What do neuroanatomical networks reveal about the ontology of human cognitive abilities?
summary
The gist
What do neuroanatomical networks reveal about the ontology of human cognitive abilities? The present study explores the ontological structure derived from neuroanatomical networks associated with…
In short
The study compared a brain-derived (neurometric) ontology with a performance-derived (psychometric) ontology using 15 cognitive tasks. Results showed partial consistency, meaning brain anatomy partly reflects cognitive ability patterns. While ability clusters are spatially segregated, they are integrated by dense structural connections, suggesting neural clusters support general intelligence.
Key concepts
- Neurometric Ontology
- This ontology is derived from anatomical brain properties like cortical thickness and myelination. It attempts to define cognitive abilities based on the physical characteristics of specific brain regions rather than just task performance.
- Psychometric Ontology
- This ontology is derived from the covariances between performance measures across various cognitive tasks. It defines abilities based on how well individuals perform on different types of mental tasks.
- Spatial Segregation
- This refers to the finding that cortical areas associated with different inferred abilities show little overlap. This means the brain regions supporting one specific ability tend to be physically located in distinct, non-overlapping parts of the cortex.
- Structural Integration
- Despite spatial segregation, these ability-related cortical areas are linked by dense white matter structural connections. This integration suggests that while abilities are specialized in location, they work together through strong physical wiring.
Terminology used across episodes
This episode discusses
- What do neuroanatomical networks reveal about the ontology of human cognitive abilities? · Paper Radio
The paper
What do neuroanatomical networks reveal about the ontology of human cognitive abilities? · Read on arXiv
Daniel Kristanto, Xinyang Liu, Werner Sommer, Andrea Hildebrandt, #Changsong Zhou
Department of Physics, Centre for Nonlinear Studies and Beijing-Hong Kong-Singapore Joint Centre for Nonlinear and Complex Systems (Hong Kong) · Carl von Ossietzky Universität Oldenburg · Shanghai Key Laboratory of Brain Functional Genomics (Ministry of Education) · Humboldt University at Berlin · Zhejiang Normal University · Zhejiang University
DOI: 10.1016/j.isci.2022.104706
Transcript
Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.
Ines: We're looking at the paper: "What do neuroanatomical networks reveal about the ontology of human cognitive abilities?".
Marcus: Here's the gist of the paper. What do neuroanatomical networks reveal about the ontology of human cognitive abilities?
Ines: The present study explores the ontological structure derived from neuroanatomical networks associated with performance in 15 cognitive tasks indicating various abilities and suggests that brain-derived (neurometric) ontology is partly consistent with the cognitive performance-derived (psychometric) ontology while revealing interpretable differences Results suggest that the brain-derived (neurometric) ontology is partly consistent with the cognitive performance-derived (psychometric) ontology while revealing interpretable and complementary differences Pruning redundant reasoning tokens The study aimed to establish a formal cognitive ontology by adopting a bottom-up approach, exploring the cognitive ontology as derived from anatomical cortical properties frequently employed in brain-behavior studies, referred to as neurometric ontology The researchers compared this neurometric ontology with the psychometric one derived from covariances of performance measures in the same tasks They investigated the extent to which the derived neurometric ontology is consistent with or distinguishable from the established psychometric ontology at stratum II of the CHC model This comparison was necessary in order to substantiate the ontology of human cognition, which should accommodate the covariance structures of both task performance and the brain systems supporting these abilities How the training signal is built The neurometric ontology was derived from several resampling realizations where task-related brain areas were identified based on linear associations between brain properties (thickness, myelination, curvature, and sulcus depth) and task performance across individuals The overlap between task-related brain areas with these different anatomical properties was quantified by creating matrices of the Intersection Over Union (IOU) index Exploratory factor analysis (EFA) was then applied to obtain a neurometric ontology The most robust neurometric ontology observed across subsamples implied a 3-factorial structure in 47 out of 60 subsamples This structure was consistent with the one derived from the whole sample which reflected a 3-factorial structure Where the gains come from The comparison between the neurometric ontology and the psychometric ontology revealed that while there was partial consistency, they also dissociated in other respects The most evident consistency was found between the Gs and EF entities being present in both ontologies However, differences were observed where the IOU matrix did not show a separation between Gf and Gc, leading to the overarching ontological entity of reasoning and comprehension knowledge (G) Furthermore, the ability-related cortical areas demonstrated competing features: they were spatially segregated but also integrated via structural fiber connections that were denser than the average brain connectome Segregation and integration of neural clusters The ability-related cortical areas showed little to no overlap between entities, such as a small overlap between G and EF (.07) and between EF and Gs (.03) This segregation suggests that the stratum II entities in the psychometric structure of cognition are partially independent Conversely, the structural connectivity between core areas was relatively dense at 50% connectivity thresholds, indicating that segregated ability-related cortical areas are integrated by fiber tracts This integration feature appears to support the positive manifold of abilities that produces general intelligence at stratum III in the CHC ontology Clinical and theoretical implications The findings have the potential to prompt further psycho-neuro-informatics studies to explore the neurometric ontology using a broader range of anatomical, biochemical, genomic and functional properties of the brain The comparative