bibtex

@article{ Vogels_etal89, author = {Vogels, R. and Spileers, W. and Orban, G.A.}, title = {The response variability of striate cortical neurons in the behaving monkey}, journal = {Exp. Brain Res.}, year = {1989}, volume = {77}, pages = {432-436}, en_number = {1.16:35}, keywords = {Visual cortex, single cell, response variability, behaving monkey}, summary = {They record in V1 of awake behaving monkey doing an orientation discrimination task (183 cells). Again, they find power laws in the variance vs. mean plots for about 2 decades (1 to 100 Hz). The variance is between recordings between different cells. The exponents are reasonably close to one, being 1.03 in one monkey and 1.20 in the other one (average of all 183 cells: 1.11). They also computed the variance for the same cell but between different orientations and found that the slope varies quite a bit (since they have only few data points for each cell) but it varies around 1.1. They say that the mean slope (for all cells) is nearly the same as before (1.10). In fact, I wonder if that is not somewhat tautological: OF course, the mean will be the same as before, since they average over the same set of data; they only sum them up in a different order. In fact, in the limit of large datasets they should obtain the IDENTICAL result (i.e. 1.11 not 1.10), and the difference can only be due to the fact that they first determine the slopes, each time for a few points only, and then take the average of all slopes, rather than computing one slope for ALL points. But this is probably not important, the relevant thing is that (a) they find power laws over about 2 decades and (b) the power is pretty close to unity. One more thing: Vogels et al argue that the power is close to one, but that the proportionality factor between mean and variance is not one, as one would expect for a Poisson process, but closer to two (they find 1.9, Dean found 1.54 and Tolhust et al 1983 found 2.8). So, therefore, it cannot be a Poisson process. } }

Downloads: 0