Anatomy of hierarchy: Feedforward and feedback pathways in macaque visual cortex

Nikola T. Markov, Julien Vezoli, Pascal Chameau, Arnaud Falchier, René Quilodran, Cyril Huissoud, Camille Lamy, Pierre Misery, Pascale Giroud, Shimon Ullman, Pascal Barone, Colette Dehay, Kenneth Knoblauch, Henry KennedyThe Journal of Comparative Neurology 2013 · DOI

TL;DR

This study reveals a fundamental organizational principle of the primate visual cortex: distinct "counterstreams" of feedforward (bottom-up) and feedback (top-down) connections reside in different layers of the cortex. This laminar segregation, particularly in supragranular layers, underpins a hierarchical organization that is more complex and compartmentalized than previously understood, enabling precise point-to-point communication in both directions.

What Problem Does This Paper Address?

The brain's visual system is organized hierarchically, with information flowing from lower-level areas (like V1) to higher-level areas, and feedback signals flowing in the opposite direction. While this general principle is accepted, the precise anatomical wiring and functional principles governing these feedforward (FF) and feedback (FB) pathways remain poorly understood. Specifically, how the laminar organization of neurons within cortical areas dictates these pathways and contributes to the overall hierarchy was a significant gap. This paper aims to provide a quantitative anatomical basis for understanding these FF and FB pathways and their role in cortical hierarchy.

Methods

The researchers employed a combination of sophisticated techniques:

  • Retrograde Tracing: They injected two different fluorescent tracers (Fast Blue and Diamidino Yellow) into various visual areas of macaque monkeys. By analyzing where these tracers were taken up by neurons and transported back to their cell bodies, they could map the origins of connections to specific areas.
  • Quantitative Analysis of Neuronal Location (SLN): A key innovation was quantifying the "percentage of supragranular labeled neurons" (SLN). This metric, representing the proportion of neurons projecting to a target area that reside in the upper cortical layers (supragranular), was used as an index of hierarchical distance. A high SLN suggests a feedforward connection, while a low SLN suggests a feedback connection.
  • Cell Filling and Morphology: To understand the physical structure of neurons involved in FF and FB pathways, they filled retrogradely labeled neurons with biocytin and examined their morphology, including dendritic arborization.
  • Statistical Modeling: They used statistical models, including a beta-binomial model, to analyze the SLN data and estimate hierarchical levels, accounting for overdispersion in the data.

Datasets

Not discussed in the paper. The study relies on experimental data collected from macaque monkeys.

Key Insights and Ideas

  • Laminar Segregation is Key: The most striking insight is that FF and FB pathways are not just defined by their direction but also by their laminar location. Supragranular layers are primarily associated with FF pathways, exhibiting point-to-point connectivity, while infragranular layers are more involved in diffuse FF and FB streams.
  • SLN as a Hierarchical Ruler: The SLN metric proved to be a powerful tool for constraining models of cortical hierarchy, revealing previously unsuspected relationships between areas. It acts as a quantitative index of hierarchical distance.
  • Combinatorial Distance Rule: The SLN values suggest a combinatorial distance rule, where connectivity patterns are influenced by both the direction of the pathway (FF/FB) and the physical distance between connected areas.
  • Compartmentalized Counterstreams: The visual cortex appears to be organized into distinct "counterstreams" of FF and FB connections, with supragranular layers housing highly segregated streams and infragranular layers housing more diffuse ones. This compartmentalization is crucial for precise information processing.
  • FF vs. FB Asymmetry: FF pathways are generally more numerous, have higher weights, and cross fewer hierarchical levels compared to FB pathways.

Key Results

  • Hierarchical Constraints: SLN values provided strong constraints on models of cortical hierarchy, revealing new areal relationships.
  • Supragranular Segregation: Supragranular layers showed highly segregated FF (bottom-up) and FB (top-down) streams with point-to-point connectivity.
  • Infragranular Diffusion: Infragranular layers contained more diffuse FF and FB streams.
  • Cell Filling Evidence: Cell filling experiments confirmed the compartmentalization of FF and FB neurons, with distinct dendritic arborizations.
  • Pathway Incidence: FB pathways were found to be more numerous than FF pathways, but FF pathways had higher average weights.
  • Distance Matters: Connectivity strength (FLN) decreased with increasing physical distance, with different patterns for FF and FB pathways.

Limitations and Future Work

  • Model Limitations: While the SLN model provided valuable insights, the authors acknowledge that it is restricted to the interval (0, 1), which can lead to foreshortening of distances. They also note that the beta-binomial model, while robust, is still a statistical approximation.
  • Functional Interpretation: While the study provides a detailed anatomical framework, further research is needed to fully elucidate the functional roles of these distinct FF and FB counterstreams and how they support cognitive processes like attention and prediction.
  • Molecular Characterization: The authors suggest extending this work by combining anatomical data with molecular characterization of parent neurons to gain a deeper understanding of pathway specificity.
  • Anterograde Tracing: Future studies using anterograde tracers are recommended to examine the laminar integration of interareal connectivity and synaptic-level details.
  • Dynamic Hierarchy: The possibility of a dynamic, task-dependent hierarchy, constrained by the structural hierarchy, is an avenue for future exploration.

Summary updated

Abstract

The laminar location of the cell bodies and terminals of interareal connections determines the hierarchical structural organization of the cortex and has been intensively studied. However, we still have only a rudimentary understanding of the connectional principles of feedforward (FF) and feedback (FB) pathways. Quantitative analysis of retrograde tracers was used to extend the notion that the laminar distribution of neurons interconnecting visual areas provides an index of hierarchical distance (percentage of supragranular labeled neurons [SLN]). We show that: 1) SLN values constrain models of cortical hierarchy, revealing previously unsuspected areal relations; 2) SLN reflects the operation of a combinatorial distance rule acting differentially on sets of connections between areas; 3) Supragranular layers contain highly segregated bottom-up and top-down streams, both of which exhibit point-to-point connectivity. This contrasts with the infragranular layers, which contain diffuse bottom-up and top-down streams; 4) Cell filling of the parent neurons of FF and FB pathways provides further evidence of compartmentalization; 5) FF pathways have higher weights, cross fewer hierarchical levels, and are less numerous than FB pathways. Taken together, the present results suggest that cortical hierarchies are built from supra- and infragranular counterstreams. This compartmentalized dual counterstream organization allows point-to-point connectivity in both bottom-up and top-down directions.
BibTeX
@article{markov2013anatomy,
  author = {Nikola T. Markov and Julien Vezoli and Pascal Chameau and Arnaud Falchier and René Quilodran and Cyril Huissoud and Camille Lamy and Pierre Misery and Pascale Giroud and Shimon Ullman and Pascal Barone and Colette Dehay and Kenneth Knoblauch and Henry Kennedy},
  title = {{Anatomy of hierarchy: Feedforward and feedback pathways in macaque visual cortex}},
  journal = {The Journal of Comparative Neurology},
  year = {2013},
  doi = {10.1002/cne.23458},
}
Figure