Six weeks ago I pulled most ultra-processed food out of my stack. Not for weight. My glucose was already dialed. I did it because my mood scores on my Oura app had been flat for months despite everything else trending up, and I wanted to isolate the variable. My HRV averaged 71 this month, up from 63 in January. My self-reported mood scores climbed 18% by week 4. N=1, obviously. But the direction matched exactly what a March 2026 editorial in Frontiers in Nutrition synthesized across 14 separate studies: frequent ultra-processed food consumption links consistently to poorer mental functioning and higher psychological distress.
That editorial is worth taking seriously. Fourteen contributions, multiple countries, converging on the same signal. The mechanism isn't mysterious. Ultra-processed foods elevate dietary inflammatory potential, and inflammation has a well-documented relationship with depressive and anxiety outcomes. The gut-brain axis is real. Emotional regulation is partly metabolic. Stanford psychiatrist Shebani Sethi has been making this case in what she calls metabolic psychiatry: mental health is not solely a neurotransmitter story. Other bodily pathways matter. That framing is not fringe anymore.
The Causality Problem Is Real, But It Doesn't Kill the Argument
Dr. Alex Chen will correctly point out that most of this evidence is cross-sectional. You cannot prove causality from a snapshot. People who eat worse may also sleep worse, move less, earn less, and live in food environments that make good choices structurally harder. The editorial authors acknowledge this directly, calling for longitudinal and interventional research to confirm the direction of effect.
Fair point. I hold it for about 10 seconds before dismissing it as a reason to wait.
The downside risk of reducing ultra-processed food is essentially zero. The cost is friction, not harm. When the mechanistic data is plausible, the population signal is consistent across 14 studies, and the intervention carries no toxicity profile, I do not need a randomized controlled trial to run the experiment on myself. That is the entire logic of low-risk optimization: act on strong signals before the meta-analysis arrives.
What I will not do is follow Robert F. Kennedy Jr. off that cliff. His February 2026 claim that diet can "cure" schizophrenia is not supported by current evidence, and researchers were right to push back hard. Metabolic psychiatry is a serious adjunctive framework. It is not a replacement for pharmacological treatment in severe psychiatric illness. Overstating the case poisons the well for everyone doing legitimate work in this space.
The Actual Protocol Change Worth Making
If you are managing anxiety or low-grade depression and you have not touched your diet, you are leaving a cheap intervention on the table. Not cheap as in easy. Cheap as in low-cost, low-risk, and mechanistically grounded. The 2026 research also shows that pairing dietary changes with physical activity amplifies the benefit, which makes sense: both interventions work partly through the same inflammatory and metabolic pathways.
My specific move: I cut packaged snacks, most fast food, and anything with an ingredient list longer than 5 items. I did not go full elimination. I tracked mood and HRV daily. The signal showed up in week 3 and held through week 6. I am not calling it a cure. I am calling it a 18% mood improvement with zero downside and a plausible mechanism behind it.
That ROI is hard to argue with. Start the experiment before the perfect study arrives, because it will take years, and your brain is running right now.
