← Back

Focus and memory

Everything in the library studied for it, medical and holistic together. Within each kind, what was shown to help comes first, strongest evidence first. What was tested and not shown to help stays on the list.

Strong several reviews agree · Moderate one review or consistent trials · Limited one or a few small studies · Untested no study in people, listed with what it rests on

Medication19 Peptides and SARMs2 Supplements36 Therapies2 Practices11 Daily life12 Food and drink5 Diets1 Lab markers7

Medication

ADHD stimulantsStrong

ADHD symptoms in children and adolescents

Shown to reduce ADHD symptoms more than placebo in children and adolescents over about 12 weeks, with moderate to large effects for amphetamines and methylphenidate.

The numbers and sources

Effect: In a network meta-analysis, clinician-rated ADHD symptoms closest to 12 weeks improved more than placebo with amphetamines (SMD -1.02, 95% CI -1.19 to -0.85) and methylphenidate (SMD -0.78, 95% CI -0.93 to -0.62). On teacher ratings, methylphenidate was better than placebo (SMD -0.82, 95% CI -1.16 to -0.48) [1]. A Cochrane review of methylphenidate found teacher-rated symptoms improved (SMD -0.74, 95% CI -0.88 to -0.61), equal to 10.58 points on a 72-point rating scale where 6.6 points is considered the smallest clinically relevant change [2]. A Cochrane review of amphetamines found improvement on parent ratings (SMD -0.57, 95% CI -0.86 to -0.27), teacher ratings (SMD -0.55, 95% CI -0.83 to -0.27) and clinician ratings (SMD -0.84, 95% CI -1.32 to -0.36), with more children rated as responders (RR 3.36, 95% CI 2.48 to 4.55) [3].
Population: 10,068 children and adolescents in the efficacy analysis of 133 double-blind trials [1]; 16,302 participants aged 3 to 18 in 212 trials, mean treatment length 28.8 days [2]; 2,675 children aged 3 to 17 in 23 trials [3].
Sources: [1], [2], [3]
Caveats: Both Cochrane reviews rated the evidence low to very low certainty, and [2] judged every trial at high risk of bias once unblinding by side effects is counted. Of the methylphenidate trials, 41% had industry funding. Too few data existed for 26 or 52 weeks [1]. Methylphenidate did not change quality of life (SMD 0.40, 95% CI -0.03 to 0.83) [2].

ADHD stimulantsStrong

ADHD symptoms in adults

Shown to reduce ADHD symptoms more than placebo in adults in short trials, with larger effects for amphetamines than for methylphenidate.

The numbers and sources

Effect: Clinician-rated symptoms improved more than placebo with amphetamines (SMD -0.79, 95% CI -0.99 to -0.58) and methylphenidate (SMD -0.49, 95% CI -0.64 to -0.35) [1]. A Cochrane review of amphetamines found reduced symptom severity rated by clinicians (SMD -0.90, 95% CI -1.04 to -0.75) and by patients (SMD -0.51, 95% CI -0.75 to -0.28). By drug, lisdexamfetamine (SMD -1.06, 95% CI -1.26 to -0.85) and mixed amphetamine salts (SMD -0.80, 95% CI -0.93 to -0.66) reduced symptoms, while one study of dexamphetamine found no evidence that it did (SMD -0.24, 95% CI -0.80 to 0.32) [4]. A Cochrane review of extended-release methylphenidate found a small to moderate improvement in self-rated symptoms (SMD -0.37, 95% CI -0.43 to -0.30) and investigator-rated symptoms (SMD -0.42, 95% CI -0.49 to -0.36), and no effect on days missed at work [5].
Population: 8,131 adults in the efficacy analysis [1]; 2,521 adults in 19 studies with a mean length of 5.3 weeks [4]; 5,066 adults in 24 trials with a median length of 8 weeks [5].
Sources: [1], [4], [5]
Caveats: Certainty was rated low to very low in both Cochrane reviews. Amphetamines did not improve retention in treatment (RR 1.06, 95% CI 0.99 to 1.13) [4]. Trials often excluded people with depression or anxiety or enrolled only earlier responders, which may overstate benefit [5].

Atomoxetine and viloxazineStrongStudied in children

ADHD in children, atomoxetine

Shown to reduce ADHD symptoms in children and adolescents more than placebo, a medium effect size of about 0.6.

Compared with placebo [1][2]

The numbers and sources

Effect: In a network meta-analysis, atomoxetine lowered clinician-rated core ADHD symptoms in children and adolescents at about 12 weeks (SMD -0.56, 95% CI -0.66 to -0.45). For teacher ratings, only methylphenidate and modafinil separated from placebo among the drugs with data [1]. Across 25 trials, atomoxetine outperformed placebo on overall ADHD symptoms (effect size -0.64, 95% CI -0.56 to -0.71), and the authors noted about 40% kept significant symptoms [2].
Population: 10,068 children and adolescents in the efficacy analysis of 81 trials [1]; 3,928 children and adolescents in 25 trials [2].
Sources: [1], [2]
Caveats: Short-term trials. There were too few data at 26 and 52 weeks [1].

Atomoxetine and viloxazineStrong

ADHD in adults, atomoxetine

Shown to reduce ADHD symptoms in adults more than placebo, a small to moderate effect of about 0.4, with more people stopping for side effects.

Compared with placebo [1][3]

The numbers and sources

Effect: Atomoxetine lowered clinician-rated ADHD symptoms in adults (SMD -0.45, 95% CI -0.58 to -0.32) and was less well tolerated than placebo (OR 2.33) [1]. Across 12 trials, symptoms fell on clinician (SMD -0.40) and self ratings (SMD -0.33). Stopping for any reason (OR 1.39) and for adverse events (OR 2.57) was more frequent than on placebo [3].
Population: 8,131 adults in the efficacy analysis of 51 adult trials [1]; 3,375 adults in 12 trials [3].
Sources: [1], [3]
Caveats: The authors of [3] described the balance of benefit and risk in adults as poor.

Atomoxetine and viloxazineStrongStudied in children

ADHD in children, viloxazine

Shown to raise response in children and adolescents with ADHD about 1.6 times over placebo, about 5.5 points on a symptom scale, with benefit leveling off above 400 mg.

Compared with placebo [4][5]

The numbers and sources

Effect: Viloxazine raised the share with a 50% drop on the ADHD Rating Scale (RR 1.62, 95% CI 1.36 to 1.93), and raised adverse events and sleepiness [4]. In fixed-dose trials, symptoms improved 5.47 points more than placebo (95% CI 4.03 to 6.91). The dose curve was bell-shaped, with no added benefit above about 400 mg, and effects leveled off at 4 to 6 weeks [5].
Population: 1,605 patients in 5 trials [4]; 1,560 children and adolescents, mean age 10.6, in 5 trials [5].
Sources: [4], [5]
Caveats: The two reviews share the same five trials.

ClonidineStrongStudied in children

ADHD in children and teenagers

Shown to reduce ADHD symptoms in children and teenagers against placebo, by a standardized difference of about 0.6 to 0.7 for clonidine and guanfacine.

Compared with placebo [4][5][6][7]; amphetamines, methylphenidate, atomoxetine and modafinil, in network comparisons [4]; placebo added to stimulants [5]

The numbers and sources

Effect: In a network meta-analysis at about 12 weeks, clinician-rated ADHD symptoms improved more than placebo with clonidine (SMD -0.71, 95% CI -1.17 to -0.24) and guanfacine (SMD -0.67, 95% CI -0.85 to -0.50). Teacher ratings for guanfacine did not clearly differ from placebo (SMD -0.63, 95% CI -1.62 to 0.35). Guanfacine was less well tolerated than placebo (OR 2.64, 95% CI 1.20 to 5.81) [4]. Another meta-analysis found alpha-2 agonists alone reduced overall symptoms (SMD -0.59) and, added to stimulants, by a smaller amount (SMD -0.36). Alone, they had fewer all-cause discontinuations than placebo (RR 0.70, NNT 10) [5].
Population: 10,068 children and adolescents in the efficacy analysis of 133 trials, with no adult trials of clonidine or guanfacine found [4]; 2,276 young people in 12 trials [5].
Sources: [4], [5]
Caveats: Evidence quality was moderate for both drugs against placebo [4]. Amphetamines (SMD -1.02) and methylphenidate (SMD -0.78) had larger point estimates against placebo in the same analysis [4].

Modafinil and armodafinilStrong

Focus and memory in rested healthy adults

Shown to improve cognitive test scores in rested healthy adults by a very small amount, an effect size near 0.1, with no clear advantage in any one domain.

Compared with placebo [9][10][11]

The numbers and sources

Effect: Across 19 placebo-controlled trials of a single dose, modafinil improved cognitive performance over placebo by g = 0.10 (95% CI 0.05 to 0.15). Domains did not differ, and 200 mg did no better than 100 mg [9]. A second meta-analysis found an overall effect of SMD 0.12, driven by memory updating (SMD 0.28), and its authors wrote that the belief among users that these drugs are strong cognitive enhancers is not supported [10]. A systematic review found most basic tests showed better executive function, half showed better attention and memory, and a few showed worse divergent creative thinking. It found no excess of side effects or mood changes [11].
Population: 19 trials and 67 effect sizes in adults who were not sleep deprived [9]; 14 modafinil studies with 64 effect sizes in healthy adults who were not sleep deprived [10]; primary studies in healthy adults from 1990 to 2014 [11].
Sources: [9], [10], [11]
Caveats: Most trials tested one dose in a lab, not repeated use as people take it [10]. An effect size of 0.1 is about a tenth of a standard deviation on a test score. [9] included some psychiatric samples and found no difference from nonpsychiatric ones.

ADHD stimulantsModerate

Amphetamines compared with methylphenidate

Shown to reduce clinician-rated ADHD symptoms more with amphetamines than with methylphenidate in head-to-head comparisons, in children and in adults.

The numbers and sources

Effect: In head-to-head comparisons, amphetamines reduced clinician-rated symptoms more than modafinil, atomoxetine and methylphenidate, with SMDs from -0.46 to -0.24 in children and adolescents and from -0.94 to -0.29 in adults. Amphetamines were less well tolerated than placebo in children (OR 2.30, 95% CI 1.36 to 3.89) and adults (OR 3.26, 95% CI 1.54 to 6.92), and methylphenidate less well tolerated than placebo in adults only (OR 2.39, 95% CI 1.40 to 4.08). Weighing both, the authors' data favored methylphenidate in children and amphetamines in adults [1].
Population: 133 double-blind trials, 81 in children and adolescents and 51 in adults [1].
Sources: [1]
Caveats: The SMD ranges cover three comparator drugs, and the abstract does not give the figure for methylphenidate alone. Confidence was low or very low for most indirect comparisons.

ADHD stimulantsModerateStudied in children

School work in children with ADHD

Shown to raise math output by 7.8% and math accuracy by 3.0% on methylphenidate in children with ADHD, gains the authors called small.

The numbers and sources

Effect: Across 34 studies, methylphenidate increased math productivity by 7.8%, math accuracy by 3.0% and reading speed (SMD 0.47), but not reading accuracy. Academic gains were small compared with symptom gains [6].
Population: Children with ADHD in 34 studies of methylphenidate [6].
Sources: [6]
Caveats: The evidence is for methylphenidate. The studies measured classroom tasks over short periods, not grades or long-term attainment.

BupropionModerate

ADHD

Shown to reduce ADHD symptoms in adults by a moderate amount in a Cochrane review of six small trials rated low quality. Not an approved use.

