Skip to content
The Longevity Archive
browse 192

The biology of fasting

BDNF and the brain

Also called: Brain derived neurotrophic factor

Written by Reviewed Sep 2026

The brain's fuel switch during a fast is directly measured and solid in humans; the claim that fasting raises BDNF was tested directly in people and did not happen.

What it is

BDNF is a secreted protein of the neurotrophin family. It binds the receptor TrkB, and to a lesser extent p75NTR, and it supports neuronal survival, dendritic and axonal growth, synapse formation, and the long term potentiation that underlies learning and memory. It is activity dependent: neurons firing increases its transcription and its release. It is one of the most studied molecules in the neurobiology of exercise, environmental enrichment and antidepressant action.

The proposed link to fasting is chemical. Beta-hydroxybutyrate inhibits class I histone deacetylases, which would open chromatin at the BDNF promoter, so a metabolic state would translate into synaptic plasticity. It is a clean hypothesis and there is a large rodent literature behind it.

This page separates two questions that get merged. Does the human brain change fuel during a fast? Yes, measured three different ways. Does fasting raise BDNF in a human? That was tested directly, and it did not.

The machinery

The fuel part first, because it is the part that holds. The brain is roughly 2% of body mass and consumes on the order of 20% of resting energy. It stores essentially no glycogen and cannot oxidise long chain fatty acids at any useful rate, because they do not cross the blood brain barrier efficiently and because neuronal and astrocytic beta-oxidation capacity is low. Ketone bodies solve that: small, water soluble, carried across on monocarboxylate transporters, oxidised by neurons. After 3.5 days of starvation, cortical grey matter glucose metabolism falls 26%, cerebral blood flow does not change, net brain uptake of beta-hydroxybutyrate rises 13-fold, and ketones supply roughly a quarter of the brain's energy. By 5 to 6 weeks they are the predominant fuel.

The BDNF part is where the mechanism stops being demonstrated in people. The rodent work measured BDNF in brain tissue under dietary restriction. The human work measures BDNF in peripheral blood, which is not the same quantity. Most human BDNF is stored in platelets and released when blood clots, which is why serum BDNF greatly exceeds plasma BDNF and why platelet count is a confounder. And the relationship between circulating BDNF and brain BDNF in humans is not well established. Those are real measurement limits, and they are why the human null below is a null about circulating BDNF specifically rather than about the brain.

What switches it on

In the rodent literature, dietary restriction. In humans, the intervention that reliably moves circulating BDNF is exercise, and specifically high intensity exercise.

On the timing
There is one direct human test of the timing question and it is a null. A 20 hour fast lowered glucose and raised ketones, so the metabolic intervention demonstrably worked, and it had no effect on circulating BDNF. No human study has measured BDNF in brain tissue during a fast, because there is no way to do that in a living person.

What has been measured

In people

The ketone fuelling of the human brain is among the best measured things in the whole fasting literature. The BDNF claim has been tested twice in humans, once by fasting and once by giving exogenous ketones at a dose that demonstrably changed cerebral blood flow, and both were null. The cognitive evidence is thin and mixed. All of that belongs on the page together.

