coding in this way may not actually be possible.Discussion●Previous research–Single neurons have high precision–One stimulus dimension limit●Temporal coding in more complex systems –more than two factors allowed to vary●Continued studies –more stimulus types –greater numbers of neurons–distributions throughout V1–possibly higher order


View Full Document

MIT 9 02 - Lecture Notes

Download Lecture Notes
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Lecture Notes and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Lecture Notes 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?