Warm AI chatbots can feel more supportive while becoming less reliable. In a 2026 Nature experiment across 5 language models, warmth fine-tuning raised wrong-answer rates by 7.43 percentage points on average, and the gap was largest when users sounded sad.1 Research Highlights Warmth came with an accuracy cost. Models trained to sound warmer made more …
The “men have shallower friendships” story is too blunt. In a 2026 Sex Roles analysis of 1,899 young adults, the familiar gender friendship gap was largely driven by one group: white men reported the lowest closeness to best friends, while Black men, Latino men, and women across ethnoracial groups looked much more similar after adjustment.1 …
The most-quoted figure on veteran suicide is “22 a day,” produced from 21-state death-certificate data for 1999–2010.2 The VA’s 2024 Annual Report — with mortality data through 2022 — puts the current daily count at 17.6 Veteran suicides, but the per-Veteran rate has gone the other way.1 Research Highlights Daily count was 17.6: 6,407 Veteran …
Clozapine has been the most effective antipsychotic in schizophrenia for almost 40 years, yet it is still routinely held back until patients have failed multiple alternatives over many years — a “third-line, after years of failure” default that a 2026 Current Opinion review by Davani et al. argues is no longer supported by the evidence.1 …
Adding music to ketamine psychotherapy failed to provide extra antidepressant benefit over matched non-music support in a 2026 randomized clinical trial (RCT; a study that assigns participants to groups by chance), even though the full 6-infusion ketamine-psychotherapy course was followed by large depression-score reductions. Research Highlights Music added no confirmed MADRS benefit: in the MUSIK …
A 2026 systematic review and meta-analysis of randomized trials found no broad superiority for phase-based complex PTSD care across most outcomes. Affect dysregulation was the clearest exception, improving more when treatment was phase-based, multi-phase, and exposure-containing. Research Highlights Most therapy formats helped: across waitlist and treatment-as-usual comparisons, psychological interventions reduced PTSD symptoms (k = 9; …
A 2025 preregistered meta-analysis found that temporary social media abstinence did not significantly improve positive affect, negative affect, or life satisfaction across 10 adult experiments involving 4,674 participants.1 The result does not prove that every break is useless, but it challenges the generic “digital detox” claim that simply logging off reliably improves mood. Research Highlights …