Overcoming Corticocentric Bias in Whole-Brain Modeling

This direction is split into other directions: cortical-subcortical integration. Its last version stays readable and citable here.

Version 1 of 1103 papers110 conceptsfrozen 2026-10-04

For decades, the field of neuroscience has been dominated by "cortical chauvinism"—a term coined by Peter Dayan to describe the tendency of theoretical models to focus almost exclusively on the wrinkly outer layer of the brain, the neocortex. Because the cortex is the seat of high-level cognition, language, and perception, it has become the primary target for neuroimaging, anatomical atlases, and deep learning models.

However, a growing body of research suggests that treating the cortex as a "monolith" of intelligence is a mistake. By ignoring the subcortical "inner engine"—structures like the thalamus, basal ganglia, and cerebellum—we are missing the essential machinery that translates sensory predictions into motor actions and value-based decisions.

The Anatomical Blind Spot

The bias toward the cortex is often a matter of technical convenience. Many of our most powerful tools are designed specifically for the cortical surface. For example, CortexMAE, a foundation model for fMRI developed by Connor Lane and Mihir Tripathy, treats brain activity as a 2D "video" using cortical flat maps. While efficient, this representation inherently excludes the deep-brain structures that reside beneath the surface.

Similarly, deep learning frameworks like TopoFit and BrainSurfCNN reconstruct and analyze the brain as a 3D mesh. These models excel at capturing the complex folds of the cortex but often fail to encompass the complex internal geometries of the thalamus or hippocampus. As Christian F. Beckmann notes in his work on Masked Mesh Learning, restricting analysis to surface meshes prevents a holistic understanding of pathologies that involve both the surface and deeper anatomical components.

The Functional Necessity of Subcortical Loops

Why does this exclusion matter? The subcortex is not merely a passive relay station; it is a critical regulator of cortical activity. Madhura R. Joglekar and Xiao‐Jing Wang have shown that signal propagation across the primate cortex relies on "Global Balanced Amplification," a mechanism that likely requires subcortical structures like the thalamus and pulvinar to regulate attention and modulate information flow between distant cortical regions.

Furthermore, our understanding of human intelligence is incomplete without the subcortex. Yunan Wu and colleagues used geometric deep learning to show that the 3D shapes of the amygdala and nucleus accumbens—regions involved in emotion and reward—are strong predictors of fluid intelligence. This suggests that abstract reasoning is not just a "frontal lobe" task but an integrated process involving deep-brain systems.

Clinical Implications of the Inner Engine

In clinical research, the corticocentric bias can lead to "missing" the primary drivers of disease. In Alzheimer’s Disease, research traditionally targets the cortex and hippocampus. However, Omar Kana and Brian Long recently discovered that the caudate nucleus exhibits unique pathology that does not involve neuronal loss, potentially explaining non-cognitive symptoms like motor deficits that cortical models miss.

Similarly, in schizophrenia, Huaning Wang and Ling‐Li Zeng identified a dissociation pattern in the Default Mode Network (DMN). They found that subcortical regions are often excluded from these models due to low signal-to-noise ratios, yet these deep structures may be central to the pathophysiology of the condition.

Toward a Unified Whole-Brain Taxonomy

To solve this, the field is moving toward a standardized, whole-brain approach. The "Tower of Babel" problem—where different researchers use inconsistent names for brain networks—is being addressed by the Network Correspondence Toolbox (NCT), developed by Ru Kong and others. The NCT provides a quantitative bridge between new fMRI results and established atlases, with a specific goal of expanding support for subcortical and cerebellar regions.

Researchers like Lucina Q. Uddin are also pushing for a unified nomenclature that explicitly includes the thalamus, striatum, and cerebellum. By integrating these structures into our hierarchical models (Teddy J. Akiki) and spatiotemporal atlases (Ya Wang), neuroscience is finally looking beneath the surface to understand the brain as a truly integrated system.

Papers in this version

Showing 60 of 103 papers.

History & citationVersion 1 · 2026-10-04

Split into other directions · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_overcoming_corticocentric_bias_in_whole__9f902b05,
  title = {Overcoming Corticocentric Bias in Whole-Brain Modeling},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/overcoming-corticocentric-bias-in-whole-brain-modeling/rdr_9f902b05ebb744afa1d49e7995dde7e8},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_overcoming_corticocentric_bias_in_whole__9f902b05_v1,
  title = {Overcoming Corticocentric Bias in Whole-Brain Modeling},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/overcoming-corticocentric-bias-in-whole-brain-modeling/rdr_9f902b05ebb744afa1d49e7995dde7e8/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}