The numbers and sources

Effect: Long-acting bupropion lowered the severity of ADHD symptoms against placebo (SMD -0.50, 95% CI -0.86 to -0.15, 3 studies, 129 participants) and raised the share with clinical improvement (RR 1.50, 95% CI 1.13 to 1.99). Withdrawal for adverse effects was similar to placebo (RR 1.20, 95% CI 0.35 to 4.10) [12]. In children and adolescents, three head-to-head trials found bupropion comparable to methylphenidate, while a large placebo-controlled study found smaller effects than methylphenidate [13].
Population: 438 adults in 6 studies of 6 to 10 weeks [12]; open and randomized trials in children and adolescents [13].
Sources: [12], [13]
Caveats: Four of six adult studies were industry funded, none was judged free of bias, and samples were small [12]. The pediatric review calls its database very limited [13].

Modafinil and armodafinilModerate

Alertness and performance during sleep deprivation

Shown to keep sleep-deprived healthy people awake and to hold memory and executive function better than placebo, though repeated doses did not prevent decline over longer deprivation.

Compared with placebo [12]

The numbers and sources

Effect: In sleep-deprived individuals, modafinil maintained wakefulness, memory and executive function to a significantly greater degree than placebo. Over longer sleep deprivation, repeated doses did not prevent cognitive performance from deteriorating, while wakefulness was maintained and people possibly became overconfident about their own performance [12].
Population: Healthy adults in single- or double-blind randomized trials [12].
Sources: [12]
Caveats: One meta-analysis, and the abstract does not give pooled effect sizes or head counts [12].

Atomoxetine and viloxazineLimited

ADHD in adults, viloxazine

One trial found viloxazine lowered adult ADHD symptoms 3.7 points more than placebo over 6 weeks.

Compared with placebo [6]

The numbers and sources

Effect: On a flexible dose of 200 to 600 mg, symptom scores fell 15.5 points against 11.7 on placebo, a difference of 3.7 (95% CI 1.2 to 6.2) [6].
Population: 374 adults aged 18 to 65, 354 analyzed [6].
Sources: [6]
Caveats: One 6-week trial, with 107 of 374 not completing [6].

Atomoxetine and viloxazineLimited

Focus and self-control in healthy adults

Small trials found a single dose of atomoxetine improved stopping a started action in healthy volunteers, mostly in those who did worst on placebo, with no effect on learning.

Compared with placebo, in crossover designs [7][8]

The numbers and sources

Effect: In healthy volunteers, atomoxetine improved response inhibition on a stop-signal task but had no effect on probabilistic learning [7]. In a later crossover trial, a single 40 mg dose sped responses and improved stopping efficiency only in proportion to how poorly a person stopped on placebo [8].
Population: Healthy volunteers [7]; 28 control volunteers and 26 people with cocaine use disorder [8].
Sources: [7], [8]
Caveats: Single doses on lab tasks. No trial tested daily use for focus at work or study in people without ADHD.

Low-dose naltrexoneLimited

Long COVID

Limited evidence: fatigue, brain fog, sleep and pain improved from baseline in uncontrolled studies of long COVID. No randomized trials were found.

The numbers and sources

Effect: A 2026 systematic review screened 397 records and found no randomized trials. In 4 pre-post studies (155 people) at 1 to 6 mg a day, fatigue (Hedges' g -0.74, 95% CI -1.11 to -0.37), brain fog (g -0.53), sleep (g -0.60), pain (g -0.93) and daily function (g -0.93) improved from baseline [12].
Population: 155 people with long COVID in the US and Ireland in uncontrolled studies [12].
Sources: [12]
Caveats: The review rated certainty as low, and its studies had no control group [12]. No peer-reviewed randomized trial was located.

Methylene blueLimited

Memory and attention in healthy adults

Small trials found one low dose raised brain activity and memory retrieval 7% in healthy adults. No trial of repeated use was sourced.

The numbers and sources

Effect: One hour after a low oral dose, methylene blue raised functional MRI activity during attention and short-term memory tasks against placebo, and correct responses during memory retrieval rose 7% (P = 0.01) [3]. In adults with claustrophobia, 260 mg given after a fear-exposure session improved contextual memory at follow-up. People with low fear at the end of the session had less fear 1 month later on methylene blue, while people with moderate to high fear tended to do worse [4].
Population: 26 healthy adults aged 22 to 62 [3]; 42 adults with marked claustrophobic fear [4].
Sources: [3], [4]
Caveats: Two small single-dose trials, one measuring brain imaging more than behavior [3]. No trial of repeated use in healthy people was sourced.

ADHD stimulantsModerate

Focus and studying in people without ADHD

Mixed results: healthy adults showed small gains on some memory and attention tests, no effect of d-amphetamine in one meta-analysis, and lower-quality effort on a complex task.

The numbers and sources

Effect: A meta-analysis of 48 studies found small but significant effects of methylphenidate and amphetamine on inhibitory control and short-term episodic memory, and medium effects on delayed episodic memory. The authors concluded the overall effect was probably modest after accounting for publication bias [13]. A second set of meta-analyses found an overall effect of methylphenidate (SMD 0.21), driven by recall (SMD 0.43), sustained attention (SMD 0.42) and inhibitory control (SMD 0.27), and no effects for d-amphetamine [14]. In a trial using a complex optimization task, methylphenidate and dextroamphetamine lowered the value of the solutions reached against placebo. Time and steps spent rose while the quality of effort fell [15]. A systematic review of nonmedical use found little evidence that it improves academic performance in people without ADHD [16].
Population: 1,409 healthy participants in 48 studies [13]; healthy adults who were not sleep deprived in 24 methylphenidate studies and 10 d-amphetamine studies [14]; adults in one placebo-controlled experiment [15]; 111 studies of nonmedical use, mostly in college students [16].
Sources: [13], [14], [15], [16]
Caveats: Laboratory tests after single doses, which the authors of [14] said do not reflect how the drugs are used. Not an approved use. See the boxed warning and the misuse block below.

Atomoxetine and viloxazineModerate

Atomoxetine against stimulants

Mixed results: in head-to-head comparisons, amphetamines reduced clinician-rated ADHD symptoms more than atomoxetine in both children and adults.

Compared with amphetamines and methylphenidate [1]

The numbers and sources

Effect: In head-to-head comparisons, the only efficacy differences favored amphetamines over atomoxetine, methylphenidate and modafinil, with SMDs of -0.46 to -0.24 in children and -0.94 to -0.29 in adults [1].
Population: 133 trials in children, adolescents and adults [1].
Sources: [1]
Caveats: Most comparisons were indirect and rated low or very low confidence [1].

Modafinil and armodafinilModerate

ADHD

Mixed results: in a network meta-analysis, modafinil reduced ADHD symptoms in children on clinician and teacher ratings but did not differ from placebo in adults.

Compared with placebo, and other ADHD drugs in a network analysis [17]

The numbers and sources

Effect: In children and adolescents, every drug in the analysis, modafinil included, beat placebo on clinician ratings, and modafinil was one of two drugs that beat placebo on teacher ratings (SMD -0.76, 95% CI -1.15 to -0.37). In adults, modafinil did not differ from placebo (SMD 0.16, 95% CI -0.28 to 0.59) and was less well tolerated than placebo (OR 4.01). Amphetamines outperformed modafinil head to head [17].
Population: 133 double-blind trials in children, adolescents and adults with ADHD [17].
Sources: [17]
Caveats: Most comparisons were indirect and rated low or very low confidence [17]. Modafinil is not FDA-approved for ADHD.

Methylene blueModerate

Alzheimer's disease

Not shown to change cognition or daily activities in Alzheimer's disease in two trials of a related compound. Later gains came only in a subgroup.

The numbers and sources

Effect: Over 15 months, LMTM, a reduced form of the methylthioninium molecule, at 75 mg or 125 mg twice daily did not differ from a 4 mg control on cognition (ADAS-Cog differences -0.02 and -0.43) or daily activities (ADCS-ADL differences -0.93 and -0.34) [1]. In a second phase 3 trial, the same compound under the name HMTM, at 8 or 16 mg per day, did not differ from control on its co-primary cognitive and daily-activity endpoints at 52 weeks. In the subgroup with mild cognitive impairment, starting 16 mg a year earlier was followed by less cognitive decline at 78 weeks (P 0.0291) and 104 weeks (P 0.0308). Plasma neurofilament light, a marker of nerve damage, rose less at 52 weeks (P 0.0291) [2].
Population: 891 people under 90 with mild to moderate Alzheimer's disease in 16 countries [1]; 598 people with amyloid on PET, 263 with mild cognitive impairment and 335 with mild to moderate dementia, at 82 centers [2].
Sources: [1], [2]
Caveats: LMTM is related to but not the same as methylene blue. In both trials the control arm received a low dose of methylthioninium to mask urine discoloration, and the authors of [2] attribute the lack of separation to activity in that arm. The later benefits in [2] come from a subgroup and a delayed-start comparison after all participants moved to 16 mg. Funded by TauRx Pharmaceuticals, the developer [1].

Peptides and SARMs

SemaxGrey marketLimited

Focus, memory and mental performance

Limited evidence: one placebo-controlled study found a nasal dose changed resting brain networks on fMRI in healthy adults. No memory or attention results were sourced.

The numbers and sources

Effect: In 24 healthy volunteers, a single nasal dose of semax enlarged the medial frontal part of the brain's default mode network on resting-state fMRI, against placebo [9]. A 2007 hypothesis paper states that semax has been found to stimulate memory and attention in rodents and humans after nasal use [10]. FDA found that the human studies of its nootropic effects were in healthy adults or written in Russian, and did not evaluate the Russian-language reports [3].
Population: 24 healthy adults, mean age 43.9, 14 on semax and 10 on placebo [9].
Sources: [9], [10], [3]
Caveats: The fMRI study measured brain activity at rest, not focus, memory or performance [9]. The human memory and attention reports could not be read in full, and the hypothesis paper gives no figures [10].

SemaxGrey marketUntested

ADHD

Not studied in people. A Russian expert council lists semax as an ADHD add-on with a limited evidence base, and animal work shows dopamine and BDNF effects.

No study in people. This rests on documented traditional use, animal studies, how it acts in the body.

The numbers and sources

Effect: In 2026 an expert council of Russian pediatric neurologists placed semax in its group of add-on ADHD medicines with a limited evidence base, used in Russian practice, below a group backed by a double-blind trial. It stated that add-on therapy is not a substitute for atomoxetine in eligible patients [13]. A 2007 hypothesis paper proposed semax for ADHD because in animal studies it increased the effect of stimulants on brain dopamine release and raised brain BDNF [10].
Population: Clinical practice in Russia, as assessed by an expert council [13]. Animal studies, reviewed [10].
Sources: [13], [10]
Caveats: A consensus statement and a hypothesis paper, not trial results. No trial of semax in ADHD was located.

Supplements

CaffeineStrong

Attention and alertness under sleep loss

Shown to improve attention, reaction time and driving lane control in sleep-deprived adults, compared with sleep loss without caffeine.

The numbers and sources

Effect: Across 45 publications in sleep-deprived or sleep-restricted people, caffeine improved attention response time (g 0.86), attention accuracy (g 0.68, 95% CI 0.48 to 0.88), executive function (g 0.35, 95% CI 0.15 to 0.55), reaction time (g 1.11, 95% CI 0.75 to 1.47) and driving lane control (g 1.67, 95% CI 1.32 to 2.02) [7]. In 7 placebo-controlled trials in military personnel under sleep loss, caffeine improved vigilance and response performance (SMD 0.73, 95% CI 0.50 to 0.96), complex cognitive performance (0.54, 95% CI 0.16 to 0.93) and military task performance (0.45, 95% CI 0.19 to 0.71) [8].
Population: Sleep-deprived or restricted adults, 327 effect estimates [7]. Military personnel in 7 randomized trials [8].
Sources: [3], [7], [8]
Caveats: The pooled effects are against a sleep-deprived baseline, not a rested one. The abstract's confidence interval for attention response time (0.53 to 0.83) does not contain its point estimate (0.86), which suggests a typo in the source.

