Alexander Sher

Spatio-temporal correlations and visual signalling in a complete neuronal population

Statistical dependencies in the responses of sensory neurons govern both the amount of stimulus information conveyed and the means by which downstream neurons can extract it. Although a variety of measurements indicate the existence of such dependencies, their origin and importance for neural coding are poorly understood. Here we analyse the functional significance of correlated firing in a complete population of macaque parasol retinal ganglion cells using a model of multi-neuron spike responses. The model, with parameters fit directly to physiological data, simultaneously captures both the stimulus dependence and detailed spatio-temporal correlations in population responses, and provides two insights into the structure of the neural code. First, neural encoding at the population level is less noisy than one would expect from the variability of individual neurons: spike times are more precise, and can be predicted more accurately when the spiking of neighbouring neurons is taken into account. Second, correlations provide additional sensory information: optimal, model-based decoding that exploits the response correlation structure extracts 20% more information about the visual scene than decoding under the assumption of independence, and preserves 40% more visual information than optimal linear decoding. This model-based approach reveals the role of correlated activity in the retinal coding of visual stimuli, and provides a general framework for understanding the importance of correlated activity in populations of neurons.

A maximum entropy model applied to spatial and temporal correlations from cortical networks in vitro.

Multineuron firing patterns are often observed, yet are predicted to be rare by models that assume independent firing. To explain these correlated network states, two groups recently applied a second-order maximum entropy model that used only observed firing rates and pairwise interactions as parameters (Schneidman et al., 2006; Shlens et al., 2006). Interestingly, with these minimal assumptions they predicted 90–99% of network correlations. If generally applicable, this approach could vastly simplify analyses of complex networks. However, this initial work was done largely on retinal tissue, and its applicability to cortical circuits is mostly unknown. This work also did not address the temporal evolution of correlated states. To investigate these issues, we applied the model to multielectrode data containing spontaneous spikes or local field potentials from cortical slices and cultures. The model worked slightly less well in cortex than in retina, accounting for 88 ± 7% (mean ± SD) of network correlations. In addition, in 8 of 13 preparations, the observed sequences of correlated states were significantly longer than predicted by concatenating states from the model. This suggested that temporal dependencies are a common feature of cortical network activity, and should be considered in future models. We found a significant relationship between strong pairwise temporal correlations and observed sequence length, suggesting that pairwise temporal correlations may allow the model to be extended into the temporal domain. We conclude that although a second-order maximum entropy model successfully predicts correlated states in cortical networks, it should be extended to account for temporal correlations observed between states.

Photovoltaic retinal prosthesis with high pixel density

Retinal degenerative diseases lead to blindness due to loss of the “image capturing” photoreceptors, while neurons in the “image processing” inner retinal layers are relatively well preserved. Electronic retinal prostheses seek to restore sight by electrically stimulating surviving neurons. Most implants are powered through inductive coils, requiring complex surgical methods to implant the coil-decoder-cable-array systems, which deliver energy to stimulating electrodes via intraocular cables. We present a photovoltaic subretinal prosthesis, in which silicon photodiodes in each pixel receive power and data directly through pulsed near-infrared illumination and electrically stimulate neurons. Stimulation was produced in normal and degenerate rat retinas, with pulse durations from 0.5 to 4 ms, and threshold peak irradiances from 0.2 to 10 mW/mm2, two orders of magnitude below the ocular safety limit. Neural responses were elicited by illuminating a single 70 μm bipolar pixel, demonstrating the possibility of a fully-integrated photovoltaic retinal prosthesis with high pixel density.

Functional connectivity in the retina at the resolution of photoreceptors.

To understand a neural circuit requires knowledge of its connectivity. Here we report measurements of functional connectivity between the input and ouput layers of the macaque retina at single-cell resolution and the implications of these for colour vision. Multi-electrode technology was used to record simultaneously from complete populations of the retinal ganglion cell types (midget, parasol and small bistratified) that transmit high-resolution visual signals to the brain. Fine-grained visual stimulation was used to identify the location, type and strength of the functional input of each cone photoreceptor to each ganglion cell. The populations of ON and OFF midget and parasol cells each sampled the complete population of long- and middle-wavelength-sensitive cones. However, only OFF midget cells frequently received strong input from short-wavelength-sensitive cones. ON and OFF midget cells showed a small non-random tendency to selectively sample from either long- or middle-wavelength-sensitive cones to a degree not explained by clumping in the cone mosaic. These measurements reveal computations in a neural circuit at the elementary resolution of individual neurons.

Direction Selectivity in the Retina Is Established Independent of Visual Experience and Cholinergic Retinal Waves

Direction selectivity in the retina requires the asymmetric wiring of inhibitory inputs onto four subtypes of On-Off direction-selective ganglion cells (DSGCs), each preferring motion in one of four cardinal directions. The primary model for the development of direction selectivity is that patterned activity plays an instructive role. Here, we use a unique, large-scale multielectrode array to demonstrate that DSGCs are present at eye opening, in mice that have been reared in darkness and in mice that lack cholinergic retinal waves. These data suggest that direction selectivity in the retina is established largely independent of patterned activity and is therefore likely to emerge as a result of complex molecular interactions.

