A high frequency resonance in the responses of retinal ganglion cells to rapidly modulated stimuli: A computer model

Abstract

Brisk Y-type ganglion cells in the cat retina exhibit a high frequency resonance (HFR) in their responses to large, rapidly modulated stimuli. We used a computer model to test whether negative feedback mediated by axon-bearing amacrine cells onto ganglion cells could account for the experimentally observed properties of HFRs. Temporal modulation transfer functions (tMTFs) recorded from model ganglion cells exhibited HFR peaks whose amplitude, width, and locations were qualitatively consistent with experimental data. Moreover, the wide spatial distribution of axon-mediated feedback accounted for the observed increase in HFR amplitude with stimulus size. Model phase plots were qualitatively similar to those recorded from Y ganglion cells, including an anomalous phase advance that in our model coincided with the amplification of low-order harmonics that overlapped the HFR peak. When axon-mediated feedback in the model was directed primarily to bipolar cells, whose synaptic output was graded, or else when the model was replaced with a simple cascade of linear filters, it was possible to produce large HFR peaks but the region of anomalous phase advance was always eliminated, suggesting the critical involvement of strongly non-linear feedback loops. To investigate whether HFRs might contribute to visual processing, we simulated high frequency ocular tremor by rapidly modulating a naturalistic image. Visual signals riding on top of the imposed jitter conveyed an enhanced representation of large objects. We conclude that by amplifying responses to ocular tremor, HFRs may selectively enhance the processing of large image features.

See globally, spike locally: oscillations in a retinal model encode large visual features

“…Total luminance (dashed light gray line), which grows linearly with pixel density, cannot account for the sharp transition in gamma activity. There were slight differences in the implementation of the retinal circuit model used for these experiments (Miller et al 2006), but these had no qualitative effect on our conclusions (Fig. 13c, dashed light gray line).…”

Extreme synergy: Spatiotemporal correlations enable rapid image reconstruction from computer-generated spike trains

“…Moreover, high-frequency retinal oscillations are strongly size dependent (Ishikane et al, 1999;Neuenschwander et al, 1999;Stephens et al, 2006), precluding the use of standard visual stimuli that are matched to the preferred spatial frequency of the cells under study. In vitro preparations also lack long-range spatiotemporal correlations likely to result from highfrequency fixational eye movements (Martinez-Conde et al, 2004), which theoretical models indicate may tap into high-frequency resonant circuitry (Miller, Denning, George, Marshak, & Kenyon, 2006). Even in the absence of a contribution from resonance circuitry in the inner retina, the present findings suggest that the spatiotemporal correlations generated by fixational eye movements are by themselves sufficient to convey precise pixel-by-pixel intensity information in an extremely synergistic manner.…”

Extracting Number-Selective Responses from Coherent Oscillations in a Computer Model

“…Coherent, object-specific, gamma-band oscillations produced by the common input model encoded numerosity in a manner that was very similar to the dynamic feedback circuit model. Connors, 2002), nonlinear dendritic processing (Polsky et al, 2004), and dynamical resonances (Miller et al, 2006;Frishman et al, 1987;Rager & Singer, 1998;Vigh, Solessio, Morgans, & Lasater, 2003) might confer a differential sensitivity to oscillatory input.…”

“…By increasing the overall delay of the axon-mediated negative feedback loop, the above changes tended to strengthen the coherent oscillations in the retinal model, but this effect was compensated for by reducing the strength of the feedback inhibition onto the model ganglion cells by a factor of 0.44. Overall, the above changes made no qualitative difference in the behavior of the coherent oscillations in the dynamic feedback circuit but were implemented to make the model more realistic and provided a better fit to the high frequency resonance in the responses of cat Y ganglion cells to rapidly modulated stimuli (Frishman, Freeman, Troy, Schweitzer-Tong, & Enroth-Cugell, 1987;Miller, Denning, George, Marshak, & Kenyon, 2006).…”