CaffeineStrong

Attention when rested

Shown to improve attention accuracy and reaction time slightly in rested adults, more at doses of 200 mg or more.

The numbers and sources

Effect: Across 31 trials (1,455 participants) in the general population, caffeine improved attention accuracy (g 0.27) and reaction time (g 0.28). Doses of 200 mg or more improved both more than lower doses. Reaction time kept improving with dose, while accuracy rose and then fell [9]. In 14 studies (250 participants) aged 20 or younger, caffeine improved cognitive task performance (SMD 0.41, 95% CI 0.28 to 0.55) [10].
Population: Adults from the general population [9]. Children, adolescents and young adults up to age 20 [10].
Sources: [9], [10]
Caveats: Effects were small. The youth meta-analysis rests on small-sample studies [10]. Habitual caffeine intake did not change the effect in either review [9][10].

CreatineStrong

Memory and cognition

Shown to improve memory scores modestly, most in older adults, with no change in people aged 11 to 31.

The numbers and sources

Effect: Pooled across 8 RCTs, memory scores improved over placebo with an SMD of 0.29 (95% CI 0.04 to 0.53, I² 66%). In adults aged 66 to 76 the SMD was 0.88 (95% CI 0.22 to 1.55, I² 83%). In people aged 11 to 31 it was 0.03 (95% CI -0.14 to 0.20) [9]. Across 16 RCTs (492 participants), creatine improved memory (SMD 0.31, 95% CI 0.18 to 0.44, moderate certainty), attention time and processing speed time, and did not change overall cognition or executive function [10]. A systematic review of 6 studies in 1,542 adults aged 55 and over found 5 reported a positive link between creatine and cognition, mostly memory and attention [11].
Population: Healthy people, 10 RCTs reviewed and 8 pooled, doses 2.2 to 20 g a day for 5 days to 24 weeks [9]; adults aged 20.8 to 76.4, healthy and with specific diseases [10]; older adults, 2 supplement trials and 4 diet surveys [11].
Sources: [9], [10], [11]
Caveats: Heterogeneity was high in [9]. The reviews disagree on who benefits: [9] found the effect in older adults and none at 11 to 31, while [10] found more benefit at 18 to 60, in women and in people with diseases. Only memory reached moderate certainty, and the other domains were low [10]. In the older adult review, 4 of 6 studies were cross-sectional diet surveys and 3 were rated poor quality [11].

GinsengStrong

Cognition in cognitive impairment or dementia

Shown to raise cognition scores slightly in people with cognitive impairment or dementia, below the threshold for clinical significance.

The numbers and sources

Effect: In 8 RCTs of people with subjective memory impairment, mild cognitive impairment or Alzheimer's disease, ginseng raised MMSE scores by 0.68 points (95% CI 0.03 to 1.32, I² 66%) and lowered ADAS-cog scores by 1.10 points (95% CI -1.82 to -0.38, I² 0%) [1]. In 9 studies of 617 older adults with mild cognitive impairment, Alzheimer's disease or dementia, MMSE rose 0.38 points (95% CI 0.03 to 0.74) and ADAS-cog fell 2.21 points (95% CI -3.97 to -0.44) [2].
Population: Adults with memory or cognitive impairment [1]. 617 older adults with mild cognitive impairment, Alzheimer's disease or dementia [2].
Sources: [1], [2]
Caveats: The authors of [2] report that the gains did not reach the threshold for clinical significance, and note few studies and moderate heterogeneity. MMSE heterogeneity in [1] was 66%. Review [2] searched Chinese databases.

L-TheanineStrong

Attention and reaction time

Shown to improve choice reaction time in the hour after a single 200 mg dose in adults. Simple reaction time and the Stroop test did not change.

The numbers and sources

Effect: Across 31 RCTs (1,168 participants), a single 200 mg dose taken 30 to 60 minutes before testing improved choice reaction time (SMD 0.51, 95% CI 0.25 to 0.77) [1]. A separate meta-analysis of healthy participants found theanine alone improved choice reaction time in the first hour (SMD -0.35, 95% CI -0.61 to -0.10), and theanine plus caffeine improved some attention-task accuracy in the second hour [2]. A third meta-analysis of 5 RCTs (148 healthy adults) found faster rapid visual information processing and recognition reaction time (MD -15.20 ms, 95% CI -28.99 to -1.41), and no difference in simple reaction time or the Stroop test [3].
Population: Healthy and clinical populations [1]. Healthy participants only [2][3].
Sources: [1], [2], [3]
Caveats: The second review notes the confidence intervals often left the direction and size of effects uncertain [2]. The third pooled only 5 trials of varied products, and simple reaction time and Stroop results did not differ [3]. The reviews agree on choice and recognition reaction time, not on every attention test. Many trials pair theanine with caffeine.

IronModerate

Mood and thinking

Shown to improve anxiety, memory and intelligence scores modestly in iron-deficient children and menstruating adults, with no effect in those without deficiency.

The numbers and sources

Effect: In RCTs of non-anemic children, adolescents and menstruating adults, iron improved anxiety symptoms (d 0.34), physical well-being (d 0.42), cognitive intelligence scores (d 0.46) and short-term memory (d 0.53), and did not change attention or depression [5]. An umbrella review of 17 systematic reviews in children found modest gains in intelligence, memory and attention in anemic children, negligible or uncertain effects in children without deficiency, and no consistent change in broader mental development or school achievement [8].
Population: 1,408 participants in 18 studies [5]. Children in 17 systematic reviews, 3 rated high confidence [8].
Sources: [5], [8]
Caveats: The two reviews studied different groups. The 2025 abstract gives effect sizes without confidence intervals, and the effects were absent in participants without iron deficiency [5].

MultivitaminModerate

Memory and cognition in older adults

Shown to improve global cognition and memory slightly over 3 years in US adults 60 and older, all in one trial of one product.

The numbers and sources

Effect: In COSMOS-Mind (2,262 adults, mean age 73), a daily multivitamin improved a global cognition composite over 3 years (mean z 0.07, 95% CI 0.02 to 0.12) [7]. In COSMOS-Web (3,562 adults), immediate recall was better at 1 year and across 3 years, and secondary outcomes did not change [8]. Pooling three COSMOS substudies, global cognition improved by 0.07 SD units (95% CI 0.03 to 0.11) and episodic memory by 0.06 SD units (95% CI 0.03 to 0.10), which the authors equate to about 2 years of cognitive aging [9].
Population: US adults aged 60 and older in the COSMOS trial [7][8][9].
Sources: [7], [8], [9]
Caveats: All three come from one parent trial and one product. COSMOS-Web gave Centrum Silver [8]. Funder lists in Europe PMC include Pfizer [9] and BASF [8]. Participants were 89% non-Hispanic White in COSMOS-Mind [7].

Alpha-GPC and choline supplementsLimited

Cognition in dementia and mild cognitive impairment

Limited evidence: citicoline and choline alfoscerate improved cognitive scores in people with dementia or mild cognitive impairment, in studies of poor quality with risk of bias.

The numbers and sources

Effect: Across 6 pooled studies, citicoline improved cognitive status with SMDs of 0.56 (95% CI 0.37 to 0.75) to 1.57 (95% CI 0.77 to 2.37) depending on the sensitivity analysis [1]. In 261 patients with mild to moderate Alzheimer's disease, choline alfoscerate 400 mg three times a day for 180 days lowered ADAS-Cog by 3.20 points from baseline while placebo rose by 2.90 points [2]. In 3 head-to-head trials (358 participants), choline alfoscerate scored better than citicoline on a geriatric clinical scale (WMD -3.92, 95% CI -7.41 to -0.42), with no difference in memory or word fluency [3].
Population: Mild cognitive impairment, Alzheimer's disease and post-stroke dementia [1]; Alzheimer's disease, aged 60 to 80 [2]; dementia disorders [3].
Sources: [1], [2], [3]
Caveats: Study quality in [1] was poor, with significant risk of bias in favor of the intervention. The head-to-head review compares two choline compounds and has no placebo arm [3].

Alpha-GPC and choline supplementsLimited

Memory and attention in healthy adults

Small trials found citicoline raised a memory score in older adults with memory impairment, and alpha-GPC raised one attention score in young men.

The numbers and sources

Effect: In 100 adults aged 50 to 85 with age-associated memory impairment, citicoline 500 mg a day for 12 weeks raised a composite memory score by 3.78 versus 0.72 on placebo, a secondary outcome [7]. In 20 resistance-trained men, single doses of 315 mg and 630 mg alpha-GPC raised Stroop scores over placebo (13.0 versus 5.2 at the higher dose). Flanker and N-back tests did not differ [8].
Population: Healthy older adults [7]; young men [8].
Sources: [7], [8]
Caveats: The memory gains in [7] were secondary outcomes, using a branded product (Cognizin). The alpha-GPC trial was funded by the ingredient maker and retrospectively registered [8].

BacopaLimited

Memory in older adults

Small trials found 300 mg a day for 12 weeks improved delayed word recall in adults over 55, while other memory tasks did not change.

The numbers and sources

Effect: In 54 adults aged 65 or older, 300 mg a day for 12 weeks raised delayed word recall and Stroop scores against placebo. Divided attention and working memory tasks did not change [6]. In 98 healthy Australians over 55, 300 mg a day of BacoMind for 12 weeks improved verbal learning and delayed recall. Visual memory, trail making and subjective memory did not differ [7].
Population: 54 older adults in Oregon [6]. 98 adults over 55, 81 completers [7].
Sources: [6], [7]
Caveats: Two small trials. Trial [7] tested BacoMind, made by Natural Remedies Pvt. Ltd., and its abstract does not report funding.

IodineLimitedStudied in children

Cognition in iodine-deficient children

Limited evidence: iodine slightly raised cognitive scores in iodine-deficient children, with each result resting on one trial.

The numbers and sources

Effect: After 24 weeks, iodine improved 4 of 7 cognitive and motor tests against placebo, including rapid target marking, symbol search, rapid object naming and Raven's Coloured Progressive Matrices. Total thyroxine rose about 40% and low thyroxine fell below 1% [1]. A 2026 Cochrane review of 8 RCTs (2,528 children and adolescents) found a slight rise in cognitive function against placebo (SMD 0.19, 95% CI 0.04 to 0.34), from 1 study of 166 participants rated high certainty [2].
Population: 310 children aged 10 to 12 in rural Albania, median urinary iodine 43 µg/L at baseline, 87% with goiter [1]. Trials in seven countries from 1982 to 2009 [2].
Sources: [1], [2]
Caveats: Each cognition result rests on one trial in children with deficiency. The Albanian trial used a single 400 mg dose of iodized oil [1]. The Cochrane trials were old and mostly gave a single large dose [2].

L-TheanineLimited

Symptoms of mental disorders

Limited evidence: a review of 11 trials reported fewer symptoms in schizophrenia, anxiety disorders and ADHD, often alongside medication. One anxiety trial found no effect.

The numbers and sources

Effect: A systematic review of 11 RCTs from six countries reported that theanine reduced psychiatric symptoms more than control in schizophrenia, anxiety disorders and ADHD [8].
Population: Patients with schizophrenia, ADHD, OCD, major depression, sleep disorders, GAD or Tourette syndrome, often taking it alongside their medication [8].
Sources: [8]
Caveats: No pooled effect sizes are given in the abstract. Results differ by disorder, and an adjunctive GAD trial found no anxiety effect [9].