  • The direct test, 12 volunteers, examining the isolated and interactive effects of a 20 hour fast, 90 minutes of light exercise, and high intensity exercise on peripheral venous BDNF. Fasting for 20 hours decreased glucose and increased ketones (P less than or equal to 0.0157) but had no effect on BDNF (P greater than or equal to 0.4637). Light cycling raised serum BDNF by 6 plus or minus 8%, independent of fed or fasted state, and that rise was mediated by a 7 plus or minus 6% increase in platelets. Six 40 second intervals at 100% of peak oxygen uptake raised plasma and serum BDNF and the BDNF-per-platelet ratio 4 to 5 fold more than light exercise. Plasma BDNF correlated with circulating lactate during the intervals. Changes in cerebral shear stress did not correspond with BDNF changes (Journal of Physiology, 2023).
  • The second null, and it removes the obvious objection that the fast was too short to raise ketones enough. Placebo controlled double blind crossover, 14 adults with obesity, mean age 56, given 30 mL of a beta-hydroxybutyrate ketone monoester or placebo three times daily for 14 days. Common carotid flow rose 12%, vertebral artery flow 11%, vertebral cerebrovascular conductance 12% and vertebral shear rate 10%. Digit symbol substitution test performance improved by 2.7 correct responses and that improvement correlated with the cerebrovascular changes. Contrary to the study's own hypothesis, beta-hydroxybutyrate did not change fasting serum or plasma BDNF (Journal of Physiology, 2021).
  • Brain fuel accounting at 3.5 days, 9 healthy volunteers. Cortical grey matter glucose metabolism fell 26%, white matter 27%, cerebral blood flow was unchanged, net brain BHB uptake rose 13-fold, and ketones supplied approximately one quarter of cerebral energy requirements (Journal of Cerebral Blood Flow and Metabolism, 1994).
  • Brain ketone concentration measured directly in the occipital lobe by 4-tesla magnetic resonance spectroscopy: 0.05 plus or minus 0.05 mmol/L non-fasted, 0.60 plus or minus 0.26 after two days, 0.98 plus or minus 0.16 after three, with brain lactate also rising (Journal of Cerebral Blood Flow and Metabolism, 2000).
  • The founding measurement, cerebral catheterisation in three obese patients through 5 to 6 weeks of starvation, showing ketones had replaced glucose as the predominant brain fuel (Journal of Clinical Investigation, 1967).
  • An 8 week randomised trial with real brain endpoints. 40 cognitively intact older adults with insulin resistance randomised to 5:2 intermittent fasting or a healthy living diet. Intermittent fasting produced greater weight loss. The two diets had comparable effects on neuronal insulin signalling biomarkers in neuron-derived extracellular vesicles, on the MRI brain-age-gap estimate, on brain glucose by magnetic resonance spectroscopy, and on blood carbohydrate and lipid markers. Cerebrospinal fluid Alzheimer's biomarkers changed minimally. Both diets improved executive function and memory, with intermittent fasting better on certain specific measures. The comparator was another healthy diet, not a no-intervention control (Cell Metabolism, 2024).
  • A well designed null on harm, which is a different claim from benefit. Randomised, waitlist controlled, 122 participants adapting to 16 hour breakfast-skipping intermittent fasting over 10 days, with repeated online cognitive tasks and twice daily mood and concentration measures analysed by Bayesian multilevel modelling. Fasting participants did not have lower cognitive performance or mood than controls. They did feel less concentrated while fasting before noon compared with after breaking the fast in the afternoon (Journal of Health Psychology, 2026).

In other species and in cell culture

Essentially every public claim that fasting raises BDNF traces to two rat studies from 2000 and 2001, which measured BDNF in rat brain tissue under dietary restriction. They are real findings about rat brain. The corresponding human measurement was attempted and did not reproduce the result in circulating BDNF.

  • Rat brain: dietary restriction stimulates BDNF production in the brain and thereby protects neurons against excitotoxic injury (Journal of Molecular Neuroscience, 2001).
  • Rat dentate gyrus: dietary restriction increases the number of newly generated neural cells and induces BDNF expression in the dentate gyrus (Journal of Molecular Neuroscience, 2000).
  • The two possibilities behind the species mismatch are both live and neither has been resolved. It may be a genuine species difference. It may be that peripheral blood BDNF does not report on brain BDNF, which is the measurement limit described above. It may be both.
  • Mouse brain autophagy during fasting is worth noting alongside this, because the two most relevant mouse studies disagree with each other. Whole-animal GFP-LC3 imaging found essentially no autophagy induction in brain after 24 to 48 hours of food deprivation, while a later study using a different detection approach found profound neuronal autophagy at the same durations. Any claim about fasting, the brain and cellular housekeeping should carry that disagreement.
  • The influential syntheses connecting the rodent mechanisms to proposed human benefit are reviews, and much of the mechanistic content they synthesise is rodent work (Nature Reviews Neuroscience, 2018).

Why it matters

This is the page where the split between human and other-species measurement does the most work, because the popular version blends them into one confident paragraph and the underlying facts point in two directions.

What is solid, human, and directly measured: the brain switches fuel during a fast. Three methods, arteriovenous catheterisation, positron emission tomography and magnetic resonance spectroscopy, agree. Ketones supply about a quarter of cerebral energy by 3.5 days and become the predominant fuel over weeks.

What is not: that fasting raises BDNF in a person. It was tested directly at 20 hours, with the metabolic intervention verified by glucose falling and ketones rising, and nothing happened to BDNF. Exogenous ketones at a dose that measurably changed cerebral blood flow and one cognitive test also did nothing to BDNF.

And the comparison that matters if BDNF is what you care about: in the same subjects, in the same study, high intensity exercise moved BDNF 4 to 5 fold more than light exercise. Intensity of exercise is doing that work, not the fast.

A claim you will see repeated

Fasting increases BDNF is one of the most repeated statements in this whole area. Its source is rat brain tissue under dietary restriction, in work from 2000 and 2001. That work is real and it is about rats. The human result is a null, and it is a strong null rather than a failed experiment: glucose fell and ketones rose, so the fast worked, and BDNF did not move. A second human study delivered ketones directly at a haemodynamically active dose and BDNF still did not move. Saying so is not a claim that fasting does nothing to the human brain. Brain BDNF has never been measured in a living, fasting person, and it cannot currently be. What can be said is that the circulating measurement that people cite as evidence has been made, and it was negative.

