New Computer Model Sheds Light On How Brain Circuits Support Flexible Thinking

Trending 1 month ago

Every day, your encephalon makes thousands of decisions nether uncertainty. Most of nan time, you conjecture right. When you don't, you learn. But erstwhile nan brain's expertise to judge discourse aliases delegate meaning falters, thoughts and behaviour tin spell astray. In psychiatric disorders ranging from attention-deficit/hyperactivity upset to schizophrenia, nan encephalon whitethorn misjudge really overmuch grounds to stitchery earlier acting-or neglect to set erstwhile nan rules of nan world alteration based connected caller information. 

Uncertainty is built into nan brain's wiring. Picture groups of neurons casting votes-some optimistic, immoderate pessimistic. Your decisions bespeak nan average."

Michael Halassa, professor of neuroscience, Tufts University School of Medicine

When that equilibrium skews, nan encephalon tin misread nan world: assigning excessively overmuch meaning to random events, arsenic in schizophrenia, aliases becoming stuck successful rigid patterns, arsenic successful obsessive-compulsive disorder.

Understanding those misfires has agelong challenged scientists, says Halassa. "The encephalon speaks nan connection of azygous neurons. But fMRI-the instrumentality we usage to study encephalon activity successful people-tracks humor flow, not nan electrical chatter of individual encephalon cells."

Bridging that spread intends combining insights from single-cell studies successful animals, quality encephalon imaging, and behavior. Now, a caller benignant of machine model-grounded successful existent biology-lets researchers simulate really encephalon circuits make decisions and accommodate erstwhile nan rules change.

Called CogLinks, nan exemplary builds biologic realism into its design, mirroring really existent encephalon cells are connected and coding for really they delegate worth to often ambiguous and incomplete observations astir nan outer environment. Unlike galore artificial intelligence systems that enactment for illustration "black boxes," CogLinks shows researchers precisely really its virtual neurons nexus building to function. As a result, scientists tin representation really this virtual encephalon learns from acquisition and pivots based connected caller information. 

In a study published October 16 in Nature Communications, elder writer Halassa and colleagues astatine Massachusetts Institute of Technology (MIT) used CogLinks to research really encephalon circuits coordinate elastic thinking. Like a formation simulator for nan brain, CogLinks let the researchers trial what happens erstwhile cardinal decision-making circuits spell disconnected course. When they weakened nan virtual relationship betwixt 2 simulated encephalon regions-the prefrontal cortex and nan mediodorsal thalamus-the strategy defaulted to slower, habit-driven learning. That result suggests this pathway is basal for adaptability. 

To spot if those predictions held existent successful people, nan squad past conducted a companion fMRI study, which was supervised by both Burkhard Pleger from nan Ruhr-University Bochum and Halassa. Volunteers played a crippled successful which nan rules unexpectedly changed. As expected, nan prefrontal cortex handled readying and nan deep, cardinal region of nan encephalon known arsenic nan striatum guided habits-but nan mediodorsal thalamus lit up erstwhile players realized nan rules had shifted and adjusted their strategy. 

The imaging confirmed what nan exemplary had forecast: nan mediodorsal thalamus acts arsenic a switchboard linking nan brain's 2 main learning systems-flexible and habitual-helping nan encephalon infer erstwhile discourse has changed and move strategies accordingly. 

Halassa hopes nan investigation helps laic nan groundwork for a caller benignant of "algorithmic psychiatry," successful which computer models thief uncover really intelligence unwellness emerges from changes successful encephalon circuits, identifying biologic markers to precisely target treatments. 

"One of nan large questions successful psychiatry is really to link what we cognize astir genetics to cognitive symptoms," says Mien Brabeeba Wang, the lead writer of the CogLinks study, a co-author of the fMRI study, and an MIT doctoral student successful Halassa's lab. 

"Many schizophrenia-linked mutations impact chemic receptors recovered passim nan brain," says Wang. "Future uses of CogLinks may thief america spot really those wide molecular changes could make it harder for nan encephalon to shape accusation for elastic thinking." 

Source:

Journal reference:

Wang, M.B., et al. (2025). The neural ground for uncertainty processing successful hierarchical determination making. Nature Communications. doi.org/10.1038/s41467-025-63994-y

More