Michael Halassa Lab School of Medicine Tufts University
Boston
Joined October 2024
- Tweets44
- Following20
- Followers12
- Likes3
Check out this preprint from work with our fantastic colleagues @PalvaLab in Helsinki
biorxiv.org/content/10.1101/…
Thanks for a great collaboration!
Congratulations to Norman and Arghya, who have been awarded with this years "Neuroscience Chair's Award for Postdoctoral Research" @TuftsMedSchool honoring their work in the recent discovery of a prefrontal transthalamic role in decision making under uncertainty. #Tupaia, #Nature
The promise is transformative: treatments designed with the whole ‘hardware’ in mind, and guided by interfacing with its ‘software’.
But combining a drug that enhances a neurochemical feature broadly with targeted neurostimulation that augments these benefits to specific circuits could. Pairing such an approach with appropriately-timed behavioral therapy may allow for the brain to be re-wired for recovery.
(2) lower overconfidence in rigid memories to improve flexibility, and (3) maintain predictive efficiency overall. No single intervention is likely to achieve all of this.
From this perspective, success means not just symptom reduction, but recalibrating the internal computations that generate symptoms. For instance, an ideal intervention might: (1) enhance predictions about the sensory world to reduce hallucinations,
This is done by combining behavioral data (e.g., choices in structured tasks) with neural signals (from EEG, fMRI, or intracranial recordings). These latent variables become the true targets of treatment.
Algorithmic models allow us to infer hidden variables in the brain’s operating system such as its belief-updating algorithms— variables like confidence in predictions, sensitivity to new information, and flexibility in adjusting beliefs.
Our work introduces algorithmic psychiatry as a solution—computational "flight simulators" that model how perturbations at any level (molecular, circuit, cognitive) propagate through the system.
This is psychiatry's fundamental challenge: we design treatments for one level of organization while their consequences unfold across all levels.
A schizophrenia drug might target dopamine receptors, but can't predict how its effects cascade through neural circuits to alter decision-making. A cognitive therapy trains flexible thinking, but doesn't account for how molecular deficits may limit its benefits.
Every mental health treatment—whether a pill, therapy, or brain stimulation—sends ripples through a patient's entire biological and cognitive system. Yet today's interventions are designed while only seeing part of the picture.
But even this breakthrough highlights a deeper issue: we still lack a framework for predicting how interventions at the molecular level will impact cognition and behavior.
New therapies like xanomeline/trospium—a recently FDA-approved drug that targets muscarinic receptors instead of dopamine—offer fresh hope, especially for negative symptoms long considered untreatable.
Despite decades of research and billions in global spending, nearly one-third of people with schizophrenia remain treatment-resistant, and fewer than 15% experience functional recovery.
Why Mental Health Treatments Need Algorithmic "Flight Simulators" - read more at #cellreportsmedicine
cell.com/cell-reports-medici…
Congratulations to Arghya for receiving an Early Career Investigator Travel Fellowship Award from the @SOBP Society of Biological Psychiatry!
Huiwen and Navi are going to present their amazing work at #Cosyne2025. Go check out their work if you get the chance:
"A direct link between prefrontal E/I balance and executive dysfunction in schizophrenia" and "A prefrontal to amygdala pathway for adaptive sleep interruption"
Mengxing received a Post-Doctoral Fellow Award from the Cognitive Neuroscience Society. #CNS2025 Congratulations Mengxing!