Section oneThe interventions converge on a few switches

Energy stress activates AMPK and, through PGC-1α, mitochondrial biogenesis. Mechanical loading activates mTORC1 for muscle protein synthesis. Exercise and fasting both produce metabolites (lactate, β-hydroxybutyrate) that signal to the brain. Cold recruits the noradrenergic system, and melatonin sets the timing of the whole sequence.

Convergence cuts both ways. Inputs to a saturating node such as AMPK and PGC-1α are likely to be sub-additive, so stacking several energy-stress signals will probably not multiply the benefit.

Section twoAerobic training: the strongest hard-outcome evidence

In 122,007 patients who had a treadmill test, mortality fell stepwise with fitness and showed no upper limit of benefit: adjusted HR 0.20 (0.16 to 0.24) for elite against low fitness.^[1] Each additional MET was associated with 13% lower all-cause mortality in a meta-analysis of 33 studies.^[2] The dose-response for intensity is an inverted U: at the highest weekly volume of HIIT, healthy volunteers showed impaired mitochondrial function and glucose tolerance.^[30] That is the physiological logic of the polarized 80/20 pattern, mostly zone 2 with one or two hard sessions.

Section threeResistance training: small dose, large return

A systematic review found any resistance training was associated with 15% lower all-cause mortality (RR 0.85, 0.77 to 0.93).^[3] A dose-response meta-analysis of 16 cohorts found a J-shaped relationship with the largest reduction at about 30 to 60 minutes a week, independent of aerobic activity.^[4] All of this is observational, but the dose is small and the biological rationale strong, so I think the asymmetry favours doing it.

Section fourCreatine: an energy buffer, strongest for muscle

Phosphocreatine regenerates ATP in milliseconds and shuttles energy between mitochondria and the sites that spend it. The brain uses the same system, but uptake across the blood-brain barrier is transporter-limited, which caps how far oral supplementation can raise brain levels.^[9] The clearest cognitive signal is under stress: a single high dose during a night of sleep deprivation protected brain PCr and improved processing speed.^[8] In healthy adults the cognition literature is small effects of uncertain size.^[7]

Section fiveFasting: richer mechanism than outcome data

As feeding stops, insulin falls, glycogen is drawn down and ketones rise, lowering mTOR activity and activating AMPK and autophagy.^[16] In human trials, though, time-restricted eating does no better than ordinary calorie restriction when calories are matched.^[20,21] In one study, 20 hours of fasting did not change circulating BDNF, whereas six 40-second intervals raised it four to five times more than light exercise.^[18]

Section sixCold exposure: a sympathetic jolt and a small adaptation

One hour of immersion at 14 °C raised metabolic rate by 350% and plasma noradrenaline by 530% without raising cortisol.^[10] Those are peripheral plasma values, so claims about brain dopamine go beyond the data. In the only large randomized trial of cold showers, sickness absence fell 29% but illness days did not change.^[11] Brown fat adds only about 1 to 5% of basal metabolic rate, so cold is not a weight-loss tool.^[34] The one clear conflict in the stack is cold after lifting, which blunted muscle gain and anabolic signalling over 12 weeks.^[12]

Section sevenMelatonin: a timing signal with an open safety question

At low doses (about 0.3 to 0.5 mg) melatonin shifts circadian phase according to when it is taken. At higher doses it accumulates in mitochondria and acts as an antioxidant.^[26] Those are different pharmacologies. A 2025 observational abstract reported that adults with insomnia who used melatonin for a year or more had higher five-year rates of heart failure (4.6% against 2.7%), heart-failure hospitalization and all-cause mortality.^[22] The authors stressed it cannot show causation, and confounding by insomnia severity is likely. I treat it as a signal to track, not a verdict, and low-dose, well-timed use is the pattern that matches both the mechanism and the benign short-term record.

Section eightFifteen pairs, two synergies, one conflict

Only two pairings earn the word synergy on current evidence: aerobic with resistance, and resistance with creatine. One is a clear conflict, cold after lifting. Most of the rest are timing problems, which are solvable.

Section nineWhat might they do together?

• H1. Foundation plus modulators. Aerobic fitness and muscle set the ceilings that dominate mortality data; the other four act on the approach to them. Someone who adds all four but under-trains the foundation should do worse than someone who trains well and adds none.
• H2. Pulsatile signalling. Fasting and zone 2 push toward AMPK and autophagy; feeding plus lifting pushes toward mTORC1. Cycling the two within a day may give more of both than a constant fed state gives of either.
• H3. Brain plasticity convergence. Lactate, ketones, noradrenaline and creatine under stress reach the same neural machinery by different routes. The convergence may mostly mean redundancy rather than addition.
• H4. Thermo-metabolic additivity. Cold, time-restricted eating and aerobic training might improve glucose disposal more together, with cold contributing the least.
• H5. A circadian scaffold. Timed exercise, restricted feeding and low-dose melatonin are all zeitgebers. Aligned they may reinforce each other, scattered they may fight. The missing zeitgeber is morning light.
• H6. The hormesis budget. Cold, fasting, intervals and antioxidant doses are stressors with an inverted U. Resting heart rate, HRV and sleep quality should drift before performance does.

Section tenOne possible day

A template that applies the interaction rules: zone 2 fasted in the morning, cold well away from lifting, creatine with the first meal, lifting in the afternoon, and low-dose melatonin well after dinner. It is something to adapt, not a prescription.

Section elevenWhat this cannot settle

Most hard-outcome data are observational, so healthy-user bias applies in both directions. The mechanisms rest heavily on rodent and in vitro work. The combinations are untested: the fair summary of the combined benefit is plausible, probably partly redundant, and more sensitive to timing than to inclusion. The scorecard grades and schematics are my judgment. Individual risks (cardiovascular disease, medications, eating-disorder history, mood disorders, kidney disease) change the calculus, and a clinician who knows the person should have the final say.

The full text, with every citation and the measurement dashboard, is in the PDF.