approach is valuable for exploring whether ontological entities derived from psychometrics translate into brain anatomy and function This translation is relevant for neuropsychological diagnostics and therapy Methodological details Anatomical measures considered included cortical thickness, myelination, curvature, and sulcus depth derived from structural MRI Structural connectivity was measured using probabilistic tractography applied to diffusion MRI data between brain areas in the MMP atlas Cognitive tasks included 15 tasks covering various domain-specific abilities at stratum II of the CHC model Performance indicators used were performance accuracies for some tasks and inverted reaction times for others The analysis compared the neurometric ontology with the psychometric ontology using both Exploratory Factor Analysis (EFA) on IOU matrices and Confirmatory Factor Analysis (CFA) on performance correlation matrices The results showed that the CFA of the IOU matrix from the whole sample had a CFI of 190.036 Task specifics The 15 cognitive tasks covered various abilities including Reasoning (Gf), Spatial Orientation Processing (Gf2), List-Sorting Working Memory (Gf3), Comprehension Knowledge (Gc1, Gc2), Verbal Episodic Memory (Gm1, Gm2), Executive Function (EF) across different dimensions like Dimensional Change Card Sorts and Flanker tasks Mental speed was assessed using Pattern Comparison Processing Speed (Gs1) and Sustained Attention (Gs2) Relational Processing 1 (Gs3) and Relational Processing 2 (Gs4) were also included in the set of tasks The task-related cortical areas covered approximately 80% of the entire cortex The analysis used a liberal p-value threshold of p <.05 to select relevant cortical areas associated with a given task The results revealed that the entities of the neurometric ontology partly reflect the current psychometric view and that brain areas related to different neurometric ontological entities were spatially segregated Although cognitive entities were segregated, their core brain areas were relatively dense in structural connections, which seems to explain moderate correlations of psychometric abilities at stratum II and the positive manifold captured at stratum III The study confirms that the segregation and integration configuration of brain networks can be understood in terms of underlying neural clusters in regional neuroanatomical properties and fiber network projections Data availability The datasets supporting the findings are available at http://www.humanconnectome.org/study/hcp-young-adult The codes used in this study are available at https://github.com/kristantodan12/Mining-the-brain References Alvarez, I., Parker, A.
Marcus: J., & Bridge, H.
Ines: (2019).
Paper summary: Ines: To summarize this paper, they are using a bottom-up approach to explore cognitive ontology by looking at brain properties like cortical thickness and myelination and comparing them to the performance-based psychometric ontology.
Marcus: They found that the brain-derived neurometric ontology is partly consistent with the cognitive performance derived psychometric ontology, but there are also interpretable differences between these two structures.
Yuki: The study specifically looked at fifteen cognitive tasks to see what neuroanatomical networks they revealed, and it highlighted how cortical areas associated with different inferred abilities are segregated in the brain.
Ines: And even though those ability-related cortical areas are segregated, they are integrated through denser structural fiber connections than the general brain connectome, which supports the idea of a positive manifold for general intelligence at stratum III of their model.
Marcus: Basically, they found that these ability-related cortical networks form a neural counterpart to the psychometric construct of general intelligence in the CHC model, while also showing how performance measures and brain anatomy relate to each other.
Yuki: It’s important because it moves beyond just observing which areas are active during a task; it tries to define *what* those networks represent ontologically.
Ines: And this comparison between psychometric and neurometric ontologies is presented as crucial knowledge for theory building, clinical diagnostics, and treatment.
Marcus: They are essentially trying to substantiate the ontology of human cognition by accommodating the covariance structures of both task performance and the brain systems supporting those abilities.
Conclusion: Ines: So what does this paper really tell us about the title "What do neuroanatomical networks reveal about the ontology of human cognitive abilities?" and its authors? It's an effort to create a formal framework rooted in biology.
Marcus: It’s not just saying brain structure correlates with performance; it’s arguing that these structures form specific, segregated networks that still work together in a way that mirrors high-level cognitive constructs.
Yuki: For us, this means we can start thinking about how to link population genetic findings—the traits—to specific neuroanatomical correlates rather than just general brain regions.
Ines: The implication is that we have a potential map of cognitive abilities derived from physical properties, which could guide future research into whether the psychometric structure is truly universal or just cohort-specific.
Marcus: They flag that the method they used only explored what the brain does based on activation patterns, not necessarily what the brain *is* structurally in isolation. That's a limitation they admit to having.
Yuki: And that limitation is important; we need more than just correlation; we need mechanisms showing how those physical properties actually cause those cognitive differences across the whole species.
More episodes
- 2610.33881-Sensorimotor behavior from an untrained brain-and-nerve-cord connectome
- 2610.40344-Morphological Features of Murine Pulmonary Arterial Networks for Diagnosing Vascular Disease
- Daily Summary for 2026-11-19
- 2609.10193-Are You Learning Biological Signal or Shortcuts? Auditing and Mitigating Bias in Protein-Protein Interaction Datasets
- 2610.38236-Generative discovery of chemical reaction networks
- 2507.11176-An Interpretable AI framework Quantifying Traditional Chinese Medicine Principles Towards Enhancing and Integrating with Modern Biomedicine
- The day in conversation for 2026-11-19
- Daily Summary for 2026-11-19
- 2504.03590-Towards a Unified Framework for Determining Conformational Ensembles of Disordered Proteins
- 2603.15217-A multiscale discrete-to-continuum framework for structured population models