#12 — At the Intersections — July 25, 2026

Could the physical complexity of the brain's surface have a direct link to the complexity of a person's language? That is one of several intersections we are exploring this week on ComplexityPod, where we connect research from different domains. We start with the brain and psychosis. We're starting with a connection between psychosis and language, specifically looking at whether the physical structure of the brain relates to how complex someone's speech is. So, not just brain activity, but the literal shape of the brain. The research looked at the cortex, the brain's folded outer layer. Exactly. Researchers, including Lena Palaniyappan, compared three groups: healthy controls, people at high risk for psychosis, and people having their first episode. They used MRI scans to measure the structural complexity of the cortex—a method called fractal dimension. And they compared that to speech samples they collected? Right. They analyzed the speech for things like vocabulary, grammar, and the density of ideas. And one finding held true across all participants: a lower density of ideas in speech was associated with a less complex cortex. So that's a general link between brain shape and speech. But what about the differences between the groups? That's where the psychosis connection would show up. That came out when they looked at grammar. In the healthy controls, there was a negative relationship—more complex grammar was linked to a *less* complex brain structure. Which sounds counterintuitive, but suggests some kind of efficiency. And what about the other two groups? That relationship was gone. It didn't exist for the high-risk or first-episode psychosis groups. The suggestion is that this typical link between brain structure and language is disrupted, and it might point to an underlying neurodevelopmental issue. It's interesting that the same researcher, Palaniyappan, was involved in another review that moves from brain *structure* to brain *activity* in conditions like schizophrenia, depression, and bipolar disorder. Yes, this work compiled findings from 62 studies that all used magnetoencephalography, or MEG. It’s a technique that measures the magnetic fields from brain activity with very high precision, and they were looking at patterns when the brain is at rest. And they found some common patterns across these different illnesses? They found three things. First, beta wave oscillations seem to show up across all three disorders and correlate with symptoms. Second, neural complexity changes differently with age. In schizophrenia and bipolar disorder, it declines with age, which is the opposite of the healthy pattern. But in depression, the normal trajectory is preserved. And the third finding? Slow-wave activity diverges. It increases in schizophrenia and decreases in mood disorders. Based on all this, the researchers propose a new model. A kind of spectrum. They call it the “temporal disorganisation spectrum.” Right. With schizophrenia at one end, characterized by fragmented brain activity, and depression at the other, with what they call pathological rigidity. Bipolar disorder falls in the middle. And the point of a model like this would be to help create biomarkers—ways to track illness severity or identify targets for treatment. Switching from brain patterns to social ones, another study looks at how university students seek support after gender-based violence. This is an age group with one of the highest rates of this kind of violence. And getting support, whether from formal or informal sources, can really mediate the negative outcomes. So the research aimed to figure out what students actually do. Yes, Tara Mantler and her team conducted interviews with Canadian undergraduate students to map out their choices. They identified a few types of resources: services at the university, community resources, and informal networks like friends and family. But just knowing the resources exist isn't the whole story. The study also looked at what helps or hinders students from actually accessing that support. The goal being that the findings can serve as a roadmap for institutions to build more effective, survivor-centered responses. From support for students to support for patients in an ICU—there's a piece of research looking at seizure detection. The standard is continuous EEG, but there can be delays in an expert reviewing the data. So this systematic review, with Rishi Ganesan, examined a different tool, quantitative EEG, or qEEG. The question was whether it could be used as a rapid screening tool by non-specialists, like ICU caregivers. So, could an ICU nurse use this to flag a potential seizure for an expert to review? How accurate was it? The results were mixed. The review looked at twelve studies and found sensitivities ranged from 64 to 100 percent, but specificities were often low. That means a lot of false positives. Which is a problem. You don't want caregivers chasing down alarms that turn out to be nothing. True, but they found performance improved when caregivers used multiple qEEG trends together. And it was especially better when they could use a patient-specific template—a confirmed pattern from that patient's own first seizure. So while it’s not a replacement for an expert, it could still be a valuable screening tool to speed things up. It can get an expert's attention on the right patient, faster. That concludes this week's issue. We'll return with more research at the intersections of science next week. From ComplexityPod, thanks for listening.