Lion's ManeLimited

Cognition in older adults with impairment

Small trials found higher cognition scores than placebo in older adults with mild cognitive impairment or mild Alzheimer's, with scores falling again after stopping.

The numbers and sources

Effect: In 30 Japanese adults aged 50 to 80 with mild cognitive impairment, four 250 mg tablets of 96% dried mushroom powder three times a day for 16 weeks raised scores on the revised Hasegawa dementia scale compared with placebo at weeks 8, 12 and 16. Scores fell again 4 weeks after stopping [1]. In a 12-week trial, MMSE scores improved with fruiting-body supplements while two other memory tests did not differ [2]. In 49 weeks of erinacine A-enriched mycelium in mild Alzheimer's disease, MMSE and daily-living scores were higher than placebo, and placebo CASI scores fell [3]. A systematic review pooling one RCT and one pilot trial reported a weighted mean MMSE increase of 1.17 [4].
Population: Older adults with mild cognitive impairment or mild Alzheimer's disease [1][2][3][4].
Sources: [1], [2], [3], [4]
Caveats: Small trials using different preparations. The Alzheimer's trial was supported by Grape King Bio, which paid salaries for five authors and supplied research materials [3].

Vitamin B6LimitedStudied in women

Memory in postmenopausal women

One trial without placebo found memory scores rose from baseline over 6 months in postmenopausal women with mild cognitive impairment in India.

The numbers and sources

Effect: Over 6 months, scores on the Montreal Cognitive Assessment rose from baseline on 25 mg of pyridoxine a day (p = 0.0323), and homocysteine fell. The abstract reports change from baseline, not a comparison with the untreated group [7].
Population: 180 postmenopausal women with mild cognitive impairment in India, randomized to pyridoxine, 100 mg of vitamin C, or no treatment [7].
Sources: [7]
Caveats: One trial with no placebo. Menopause quality of life scores did not change overall [7].

Vitamin ELimited

Alzheimer's disease

One trial found daily living ability declined about 19% more slowly per year on 2,000 IU a day in veterans with mild to moderate Alzheimer's disease.

The numbers and sources

Effect: Over a mean 2.27 years, scores on a daily living scale declined 3.15 units less on 2,000 IU a day than on placebo (95% CI 0.92 to 5.39), which the authors equate to a 19% slower decline per year. Caregiver time rose least in the vitamin E group [10].
Population: 613 veterans with mild to moderate Alzheimer's disease taking a cholinesterase inhibitor, 561 analyzed [10].
Sources: [10]
Caveats: One trial at 14 Veterans Affairs centers. 52 participants with no follow-up data were left out of the analysis [10].

BacopaModerate

Memory and attention in healthy adults

Mixed results: reviews found faster reaction times and better memory in adults on standardized extracts, while a 12-week trial in adults with memory complaints found no change.

The numbers and sources

Effect: In 9 RCTs of standardized extracts taken for 12 weeks or more (518 subjects), Trail B time shortened by 17.9 ms (95% CI -24.6 to -11.2) and choice reaction time by 10.6 ms (95% CI -12.1 to -9.2) [1]. A systematic review of 6 trials found bacopa improved 9 of 17 tests of memory free recall, with little evidence in other cognitive domains [2]. In one 90-day trial (62 completers), KeenMind at 2 × 150 mg improved spatial working memory accuracy [3]. A 2026 network meta-analysis found that 600 mg a day or more improved working memory against placebo (SMD 2.03, 95% CI 1.28 to 2.78), and 300 to under 600 mg a day improved delayed memory. Sustained attention, selective attention and processing speed did not differ [4]. In adults with self-reported memory problems, 300 mg a day of Bacumen for 12 weeks did not change verbal learning (p = 0.391), attention (p = 0.713) or working memory (p = 0.610) against placebo [5].
Population: Adults without dementia [1][2][3]. 2,107 healthy adults in 29 RCTs of bacopa or ginkgo [4]. 87 completers aged 40 to 70 [5].
Sources: [1], [2], [3], [4], [5]
Caveats: Trials were small. Three different extracts were used at 300 to 450 mg a day [2]. The reviews disagree on attention [1][4], and the network analysis had few direct comparisons [4]. The abstracts of [1], [2] and [3] do not report funding. The lead author of [5] directs a contract research organization that has received nutraceutical company funding.

GinkgoModerate

Symptoms of dementia

Mixed results: some reviews found small cognition gains in dementia, mostly with one branded extract, while others found inconsistent or no benefit.

The numbers and sources

Effect: A Cochrane review of 36 trials found inconsistent results for cognition, daily living, mood and carer burden. Of the 4 most recent trials, 3 found no difference from placebo and 1 found very large effects. In 925 Alzheimer's patients from 9 trials there was no consistent benefit [1]. A meta-analysis of 9 EGb 761 trials (2,372 patients) found cognition favored ginkgo (SMD -0.58, 95% CI -1.14 to -0.01) and daily living did not differ significantly (SMD -0.32, 95% CI -0.66 to 0.03) [2]. A 2014 meta-analysis of 9 trials (2,561 patients, 22 to 26 weeks) reported cognition (WMD -2.86) and daily living (SMD -0.36) favored EGb 761, mainly at 240 mg a day [3]. A 2026 Cochrane review of 13 placebo-controlled dementia trials (3,288 participants) found cognition at 6 months may be better on ginkgo (MD -1.86 on the Syndrom-Kurztest, 95% CI -3.48 to -0.24, 9 studies, 2,801 participants) and daily living slightly better (MD -0.19, 95% CI -0.35 to -0.03, 8 studies), both at low certainty [4].
Population: People with Alzheimer's, vascular or mixed dementia, or cognitive impairment [1][2][3][4].
Sources: [1], [2], [3], [4]
Caveats: The reviews disagree. Heterogeneity was high [2], with I² of 91% to 96% for cognition and daily living in [4]. Only 4 of 82 studies in [4] were at low risk of bias in all domains. The Cochrane authors note many early trials were small, used unsatisfactory methods, and publication bias cannot be excluded [1]. Most trials tested one branded extract, EGb 761 [1][2][3].

TurmericModerate

Memory and thinking

Mixed results: one review found better overall thinking in older adults and longer trials, while another found no change in overall cognition.

The numbers and sources

Effect: In 9 RCTs (501 participants), curcumin improved global cognitive function against placebo (SMD 0.82, 95% CI 0.19 to 1.45). The effect was significant only in trials of 24 weeks or more, in participants aged 60 or older, or in Asian participants [38]. Another meta-analysis of 10 human studies (531 participants) found no effect on global cognition (SMD 0.14, 95% CI -0.78 to 1.07), with gains in working memory (SMD 1.01) and processing speed (SMD 0.37) in subgroups [39].
Population: Adults in RCTs, many older [38][39].
Sources: [38], [39]
Caveats: The two reviews disagree on global cognition. Domain gains come from subgroup analyses [39].

Fisetin and quercetin (senolytics)Limited

Physical function and cognition

Small trials found mixed results: walking improved in an uncontrolled pulmonary fibrosis study, cognition did not change in Alzheimer disease, and quercetin results in older adults varied.

The numbers and sources

Effect: In idiopathic pulmonary fibrosis, six-minute walk distance, 4 m gait speed and chair-stand time improved significantly after 3 weeks of intermittent dosing [5]. In mild Alzheimer disease, cognitive and brain imaging outcomes did not differ from baseline after 12 weeks [6]. With quercetin alone, 26 older adults doing 6 weeks of resistance training on 200 mg a day of quercetin glycosides gained more knee extension force than on placebo (115.1% against 105.3% of baseline), with no change in muscle mass [7]. In 38 older adults, a single 500 mg dose before low or moderate intensity resistance exercise produced a larger drop in electrically elicited muscle torque afterward than placebo, and not at high intensity [8].
Population: 14 adults with idiopathic pulmonary fibrosis [5]; 5 adults with early Alzheimer disease, mean age 76 [6]; healthy older adults [7][8].
Sources: [5], [6], [7], [8]
Caveats: The dasatinib plus quercetin studies were open-label with no placebo group, built to test feasibility [5][6]. The two quercetin-only trials are small and from one research group [7][8].

ResveratrolLimited

Memory and mood

Limited evidence: of 10 studies, 3 found some cognitive gains and 5 no effect, with small pooled gains in recognition and mood.

The numbers and sources

Effect: Of 10 studies, 3 found improvement in some cognitive measures, 2 mixed findings and 5 no effect. Pooled data showed better delayed recognition (SMD 0.39, 95% CI 0.08 to 0.70) and lower negative mood (SMD -0.18, 95% CI -0.31 to -0.05) [10].
Population: 10 RCTs. The pooled analyses each had 3 studies and about 165 participants [10].
Sources: [10]
Caveats: The authors call the literature inconsistent and limited [10].

B-complex vitaminsModerate

Memory and cognitive aging

Not shown to change memory or thinking across 11 trials. One trial in older adults with mild cognitive impairment found slower brain shrinkage.

The numbers and sources

Effect: Across 11 trials, B vitamins lowered homocysteine by 26% to 28% and did not change memory, speed, executive function or global cognition (z score difference 0.00, 95% CI -0.05 to 0.06) [3]. In one trial in mild cognitive impairment, brain atrophy was 0.76% a year on B vitamins and 1.08% on placebo (P = 0.001) [4].
Population: 22,000 participants in 11 trials, mean 2.3 to 5 years [3]; 168 people over 70 with mild cognitive impairment who completed MRI scans, 24 months [4].
Sources: [3], [4]
Caveats: The atrophy trial measured brain volume, not cognition. Its effect was concentrated in people with homocysteine above 13 µmol/L [4].

Fish oilModerate

Cognition in Alzheimer disease

Not shown to change cognition scores in people with Alzheimer disease across 4 trials.

The numbers and sources

Effect: Across 4 RCTs (702 patients with Alzheimer disease), ADAS-Cog scores did not differ significantly from placebo (MD 1.37, 95% CI 0.00 to 2.73, p 0.05, I² 35%) [31].
Population: 376 patients on omega-3 and 326 on placebo [31].
Sources: [31]
Caveats: Few trials. Higher ADAS-Cog scores mean worse cognition, and the interval reaches no difference [31].

FolateModerate

Memory and cognitive decline

Not shown to change memory or cognitive scores in adults across 9 placebo-controlled trials.

The numbers and sources

Effect: Across 9 placebo-controlled trials (2,835 participants, median 6 months), folic acid with or without other B vitamins changed cognitive scores by an SMD of 0.01 (95% CI -0.08 to 0.10). Memory, speed, language and executive function each showed no effect [6].
Population: Adults in 9 RCTs [6].
Sources: [6]
Caveats: Trials ran up to 3 years. None recorded dementia incidence [6].

GinkgoModerate

Mild cognitive impairment

Not shown to change cognition, daily living or global status at 6 months in adults with mild cognitive impairment.

The numbers and sources

Effect: In a 2026 Cochrane review, 12 placebo-controlled trials (1,913 participants) tested ginkgo in mild cognitive impairment. At 6 months, ginkgo probably had little to no effect on global clinical status (MD -0.03, 95% CI -0.06 to 0.01, 3 studies, 631 participants), cognition (ADAS-cog MD -0.07, 95% CI -0.67 to 0.51, 2 studies, 508 participants) or daily living (1 study, 350 participants), at moderate certainty [4].
Population: Adults with mild cognitive impairment [4].
Sources: [4]
Caveats: Each pooled outcome draws on 1 to 3 of the 12 trials [4].

