Brains did not evolve to make you serene; they evolved to keep the roof from caving in when something big, toothy, angry, or bill-shaped showed up. A little negative emotion can be useful. It gets the crew moving. The trouble starts when the alarm system keeps pulling the fire handle because the toaster looked at it funny.

That is the territory of neuroticism, a broad personality trait tied to worry, anxiety, low mood, and sensitivity to stress. Scientists have long suspected that the amygdala, the brain's famous threat detector, was the foreman barking orders from a tiny emotional job site. The salience network, which helps decide what deserves attention, also got prime suspect status. It looked like an open-and-shut case.

Then Sicorello and colleagues brought a bigger clipboard.

The Usual Suspect Got Questioned

In a new Nature Communications study, the researchers tested whether people's tendency toward negative emotions shows up in brain responses during two workhorse fMRI tasks: looking at emotional scenes and emotional faces. They used preregistered tests, Bayes factors, machine learning, and a "multiverse" analysis across 14 trait constructs and 1,176 models. That is not poking one wall with a screwdriver. That is a full structural inspection with helmets, permits, and somebody named Gary yelling about load-bearing beams.

The main result: broad neuroticism did not map onto the amygdala. It did not map onto the salience network. It did not map onto any single region, broad network, validated affective signature, or machine-learning pattern in a meaningful way Sicorello et al., 2026.

For years, the working cartoon was simple: more neuroticism, louder amygdala. Nice cartoon. Easy to draw. But biology loves taking easy cartoons and dropping a toolbox on them.

Stress Vulnerability Was the Live Wire

The twist is that one narrower piece of neuroticism did show up: stress vulnerability. A whole-brain machine-learning pattern predicted this facet modestly in one sample and replicated in another, with correlations around r = .21 and r = .19. That is not a crystal ball. It is more like spotting a hairline crack in a foundation before the basement floods. Useful, but nobody should start selling "brain scan says you hate Mondays" kiosks at the mall.

Even better, the signal did not lean hardest on classic emotion regions. It depended most on visual and somatomotor networks. In plain English: the pattern looked less like raw feelings blasting from an emotional furnace, and more like the brain's perception-action system sizing up ugly scenes and getting the body ready to do something about them.

That makes practical sense. Stress is not just a feeling. It is preparation. Where is the exit? Who needs help? Can I lift that? Why is the ladder on fire again?

Why This Matters

This paper pushes against a common neuroscience habit: taking a famous brain region, handing it a hard hat, and blaming it for the whole construction project. The amygdala matters. The salience network matters. Recent work still links salience-network organization to depression, including a 2024 Nature study reporting that the frontostriatal salience network was much larger in many people with depression Lynch et al., 2024. A 2023 review also found resting-state neuroimaging correlates of neuroticism across studies Lin et al., 2023.

But the new study says task, trait, and measurement choice matter. "Negative emotions" is not one pipe in the wall. It is plumbing, wiring, ventilation, and one mysterious switch nobody admits installing.

Clinically, that matters because neuroticism and negative affectivity cut across depression, anxiety, trauma-related problems, and other forms of emotional dysfunction. The HiTOP framework, for example, treats negative affect and neuroticism as part of a broader emotional dysfunction superspectrum Kotov et al., 2022. If the biology differs by subcomponent, then lumping everything together may bury the useful signals under rubble.

The Real-World Payoff, If It Holds

If this stress-vulnerability pattern keeps replicating, it could sharpen how researchers study risk. Not diagnose you from a scan. Not replace therapy. Not tell your boss you are biologically unsuited for Tuesday meetings, tempting as that may be.

The payoff would be more basic and more useful: better targets for prevention, better task design, and cleaner biomarkers for stress sensitivity. Other stress-vulnerability work is already moving in that direction, including single-cell studies in mouse prefrontal cortex that connect stress-related brain networks to cell types and gene expression Hing et al., 2024.

The big lesson is sturdy: the brain does not keep one neat "negative emotion room" behind a labeled door. It builds distributed systems. Some watch. Some move. Some prepare. Some panic and knock over the paint cans.

This study is interesting because it says the old blueprint was too tidy. The amygdala may still be on the crew, but it is not always the boss. Sometimes the action is in how the brain sees the scene, braces the body, and gets ready for impact. That is less cinematic than a tiny fear button. It is also probably closer to how the building actually stands.

References

Sicorello M, Gianaros PJ, Wright AGC, Petre B, Kraynak TE, Manuck SB, Schmahl C, Wager TD. The functional neurobiology of dispositions towards negative emotions. Nature Communications. 2026;17. doi: 10.1038/s41467-026-74565-0

Lynch CJ, et al. Frontostriatal salience network expansion in individuals in depression. Nature. 2024;633(8030):624-633. PMCID: PMC11410656. doi: 10.1038/s41586-024-07805-2

Lin J, Li L, Pan N, Liu X, Zhang X, Suo X, Kemp GJ, Wang S, Gong Q. Neural correlates of neuroticism: A coordinate-based meta-analysis of resting-state functional brain imaging studies. Neuroscience & Biobehavioral Reviews. 2023;146:105055. doi: 10.1016/j.neubiorev.2023.105055

Kotov R, et al. Validity and utility of Hierarchical Taxonomy of Psychopathology (HiTOP): III. Emotional dysfunction superspectrum. World Psychiatry. 2022;21(1):26-54. PMCID: PMC8751579. doi: 10.1002/wps.20943

Hing B, et al. Transcriptomic Evaluation of a Stress Vulnerability Network Using Single-Cell RNA Sequencing in Mouse Prefrontal Cortex. Biological Psychiatry. 2024;96(11):886-899. PMCID: PMC11524784. doi: 10.1016/j.biopsych.2024.05.023

Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.