Competition Between Brain Systems Improves Digital Twins, Study Finds

Researchers describe a shift in digital brain twins from simplistic cooperative models to ones that include competition between neural systems. In a Nature Neuroscience study, realistic digital twins preserve individual brain fingerprints only when long-range competition is allowed, reducing generic predictions. The work compares cooperative versus competitive models across humans, macaques and mice, finding that competitive interactions yield activity patterns that more faithfully reflect cognitive circuits. The authors argue that this approach could improve preclinical testing, cross-species translation, and even inform next-generation AI in ways that align more closely with the human mind. Original publisher: The Conversation.

Overview

The article discusses the potential of personalised digital twins of the brain and body, explaining that whole-brain computer models aim to simulate how brain regions interact under stimulation, disease or medication. It highlights that brain networks are wired differently in every person, producing unique “brain fingerprints.”

Competition and Cooperation in Brain Networks

Historically, many brain simulations have forced neighbouring regions to cooperate, which can push the model into overly synchronised states that do not match real brains. The authors argue that everyday cognitive tasks reveal inherent competition for limited neural resources, with some regions increasing activity during tasks (like the intraparietal sulcus) while others decrease (such as the posterior cingulate cortex and medial prefrontal cortex).

Quote 1

"Crucially, models with competitive interactions were not only more accurate but also more individual-specific." - Luppi et al., Nature Neuroscience.

Two Modelling Approaches and Key Findings

The study compared two brain models: one where all interactions were cooperative, and another where regions could excite or suppress each other. Across humans, macaques and mice, models that included competitive interactions outperformed cooperative-only models. A large analysis of over 14,000 neuroimaging studies showed that spontaneous activity in competitive models better reflected cognitive circuits involved in attention and memory, suggesting competition enables flexible activation of appropriate region combinations — a hallmark of intelligent behaviour.

Quote 2

"The human brain is never static." - The Conversation.

Cross-Species Translation and Translational Neuroscience

Because the findings hold across humans and other mammals, the work has potential implications for translational neuroscience. It addresses the common problem of species differences limiting the translation of animal-model results to humans, noting that around 90% of treatments for neuropsychiatric disorders fail in human trials despite promising animal data. The authors propose that integrating human brain imaging with whole-brain modelling could bridge this gap and create a cross-species framework for testing therapies.

Quote 3

"This could have major implications for translational neuroscience." - The Conversation.

Implications for AI and the Next Generation of Brain Modelling

The authors discuss how these principles might inform future AI systems and the design of digital twins that better reflect human brain organisation. They suggest that competitive brain interactions contribute to energy efficiency and social-cognitive flexibility, offering a path toward more faithful AI models that echo human neural dynamics.

Quote 4

"competition is crucial for enabling the brain to flexibly activate appropriate combinations of regions" - Luppi et al., Nature Neuroscience.

Conclusion

By embracing competitive inter-regional interactions, whole-brain models can become more individual-specific and translationally relevant, potentially improving both clinical interventions and AI systems inspired by brain function.