Spatial-Temporal Patterns of Retinal Waves Underlying Activity-Dependent Refinement of Retinofugal Projections

During development, retinal axons project coarsely within their visual targets before refining to form organized synaptic connections. Spontaneous retinal activity, in the form of acetylcholine-driven retinal waves, is proposed to be necessary for establishing these projection patterns. In particular, both axonal terminations of retinal ganglion cells (RGCs) and the size of receptive fields of target neurons are larger in mice that lack the beta2 subunit of the nicotinic acetylcholine receptor (beta2KO). Here, using a large-scale, high-density multielectrode array to record activity from hundreds of RGCs simultaneously, we present analysis of early postnatal retinal activity from both wild-type (WT) and beta2KO retinas. We find that beta2KO retinas have correlated patterns of activity, but many aspects of these patterns differ from those of WT retina. Quantitative analysis suggests that wave directionality, coupled with short-range correlated bursting patterns of RGCs, work together to refine retinofugal projections.

High sensitivity rod photoreceptor input to the blue-yellow color opponent pathway in macaque retina

Small bistratified cells (SBCs) in the primate retina carry a major blue-yellow opponent signal to the brain. We found that SBCs also carry signals from rod photoreceptors, with the same sign as S cone input. SBCs exhibited robust responses under low scotopic conditions. Physiological and anatomical experiments indicated that this rod input arose from the AII amacrine cell–mediated rod pathway. Rod and cone signals were both present in SBCs at mesopic light levels. These findings have three implications. First, more retinal circuits may multiplex rod and cone signals than were previously thought to, efficiently exploiting the limited number of optic nerve fibers. Second, signals from AII amacrine cells may diverge to most or all of the ∼20 retinal ganglion cell types in the peripheral primate retina. Third, rod input to SBCs may be the substrate for behavioral biases toward perception of blue at mesopic light levels.

A non-canonical pathway for mammalian blue-green color vision

The dynamic range of visual coding is extended by having separate ganglion cell types that respond to light increments and decrements. Although the primordial color vision system in mammals contains a well-characterized ganglion cell that responds to blue light increments (a blue On center cell), less is known about ganglion cells that respond to blue light decrements (blue Off center cells). We identified a regular mosaic of blue Off center ganglion cells in the ground squirrel. Contrary to the standard scheme, blue Off responses came from a blue On bipolar and inverting amacrine cell.

Restoration of Retinal Structure and Function after Selective Photocoagulation

CNS neurons change their connectivity to accommodate a changing environment, form memories, or respond to injury. Plasticity in the adult mammalian retina after injury or disease was thought to be limited to restructuring resulting in abnormal retinal anatomy and function. Here we report that neurons in the mammalian retina change their connectivity and restore normal retinal anatomy and function after injury. Patches of photoreceptors in the rabbit retina were destroyed by selective laser photocoagulation, leaving retinal inner neurons (bipolar, amacrine, horizontal, ganglion cells) intact. Photoreceptors located outside of the damaged zone migrated to make new functional connections with deafferented bipolar cells located inside the lesion. The new connections restored ON and OFF responses in deafferented ganglion cells. This finding extends the previously perceived limits of restorative plasticity in the adult retina and allows for new approaches to retinal laser therapy free of current detrimental side effects such as scotomata and scarring.

Photodiode Circuits for Retinal Prostheses

Photodiode circuits show promise for the development of high-resolution retinal prostheses. While several of these systems have been constructed and some even implanted in humans, existing descriptions of the complex optoelectronic interaction between light, photodiode, and the electrode/electrolyte load are limited. This study examines this interaction in depth with theoretical calculations and experimental measurements. Actively biased photoconductive and passive photovoltaic circuits are investigated, with the photovoltaic circuits consisting of one or more diodes connected in series, and the photoconductive circuits consisting of a single diode in series with a pulsed bias voltage. Circuit behavior and charge injection levels were markedly different for platinum and sputtered iridium-oxide film (SIROF) electrodes. Photovoltaic circuits were able to deliver 0.038 mC/cm2 (0.75 nC/phase) per photodiode with 50- μm platinum electrodes, and 0.54-mC/cm2 (11 nC/phase) per photodiode with 50-μ m SIROF electrodes driven with 0.5-ms pulses of light at 25 Hz. The same pulses applied to photoconductive circuits with the same electrodes were able to deliver charge injections as high as 0.38 and 7.6 mC/cm2 (7.5 and 150 nC/phase), respectively. We demonstrate photovoltaic stimulation of rabbit retina in-vitro, with 0.5-ms pulses of 905-nm light using peak irradiance of 1 mW/mm2. Based on the experimental data, we derive electrochemical and optical safety limits for pixel density and charge injection in various circuits. While photoconductive circuits offer smaller pixels, photovoltaic systems do not require an external bias voltage. Both classes of circuits show promise for the development of high-resolution optoelectronic retinal prostheses.