direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

Analysis of Neural Data

Selected Abstracts

Contrast respones of p-cells and V1 neurons are optimized for a winner-takes-all encoding
Citation key Martin2009
Author R. Martin and Y. Tadmor and K. Obermayer
Title of Book Computational and Systems Neuroscience
Year 2009
Abstract The principle of histogram equalization/infomax predicts neuronal response properties of contrast sensitive neurons from the distribution of contrasts in natural images. This has been successfully demonstrated for the fly large monopolar cell [1], cat X- and Y-retinal ganglion cells (RGC) and LGN neurons, cat V1 neurons, and macaque retinal and LGN M-cells [2,3]. However, responses of macaque P-RGC and -LGN neurons and V1 neurons are less sensitive to contrast than histogram equalization predicts. Here, we determine the distribution of contrasts RGC, LGN neurons and V1 neurons encounter in natural images. Using 70 Difference-of-Gaussian (DoG) and 80 Gabor contrast operators, both biologically plausible and representative of the range of macaque neurons in LGN and V1, we sample a set of calibrated greyscale natural images [4]. The full contrast distribution (each position sampled by all contrast operators) reconfirms the above-cited results regarding histogram equalization. However, sampling each location only with the contrast operator that shows the strongest response at this location (for each, DoG and Gabor operators separately) produces a contrast distribution that predicts the contrast response function of both, macaque P- and V1 neurons. We then compare the performance of the histogram equalization-based (referred to as “M-like”) encoding strategy with that of a top-response-based (“P-like”) strategy, for the stages of LGN contrast coding and of V1 contrast coding. Using both, mutual information and rate distortion theory we find that for each individual neuron, M-like encoding outperforms P-like encoding with respect to the information conveyed about natural image contrast. However, considering the mutual information between neurons with similar receptive field characteristics (e.g., neighboring preferred spatial frequencies) shows that P-like encoding is accompanied by a pronounced decrease in the mutual information between different output channels. Taken together these findings suggests that macaque P- and V1 neurons employ a contrast coding strategy intrinsically different from M-cells and cat visual neurons. Unlike the latter, P and V1 contrast responses are not optimized for maximizing the contrast information transmitted by individual cells, but also to reduce the redundancy between different transmission channels, thus providing a trade-off between the desire of maximizing the transmitted information and reducing the redundancy at the population level. In doing so, their contrast response is in fact optimized for encoding the contrast of spatially optimal features. Such a strategy would imply a Winner-Takes-All-like encoding to operate at the readout stage. And indeed, we can also show that the lower contrast sensitivity of the P-like encoding aids any winner-takes-all mechanism: For both, LGN and V1, the ratio between highest and second-highest response in the P-like encoding is substantially larger than in the M-like encoding. [1] Laughlin, 1981 Zeitschrift f ¨ ur Naturforschung 36: 910 – 912 [2] Tadmor and Tolhurst, 2000 Vision Research 40: 3145 – 3157 [3] Clatworthy et al, 2003 Vision Research 43: 1983 – 2001 [4] van Hateren and van der Schaaf, 1998 Proceedings of the Royal Society 265: 359-366 doi:10.3389/conf.neuro.06.2009.03.076
Link to publication Download Bibtex entry

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions