#21 — At the Intersections — September 5, 2026

How does a model for wind striking a building's roof connect to carbon cycles in soil or conversational patterns in psychosis? On ComplexityPod, we look for these connections across scientific research. We begin with a study of aerodynamic effects on structures. This edition connects models of how physical and natural systems respond to external forces. Starting with the physical. There's a new method for predicting how wind affects low-rise buildings. Right, existing models have limitations. This research, with member Gregory Kopp, aims for more accuracy by factoring in something that's often overlooked: the vertical angle of the wind. So not just the wind hitting the side of the building, but the angle at which it comes down from above? Exactly. The approach models that variation in the vertical wind angle as an effective change in the building's roof slope. It draws an analogy between the two. That's a clever way to reframe the problem. Instead of a complex wind calculation, you treat it as if the building itself has a different shape. And it seems to work. After they filtered the data to remove the effects of small-scale turbulence, the model was effective at predicting pressure changes in specific areas. Specifically where there's strong flow separation from the building. So, from structural forces on buildings to ecological ones in the soil. A study with Zoë Lindo is looking at the uncertainties in soil food web models, which are used to connect animal diversity to carbon and nutrient cycling. And there's a specific question here: When soil animals have enough nutrients to grow, what do they do with excess carbon? How do we model that? The researchers developed a new model using data from tree plantations in Brazil. They tested two different ways to represent what happens. In one, the animals just release the extra carbon through increased respiration. And in the other? They become less efficient at absorbing carbon in the first place. So they tested these two mechanisms, and the result was that the uncertainty in these models is enormous. It’s a huge range. Predictions from the best-fitting models varied from half a percent to two thousand percent of the predicted mean. What's driving that uncertainty? Is it the model structure itself? That's secondary. The biggest factors were parameters governing animal diet and their basic physiological traits. But the choice of model structure still mattered. For example, just switching between those two mechanisms for handling nutrient limitation changed the contribution of animals to carbon loss by between 16 and 37 percent. And while microbes still dominate the process, the contributions from soil macrofauna—termites, earthworms, beetles—were significant. The study ends by calling for more targeted measurements of animal diet and physiology to narrow down these uncertainties. Now we turn from ecosystems to tools for observing human biology directly. A new technique for measuring two key brain metabolites at the same time. Yes, a study with member Lena Palaniyappan introduces a magnetic resonance spectroscopy method for simultaneously measuring glutamate and glutathione. They share a regulated cycle, and imbalances can be a sign of neurocognitive disorders. And current techniques have drawbacks—they can be slow or produce noisy signals. This new method, called Double DANTE-PRESS, is a single-shot technique that avoids some of those issues. It uses narrowband refocusing pulses to isolate the signals from just those two metabolites. So it actively filters out the unwanted signals, which would mean it produces a much cleaner spectrum to analyze. And that simplifies quantification and increases the precision of the measurements. They tested it on a 7 Tesla MR scanner, first with lab phantoms and then on five healthy volunteers. And in the human subjects, they were able to get high-precision measurements from the dorsal anterior cingulate cortex, with low variability between the individuals. It seems to be a solid proof-of-concept. From the chemical level of the brain to the behavioral, we look at how conversational patterns can reflect psychopathology. This also involves member Lena Palaniyappan. The focus is on something called Linguistic Style Matching, or LSM. It’s the way people in a conversation unconsciously start to align their use of function words—like pronouns and articles. It's considered an index of their interpersonal attunement. The question here was, for individuals with psychosis, is a change in this alignment a stable trait of the patient, or does it fluctuate with the severity of their symptoms? To find out, they followed first-episode psychosis patients and a control group for 12 months, analyzing interviews. They found that LSM is not a fixed trait at all. The groups were similar at the start, but they diverged over the year. Patients showed less alignment with the interviewer than the controls did. And this was specific to when patients were talking about themselves. But here's the unexpected part. Within individual patients, as their symptoms of formal thought disorder and their negative symptoms got worse, their alignment with the interviewer actually increased. So there’s a contradiction. They align less when discussing their own experience, but align more as their own ability to self-organize their thoughts fails. The interpretation is that they're shifting toward a conversational style that is more scaffolded by their partner. When their own internal structure is failing, they lean more heavily on the structure provided by the other person. Meaning that LSM, this linguistic matching, might serve as an index for a disordered sense of self-anchoring in the context of psychosis. That's it for this issue. We will return next week with more research summaries from across different fields. Until then, thanks for listening to ComplexityPod.