Engineering the Visual System

A Special Invited Session on Engineering the Visual System on 8 February 2008 will feature a Keynote Address by Robert G. Smith, Ph.D., Department of Neuroscience, School of Medicine, University of Pennsylvania, Philadelphia, PA.

Sensitivity and Coding in the Retinal Ganglion Cell

The retinal ganglion cell collects signals from several pathways, which serve different ranges of light intensity. The circuit for each pathway differs according to its adaptational requirements and specific neural code. The rod bipolar pathway collects signals in the range of starlight to moonlight which are transduced as single photon signals generated by a photoreceptor. Intrinsic thermal noise and synaptic noise would limit the ultimate sensitivity of this pathway, which therefore requires special nonlinear processing to maintain signal quality. A model of the rod pathway shows the essential processing steps: each single photon signal passes through a threshold nonlinearity to remove continuous dark noise, and an array of AII amacrine cells collects and averages the single photon signals.

Signals in the cone pathway are processed by two layers of negative feedback that modulate synaptic release for highest signal quality. The ganglion cell is responsible for generating spikes sent to the brain, but the dynamic range of its axon is very limited, and its contrast sensitivity is limited by the efficiency of its spike generator. Using an ideal observer to measure contrast sensitivity, we compared signal quality in the real ganglion cell to a simple model to show the effect of synaptic noise from the presynaptic layers. The results show that although several synaptic stages are necessary to provide enough gain, the synapses, being noisy, reduce the ganglion cell's contrast sensitivity. Overall, the ganglion cell is exquisitely sensitive to fine contrasts over 10 log units of background intensity, and each synaptic processing mechanism, whether its role is to increase or modulate gain, adds noise and reduces performance.

Portrait of Dr. Rob Smith

Rob Smith's laboratory studies how retinal circuitry processes visual signals. He analyzes what is known about a circuit, constructs a biophysically realistic model of it, and through simulation attempts to reconcile the circuit's known physiological properties with the function of its neural components. This allows him to suggest a functional interpretation for biophysical features such as dendritic branching, density of voltage-gated membrane channels, and specific location, strength, and properties of synaptic inputs. By including the noise properties of membrane channels and synaptic vesicle release, he can generate realistic noise properties that can be compared directly with recordings from the live neurons. Rob's laboratory is currently focusing on 3 circuits:

  1. the cone photoreceptor to horizontal cell network,
  2. the pathway from rod photoreceptors to ganglion cells used during dark adaptation and
  3. the ganglion cell, its spike generator, and its presynaptic circuitry.