今年早些时候发表的两项荟萃分析发现，运动可将抑郁和焦虑症状降低到与常规治疗大体相当的程度。1 月份的 Cochrane 回顾整合了 69 项随机对照试验，2 月份的元-元分析则纳入了超过 1,000 项试验、近 80,000 名参与者。

然而，两组证据都存在明显限制：运动研究无法对参与者实施盲法，导致自我报告情绪容易受期待效应影响，且 Cochrane 评估所有入组试验均为高偏倚风险。第二项分析也没有把运动与其他干预在同一试验中直接比较，而是与独立的抗抑郁药与心理治疗试验结果对照，这些随机对照试验通常盲法更好、安慰剂效应更强，因此 UCL 神经科学教授 Jonathan Roiser 认为这种比较并不公平。

机制仍不明朗：2021 年的一项研究显示，阻断跑步者的阿片受体并未减少其运动后欣快感或焦虑下降，削弱了“内啡肽促使情绪好转”的解释。研究者转而关注其他途径，包括可能的内源大麻素作用、炎症下降、脑可塑性改善、以及与努力—奖励权衡相关的多巴胺传递增强，同时也有成就感、掌控感和掌握感等心理收益；运动中有氧形式（跑步、步行、骑行）通常最有益，抑郁在团体/监督式训练中的收益更强且多月累积，焦虑则更适合低强度运动。

Two meta-analyses published earlier this year found exercise reduces symptoms of depression and anxiety to roughly the same degree as conventional treatment. The January Cochrane review pooled 69 randomized controlled trials, while the February meta-meta-analysis included over 1,000 trials with nearly 80,000 participants.

However, both evidence sets have caveats: the exercise studies could not blind participants, so self-reported mood was vulnerable to expectation effects, and the Cochrane review rated all included trials as high risk of bias. The second analysis also did not directly compare exercise against other interventions, instead contrasting results from separate antidepressant and therapy trials that were generally better blinded and had stronger placebo effects, which Jonathan Roiser says can make the comparison unfair.

The mechanism remains uncertain: a 2021 study showed that blocking runners’ opioid receptors did not reduce post-run euphoria or anxiety reduction, weakening the endorphin-based explanation. Researchers now point to other pathways—including possible endocannabinoid effects, reduced inflammation, improved brain plasticity, and increased dopamine transmission tied to effort-reward balance—plus psychological gains such as achievement, agency, and mastery; aerobic exercise (running, walking, cycling) is generally most beneficial, group or supervised formats are stronger for depression and build over several months, while lower-intensity activity appears best for anxiety.

Source: Is exercise as good as antidepressants?

Subtitle: Some big studies say yes. Many experts have reservations

Dateline: The Economist May 2nd 2026