Summary: Researchers at the University of Birmingham have developed a novel, higher-order connectomics method to map brain activity and interactions. Unlike traditional pairwise network models, this approach extracts multi-region interaction patterns from fMRI signals to reveal how complex cognitive functions—such as language, attention and thought—emerge from coordinated activity across groups of brain regions. The technique improves task decoding, individual identification from brain data, and the association of neural activity with behavioral traits, and may offer new tools for studying neurodegenerative conditions.
This method draws on fMRI data from the Human Connectome Project and uses statistical processing to turn noisy neuroimaging signals into robust models of higher-order interactions. Validated on 100 unrelated subjects, the approach demonstrates practical advantages over pairwise models and opens new possibilities for both basic neuroscience and clinical research.
Key Facts:
- Models higher-order interactions among three or more brain regions, extending beyond traditional pair-based connectivity methods.
- Validated on fMRI data from 100 unrelated participants drawn from the Human Connectome Project.
- Demonstrated capabilities include task decoding, individual “brain fingerprinting,” and stronger links between brain network features and behavior.
- Published in Nature Communications and led by Enrico Amico with first author Andrea Santoro.
- Potential application in tracking brain changes in neurodegenerative diseases, and in identifying early or pre-clinical signs of conditions such as Alzheimer’s.
Traditional network representations of brain function typically capture pairwise connections between two regions. While informative, pairwise models miss complex interactions that arise when multiple regions engage together. The new higher-order connectomics approach overcomes this limitation by inferring multi-region dependencies directly from temporal fMRI time series and correcting for measurement noise using principled statistical techniques.
Lead researcher Dr Enrico Amico explained the motivation: “Complex systems like the brain depend on interactions between groups of regions, not just between pairs of regions. Although we know – in theory – that this is the case, until now we have not had the processing power required to model this.” Using modern computational resources and rigorous signal-processing methods, the team translated noisy neuroimaging signals into clearer maps of how regional groups contribute to cognition and behavior.
The team used fMRI recordings from the Human Connectome Project, a large consortium effort designed to link brain structure, function and behavior. From that dataset they selected 100 unrelated participants and built detailed models of higher-order interactions. These models were evaluated across three targeted tests to demonstrate the method’s utility and robustness.
First, the researchers showed improved task decoding: the higher-order models could more accurately infer which experimental task a participant was performing during a scan. Second, the models yielded distinctive individual signatures, allowing researchers to identify a person from their brain activity patterns—what the study describes as a brain “fingerprint.” Third, the team separated higher-order signals from lower-order ones and showed that the higher-order components correlated more strongly with individual behavioral measures, enhancing the link between brain dynamics and personal traits.
Dr Andrea Santoro, first author of the study and affiliated with CENTAI Institute in Italy, summarized the significance: “Our approach, validated using data from healthy individuals, demonstrates the substantial advantages that this method can offer to neuroscience research. In the future, this method could also be used to help model interactions in individuals with neurodegenerative diseases, such as Alzheimer’s, where they could give valuable insights into how brain function is changing over time, or even to identify pre-clinical symptoms of these conditions.”
About this brain mapping and behavioral neuroscience research news
Author: Beck Lockwood
Source: University of Birmingham
Contact: Beck Lockwood – University of Birmingham
Image: The image is credited to Neuroscience News
Original Research: Open access. “Higher-order connectomics of human brain function reveals local topological signatures of task decoding, individual identification, and behavior” by Enrico Amico et al., published in Nature Communications.
Abstract
Higher-order connectomics of human brain function reveals local topological signatures of task decoding, individual identification, and behavior
Conventional approaches typically model the brain as a network of pairwise interactions between regions. Recent methods extend this concept to infer higher-order interactions that involve three or more regions simultaneously, but until now it has been unclear whether higher-order models provide practical advantages for fMRI analysis.
To address this, the authors analyzed fMRI time series from 100 unrelated Human Connectome Project participants. They demonstrate that higher-order connectomics substantially improves the decoding of dynamic task states, enhances the individual identification of both unimodal and transmodal functional subsystems, and strengthens the statistical association between brain activity patterns and behavioral measures.
Overall, this work illuminates the higher-order organization present in fMRI data, revealing rich topological structures and dynamic group dependencies that are not captured by traditional pairwise approaches. These findings suggest a broad, unexplored space of functional brain structures that higher-order models can uncover, with potential applications in mapping cognition and monitoring clinical change over time.