GinkgoModerate

Memory and attention in healthy people

Not shown to improve memory, attention or executive function in healthy people.

The numbers and sources

Effect: Across RCTs in healthy people, effect sizes were near zero for memory (d -0.04, 95% CI -0.17 to 0.07, 13 trials, 1,132 participants), executive function (d -0.05) and attention (d -0.08). Results did not vary with age, trial length, dose or sample size [6].
Population: Healthy individuals [6].
Sources: [6]
Caveats: Effect sizes were derived from the trials that reported usable data [6].

IodineModerate

Child development after iodine in pregnancy

Not shown to change child cognition, language or motor scores after iodine in pregnancy in women with mild to moderate deficiency.

The numbers and sources

Effect: Pooled trials found no effect on child cognitive scores (mean difference -0.18, 95% CI -1.22 to 0.87), language or motor scores [3]. In the largest trial, verbal IQ at age 5 to 6 was 89.5 on 200 µg a day and 90.2 on placebo (difference -0.7, 95% CI -2.9 to 1.5), and performance IQ 97.5 against 99.1 [4].
Population: Mildly to moderately iodine-deficient pregnant women, 10 RCTs in the review with 2 pooled for cognition [3]; 832 women in India and Thailand, 313 children analyzed [4].
Sources: [3], [4]
Caveats: Only 1 of 3 trials on child development was adequately powered [3]. These results do not describe severe deficiency. The trial was funded in part by the Nestlé Foundation [4].

TaurineModerate

Memory and cognition

Not shown to improve memory or thinking in people, alone or with exercise. One subgroup on dementia drugs scored about 3 points higher.

The numbers and sources

Effect: Taurine alone or combined with exercise did not change cognitive scores. Added to dementia drugs, MMSE scores rose by 3.09 points (95% CI 2.29 to 3.90) in a subgroup [6]. A 2026 systematic review of 8 single-dose trials, mostly at 1 to 3 g, found that most cognitive outcomes showed no effect, and the gains seen were small and inconsistent. Trials that added caffeine showed more consistent gains but could not separate taurine's part [7].
Population: 7 RCTs with 402 people, with and without cognitive impairment [6]. 8 RCTs with 244 healthy participants [7].
Sources: [6], [7]
Caveats: Few small trials. The positive subgroup is one comparison among several [6]. The 2026 review covered only single doses and did not pool results [7].

Vitamin B12Moderate

Memory and thinking in people without deficiency

Not shown to improve memory or thinking in people without B12 deficiency, alone or with folic acid.

The numbers and sources

Effect: Across 16 RCTs with 6,276 participants, neither B12 alone nor B12 combined with folic acid, with or without B6, changed any subdomain of cognitive function. Meta-regression found no factor that predicted an effect [6].
Population: People without advanced neurological disorders or overt B12 deficiency [6].
Sources: [6]
Caveats: The review excluded people with overt deficiency, so it does not describe the effect of treating a deficiency.

Vitamin B12ModerateStudied in children

Growth and thinking in children

Not shown to improve thinking or growth in children. B12 alone lowered B12 deficiency, 1.8% against 7.6% on placebo.

The numbers and sources

Effect: B12 alone against placebo made little to no difference to cognitive function such as attention (SMD 0.09, 95% CI -0.10 to 0.28, 2 trials, 438 children) or height (MD -0.19 cm, 95% CI -2.44 to 2.06). It lowered B12 deficiency, 1.8% against 7.6% (RR 0.27, 95% CI 0.11 to 0.69, 3 trials, 642 children). Added to other micronutrients, B12 made little to no difference to cognitive function (SMD 0.00, 95% CI -0.13 to 0.13, 3 trials, 956 children) or development (SMD -0.05, 95% CI -0.18 to 0.08) [7].
Population: 16 trials with 4,083 children aged 2 days to 11 years in 10 countries, given 0.9 to 150 µg a day for 28 days to 31 months [7].
Sources: [7]
Caveats: The authors rated growth and cognition evidence low certainty. Few trials reported these outcomes, and no trial of B12 alone reported adverse effects [7].

BacopaLimited

Alzheimer's dementia

Limited evidence: 5 small, high-bias trials found no difference between bacopa and placebo or donepezil in Alzheimer's dementia.

The numbers and sources

Effect: A systematic review of 5 trials found no difference between bacopa and placebo or donepezil, at very low certainty. Three of the trials combined bacopa with other herbal extracts [8].
Population: Adults with Alzheimer's dementia or mild cognitive impairment due to Alzheimer's [8].
Sources: [8]
Caveats: All 5 trials were judged at high risk of bias, with small samples and wide intervals [8].

CholineLimitedStudied in women

Memory and focus in adults (choline bitartrate or chloride)

One small pilot trial found a single 1,650 mg dose did not change working memory in 20 postmenopausal women.

The numbers and sources

Effect: A single 1,650 mg dose of choline bitartrate 3 hours before testing did not change working memory performance on an N-back task against placebo. Brain activation during the task was lower on choline in parts of the right temporal lobe [11].
Population: 20 postmenopausal women in a crossover pilot study [11].
Sources: [11]
Caveats: One small pilot trial of a single dose. Brain activation is an imaging measure, not a memory outcome [11].

GinkgoLimited

Preventing dementia

One trial found 120 mg twice a day did not lower dementia or Alzheimer's incidence over 6 years in adults aged 75 or older.

The numbers and sources

Effect: In 3,069 adults aged 75 or older followed a median 6.1 years, 120 mg twice a day did not lower dementia incidence (HR 1.12, 95% CI 0.94 to 1.33) or Alzheimer's incidence (HR 1.16, 95% CI 0.97 to 1.39). Dementia rates were 3.3 per 100 person-years on ginkgo and 2.9 on placebo. Progression from mild cognitive impairment did not change (HR 1.13) [5].
Population: 2,587 adults with normal cognition and 482 with mild cognitive impairment [5].
Sources: [5]
Caveats: One trial, funded by the NIH [5].
Why limited: This rests on one trial, of 3,069 people. Strength counts independent studies that agree, so one trial is limited at any size.

GinsengLimited

Cognition in healthy people

Limited evidence: a Cochrane review found no convincing evidence of a cognitive effect in healthy people.

The numbers and sources

Effect: A Cochrane review found 9 trials, 5 with usable data, all in healthy participants. Data could not be pooled. Some aspects of cognition, behavior and quality of life improved, and the authors found no convincing evidence of a cognitive effect in healthy people and no high-quality evidence in dementia [9].
Population: Healthy adults and one trial in age-associated memory impairment [9].
Sources: [9]
Caveats: Search date was 2009 [9].

Lion's ManeLimited

Cognition in healthy young adults

Small trials found no change in overall cognition or mood in healthy young adults, with better results on single tests such as the Stroop.

The numbers and sources

Effect: In 18 adults aged 18 to 35, a single 3 g dose of a 10:1 fruiting-body extract did not change composite cognition or mood scores at 90 minutes, with better performance on one pegboard test [5]. In 41 adults aged 18 to 45, 1.8 g a day gave a faster Stroop time 60 minutes after the first dose (p = 0.005), and 28 days showed a trend toward lower subjective stress (p = 0.051) [6].
Population: Healthy young adults [5][6].
Sources: [5], [6]
Caveats: Pilot-sized trials. The authors of the 28-day trial note null and some negative findings [6].

Nicotinamide ribosideLimited

Memory and cognition in mild cognitive impairment

Small trials found no change in cognition over 10 to 12 weeks in older adults with mild cognitive impairment.

The numbers and sources

Effect: Cognition scores did not change over 10 weeks on 1 g a day [3]. In a 12-week pilot, NR doubled blood NAD but did not improve cognitive function, the primary outcome, or total cerebral blood flow against placebo [9].
Population: 20 older adults with mild cognitive impairment [3]; 42 older adults with amnestic mild cognitive impairment [9].
Sources: [3], [9]
Caveats: Two small, short pilot trials.

Vitamin DLimited

Dementia

One trial found no clear difference in dementia over about 4 years in mostly vitamin D sufficient older adults, from few cases.

The numbers and sources

Effect: In the Finnish Vitamin D Trial, dementia was diagnosed in 18 people on placebo, 14 on 1,600 IU a day (HR 0.77, 95% CI 0.38 to 1.55) and 13 on 3,200 IU a day (HR 0.72, 95% CI 0.35 to 1.48) over a mean 4.2 years [51].
Population: 2,492 adults in Finland free of dementia, mean age 68.2 [51].
Sources: [51]
Caveats: There were 45 cases in all, so the intervals are wide. Participants were largely vitamin D sufficient, with a mean baseline of 74.8 nmol/L [51].
Why limited: This rests on one trial, of 2,492 people. Strength counts independent studies that agree, so one trial is limited at any size.

Therapies

Biofeedback and neurofeedbackModerateStudied in children

ADHD symptoms, rated by those closest to treatment

Mixed results: neurofeedback lowered ADHD ratings slightly (SMD 0.35) when rated by those closest to treatment, the least blinded raters, an effect that disappeared under blinded ratings.

Compared with control conditions in randomized trials, rated by the assessors closest to the treatment setting [6]

The numbers and sources

Effect: With ratings from the assessors most proximal to the treatment setting, the least blinded measure, neurofeedback lowered total ADHD symptoms (SMD 0.35, 95% CI 0.11 to 0.59), inattention (0.36) and hyperactivity-impulsivity (0.26). The same trials showed no significant effect when probably blinded ratings were used [6].
Population: 520 children and adolescents in 13 trials [6].
Sources: [6]
Caveats: These were the least blinded measures in the review. Only 4 studies checked whether children learned to change their brain activity [6].

Biofeedback and neurofeedbackStrong

ADHD symptoms, blinded ratings and sham-controlled trials

Not shown to reduce ADHD symptoms when rated by people who did not know the treatment, or against sham neurofeedback, in two meta-analyses and a double-blind trial.

Compared with control conditions rated by assessors probably unaware of the treatment, including active and sham controls [6][7]; sham neurofeedback using prerecorded brain-wave recordings [8]

The numbers and sources

Effect: In 38 trials, ADHD symptom reports by people probably unaware of the treatment showed no improvement (SMD 0.04, 95% CI -0.10 to 0.18, 20 trials, 1,214 participants). Limited to trials using standard protocols, a small improvement appeared (SMD 0.21, 95% CI 0.02 to 0.40). Of five lab measures of thinking, only processing speed improved slightly (SMD 0.35) [7]. An earlier meta-analysis found no significant effect with probably blinded ratings or in trials with active or sham controls, and no effect on lab measures of inhibition or attention [6]. In 144 children given theta/beta neurofeedback or a sham of equal length, parent- and teacher-rated inattention improved in both groups and did not differ at the end of treatment (d 0.01) or 13 months later (d 0.23). Responders were 61% and 54% [8].
Population: 2,472 participants aged 5 to 40 in 38 trials [7]. 520 children and adolescents in 13 trials [6]. 142 children aged 7 to 10 with moderate or severe ADHD [8].
Sources: [7], [6], [8]
Caveats: Blinding in the sham trial was rated excellent [8]. The unblinded results are in the block below.

Practices

Hyperbaric oxygen therapyModerate

Vascular dementia

Shown to raise cognitive test scores by 4 points on the Mini-Mental State Examination when added to drug treatment for vascular dementia, in one meta-analysis of Chinese trials.

The numbers and sources

Effect: Across 25 RCTs (1,954 patients), adding hyperbaric oxygen to drug and conventional treatment improved Mini-Mental State Examination scores (mean difference 4.00, 95% CI 3.28 to 4.73) and activities of daily living. Adverse event rates did not differ between groups (OR 0.85, 95% CI 0.26 to 2.78) [8].
Population: 1,954 people with vascular dementia in 25 trials, 24 of which combined hyperbaric oxygen with a drug such as oxiracetam or donepezil [8].
Sources: [8]
Caveats: The abstract does not report blinding, sham controls or risk of bias ratings. Trials were located largely through Chinese databases.

