Display Settings:

Items per page

Results: 6

Figure 3

Figure 3. From: Preserving information in neural transmission.

Significant differences did not exist between RGC and LGN filters computed as spike-triggered averages (STAs, in a), or as decorrelated STAs (b). Regularization of the decorrelated STA (c) smoothed noise out of the filters, but did not yield a prominent difference in filter shape. Although these decorrelated fitlers resembled those found previously using uncorrelated noise techniques (Chander and Chichilnisky, 2001; Reid and Shapley, 2002), they were noisier than the standard STA. Error bars are standard errors of the mean, computed from data subsets; in a and b, they were too small to be usefully plotted. RMS amplitudes of subtracted STAs averaged (0.018 ± 0.01 s.d.), significantly lower than the RMS values obtained from MID features (p < 0.001 for both filters, see Fig. 4a, b).

Lawrence C. Sincich, et al. J Neurosci. ;29(19):6207-6216.
Figure 5

Figure 5. From: Preserving information in neural transmission.

Maximally informative filters account for the increased efficacy of information transfer. (a) The combined MID filters considered as a model of spike activity manifest the higher information capacity of single LGN spikes seen experimentally across the population (Cf. Fig. 2b). Cell pairs are represented by circles, color coded by LGN neuron type. (b) The combined model accounts for most information available in the spike trains (means: RGC = 79 ± 18%; LGN = 83 ± 15%). (c) The combined filter model also exhibits the same degree of lossless information transfer rate found experimentally (Cf. Fig. 2d). Error bars are ±1 s.d, incorporating the errors when computing bits/spike (too small to plot usefully in panels a and b), and the variation in firing rate during repeated stimuli.

Lawrence C. Sincich, et al. J Neurosci. ;29(19):6207-6216.
Figure 6

Figure 6. From: Preserving information in neural transmission.

Short-term retinal spike summation improves LGN information transfer efficiency. (a) Luminance steps, indicated in white, of a repeated stimulus that resembles the MID1 filters (top, aligned to a selected spike) lead to EPSPs that summate to drive LGN spikes, evident in raw traces (middle) as well as raster plots. Spike rates and corresponding spike rate gains (bottom) show that the LGN neuron (black) exhibits greater firing rate modulation than the ganglion cell (red). (b,c) Different retinal spike trains convey different amounts of information. The population mean retinal train, showing increased EPSP probability during the 30 ms prior to an LGN spike, carries the most information (b), while the next most informative train shows inhibition over a longer period before an LGN spike (c). Shaded areas are ± 1 s.d.

Lawrence C. Sincich, et al. J Neurosci. ;29(19):6207-6216.
Figure 1

Figure 1. From: Preserving information in neural transmission.

Extracellular recording of retinal input to a single LGN neuron. (a) Segment of luminance flicker used to stimulate the receptive field center of an ON-center magnocellular LGN neuron. Center diameter = 0.5°, mean center luminance = 6.1 cd/m², surround luminance = 0.002 cd/m². (b) Raw microelectrode voltage trace recorded in response to the 5 s flicker segment shown in a. (c) Expanded view of the trace outlined in red in b, showing 2 isolated retinal EPSPs followed by 3 LGN spikes, one of which is small because it failed to propagate back into the soma and dendrites. Each spike was preceded by an EPSP (red arrows) and was verified by offline analysis to reconstruct the spike trains. (d) Peri-stimulus time histogram of the firing rates of the ganglion cell (RGC, red) and LGN neuron (black) in response to a portion of the repeated stimulus shows that the LGN rates are a variable fraction of the RGC rates. Histogram compiled from 295 stimulus repeats.

Lawrence C. Sincich, et al. J Neurosci. ;29(19):6207-6216.
Figure 2

Figure 2. From: Preserving information in neural transmission.

Increased information efficiency from retina to LGN. (a) Mean retinal ganglion cell (RGC) spike rates are higher than mean LGN spike rates during stimulus presentation. Population averages: RGC = 38.2 spikes/s; LGN = 21.1 spikes/s. Each cell pair is represented by a circle, color coded by LGN neuron type when known. Error bars are ±1 s.d., computed from repeated stimuli. (b) Individual spikes of LGN cells always carry more information about the stimulus than corresponding input RGC spikes. (c) Information capacity in redacted RGC spike trains (randomly dropping a fraction of the spikes to equal the number in the LGN spike train) does not differ from observed RGC trains. (d) Multiplying firing rate (a) by bits/spike (b) reveals that LGN neurons occasionally attain transmission rates with no loss of retinal information (data points on the unity line). Population averages: RGC = 31.9 bits/s; LGN = 21.8 bits/s. Error bars are ±1 s.d., incorporating the errors when computing bits/spike (too small to plot usefully in panels b and c), and the variation in firing rate during repeated stimuli.

Lawrence C. Sincich, et al. J Neurosci. ;29(19):6207-6216.
Figure 4

Figure 4. From: Preserving information in neural transmission.

Stimulus representation changes from retina to LGN. (a,b) Normalized MID filters for the RGC (red) and LGN neuron (black) are temporally altered, as revealed by subtracting the filters (blue). MID1 yields the most information, while MID2 represents an orthogonal stimulus dimension adding maximal information to the first. Error estimates computed over data subsets had average standard errors of the means < 0.01 for each point along all filters (not shown). (c,d) Spike rate gain measures how far above the mean firing rate (at gain = 1) any stimulus can drive the cell, plotted as a function of projection value distributions (see Methods for details). Filter combinations significantly increase the spike rate gain of the LGN neuron (d) over the RGC (c). Projection values with positive standard deviations (SDs) represent stimuli with increasing resemblance to the filters. (e) Population data comparing RGC and LGN peak spike rate gains for all filters (e.g. the peaks in Supplementary Fig. 1g–i), normalized to the peak RGC STA rate gain. The combined MID filters exhibited the highest gains. (f) Peri-stimulus time histograms of spike rate gains for 3 cells, in response to the same repeated stimuli. Stimulus segment is the same as shown in Fig. 1d. When LGN gains (black) are above the unity gain line (gray), they exceed RGC gains (red); and when below unity, they are lower than RGC gains, suggesting a greater modulation range for LGN neurons. The information transmission ratios for these cell pairs were: magno ON = 1.0; magno OFF = 0.74; parvo ON = 0.75.

Lawrence C. Sincich, et al. J Neurosci. ;29(19):6207-6216.

Display Settings:

Items per page

Supplemental Content

Recent activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...
Write to the Help Desk