Cortical hierarchy, dual counterstream architecture and the importance of top-down generative networks

Vezoli, Julien; Magrou, Loïc; Goebel, Rainer; Wang, Xiaojing; Knoblauch, Kenneth; Vinck, Martin; Kennedy, Henry

doi:10.1016/j.neuroimage.2020.117479

Cited by 91 publications

(115 citation statements)

References 169 publications

Supporting

Mentioning

109

Contrasting

Order By: Relevance

“…This article provides an overview and unified analysis of laminar cortical models of multiple types of biological intelligence, including 3D vision and figure-ground perception; attentive category learning and decision-making; speech perception; and cognitive working memory (WM), planning, and prediction. These models are consistent with descriptions and statistical analyses of cortical connectivity patterns (e.g., Gilbert and Wiesel, 1979;Felleman and Van Essen, 1991;Markov et al, 2013;Vezoli et al, 2021), as well as recent data about ''visual evoked responses in laminar recordings from six cortical areas in awake mice'' (Barzegaran and Plomp, 2021).…”

Section: Introduction: Why Does the Cerebral Cortex Have Layers?supporting

confidence: 87%

A Canonical Laminar Neocortical Circuit Whose Bottom-Up, Horizontal, and Top-Down Pathways Control Attention, Learning, and Prediction

Grossberg

2021

Front. Syst. Neurosci.

View full text Add to dashboard Cite

All perceptual and cognitive circuits in the human cerebral cortex are organized into layers. Specializations of a canonical laminar network of bottom-up, horizontal, and top-down pathways carry out multiple kinds of biological intelligence across different neocortical areas. This article describes what this canonical network is and notes that it can support processes as different as 3D vision and figure-ground perception; attentive category learning and decision-making; speech perception; and cognitive working memory (WM), planning, and prediction. These processes take place within and between multiple parallel cortical streams that obey computationally complementary laws. The interstream interactions that are needed to overcome these complementary deficiencies mix cell properties so thoroughly that some authors have noted the difficulty of determining what exactly constitutes a cortical stream and the differences between streams. The models summarized herein explain how these complementary properties arise, and how their interstream interactions overcome their computational deficiencies to support effective goal-oriented behaviors.

show abstract

Section: Introduction: Why Does the Cerebral Cortex Have Layers?supporting

confidence: 87%

A Canonical Laminar Neocortical Circuit Whose Bottom-Up, Horizontal, and Top-Down Pathways Control Attention, Learning, and Prediction

Grossberg

2021

Front. Syst. Neurosci.

View full text Add to dashboard Cite

show abstract

“…Overall, feedback amplitudes were small compared to the feedforward amplitudes 35 , and while feedback was transient, feedforward networks showed more robust evoked responses and for longer time. Anatomical studies in monkey show that feedback projections are more numerous than feedforward projections, but that feedforward projections are stronger, especially at shorter range 6 .…”

Section: Discussionmentioning

confidence: 87%

“…The second strongest network contributed 18 ± 2% of total model amplitude, for both stimulus contrast. Its laminar connectivity pattern showed L2 and L3 as both the strongest source and target layers (Figure 3a), a pattern that is typically associated with feedforward processing streams in mouse and primate 6,33–35 .…”

Section: Resultsmentioning

confidence: 99%

“…One was a supragranular network, with putative L1 as the main source and target of inter-areal connections, the other was an infragranular network with L6 as the main source and target ( Figure 3a). Anatomical studies of primate visual cortex in monkey established that besides the well-known feedback pathway over L6, another feedback pathway exists over supragranular layers L1 and L2 6,34,35 . Our findings for the first time distinguish these two feedback pathways using analysis of functional data and demonstrate their presence in mouse.…”

Section: Two Low-frequency Feedback Networkmentioning

confidence: 99%

See 1 more Smart Citation

Multiple concurrent feedforward and feedback streams in a cortical hierarchy

Barzegaran

Plomp

2021

Preprint

View full text Add to dashboard Cite

Visual stimuli evoke fast-evolving activity patterns that are distributed across multiple cortical areas. These areas are hierarchically structured, as indicated by their anatomical projections, but how large-scale feedforward and feedback streams are functionally organized in this system remains an important missing clue to understanding cortical processing. By analyzing visual evoked responses in laminar recordings from six cortical areas in awake mice, we established the simultaneous presence of two feedforward and two feedback networks, each with a distinct laminar functional connectivity profile, frequency spectrum, temporal dynamics and functional hierarchy. We furthermore identified a distinct role for each of these four processing streams, by leveraging stimulus contrast effects and analyzing receptive field convergency along functional interactions. Our results support a dynamic dual counterstream view of hierarchical processing and provide new insight into how separate functional streams can simultaneously and dynamically operate in visual cortex.

show abstract

“…Research has investigated the hierarchical organisation of brain structure and function [31][32][33][34] , which has demonstrated a common network available for the orchestration of task and rest 15 . Previous research has demonstrated that the turbulent core of brain activity is largely overlapping with brain regions known to be involved in lower level sensory processing 15 .…”

Section: Discussionmentioning

confidence: 99%

Rare long-range cortical connections enhance information processing

Deco

Perl

Vuust

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

A fundamental and unanswered question concerns the key topological features of connectivity that are critically relevant for generating the dynamics underlying efficient cortical function. A candidate feature that has recently emerged is that the connectivity of the mammalian cortex follows an exponential distance rule, which uniquely includes a small proportion of long-range high-weight anatomical connections. We investigate how these long-range connections influence whole-brain dynamics with coupled oscillators. To understand the causal function of long-range connections, we first studied these connections in simple ring structures and then in complex empirical brain architectures. A small proportion of long-range connections are sufficient for significantly improving information transmission, i.e. information cascade. Large-scale empirical neuroimaging modelling point to the immense functional benefits for information processing of a brain architecture with long-range coupling that improves the information cascade thanks to the underlying turbulent regime of brain dynamics.

show abstract

Cortical hierarchy, dual counterstream architecture and the importance of top-down generative networks

Cited by 91 publications

References 169 publications

A Canonical Laminar Neocortical Circuit Whose Bottom-Up, Horizontal, and Top-Down Pathways Control Attention, Learning, and Prediction

A Canonical Laminar Neocortical Circuit Whose Bottom-Up, Horizontal, and Top-Down Pathways Control Attention, Learning, and Prediction

Multiple concurrent feedforward and feedback streams in a cortical hierarchy

Rare long-range cortical connections enhance information processing

Contact Info

Product

Resources

About