CogLinks: A Novel Brain Model Paving the Way for Algorithmic Psychiatry

A groundbreaking computational framework named CogLinks has been unveiled by researchers from Tufts University, working in close collaboration with their counterparts at Ruhr University Bochum in Germany. This innovative tool functions much like a sophisticated “flight simulator” for neural circuits, enabling scientists to conduct in-depth investigations into the brain's complex decision-making processes and the precise mechanisms it employs to adjust behavior when external conditions shift. The details of this model were formally published on October 16, 2025, in the journal Nature Communications, outlining how CogLinks successfully simulates essential cognitive functions within neural networks, including learning, error correction, and adaptation.

Many artificial intelligence systems operate as opaque “black boxes,” making it difficult to trace their decision pathways. CogLinks, however, distinguishes itself by being a biologically validated model that precisely replicates the architecture and connectivity of actual neurons. This inherent transparency is crucial because the model is capable of demonstrating not only the successful execution of complex cognitive tasks—such as navigating novel environments or solving problems—but also the specific points at which these processes fail or become compromised. This capacity to model dysfunction offers direct and unprecedented insights into the underlying nature of mental disorders, illuminating precisely how the brain forms judgments in ambiguous scenarios—a critical requirement for flexible behavior and deliberate choice.

To validate the computational predictions of the model, the team conducted rigorous experiments involving human volunteers undergoing functional magnetic resonance imaging (fMRI). Participants were required to complete a task demanding a rapid shift in strategy following an abrupt change in the established rules. The resulting fMRI data strongly corroborated the hypotheses generated by CogLinks. Specifically, the findings indicated that the mediodorsal thalamus functions as a critical “control panel,” orchestrating flexible planning—a role typically assigned to the prefrontal cortex—with the automatic habits that are governed by the striatum.

The research group, spearheaded by Tufts University Professor of Neuroscience Michael Halassa, views this achievement as the definitive dawn of the era of “algorithmic psychiatry.” This emerging discipline relies heavily on leveraging detailed computational simulations, like CogLinks, to precisely map the biological root causes of psychiatric ailments, moving beyond symptom-based diagnoses. The ultimate goal is to develop highly targeted therapeutic interventions based on these biological maps. Professor Halassa emphasized that the overarching mission is to seamlessly integrate biology, advanced computation, and clinical practice to achieve a more accurate, actionable representation of the human mind. He stated plainly regarding the future of treatment: “If we understand how the brain deviates from the norm, we can learn how to retune it.”

Dr. Mien Brabeeeba Wang, the study’s lead author and a doctoral candidate at the Massachusetts Institute of Technology (MIT) working in the Halassa lab, highlighted the profound clinical potential of CogLinks. She noted that the model could be instrumental in clarifying how schizophrenia-related mutations, which impact receptors throughout the brain, disrupt the necessary organization of information required for flexible thinking. This significant finding provides a detailed, observable mechanism for grasping both cognitive flexibility and its dysfunction, paving the way for the development of truly personalized psychiatric care tailored to individual biological profiles.