Mental Health

Exercise RCT Meta-Analyses & Reviews

• Acute exercise has an immediate moderate effect on short-term and long-term memory, while long-term exercise had small and insignificant effects, respectively [3].
• Aerobic exercise lasting 21-30 minutes reduces anxiety (d=0.41) [4]. Moreover, these effects grow over the first several weeks and anaerobic exercise generally lacked this.
• Both cardio and resistance training improve mood (d=0.69), but should be done only 1-2 times per week for this purpose. Effects shrink over time [5]
• Exercise also improves on 8 of 10 psychological measures examined in [6], especially self-efficacy (d=0.38) and overall well-being (d=0.30). Strength training had positive effects, but they were generally smaller than cardio's (d=0.16 vs 0.26).
• Aerobic exercise also boosts positive affect (d=0.45) [7]
• Tai Chi reduces stress (d=0.66), anxiety (d=0.66), depression (d=0.56), while improving mood (d=0.45) [8]. Another meta-analysis found similar results [9].
• Exercise reduced anxiety (d=0.48) [10].
• Aerobic exercise improves executive function (D=0.34) [11], especially the preadolescent children and older adults [12] - both groups with their easily-googleable meta-analyses I won't link to here.
• Exercise improves cognitive performance [13]

Probably the biggest issue with these cardio studies is that, as far as I can tell, all the measures of these various mental health metrics were conducted within 30 minutes of the exercise stopping, so its really unclear whether running actually makes you happier long-term other than that the effect size doesn't appear to shrink over that half hour [4] - this is a problem with lots of social psychology [14].

Anaerobic exercise also has significant cognitive benefits, generally of similar magnitude as cardio [15]. However I've yet to see any literature review or meta-analysis claiming that it boosts mood or well-being.

Positive Psychology Intervention (PPI) RCT Meta-Analyses & Reviews

Positive psychology interventions improve mental health along several measures (d~0.3) [16] [17] (see also this literature review [18]).

A 2009 meta-analysis gives some pragmatic advice [16]:

Unfortunately, there simply haven't been enough high-quality trials to distinguish which particular PPIs are most effective [17] so these meta-analyses have generally grouped them all together.

What's worse, a later meta-analysis found these effects were largely driven by small sample size bias and that accounting for this resulted in effect sizes several times smaller (d~0.1) or, sometimes, statistically insignificant [19].

Moreover, as with cardio, almost all the studies on which these meta-analyses are based measured the effects soon after the intervention, which makes it difficult to determine to what extent these mental health benefits persist throughout the day.^a

Nevertheless, for completeness, here are some meta-analyses on specific interventions:

One meta-analysis of gratitude interventions found gratitude interventions improve psychological well-being (d~0.3) even compared to alternative interventions (d~0.2) [21]. This was generally confirmed a year later [22]. A meta-analysis found even stronger effects among youth [23].

Meditation is another PPI. It improves anxiety (d=0.22), depression (d=0.23), pain [24] and more generally on emotional and relationship measurers, but not much on cognitive measures [25]. Moreover, when compared to generic relaxation techniques, most of these effects remain, suggesting meditation is doing more than just relaxing people (or at least it relaxes them exceptionally well). Like exercise, the positive effects grow during the first several weeks. Meditation improves sleep quality, but other techniques are comparable [26]. Unfortunately, I could not find an analysis for how long (in minutes/hours) to meditate for most metrics. The exception is sleep, where they examined 3-42 hours of meditation and found no dose-response. The minimum time studied was 1.25 hours per week. One RCT found that meditation reduces stress even if done immediately after exercise (at least in the short-term) [27] See also [28]. TODO: [29]. The evidence is more sparse for yoga, but looks more-or-less similar [30].

Another PPI is reminiscing, which improves well-being (d~0.5) [31].

Next, there are compassion-based PPIs. TODO: [32] [33].

Finally, there's an 8-week course on various PPI techniques called "Mindfulness-based stress reduction" [34]. It has a large effect size (d~0.5) [35]. I've seen people say one individual study (not a meta-analysis) [36] shows benefits extend at least two weeks, but I've been unable to obtain that study's text.

Other Meta-Analyses & Reviews

• Psychotherapy (especially in groups) improves affect [37]
• Light therapy reduces depression (d~0.4) even among people with nominally non-seasonal depression. The balance of the evidence especially when using a bright white light in the morning for less than an hour [38]. It's also effective at treating seasonal affective disorder (d~0.4) [39].
• A literature review on insomnia [40] ended up concluding that the most consistent evidence exists for melatonin. Meditation, exercise, and hypnotherapy have weaker evidence in favor. There is minimal evidence for herbal pharmacotherapies, light exposure therapy, homeopathy, or dietary supplements. A second review also found exercise to be quite effective on a number of different sleep-related metrics , with the balance of the evidence suggesting exercising ~3 days per week of moderate intensity exercise [41].
• Vitamin B improves cognition but not antioxidants or vitamin D [42].
• Zinc supplementation improves depression [43] [44] .
• Four reviews found St John's wort improves depression [45] [46] [47] [48]. A fifth review concluded the evidence was mixed [49].
• Afternoon naps improve cognitive performance [50]

RCTs

• A famous study from 1978 found that lottery winners were no happier than their demographics would have predicted [51], but a later study found the opposite: that winning the lottery does, in fact, make you happier, even 2 years later [52].