The cortico-thalamic loop attunes competitive lateral interactions across retinotopic and orientation preference maps

Domenico Guarino, Andrew P. Davison, Yves Frégnac, Ján AntolíkbioRxiv 2022 · DOI

TL;DR

This paper presents a detailed computational model of the early visual system that demonstrates how the dense feedback loop between the cortex and thalamus is crucial for refining visual representations. By simulating inactivation of this feedback, the authors show it enhances competition, selectivity, and contextual modulation in both thalamic and cortical neurons, explaining several long-standing experimental observations.

What Problem Does This Paper Address?

The early visual system, comprising the retina, thalamus (Lateral Geniculate Nucleus - LGN, and Perigeniculate Nucleus - PGN), and primary visual cortex (V1), is responsible for initial processing of visual information. While the feedforward pathway from the retina to V1 has been extensively studied, the role of the much denser feedback projections from V1 back to the thalamus has remained less understood. This feedback loop is thought to be important, but its precise functional contributions to visual processing, such as how it shapes neuronal responses and interactions, have been difficult to disentangle experimentally. This paper aims to bridge this gap by proposing and testing a unified computational framework that explicitly models this cortico-thalamic loop and its influence on neuronal activity and competitive interactions across retinotopic and orientation maps.

Methods

The authors developed a detailed conductance-based spiking network model of the cat's early visual system. This model incorporates:

  • Retinal Ganglion Cells (RGCs): Modeled with center-surround opponency.
  • Thalamus: Includes LGN (ON and OFF cells) and PGN, with realistic connectivity statistics and synaptic properties.
  • Primary Visual Cortex (V1): Modeled with excitatory and inhibitory neurons, and connections based on Gabor filters and a probability distribution.
  • Cortico-Thalamic Feedback: Explicitly modeled, including both direct excitatory connections from V1 to LGN and indirect inhibitory pathways involving PGN.

Crucially, the model was constrained by a wide range of experimental data, including electrophysiological properties, connectivity statistics, and functional responses to various visual stimuli. They then compared three model configurations:

  1. Full Model: All feedforward and feedback pathways are present (representing an intact brain).
  2. Variable Feedback: Cortical feedback is modulated (mimicking experimental inactivation).
  3. Feedforward-Only: Cortical feedback is absent.

This comparative approach allowed them to isolate the specific contributions of the cortico-thalamic feedback.

Datasets

The model was trained and validated against a comprehensive set of existing experimental data from cat and macaque visual systems. This included data on:

  • Neuronal tuning to contrast, spatial and temporal frequency, orientation, and size.
  • Synaptic properties and connectivity statistics.
  • Responses to stimuli under intact and inactivated cortical feedback conditions.

No new datasets were introduced; rather, the model was built to reproduce and explain a wide array of previously published findings.

Key Insights and Ideas

  • Feedback is not just modulation, it's active shaping: The study strongly suggests that cortico-thalamic feedback isn't merely a passive modulator but actively shapes thalamic and cortical representations by introducing competition and enhancing selectivity.
  • A unified mechanism for diverse phenomena: The model provides a single framework that can explain seemingly disparate phenomena like intra-thalamic surround suppression, orientation tuning in LGN, and enhanced cortical selectivity.
  • Competition within the hypercolumn: The feedback loop appears to sharpen competition between different orientation preferences within a local cortical area (hypercolumn) but doesn't extend this effect to broader surrounds.
  • Selective suppression of connectivity: The feedback loop selectively suppresses facilitatory interactions between neurons with similar orientation preferences, particularly for larger stimuli, suggesting a sophisticated tuning of cortical communication.

Key Results

  • Enhanced Thalamic Selectivity: The full model showed significantly increased selectivity for stimulus size and orientation in LGN neurons compared to the feedforward-only model. For instance, the suppression index for size tuning increased by 43.3%.
  • Cortical Feedback Drives Orientation Bias in LGN: The model demonstrated that cortical feedback can induce a bias for stimulus orientation in LGN neurons, particularly when the local cortical input is orientationally homogeneous.
  • Center-Surround Opponency: Cortical feedback was shown to enhance center-surround opponency in LGN cells, with inactivation leading to reduced responses for smaller stimuli and increased responses for larger stimuli in overlapping receptive fields.
  • Increased Cortical Selectivity: The full model exhibited enhanced suppression for non-preferred stimulus sizes and orientations in V1 neurons, leading to higher overall selectivity compared to the feedforward-only model.
  • Functional Connectivity Modulation: The cortico-thalamic loop was found to selectively suppress facilitatory interactions between neurons with matching orientation preferences, while having less impact on cross-oriented neurons.

Limitations and Future Work

  • Model Simplifications: The authors acknowledge that the model is a simplification of the biological system, for example, by using a simplified representation of retinal ganglion cells and down-sampling the number of neurons and synapses. They also note that the exact synaptic weights and connection densities for feedback pathways are not fully constrained by existing literature.
  • Parameter Sensitivity: While they performed parameter searches to assess stability, the precise values of some parameters (e.g., feedback connection density) might influence the quantitative outcomes.
  • Future Work: The authors suggest their model serves as a versatile computational toolbox for testing alternative hypotheses about the cortico-thalamic loop's function, which cannot be addressed by direct lesion experiments. They also highlight the need to further elucidate the broader purpose of these computations in the context of complex visual processing.

Summary updated

Abstract

Abstract In the early visual system, corticothalamic feedback projections greatly outnumber thalamocortical feedforward projections. Extensive experimental and modeling work has been devoted to the functional impact of the feedforward pathway, but the role of its denser feedback counterpart remains elusive. Here, we propose a novel unifying framework where thalamic recurrent interactions and corticothalamic feedback act in a closed-loop fashion to attune multiple stimulus representations. At each position of the visual field, the loop puts into competition local representations of the stimulus in thalamus and cortex through direct excitation of narrow topologically-aligned portions of the thalamus, accompanied with peri-geniculate nucleus mediated broad inhibition suppressing the topological surround. We built a detailed conductance-based spiking model incorporating retinal input, lateral geniculate nucleus, peri-geniculate nucleus, primary visual cortex, and all the relevant intra-areal and feedback pathways. For the first time we perform comparative analyses between model configurations with completely or locally inactivated cortico-thalamic feedback, as in the experimental preparations. The model mechanistically explains (i) the existence of intra-thalamic surround suppression, (ii) the sensitivity of thalamic neurons to orientation tuning, (iii) the cortex-dependent center-surround opponency in thalamic cells, (iv) the cortical increase of size and orientation selectivity, (v) the cortically enhanced competition between cross-oriented domains within the hypercolumn, and (vi) the selective suppression of cortical functional connectivity. Our results integrate decades of experimental and theoretical research, supporting the hypothesis that cortico-thalamic loop exerts competitive influence between neighboring regions in the thalamus and cortex, complementing the lateral intra-V1 interactions in center-surround contextual modulation.
BibTeX
@misc{guarino2022cortico,
  author = {Domenico Guarino and Andrew P. Davison and Yves Frégnac and Ján Antolík},
  title = {{The cortico-thalamic loop attunes competitive lateral interactions across retinotopic and orientation preference maps}},
  year = {2022},
  doi = {10.1101/2022.12.19.521069},
  howpublished = {bioRxiv},
}
Figure