MeditationModerate

Focus, memory and thinking skills, against no program

Shown to improve thinking skills very slightly in one meta-analysis, mainly executive function and working memory, against inactive comparisons such as a waiting list.

Compared with inactive comparators [19]

The numbers and sources

Effect: Pooling all cognitive domains, mindfulness programs outperformed comparators (g = 0.15, 95% CI 0.05 to 0.24). By domain, effects appeared for executive function (g = 0.15) and working memory (g = 0.23) only. Programs outperformed inactive comparators but not active ones [19].
Population: 2,931 adults in 56 randomized studies [19].
Sources: [19]
Caveats: Most trials had unclear risk of bias, and some p-values were unreliable [19].

Red light therapyModerate

Thinking and memory

Shown to improve cognitive test scores in older adults with cognitive impairment and in young adults in two meta-analyses of small trials. Home-use devices showed no significant effect.

The numbers and sources

Effect: Across 11 RCTs in age-related cognitive impairment, photobiomodulation improved global cognition (SMD 0.51, 95% CI 0.162 to 0.864). The effect was significant in clinical settings (SMD 0.468) and not with home-use devices (SMD 0.616, 95% CI -0.121 to 1.354) [11]. In young healthy adults, transcranial light improved cognition-related outcomes (SMD 0.833, 95% CI 0.458 to 1.209, 14 comparisons from 6 publications) [12]. A systematic review found 29 of 35 human studies reported cognitive improvement, and noted that stroke trials were stopped at phase III for lack of statistical significance [13].
Population: Older adults with cognitive impairment in 11 RCTs [11]. Healthy adults aged 17 to 35 [12]. Healthy adults and people with cognitive disorders in 35 studies [13].
Sources: [11], [12], [13]
Caveats: In the young adult analysis, funnel plot tests indicated publication bias, heterogeneity was high and study quality was rated modest [12]. Only half of the clinical trials in the wider review were randomized [13]. Light was applied to the head, mostly near-infrared at 810 nm.

Cold water immersionLimited

Alertness, noradrenaline and dopamine

Small studies found plasma noradrenaline rose 530% and dopamine 250% during an hour in 14°C water, and people reported feeling more alert after a 5-minute immersion.

The numbers and sources

Effect: During 1-hour head-out immersion at 14°C, plasma noradrenaline rose by 530% and dopamine by 250%, with adrenaline unchanged. Immersion at 20°C or 32°C did not produce these rises [16]. After a 5-minute cold bath, 33 adults reported feeling more active, alert and attentive [19].
Population: A group of young men, number not given in the abstract [16]. 33 adults [19].
Sources: [16], [19]
Caveats: Hormones were measured in blood during a 1-hour immersion, far longer than a typical plunge, and blood levels are not brain levels. Neither study had a control group for the self-reported feelings. No trial measured attention or cognitive performance.

Hyperbaric oxygen therapyLimited

Thinking and memory in healthy older adults

One trial found better global cognitive scores after 3 months of sessions in 63 healthy adults over 64, compared with a control group that received no treatment.

The numbers and sources

Effect: Global cognitive function improved more in the hyperbaric group than the control group (group by time interaction p = 0.0017), with the largest gains in attention (net effect size 0.745) and information processing speed (net effect size 0.788). Brain blood flow rose in several frontal and parietal regions [10]. A systematic review of 42 studies of hyperbaric oxygen and cognition across neurological conditions described the results as controversial [11].
Population: 63 healthy adults over 64, 33 assigned to hyperbaric oxygen and 30 to control [10]. 42 studies in various neurological disorders [11].
Sources: [10], [11]
Caveats: One trial. The control group had no sham sessions, so participants knew whether they were treated. No follow-up beyond the end of treatment is reported in the abstract.

Hyperbaric oxygen therapyLimited

Alzheimer's disease

Limited evidence: cognitive test scores were higher 1 month after 20 sessions in 42 people with Alzheimer's disease than in 30 untreated patients, in a non-randomized study.

The numbers and sources

Effect: After 40 minutes of hyperbaric oxygen a day for 20 days, Mini-Mental State Examination and Montreal Cognitive Assessment scores were significantly higher at 1 month in treated patients than in untreated controls. Activities of daily living improved at 1 and 3 months against baseline [19].
Population: 42 people with Alzheimer's disease and 11 with amnestic mild cognitive impairment who were treated, and 30 untreated people with Alzheimer's disease [19].
Sources: [19], [1]
Caveats: Not randomized and not blinded, with the between-group difference reported only at 1 month. FDA lists Alzheimer's disease among conditions that hyperbaric devices are not proven to cure [1].

SaunaLimited

Dementia and Alzheimer's disease

Linked to fewer dementia diagnoses in two Finnish cohorts, with hazard ratios of 0.34 and 0.81 for the most frequent users in each.

The numbers and sources

Effect: In 2,315 men followed for 20.7 years, 204 developed dementia and 123 Alzheimer's disease. Against one session a week, hazard ratios for dementia were 0.78 (95% CI 0.57 to 1.06) at 2 to 3 sessions and 0.34 (95% CI 0.16 to 0.71) at 4 to 7 sessions. For Alzheimer's disease they were 0.80 (95% CI 0.53 to 1.20) and 0.35 (95% CI 0.14 to 0.90) [15]. In 13,994 men and women followed for 39 years with 1,805 dementia diagnoses, sauna bathing 9 to 12 times a month against under 4 times carried a hazard ratio of 0.47 (95% CI 0.25 to 0.88) in the first 20 years and 0.81 (95% CI 0.69 to 0.97) over the whole follow-up [16].
Population: Finnish men aged 42 to 60 [15]. Finnish men and women aged 30 to 69 in the Mobile Clinic Follow-up Survey [16].
Sources: [15], [16]
Caveats: Observational. The two cohorts are independent of each other, and the association weakened over longer follow-up in the larger one [16].

Whole-body cryotherapyLimited

Cognitive function in mild cognitive impairment

One trial found better immediate recall and orientation scores after 10 sessions in 62 people with mild cognitive impairment, compared with a sham.

The numbers and sources

Effect: After the 10th session, immediate recall on the DemTect and orientation on the Test Your Memory differed between groups in favor of cryotherapy. Self-rated mood improved (p = 0.02 after the sessions, p = 0.07 two weeks later) [15].
Population: 62 people with mild cognitive impairment, 33 at -110°C to -160°C and 29 in a sham group [15].
Sources: [15]
Caveats: One trial. Two subtests of several differed, and the difference was measured straight after the course.

Hyperbaric oxygen therapyModerate

Persistent post-concussion symptoms

Mixed results: three sham-controlled military trials found no benefit over sham compression, while a review that counted other control designs reported improvement at 1.5 ATA.

The numbers and sources

Effect: In 72 service members, a clinically significant symptom improvement occurred in 52% with 40 hyperbaric oxygen sessions at 1.5 ATA, 33% with sham air sessions and 25% with usual care alone (P = .24). Both chamber groups improved more than usual care, and oxygen did not differ from sham (P = .70) [2]. In 50 service members, 30 sessions at 2.4 ATA produced no difference from sham on symptom or PTSD scores [3]. In 60 service members, 40 sessions at oxygen doses equal to 1.5 or 2.0 ATA produced no difference from sham [4]. A pooled analysis of 254 participants in four Department of Defense studies found non-significant trends on symptoms (Rivermead total -2.3, 95% CI -5.6 to 1.0) and a gain in verbal memory (3.8 points, 95% CI 1.0 to 6.7) [5]. A systematic review of 11 studies concluded that 40 sessions at 1.5 ATA produced symptom or cognitive improvements in four randomized trials [6].
Population: US military service members with symptoms at least months after mild traumatic brain injury [2][3][4][5]. Adults in 6 randomized trials and 5 smaller reports, all with fewer than 75 participants [6].
Sources: [2], [3], [4], [5], [6]
Caveats: The disagreement turns on what counts as a control. Sham sessions used pressurized air at 1.2 to 1.3 ATA, which the favorable review treats as an active dose [6]. The authors of the largest trial concluded the improvements were not oxygen mediated and may reflect placebo effects [2]. The favorable review has a single author.

Hyperbaric oxygen therapyLimited

Long COVID

Mixed results: a 40-session sham-controlled trial found better cognition, energy and sleep scores in long COVID, and a 10-session trial found no difference from sham.

The numbers and sources

Effect: In 73 patients with symptoms at least 3 months after infection, 40 daily sessions improved global cognitive function (d = 0.495, p = 0.038), attention, executive function, energy (d = 0.522), sleep, psychiatric symptoms and pain interference compared with sham [12]. In 80 patients, 10 sessions at 2.4 bar over 6 weeks did not differ from sham at 13 weeks in physical functioning (difference 0.63, 95% CI -7.04 to 8.29) or role limitations due to physical health (difference 2.35, 95% CI -5.95 to 10.66). Both groups improved [13].
Population: 73 adults, 37 treated and 36 sham [12]. 80 previously healthy adults aged 18 to 60, 40 per group [13].
Sources: [12], [13]
Caveats: The trials differ fourfold in the number of sessions and in their outcome measures. Outcomes in the positive trial were assessed 1 to 3 weeks after the last session [12].

MeditationModerate

Focus and attention, against active programs

Not shown to improve attention or thinking skills more than active comparison programs, in two meta-analyses.

Compared with active comparators [19]; active control programs [3]

The numbers and sources

Effect: Mindfulness programs did not outperform active comparators on objective cognitive tests [19]. A review limited to active controls found low evidence of no effect, or insufficient evidence, for attention [3].
Population: 2,931 adults in 56 studies [19]. 3,515 participants in 47 trials [3].
Sources: [19], [3]
Caveats: Few trials measured attention against an active control, so a small effect could be missed.

Daily life

NappingStrong

Memory and thinking after a nap

Shown to improve memory and overall test performance after an afternoon nap, a small to medium effect in three meta-analyses, similar across ages and nap lengths.

Compared with control conditions the abstract does not describe [1]; a no-nap control condition, and the same people before the nap [2]; a no-nap condition in randomized trials [3]

The numbers and sources

Effect: Across 60 samples, naps improved the aggregate of cognitive tests (Cohen's d 0.379, 95% CI 0.296 to 0.462), declarative memory (d 0.376), procedural memory (d 0.494) and speed of processing (d 0.211). Age, nap length, nap start time, habitual napping and prior sleep loss did not change the effect [1]. In working-age adults, overall cognitive performance improved after a nap (effect size 0.18, 95% CI 0.09 to 0.27), mainly up to 120 minutes after waking [2]. In young athletes and active men, a nap after a normal night improved cognitive performance (SMD 0.69, 95% CI 0.37 to 1.00) [3].
Population: 60 samples from 54 studies [1]. 381 participants in 11 studies, almost all in laboratories, mean nap 55.4 minutes [2]. 291 males aged 18 to 35 in 22 randomized trials [3].
Sources: [1], [2], [3]
Caveats: Short laboratory studies of single naps. Heterogeneity was high in the athlete trials (I² 71.5%) [3]. Results in real workplaces are scarce [2].

NappingStrong

Alertness and fatigue after a nap

Shown to raise vigilance and alertness and to reduce felt fatigue after a nap, in three meta-analyses of nap studies.

