Controversies and progress on standardization of large-scale brain network nomenclature

Lucina Q. Uddin, Richard F. Betzel, Jessica R. Cohen, Jessica S. Damoiseaux, Felipe De Brigard, Simon B. Eickhoff, Alex Fornito, Caterina Gratton, Evan M. Gordon, Angela R. Laird, Linda Larson‐Prior, Anthony R. McIntosh, Lisa D. Nickerson, Luiz Pessoa, Ana Lúısa Pinho, Russell A. Poldrack, Adeel Razi, Sepideh Sadaghiani, James M. Shine, Anastasia Yendiki, B.t. Thomas Yeo, R. Nathan Spreng

TL;DR

The "WHATNET" workgroup addresses the "Tower of Babel" problem in neuroimaging: the lack of a standardized naming system for large-scale brain networks. By surveying the community and reviewing technical hurdles like inter-individual variability and spatial resolution, the authors provide a roadmap and minimal reporting guidelines to move the field toward a unified taxonomy.

What Problem Does This Paper Address?

In the last two decades, "network neuroscience" has exploded, yet the field lacks a consensus on its most fundamental constructs. One researcher’s "Salience Network" is another’s "Cingulo-Opercular Network." This inconsistency isn't just a semantic annoyance; it hinders replication, data sharing, and the synthesis of findings across different labs.

The challenge is multifaceted:

  1. Functional vs. Anatomical Naming: Should we name a network by what it does (e.g., "Executive Control") or where it is (e.g., "Lateral Frontoparietal")?
  2. The Scale Problem: Brain networks are hierarchical. Depending on the resolution of your analysis, the "Default Mode Network" might appear as one giant system or three distinct subsystems.
  3. Individual Differences: Brains are like fingerprints. A "standard" network map might be displaced or shaped differently in a child, an elderly person, or a patient with a brain injury.

Methods

The authors, members of the Workgroup for HArmonized Taxonomy of NETworks (WHATNET), employed a two-pronged approach:

  • Community Survey: They conducted an online survey of 956 members of the neuroimaging community (primarily via the Organization for Human Brain Mapping). Participants were shown 93 brain network images and asked to "Please name this network."
  • Technical Analysis: They used data-driven methods—specifically Spatial Independent Component Analysis (sICA) and hierarchical clustering—to demonstrate how network definitions change across different "model orders" (levels of detail).
  • Literature Synthesis: They reviewed five critical challenges: spatial scale/hierarchy, inter-individual variability, time-varying dynamics, subcortical affiliations, and multi-modal (fMRI vs. EEG/MEG) convergence.

Datasets

  • WHATNET Survey Data: A newly introduced dataset of community responses, now publicly available via the Open Science Framework (OSF).
  • Human Connectome Project (HCP): Specifically, the HCP_PTN820 group-level functional parcellations were used to illustrate how networks fractionate as analysis resolution increases.

Key Insights and Ideas

  • The Consensus Gap: There is high agreement on "unimodal" networks (Visual and Somatomotor) and the "Default Mode Network." However, consensus collapses for "higher-order" distributed networks like the Salience or Frontoparietal systems.
  • The "Model Order" Trap: The paper highlights that a "network" is often an artifact of the researcher's choice of resolution. At low resolution, the brain looks like a few big systems; at high resolution, those systems split into "subnetworks." Without a taxonomy, it’s unclear if these should be treated as parts or wholes.
  • Subcortical Neglect: Most current network maps are "cortico-centric." The authors argue that a true taxonomy must include the thalamus, striatum, and cerebellum, which are often ignored due to signal-to-noise challenges in fMRI.
  • Function is Context-Dependent: Naming a network after a single cognitive function (e.g., "The Pain Network") is misleading because most networks are "flexible hubs" that participate in many different tasks.

Key Results

  • Survey Agreement: The "Somato network" had the highest agreement (96.97%), followed by the "Default network" (92.93%) and "Visual network" (92.08%).
  • The "Lure" Effect: Images of "lure" networks (non-canonical ensembles) received 50% or less agreement, showing that researchers are hesitant when a pattern doesn't fit a well-known template.
  • Reporting Guidelines (Box 1): The paper concludes with six concrete recommendations, including:
  • Avoid idiosyncratic cognitive labels.
  • Reference findings against existing, well-defined parcellation schemes (like the Yeo or Gordon atlases).
  • Clearly report the demographic characteristics of the sample used to define the network.

Limitations and Future Work

The authors acknowledge that a "Universal Taxonomy" is currently impossible because the field is still discovering how networks reconfigure during tasks and across the lifespan.

Future directions include: * Adversarial Collaboration: Encouraging labs with competing naming traditions (e.g., "Salience" vs. "Ventral Attention") to co-design experiments to resolve their differences. * Cross-Species Mapping: Delineating how human brain networks relate to those in non-human primates and rodents to find evolutionarily conserved "core" systems. * Automated Labeling Tools: The group is developing a tool to quantify the spatial overlap between a researcher's new findings and 16 commonly used brain atlases to ensure more objective reporting.

Summary updated

Abstract

Progress in scientific disciplines is accompanied by standardization of terminology. Network neuroscience, at the level of macroscale organization of the brain, is beginning to confront the challenges associated with developing a taxonomy of its fundamental explanatory constructs. The Workgroup for HArmonized Taxonomy of NETworks (WHATNET) was formed in 2020 as an Organization for Human Brain Mapping (OHBM)-endorsed best practices committee to provide recommendations on points of consensus, identify open questions, and highlight areas of ongoing debate in the service of moving the field toward standardized reporting of network neuroscience results. The committee conducted a survey to catalog current practices in large-scale brain network nomenclature. A few well-known network names (e.g., default mode network) dominated responses to the survey, and a number of illuminating points of disagreement emerged. We summarize survey results and provide initial considerations and recommendations from the workgroup. This perspective piece includes a selective review of challenges to this enterprise, including (1) network scale, resolution, and hierarchies; (2) interindividual variability of networks; (3) dynamics and nonstationarity of networks; (4) consideration of network affiliations of subcortical structures; and (5) consideration of multimodal information. We close with minimal reporting guidelines for the cognitive and network neuroscience communities to adopt.
BibTeX
@inproceedings{uddin2023controversies,
  author = {Lucina Q. Uddin and Richard F. Betzel and Jessica R. Cohen and Jessica S. Damoiseaux and Felipe De Brigard and Simon B. Eickhoff and Alex Fornito and Caterina Gratton and Evan M. Gordon and Angela R. Laird and Linda Larson‐Prior and Anthony R. McIntosh and Lisa D. Nickerson and Luiz Pessoa and Ana Lúısa Pinho and Russell A. Poldrack and Adeel Razi and Sepideh Sadaghiani and James M. Shine and Anastasia Yendiki and B.T. Thomas Yeo and R. Nathan Spreng},
  title = {{Controversies and progress on standardization of large-scale brain network nomenclature}},
  booktitle = {Network},
  year = {2023},
  doi = {10.1162/netn\_a\_00323},
}

Loading related content...

Figure