A single gene mutation slows the brain's ability to update beliefs - and it may explain part of schizophrenia

Imagine you have two options, and you know which one pays better. Then the rules change. The good option gets worse, slowly, and at some point the smart move is to switch. Most people make that adjustment without thinking much about it. For some people with schizophrenia, that adjustment is profoundly difficult.

This failure to update beliefs based on new information has long been hypothesized as a root mechanism behind psychosis. If your brain over-relies on what it previously believed and under-weighs what it is currently sensing, the gap between internal model and external reality widens. Eventually, the model wins - and reality recedes.

A new study from MIT, published March 18 in Nature Neuroscience, traces this problem to a specific gene, a specific brain circuit, and a specific failure mode - and demonstrates that the deficit can be reversed by activating the right neurons with light.

From genome-wide screens to a single gene called grin2a

The genetic backstory begins at the Broad Institute's Stanley Center for Psychiatric Research, where researchers have spent years cataloging gene variants associated with schizophrenia risk. Using whole-exome sequencing - a method that reads only the protein-coding regions of the genome - the Stanley Center team analyzed DNA from roughly 25,000 people with schizophrenia and 100,000 control subjects. They identified 10 genes in which mutations significantly increase schizophrenia risk.

One of those genes is grin2a. It encodes a subunit of the NMDA receptor, a protein found on the surface of neurons throughout the brain. NMDA receptors respond to the neurotransmitter glutamate and play critical roles in synaptic plasticity - the process by which connections between neurons strengthen or weaken based on experience. They are, in essence, part of the brain's learning machinery.

Guoping Feng, the James W. and Patricia T. Poitras Professor in Brain and Cognitive Sciences at MIT, and Michael Halassa, an associate professor at Tufts University, set out to understand what happens when grin2a is mutated. They created mice carrying the same type of mutation found in human schizophrenia patients and asked a straightforward question: do these mice show any of the cognitive problems associated with the disease?

The two-lever test

Studying psychosis in mice is, for obvious reasons, impossible. Hallucinations and delusions cannot be directly observed in an animal. But the cognitive impairments that accompany schizophrenia - specifically, the difficulty incorporating new information into decisions - can be modeled.

Lead author Tingting Zhou, a research scientist at MIT's McGovern Institute, designed an experiment around two levers. One lever was low-reward: push it six times to receive one drop of milk. The other was high-reward: one push for three drops. All mice quickly learned to prefer the high-reward lever.

Then Zhou gradually increased the effort required for the high-reward option. Over time, the number of presses climbed - from one to six to twelve to eighteen. At some point, the effort per drop of milk became roughly equal between the two levers. A mouse paying attention to the changing environment should switch strategies.

Healthy mice did exactly that. They began exploring both levers as the high-reward option deteriorated, and they made the permanent switch to the low-reward lever right around the point where the two options reached equal value. Their decision-making tracked the changing reality.

Mice with the grin2a mutation behaved differently. They spent more time vacillating between the two options and made the switch to the low-reward side much later - well past the point of equal value. They were not incapable of switching. They were dramatically slower to recognize that the situation had changed and act accordingly.

Zhou put it precisely: the mutant animals' adaptive decision-making was significantly slower compared to wild-type animals. They over-relied on their prior belief about which lever was better and under-weighted the current sensory evidence that the situation had shifted.

The mediodorsal thalamus keeps score

Using functional ultrasound imaging and electrical recordings, the researchers traced the neural deficit to a specific brain structure: the mediodorsal thalamus. This region connects to the prefrontal cortex to form a thalamocortical circuit involved in executive control, working memory, and decision-making.

In healthy mice, neuronal activity in the mediodorsal thalamus tracked the changing values of the two reward options in real time. The neurons effectively maintained a running comparison - a value signal that updated as conditions changed. The activity patterns also differed depending on whether the mouse was in an exploratory state (sampling both levers) or a committed state (consistently choosing one side).

In mice with the grin2a mutation, this value-tracking signal was degraded. The mediodorsal thalamus was slower to register changes in reward conditions and slower to shift its activity patterns in response. The circuit was not offline - it was sluggish. The information needed to update beliefs was arriving late and landing weakly.

Feng described the implications directly: if this circuit does not work well, you cannot quickly integrate information. The team is quite confident that this circuit represents one of the mechanisms contributing to the cognitive impairment seen in schizophrenia.

Reversing the deficit with light

To confirm that the mediodorsal thalamus was causally responsible for the behavioral deficit - not merely correlated with it - the team used optogenetics. They engineered the neurons of the mediodorsal thalamus in the mutant mice to respond to light stimulation. When those neurons were activated, the mice began behaving like their healthy counterparts, switching strategies at the appropriate time.

This is a strong result. It establishes that activating the right circuit can compensate for the genetic mutation's effects, at least in this specific behavioral context. The deficit is not hard-wired into every aspect of the brain - it is concentrated in a circuit that can, in principle, be targeted.

From mice to medicine: the caveats

The limitations of this work are real and the researchers acknowledge them. Only a very small percentage of schizophrenia patients carry mutations in grin2a specifically. Schizophrenia is a genetically heterogeneous disease with dozens of implicated genes, and most cases likely involve combinations of common variants rather than single rare mutations.

That said, the researchers suggest that dysfunction in this thalamus-cortex circuit could be a convergent mechanism - a common downstream effect of multiple different genetic causes. Different mutations in different genes might all impair the same circuit, producing similar cognitive deficits through different molecular routes. If true, targeting the circuit rather than the gene could help a broader patient population.

Optogenetics is not a clinical treatment. Shining light on genetically modified neurons works in a mouse brain but cannot be directly translated to human patients. The team is now working to identify druggable targets within the mediodorsal thalamus circuit - molecules that could be modulated with conventional pharmaceuticals to achieve a similar effect.

Schizophrenia's most debilitating cognitive symptoms - the ones that make it difficult to hold a job, maintain relationships, or navigate daily decisions - have resisted drug treatment for decades. Current antipsychotics primarily target dopamine pathways and can reduce hallucinations and delusions, but they do little for cognitive impairment. A therapy that specifically addresses the brain's ability to update beliefs in response to changing circumstances would address a gap that existing medications leave wide open.