Compared with control conditions the abstract does not describe [1]; a no-nap control condition, and the same people before the nap [2]; a no-nap condition in randomized trials [3]

The numbers and sources

Effect: Naps improved vigilance (Cohen's d 0.610, 95% CI 0.291 to 0.929) [1]. Alertness improved after a nap (effect size 0.29, 95% CI 0.10 to 0.48), with better results for naps before 1 p.m. [2]. Perceived fatigue fell after a nap following a normal night (SMD -0.76, 95% CI -1.24 to -0.28), and benefits were also seen after partial sleep deprivation [3].
Population: 60 samples from 54 studies [1]. 381 working-age adults in 11 studies [2]. 291 young males in 22 randomized trials [3].
Sources: [1], [2], [3]
Caveats: Most studies were in laboratories, short and small. Fatigue was self-rated, and heterogeneity was high (I² 89.5%) [3].

Daily stepsModerate

Dementia

Linked to 38% less dementia at 7,000 steps a day than at 2,000, in one meta-analysis of two cohorts.

Compared with 2,000 steps a day [4]

The numbers and sources

Effect: Against 2,000 steps a day, 7,000 steps was linked to a 38% lower risk of dementia (HR 0.62, 95% CI 0.53 to 0.73, I² 0%), with a non-linear pattern [4].
Population: 2 cohort studies [4].
Sources: [4]
Caveats: Observational and only two studies. Dementia develops for years before diagnosis and lowers activity in that time, so reverse causation is a strong candidate.

NappingModerate

Naps during night shifts

Shown to improve cognitive performance during night shifts after a nap, mostly from 30 minutes after waking, in one meta-analysis. Memory dipped right after waking.

Compared with control conditions without a nap [4]

The numbers and sources

Effect: Performance improved after a night-shift nap against no nap (effect size 0.25, 95% CI 0.10 to 0.41). Gains were seen mainly 30 minutes after waking. Memory got worse immediately after waking, with no overall change in global performance at that point. Earlier naps in the night tended to help more [4].
Population: 494 participants in 18 studies, 6 in real work settings and 12 in laboratories. Mean nap 41.6 minutes, at a mean time of 2:12 a.m. [4].
Sources: [4]
Caveats: One meta-analysis with few real-work studies. Safety, health and economic outcomes were not measured [4].

Sleep durationModerate

Attention and thinking

Shown to impair attention after total sleep loss in pooled experiments, and 14 nights of 6 hours or less in bed caused growing deficits in one trial.

Compared with a baseline test in the same people, or a control group [11]; 8 hours in bed a night [12]

The numbers and sources

Effect: Short-term total sleep deprivation had its largest effect on lapses in simple attention (g -0.776, 95% CI -0.96 to -0.60) and a small, nonsignificant effect on reasoning accuracy (g -0.125, 95% CI -0.27 to 0.02) [11]. Restricting time in bed to 4 or 6 hours for 14 days caused cumulative, dose-dependent deficits on all tasks, equal to up to 2 nights without sleep. Sleepiness ratings rose little after the first days, so people largely did not notice their growing deficits [12].
Population: 147 cognitive tests in 70 articles of under 48 hours of total sleep deprivation [11]. 48 healthy adults aged 21 to 38 in a laboratory [12].
Sources: [11], [12]
Caveats: Laboratory studies in healthy young adults. Total deprivation is more extreme than everyday short sleep. The restriction trial was small [12].

Volunteering and purposeModerate

Memory and thinking in older volunteers

Linked to better memory and verbal fluency in older volunteers in 16 of 17 studies in one review. One large US cohort found no link with cognitive impairment.

Compared with older adults who did not volunteer [5]; retired seniors who did not volunteer or volunteered only on and off [6]; people who volunteered 0 hours a year [4]

The numbers and sources

Effect: Sixteen of 17 longitudinal studies and trials found a positive link between volunteering and cognition, with better episodic memory, working memory and verbal fluency [5]. Seniors who volunteered continuously reported fewer cognitive complaints over time and were about 2.4 times less likely to be prescribed an anti-dementia drug [6]. In a 4-year US cohort, volunteering was not linked to cognitive impairment [4].
Population: 17 studies, mostly in high-income countries [5]. 1,001 retired Swedish seniors over 5 years [6]. 12,998 US adults over 50 [4].
Sources: [5], [6], [4]
Caveats: Observational for the most part. Early cognitive decline makes people stop volunteering, which is reverse causation. Drug prescriptions are an indirect measure of dementia [6].

Volunteering and purposeModerate

Purpose in life and dementia

Linked to 23% less dementia in people with a stronger sense of purpose in life, in one meta-analysis of six samples followed up to 17 years.

Compared with people with a lower sense of purpose in life [12]

The numbers and sources

Effect: Greater purpose was linked to lower dementia risk (HR 0.77, 95% CI 0.73 to 0.81). The link held after adjusting for depression and physical inactivity and did not differ by age, gender or education [12].
Population: 53,499 people with 5,862 new dementia cases in 6 samples [12].
Sources: [12]
Caveats: Observational. Dementia changes motivation for years before diagnosis, so low purpose may be an early sign rather than a cause.

Sleep timingLimited

Shifting night owls to an earlier schedule

One trial moved night owls' sleep and wake times about 2 hours earlier and found less depression and stress and better morning reaction time and grip strength.

Compared with a control group the abstract does not describe [1]

The numbers and sources

Effect: Earlier wake and sleep times, fixed meal times, limits on caffeine and timed exercise advanced sleep and wake timing by about 2 hours, measured by actigraphy, dim light melatonin onset and the cortisol awakening response, with no loss of sleep duration. Self-reported depression and stress improved, and reaction time and grip strength improved in the morning hours [1].
Population: Night owls with a late sleep and wake pattern, in one randomized controlled trial in a real-world setting. The abstract does not give the head count or length [1].
Sources: [1]
Caveats: One small trial. Participants knew their group, and mood and stress were self-rated. The trial changed several habits at once, so it cannot say which mattered.

Physical activityModerate

Dementia

Mixed results: one meta-analysis linked activity to 20% less dementia even with long follow-up, while another found no link once activity was measured 10 or more years before diagnosis.

Compared with people with lower activity, as each study defined it [16]; physically inactive participants [17]

The numbers and sources

Effect: Activity was linked to lower all-cause dementia (RR 0.80, 95% CI 0.77 to 0.84) and Alzheimer's disease (RR 0.86), including in follow-ups of 20 years or more [16]. Inactivity measured under 10 years before diagnosis was linked to more dementia (HR 1.40). Measured 10 or more years before, it was not (HR 1.01, 95% CI 0.89 to 1.14) [17].
Population: 257,983 adults for all-cause dementia in 58 studies [16]. 404,840 people, mean age 45.5, in 19 cohorts with 2,044 dementia cases [17].
Sources: [16], [17]
Caveats: Dementia develops for years before diagnosis and lowers activity in that time, so reverse causation is a strong candidate [17]. There were signs of publication bias [16].

Shift workModerate

Bright light, naps and other measures for sleepiness on shift

Mixed results: in a Cochrane review, bright light at night lowered sleepiness slightly in two studies, while naps, exercise and sleep education gave inconclusive results.

Compared with normal light [2]; control conditions the review abstract does not describe, for naps and other measures [2]

The numbers and sources

Effect: Bright light during the night shift lowered sleepiness by 0.83 points on the Stanford Sleepiness Scale (95% CI -1.3 to -0.36, very low-quality evidence). Another trial found no difference on another scale, and bright light during day shifts did not reduce sleepiness. A single nap opportunity changed reaction time by 11.87 ms, not significant. Exercise and sleep education showed promise without enough data to judge [2].
Population: 556 shift workers who also work nights, in 17 randomized trials [2].
Sources: [2]
Caveats: Most comparisons gave low to very low-quality evidence and studies were too varied to pool [2].

Volunteering and purposeLimited

A school volunteering program in a randomized trial

One trial of a school volunteering program found no change in a standard thinking test by intention to treat. Brain volume rose in men, against declines in controls.

Compared with a low-activity control group [7]; controls in the same trial [8]; controls offered usual volunteer opportunities [9]

The numbers and sources

Effect: On a timed attention test, intention-to-treat analyses found no significant improvement, while people with the highest attendance showed less variability in speed at 24 months [7]. Over 2 years, men in the program gained 0.7% to 1.6% in cortical and hippocampal volume while men in the control group declined, and women in the program gained 0.3% to 0.54% against declines of about 1% in controls. Only the gains in men were significant [8]. Volunteers reported higher generativity, a sense of contributing to others, at each follow-up [9].
Population: 336 older adults without dementia [7]. 111 men and women in the brain imaging study [8]. Older adults in the Baltimore Experience Corps Trial, randomized to serve as tutors in elementary schools or to control [9].
Sources: [7], [8], [9]
Caveats: One trial with small imaging and cognition subsamples. Brain volume is not a measure of memory or dementia. The cognition gains appeared only in people who attended most [7].

Food and drink

Fish and seafoodStrong

Memory, cognitive decline and dementia

Linked to 18% lower cognitive impairment and 18% lower dementia at the highest fish intake than the lowest, in pooled observational studies of adults.

Compared with the lowest fish intake [14]; one fewer serving a week, in a dose-response model [15]

The numbers and sources

Effect: Across 35 studies, the highest against the lowest fish intake was linked to lower cognitive impairment or decline (RR 0.82, 95% CI 0.75 to 0.90), dementia (RR 0.82, 95% CI 0.73 to 0.93) and Alzheimer disease (RR 0.80, 95% CI 0.67 to 0.96). Risk of cognitive impairment was up to 30% lower at 150 g a day (RR 0.70, 95% CI 0.52 to 0.95) [14]. Across 21 cohorts, each extra weekly serving of fish was linked to 5% lower dementia (RR 0.95, 95% CI 0.90 to 0.99) and 7% lower Alzheimer disease (RR 0.93, 95% CI 0.90 to 0.95) [15].
Population: 35 observational studies [14]. 181,580 participants and 4,438 cases followed 2.1 to 21 years [15].
Sources: [14], [15]
Caveats: Observational, with heterogeneity up to I² 61.1% [14] and 74.8% [15]. The authors of the newer review call for further studies on dementia and Alzheimer disease [14]. People with early memory loss may change their diet, which can run the link backward. Day-to-day focus and attention in adults were not the outcomes measured.

CoffeeLimited

Focus and alertness

One trial found caffeinated coffee helped people take in new information and offset fatigue against decaffeinated coffee, with no change in working memory or mood.

Compared with decaffeinated coffee [12]; people drinking no coffee [13]

The numbers and sources

Effect: In 144 volunteers, caffeinated coffee had no effect on initial mood or working memory, improved encoding of new information and counteracted fatigue that built up over the test session, against decaffeinated coffee. It raised blood pressure and pulse [12]. In adults 60 or older, those drinking 266.4 to 495 g of coffee a day had lower odds of a low processing speed score (OR 0.56, 95% CI 0.35 to 0.89) than non-drinkers. Decaffeinated coffee was not linked to any cognitive score [13].
Population: 144 volunteers, 72 men and 72 women, mean age 21, on one morning [12]. 2,513 US adults aged 60 or older, surveyed once [13].
Sources: [12], [13]
Caveats: One small single-morning trial, and a survey measured at one time point that cannot show cause. Trials of caffeine on attention are on the caffeine page.

Energy drinksLimited

Alertness and mental performance

A systematic review found the caffeine dose in a typical energy drink can improve mental performance, with little evidence that taurine, ginseng or other added ingredients contribute.

