We tested the hypothesis that the neural code of retinal ganglion cells is optimized to transmit visual information at minimal metabolic cost. Under a broad ensemble of light patterns, ganglion cell spike trains consisted of sparse, precise bursts of spikes. These bursts were viewed as independent neural symbols. The noise in each burst was measured via repeated presentation of the visual stimulus, and the energy cost was estimated from the total charge flow in a biophysically realistic model of ganglion cell spiking. Given these costs and noise, the theory of efficient codes predicts an optimal distribution of symbol usage. Symbols that are either noisy or metabolically costly are suppressed in this optimal code. We found excellent qualitative and quantitative agreement with the measured distribution of burst sizes for ganglion cells in the tiger salamander retina. ∗ Both authors contributed equally to this work. 1

In executing a voluntary movement, one is faced with the problem of translating a specification of the movement in task space (e.g. a visual goal) into a muscle recruitment pattern. Among

design of photoreceptor mosaics: why we don’t see color at night

have the same efficiency. Thus, patterns per second sets the coding capacity. We asked we asked, what is the maximum information rate that if different ganglion cell types make equally efficient use an ideal ganglion cell could transmit, and how closely of their coding capacity or whether efficiency depends do real cells approach this limit? on the message conveyed. Information rates were estimated by the direct method Results: We recorded spike trains from retinal ganglion [14], which counts the number of different spike patcells in an in vitro preparation of the guinea pig retina. terns, accounts for the deleterious effect of noise on By calculating, for the observed spike rate, the number information, and requires long, stable recordings. This of possible spike patterns per second, we calculated was achieved, even for small cells, by viewing the intact coding capacity, and by counting the actual number guinea pig retina in vitro with infrared-DIC optics and of pa

Abstract not found