The Structure of Large-Scale Synchronized Firing in Primate Retina

Jonathon Shlens, Greg D. Field, Jeffrey L. Gauthier, Martin Greschner, Alexander Sher, A. M. Litke, E. J. ChichilniskyJournal of Neuroscience 2009 · DOI

Abstract

Synchronized firing among neurons has been proposed to constitute an elementary aspect of the neural code in sensory and motor systems. However, it remains unclear how synchronized firing affects the large-scale patterns of activity and redundancy of visual signals in a complete population of neurons. We recorded simultaneously from hundreds of retinal ganglion cells in primate retina, and examined synchronized firing in completely sampled populations of approximately 50-100 ON-parasol cells, which form a major projection to the magnocellular layers of the lateral geniculate nucleus. Synchronized firing in pairs of cells was a subset of a much larger pattern of activity that exhibited local, isotropic spatial properties. However, a simple model based solely on interactions between adjacent cells reproduced 99% of the spatial structure and scale of synchronized firing. No more than 20% of the variability in firing of an individual cell was predictable from the activity of its neighbors. These results held both for spontaneous firing and in the presence of independent visual modulation of the firing of each cell. In sum, large-scale synchronized firing in the entire population of ON-parasol cells appears to reflect simple neighbor interactions, rather than a unique visual signal or a highly redundant coding scheme.

Introduction: The Language of Neural Synchrony

A central question in neuroscience is how populations of neurons coordinate their activity to represent information. While synchronized firing—the simultaneous spiking of multiple neurons—is a well-known phenomenon, its global structure has been difficult to characterize. Previous studies were often limited to small subsets of cells, leaving it unclear whether synchrony emerges from complex, long-range circuits or simple local interactions.

Shlens et al. (2009) investigated this by recording from nearly every ON-parasol retinal ganglion cell in a specific patch of the macaque monkey retina. This "complete mosaic" approach allowed them to move beyond pairwise correlations and observe the full spatial extent of synchronized activity across a functional population.

Methods: Mapping the Complete Mosaic

The researchers utilized a high-density 512-electrode array to record from isolated macaque retina. They focused on ON-parasol cells, which project to the magnocellular layers of the lateral geniculate nucleus (LGN) and are critical for high-speed visual processing.

To analyze the data, they employed Maximum Entropy Models, specifically a "pairwise-adjacent" model inspired by the Ising model in statistical physics. This model assumes that the state of the entire population can be predicted using only the individual firing rates and the correlations between immediate neighbors in the retinal mosaic. By comparing this model's predictions to actual recordings under both constant illumination and visual stimulation, they could determine if local rules are sufficient to explain large-scale population dynamics.

Results: Local Interactions, Global Patterns

Islands of Synchronized Activity

The recordings revealed that synchronized firing is not just a pairwise phenomenon but involves large, contiguous "islands" of neurons. These synchronized events were surprisingly extensive, sometimes involving up to 26 contiguous neurons or nearly half of the recorded population.

Spatial patterns of synchronized firing in the ON-parasol cell mosaic. Panels A and B show spontaneous activity under constant illumination, while C and D show activity during visual stimulation. Black ellipses represent neurons firing in a synchronized event.

These patterns were found to be isotropic (direction-independent) and translationally invariant across the retina, suggesting they are a fundamental structural property of the retinal circuitry rather than a response to specific visual features.

The Power of the Pairwise-Adjacent Model

The most significant finding was the accuracy of the simple pairwise-adjacent model. Despite the apparent complexity of the large-scale "islands," the model—which only considers interactions between immediate neighbors—reproduced 99% of the spatial structure of the synchronized firing.

The pairwise-adjacent model (solid line) accurately captures the distribution of synchronized firing events observed in the experimental data (bars), accounting for 99.8% of the structure.

This result indicates that long-range synchronization in the retina does not require specialized wide-field circuitry. Instead, it "propagates" through the network via local, nearest-neighbor interactions, much like ripples in a pond.

Discussion: Efficiency and Redundancy

The study also addressed the "redundancy paradox." If many neurons fire together in large blocks, one might assume the retina is sending highly redundant information, wasting neural bandwidth. However, the authors found that no more than 20% of an individual cell's firing variability could be predicted by its neighbors.

This suggests that the neural code in the ON-parasol mosaic is highly efficient. The retina manages to maintain structured, concerted activity patterns while ensuring that 80% of each neuron's signal remains independent. This balance allows for a robust population signal without sacrificing the information capacity of individual neurons. While this study focused on ON-parasol cells, it provides a foundational framework for understanding how local connectivity shapes global neural activity across the visual system.

Summary updated

Abstract

Synchronized firing among neurons has been proposed to constitute an elementary aspect of the neural code in sensory and motor systems. However, it remains unclear how synchronized firing affects the large-scale patterns of activity and redundancy of visual signals in a complete population of neurons. We recorded simultaneously from hundreds of retinal ganglion cells in primate retina, and examined synchronized firing in completely sampled populations of approximately 50-100 ON-parasol cells, which form a major projection to the magnocellular layers of the lateral geniculate nucleus. Synchronized firing in pairs of cells was a subset of a much larger pattern of activity that exhibited local, isotropic spatial properties. However, a simple model based solely on interactions between adjacent cells reproduced 99% of the spatial structure and scale of synchronized firing. No more than 20% of the variability in firing of an individual cell was predictable from the activity of its neighbors. These results held both for spontaneous firing and in the presence of independent visual modulation of the firing of each cell. In sum, large-scale synchronized firing in the entire population of ON-parasol cells appears to reflect simple neighbor interactions, rather than a unique visual signal or a highly redundant coding scheme.
BibTeX
@article{shlens2009structure,
  author = {Jonathon Shlens and Greg D. Field and Jeffrey L. Gauthier and Martin Greschner and Alexander Sher and A. M. Litke and E. J. Chichilnisky},
  title = {{The Structure of Large-Scale Synchronized Firing in Primate Retina}},
  journal = {Journal of Neuroscience},
  year = {2009},
  doi = {10.1523/jneurosci.5187-08.2009},
}
Figure