The AI Psych Brief

Reducing Ultraprocessed Foods From the Diets of Patients With Depression: A Pilot Crossover Trial

Chagwedera DN, Kim E, Chiu DT, Jin C, Lee C, Tye S, Krystal A · JAMA Psychiatry · September 2026

This UCSF pilot enrolled 20 adults with depression, high ultraprocessed food intake and at least one metabolic abnormality (mean age 44, 70% female); 19 completed. In an open-label, order-randomized crossover, each patient spent four weeks on their usual diet and four weeks on a low-UPF diet they shopped for themselves with study guidance. The primary outcomes were feasibility and acceptability: UPF screening scores fell from 10.1 to 1.4, and participants rated the diet modestly hard (5.7 of 10) but very useful (8.4 of 10).

Mood moved a lot. MADRS was about 11 points lower during the low-UPF period than during usual diet (8.1 vs 18.9), with PHQ-9 and GAD-7 also significantly lower.

Read the paper →  doi:10.1001/jamapsychiatry.2026.2320

Transformative pathways: shame and fear of compassion as potential mechanisms of change in psychotherapy

Johansen FT, Finsrud I, Flo-Groeneboom E, Kvamme IR · Psychotherapy Research · September 2026

Researchers followed 1,134 transdiagnostic inpatients at Modum Bad, a Norwegian psychiatric hospital, who completed weekly self-reports of symptom distress, shame, and fear of receiving compassion from others. Using a latent curve model with structured residuals — which separates each patient's week-to-week fluctuations from stable between-person differences — they asked what predicts what within the same patient over therapy.

Shame and distress predicted each other reciprocally, as did fear of compassion and distress. Critically, shame predicted later fear of compassion, but not the reverse. The shame–distress link held across treatment units, while the fear-of-compassion pathways varied more by setting.

Read the paper →  doi:10.1080/10503307.2026.2732028

Risperidone's prolactin problem lives outside the blood-brain barrier

Risperidone and paliperidone sit at the top of the second-generation antipsychotics for prolactin elevation, often at doses that are unremarkable for striatal D2 occupancy. The reason is anatomy plus transport. The anterior pituitary's lactotrophs sit outside the blood-brain barrier, so they see plasma drug concentrations directly. Both risperidone and its active metabolite 9-hydroxyrisperidone (paliperidone) are P-glycoprotein substrates, and P-gp efflux at the barrier holds their brain concentrations well below plasma. The result is a high pituitary-to-striatal occupancy ratio: the pituitary is blocked more thoroughly than the target. Knegtering and colleagues showed prolactin tracks the 9-hydroxy metabolite rather than the parent drug.

Two practical implications follow. Prolactin elevation on risperidone is not evidence of adequate central dosing — a patient can be hyperprolactinemic and still under-occupied in the striatum. And moving from risperidone to paliperidone does not fix it, because you have simply given the metabolite doing the work. If amenorrhea, galactorrhea, sexual dysfunction or bone loss matter, switch to a prolactin-sparing agent or add low-dose aripiprazole.

In supportive therapy, praise has to be accurate to do any work

Supportive psychotherapy (Pinsker; Winston, Rosenthal and Pinsker) is not the absence of technique. Its aims are to maintain or improve self-esteem, adaptive skills and psychological function, with the therapist lending ego functions the patient cannot yet supply. Praise and reassurance are its core tools, which is why they have to be disciplined. Reassurance must be honest, and praise must be specific and tied to something the patient actually did. Global encouragement ("you're doing great") is heard by a demoralized patient as the therapist not seeing them — or, worse, as confirmation that nobody expects much. Accurate praise does two things: it names a real capacity the patient can repeat, and it shows that the therapist is paying close enough attention to be trusted when the news is less good.