Citations

  1. Human2023
    Fasting for 20 h does not affect exercise-induced increases in circulating BDNF in humans

    Journal of Physiology

    12 volunteers, examining the isolated and interactive effects of a 20 hour fast, 90 minutes of light exercise, and high intensity exercise on peripheral venous BDNF. Fasting decreased glucose and increased ketones (P less than or equal to 0.0157) but had no effect on BDNF (P greater than or equal to 0.4637). Light cycling raised serum BDNF 6 plus or minus 8%, mediated by a 7 plus or minus 6% rise in platelets. Six 40 second intervals at 100% of peak oxygen uptake raised plasma and serum BDNF and the BDNF-per-platelet ratio 4 to 5 fold more than light exercise. Cerebral shear stress changes did not correspond with BDNF changes.

  2. Human2021
    Short-term ketone monoester supplementation improves cerebral blood flow and cognition in obesity: A randomized cross-over trial

    Journal of Physiology

    Placebo controlled double blind crossover, 14 adults with obesity, mean age 56, 30 mL of a beta-hydroxybutyrate ketone monoester or placebo three times daily for 14 days. Common carotid flow +12%, vertebral artery flow +11%, vertebral cerebrovascular conductance +12%, vertebral shear rate +10%. Digit symbol substitution improved by 2.7 correct responses, correlating with the cerebrovascular improvements. Contrary to the study's own hypothesis, beta-hydroxybutyrate did not change fasting serum or plasma BDNF.

  3. Human1994
    Brain metabolism during short-term starvation in humans

    Journal of Cerebral Blood Flow and Metabolism

    9 healthy volunteers before and after 3.5 days of starvation. Cortical grey matter glucose metabolism fell 26%, white matter 27%, uniformly across regions, confirmed at 24% by an independent Fick determination. Cerebral blood flow unchanged. Net brain uptake of beta-hydroxybutyrate rose 13-fold. Ketones account for approximately one quarter of cerebral energy requirements at 3 days.

  4. Human2000
    Human brain beta-hydroxybutyrate and lactate increase in fasting-induced ketosis

    Journal of Cerebral Blood Flow and Metabolism

    Occipital lobe beta-hydroxybutyrate by 4-tesla magnetic resonance spectroscopy in healthy adults: 0.05 plus or minus 0.05 mmol/L non-fasted, 0.60 plus or minus 0.26 after two days, 0.98 plus or minus 0.16 after three. Brain lactate rose from 0.69 plus or minus 0.17 to 1.47 plus or minus 0.22 mmol/L. Plasma and brain BHB correlated at r = 0.86.

  5. Human1967
    Brain metabolism during fasting

    Journal of Clinical Investigation

    Cerebral vessel catheterisation in three obese patients through 5 to 6 weeks of starvation, demonstrating that beta-hydroxybutyrate and acetoacetate had replaced glucose as the predominant fuel for brain metabolism.

  6. Human2024
    Brain responses to intermittent fasting and the healthy living diet in older adults

    Cell Metabolism

    40 cognitively intact older adults with insulin resistance randomised to 5:2 intermittent fasting or a healthy living diet for 8 weeks. Intermittent fasting produced greater weight loss. The two diets had comparable effects on neuronal insulin signalling biomarkers in neuron-derived extracellular vesicles, the MRI brain-age-gap estimate, brain glucose by spectroscopy, and blood carbohydrate and lipid markers. Cerebrospinal fluid Alzheimer's biomarkers changed minimally. Both diets improved executive function and memory, with intermittent fasting benefiting more on certain measures.

  7. Human2026
    Stable cognitive performance while adapting to intermittent fasting: A randomised controlled trial

    Journal of Health Psychology

    Randomised, waitlist controlled, N = 122, adapting to 16 hour breakfast-skipping intermittent fasting over 10 days with repeated online cognitive tasks and twice daily mood and concentration measures, analysed by Bayesian multilevel modelling. Fasting participants did not have lower cognitive performance or mood than controls, though they felt less concentrated while fasting before noon than after breaking the fast in the afternoon.

  8. Animal2001
    Dietary restriction stimulates BDNF production in the brain and thereby protects neurons against excitotoxic injury

    Journal of Molecular Neuroscience

    Rat brain. Dietary restriction stimulates BDNF production in the brain and thereby protects neurons against excitotoxic injury. This paper and the one below are the origin of essentially every public claim that fasting increases BDNF.

  9. Animal2000
    Dietary restriction increases the number of newly generated neural cells, and induces BDNF expression, in the dentate gyrus of rats

    Journal of Molecular Neuroscience

    Rat dentate gyrus. Dietary restriction increases the number of newly generated neural cells and induces BDNF expression in that region.

  10. Review2018
    Intermittent metabolic switching, neuroplasticity and brain health

    Nature Reviews Neuroscience

    The influential synthesis connecting fasting, the metabolic switch and neuroplasticity. It is a review, and much of the mechanistic content it synthesises is rodent work.