Compared with caffeine alone or placebo, in the studies reviewed [5]

The numbers and sources

Effect: A typical 235 ml energy drink provides 40 to 250 mg of caffeine, doses that improve cognitive performance and, at the higher levels, physical performance. Across 32 articles, apart from weak evidence for glucose and guarana extract, there was an overwhelming lack of evidence that ingredients other than caffeine enhance physical or cognitive performance [5].
Population: 32 articles testing energy drink ingredients alone or with caffeine [5].
Sources: [5]
Caveats: A systematic review without pooled numbers. Pooled trials of caffeine and attention are on the caffeine page.

Dark chocolateModerate

Focus and thinking

Mixed results: two meta-analyses found better scores on some attention and executive tasks, while a 3-year trial of cocoa extract found no change in overall thinking.

Compared with control conditions the abstract does not describe [8]; control groups [9]; placebo [10]

The numbers and sources

Effect: Regular dark chocolate or cocoa improved Trail Making Test part A (WMD -4.51) and part B (WMD -13.1) and verbal fluency (WMD 3.15), with larger effects at 500 mg of flavanols a day or more and 8 weeks or longer. Memory results were inconsistent and Mini-Mental State scores did not improve [8]. In 7 trials in healthy adults, executive function time fell (WMD -11.77, 95% CI -22.49 to -1.05) [9]. In 2,262 older adults, 3 years of cocoa extract with 500 mg of flavanols did not change global cognition (mean z 0.03, 95% CI -0.02 to 0.08) [10].
Population: 2,847 participants in 16 trials [8]. Healthy adults in 7 trials [9]. 2,262 adults with mean age 73 [10].
Sources: [8], [9], [10]
Caveats: The two meta-analyses share trials. The large trial tested an extract by telephone assessments in older adults [10].

Added sugarModerateStudied in children

Behavior and thinking in children

Not shown to change behavior or thinking in children against an artificial sweetener placebo, in a meta-analysis of 23 blinded studies.

Compared with an artificial sweetener placebo, with children, parents and staff blinded [6]

The numbers and sources

Effect: Across 14 measures of behavior and thinking, effect sizes ranged from -0.14 for direct observation to +0.30 for academic tests, and every 95% confidence interval included zero [6].
Population: Children in 23 within-subject studies from 16 reports [6].
Sources: [6]
Caveats: The authors note a small effect, or effects in some children, cannot be ruled out [6]. The review dates from 1995.

Diets

Mediterranean dietLimited

Cognition and dementia

Mixed results: one trial found better cognitive composite scores, most other trial outcomes did not differ, and two cohort meta-analyses found no significant dementia link.

Compared with a control diet of advice to reduce dietary fat [14]; an alternate diet the abstract does not describe [15]; people with the lowest Mediterranean diet adherence, in cohorts [16], [17]

The numbers and sources

Effect: In 334 PREDIMED participants retested after a median 4.1 years, the olive oil diet scored better on global and frontal composites and the nut diet on memory against a low-fat control, while all composites fell in the control group [14]. Across 5 RCTs (1,888 participants), only 8 of 66 cognitive outcomes favored a Mediterranean diet, 2 favored controls, and no trial found less cognitive impairment or dementia [15]. In cohort meta-analyses, high adherence was not significantly linked to dementia risk (HR 0.85, 95% CI 0.71 to 1.03 [16], HR 0.90, 95% CI 0.78 to 1.05 [17]).
Population: Cognitively healthy adults with a mean age of 66.9 at high cardiovascular risk [14]; adults in 5 RCTs [15]; older adults in prospective cohorts, observational [16][17].
Sources: [14], [15], [16], [17]
Caveats: The review could not pool the trials [15]. The prescribed diets varied a lot between trials [15]. Within one cohort meta-analysis, the link reached significance in studies following people 10 years or less (HR 0.72) [16].

Lab markers

HomocysteineModerate

Dementia in cohorts

Linked to more dementia at higher levels, 35% higher odds for each 5 µmol/L across 8 pooled cohorts.

The numbers and sources

Effect: In a meta-analysis of 8 cohort studies, each 5 µmol/L higher serum homocysteine had an odds ratio for new dementia of 1.35 (95% CI 1.02 to 1.79), or 1.50 (95% CI 1.13 to 2.00) after adjustment for regression dilution [10]. In the Framingham Study, each standard deviation higher log homocysteine had a relative risk of dementia of 1.4 (95% CI 1.1 to 1.9), and a level above 14 µmol/L nearly doubled the risk of Alzheimer's disease [11]. A consensus statement reports relative risks from 1.15 to 2.5 across clinical studies [2].
Population: 8,669 participants in 8 cohorts, mean ages 47 to 81, median study duration 5 years [10]. 1,092 people without dementia, mean age 76, 111 cases over a median 8 years [11].
Sources: [10], [11], [2]
Caveats: Observational. The meta-analysis authors state that the results do not provide evidence of cause and effect [10].

Folate levelLimited

Anemia and cognitive impairment when folate is high and vitamin B12 is low

Limited evidence: one US survey linked high serum folate together with low B12 status to more anemia and cognitive impairment in older adults.

The numbers and sources

Effect: Among seniors with low vitamin B12 status, serum folate above 59 nmol/L (about 26 ng/mL, the 80th percentile) was associated with anemia (OR 3.1, 95% CI 1.5 to 6.6) and cognitive impairment (OR 2.6, 95% CI 1.1 to 6.1), compared with folate at or below that level. Among those with normal B12 status, high folate was associated with less cognitive impairment (OR 0.4, 95% CI 0.2 to 0.9) [15].
Population: 1,459 seniors in the 1999 to 2002 US National Health and Nutrition Examination Survey, after folic acid fortification began [15].
Sources: [15]
Caveats: One cross-sectional survey, so it cannot show cause or order. The ng/mL figure is converted here from the study's nmol/L [15].

Holotranscobalamin (active B12)Limited

Cognitive decline and Alzheimer disease in older adults

Limited evidence: two cohorts linked lower holotranscobalamin to faster cognitive decline and to more Alzheimer disease over 7 to 10 years.

The numbers and sources

Effect: In an Oxford cohort, a doubling of holotranscobalamin from 50 to 100 pmol/L went with a 30% slower rate of cognitive decline over 10 years, and the association remained significant after adjustment for the other vitamin markers [12]. In a Finnish cohort, each 1 pmol/L higher holotranscobalamin at baseline had an odds ratio for Alzheimer disease of 0.980 (95% CI 0.965 to 0.995), and 0.977 (0.958 to 0.997) after adjustment for confounders that included age, sex, education, the APOE ε4 allele, smoking, stroke and blood pressure [13].
Population: 1,648 older adults in Oxford, United Kingdom, tested on 3 or more occasions over 10 years [12]. 271 Finnish adults aged 65 to 79 without dementia, followed for 7 years [13].
Sources: [12], [13], [14]
Caveats: Observational. In a meta-analysis of 11 trials with 22,000 participants, B vitamins given to lower homocysteine had no significant effect on global cognitive function (z score difference -0.01, 95% CI -0.03 to 0.02). Those trials did not select people by holotranscobalamin and were not trials of treating B12 deficiency [14].

Methylmalonic acid (MMA)Limited

Cognitive decline in older adults

Limited evidence: one Oxford cohort linked a doubling of methylmalonic acid to cognitive decline more than 50% faster over 10 years.

The numbers and sources

Effect: A doubling of methylmalonic acid from 250 to 500 nmol/L went with a rate of cognitive decline more than 50% faster on the Mini-Mental State Examination. The association remained significant after adjustment for the other vitamin markers. Holotranscobalamin and homocysteine also predicted decline, and folate did not [11].
Population: 1,648 older adults in Oxford, United Kingdom, tested on 3 or more occasions over 10 years [11].
Sources: [11], [12]
Caveats: Observational, from one cohort. The authors state that randomized trials are required to determine whether B12 supplementation is relevant to preventing dementia [11]. In a meta-analysis of 11 trials with 22,000 participants, B vitamins given to lower homocysteine had no significant effect on global cognitive function (z score difference -0.01, 95% CI -0.03 to 0.02). Those trials did not select people by methylmalonic acid and were not trials of treating B12 deficiency [12].

HomocysteineModerate

Cognitive decline and brain shrinkage in trials that lowered homocysteine

Mixed results: pooled trials found no effect of homocysteine-lowering B vitamins on cognitive function, while one trial in mild cognitive impairment found slower brain shrinkage.

The numbers and sources

Effect: In a meta-analysis of 11 trials, B vitamins lowered homocysteine by 26% to 28% and had no significant effect on individual cognitive domains or on global cognitive function (z score difference -0.01, 95% CI -0.03 to 0.02). The authors calculate that a 25% reduction in homocysteine equated to 0.02 years of cognitive aging per year (95% CI -0.10 to 0.13) [12]. In a 2-year trial in people over 70 with mild cognitive impairment, the yearly rate of brain atrophy was 0.76% (95% CI 0.63 to 0.90) with folic acid, vitamin B12 and vitamin B6 and 1.08% (95% CI 0.94 to 1.22) with placebo. Among participants whose homocysteine was above 13 µmol/L, the rate was 53% lower with treatment [13].
Population: 22,000 participants in 11 trials, treated for a mean of 2.3 to 5 years [12]. 271 people randomized, 168 of whom completed the brain scans [13].
Sources: [12], [13]
Caveats: The single trial measured brain volume on scans as its main outcome, in one center [13]. In 7 of the 11 pooled trials, cognitive tests were available only at the end of treatment [12].

Vitamin B12 levelModerate

Cognitive decline in older adults

Mixed results: cohort and survey data linked low B12 status to faster cognitive decline, while pooled trials of B vitamins found no effect on cognitive function.

The numbers and sources

Effect: In an Oxford cohort followed for 10 years, holotranscobalamin, methylmalonic acid and homocysteine predicted cognitive decline and folate did not. A doubling of holotranscobalamin from 50 to 100 pmol/L went with a 30% slower rate of decline, and a doubling of methylmalonic acid from 0.25 to 0.50 µmol/L with a decline more than 50% faster [6]. In a US survey of older adults, low B12 status (serum B12 below 148 pmol/L or methylmalonic acid above 210 nmol/L) was associated with cognitive impairment (OR 2.5, 95% CI 1.6 to 3.8) [7]. In a meta-analysis of 11 trials, B vitamins lowered homocysteine by 26% to 28% and had no significant effect on global cognitive function (z score difference -0.01, 95% CI -0.03 to 0.02) or on individual cognitive domains [8].
Population: 1,648 older adults in Oxford, United Kingdom, tested on 3 or more occasions over 10 years [6]. 1,459 senior participants in the 1999 to 2002 US National Health and Nutrition Examination Survey [7]. 22,000 participants in 11 trials, with treatment lasting a mean of 2.3 to 5 years [8].
Sources: [6], [7], [8]
Caveats: The cohort and survey findings are observational, and the survey measured status and cognition at one point in time. The trials tested B vitamins given to lower homocysteine and were not trials of treating B12 deficiency [8].

Folate levelModerate

Cognitive decline in older adults

Not shown to track cognitive decline: folate did not predict decline in one cohort, and trials of B vitamins containing folic acid found no cognitive effect.

The numbers and sources

Effect: In an Oxford cohort followed for 10 years, markers of vitamin B12 status and homocysteine predicted cognitive decline and serum folate did not [13]. In a meta-analysis of 11 trials, B vitamins lowered homocysteine by 26% to 28% and had no significant effect on global cognitive function (z score difference -0.01, 95% CI -0.03 to 0.02) [14].
Population: 1,648 older adults in Oxford, United Kingdom [13]. 22,000 participants in 11 trials [14].
Sources: [13], [14]
Caveats: The cohort finding is observational. The trials tested B vitamins given to lower homocysteine and were not trials of treating folate deficiency [14].