Lifestyle Choices May Shape Alzheimer’s Risk Long Before Memory Fails

David Perlmutter’s central claim is that the risk of cognitive decline is shaped not only by age and inherited susceptibility, but also by the conditions in which the brain operates over decades. He describes Alzheimer’s as primarily a metabolic condition of the brain and argues that habits affecting blood sugar, inflammation, sleep, movement and social connection may influence that risk. His emphasis is on acting before symptoms appear. He does not establish when the disease process begins in an individual; rather, he argues that waiting until forgetfulness prompts a medical evaluation is too late to begin asking about the factors that may affect brain health.

Perlmutter frames this as a question of agency. He says people are often encouraged to live as they wish and trust that neuroscience or pharmaceuticals will eventually provide a rescue. “We aren’t there yet,” he says. He would welcome an effective treatment, and says he would have used one for his father, who developed Alzheimer’s. But in his view, there is no meaningfully effective treatment to rely on today, making prevention more consequential. He is not arguing that medicine has no role; he is arguing against making future treatment the plan.

He gives a stark estimate of risk: among people who live to 85, the chance of developing dementia, generally Alzheimer’s, is in the high-40-percent range. He also says Alzheimer’s is the third leading cause of death in the United States and that women have roughly twice men’s risk. His shorthand is that “genetics loads the gun, lifestyle pulls the trigger.”

He distinguishes dementia from Alzheimer’s. Dementia, as he explains it, is a broad term for impaired cognitive function across multiple domains. Alzheimer’s is one cause; others include vascular dementia, dementia associated with Lewy bodies and frontotemporal dementia. Cognitive problems can also arise from reversible causes, such as medication effects or other medical problems. The categories differ, but Perlmutter argues that disruption of the brain’s immune system may be a mechanism across many neurodegenerative conditions.

His argument is not that lifestyle replaces clinical care. It is that waiting for a diagnosis before considering brain health is like waiting for a heart attack before looking at blood pressure, lipids, sleep and activity. He wants those questions asked in people’s 30s and 40s, not only after memory problems bring them to a doctor. He compares the timing to fixing a roof while the sun is shining: prevention is most useful before an obvious crisis forces attention.

Perlmutter’s explanation centers on microglia: immune cells that live in the brain. He describes them as both defenders and maintenance workers. In a supportive state, which he calls M2, they help maintain connections between neurons, protect the blood–brain barrier and support healthy brain function. In another state, which he calls M1, they can become destructive, damaging synapses and contributing to inflammation.

The “evil twin” is Perlmutter’s metaphor for that harmful state. M1 and M2 are not different cells in his account, but different configurations of the same kind of cell. A short-term destructive response can be useful after an injury or infection; the problem, he says, is when the cells fail to return to a supportive state. High blood sugar, inflammation, stress, sedentary behavior, poor sleep and social isolation are among the conditions he says may push microglia away from their supportive role.

He describes two interacting influences on microglia: signals from the rest of the body, including inflammatory chemicals, and the cells’ own energy state. Mitochondria—the structures involved in producing cellular energy—are central to his account. In his model, metabolic changes affect mitochondrial function, which can influence whether microglia remain supportive or become destructive. Once in a more destructive state, he says, microglia can further reduce mitochondrial function and release inflammatory signals that encourage more cells to shift. He calls this a feed-forward process, while emphasizing that the transition is not a simple on-off switch.

The downstream consequence he emphasizes is damage to synapses, the connections between neurons. Perlmutter says Alzheimer’s can be understood as a “synaptopathy”—a disease involving the loss of those connections. His account connects that cellular damage with the eventual effects on memory, judgment and a person’s ability to engage with the world.

He links this account to the developing brain, which forms and removes connections. Perlmutter says that at around one year of age, synapses are being formed at roughly one million per second, while some are also being removed through a process called synaptic pruning. Microglia participate in that normal refinement. His proposed connection is that Alzheimer’s involves a harmful reactivation of a process that is appropriate during development: the brain pruning connections when it should not.

Age remains an important risk factor in his explanation. Perlmutter says microglia tend to become more inflammatory with age, while suggesting that biological age may be more relevant than the number of birthdays alone. Chronological age cannot be changed, he says, but other inputs to the process can be influenced.

A PET image shown during the discussion compares TSPO imaging in a healthy control with an Alzheimer’s brain. Perlmutter says the scan can show activation of brain immune cells and that such activation appears across a range of neurodegenerative conditions. He also cautions against interpreting an image of brain atrophy on its own: enlarged fluid-filled spaces, for example, can have more than one explanation.

Perlmutter challenges what he describes as the traditional focus on beta-amyloid: a protein fragment that can accumulate in plaques between neurons. In his account, the idea that removing amyloid will solve Alzheimer’s is too narrow. He notes that people can have substantial plaque accumulation without dementia symptoms, and argues that amyloid is a participant in the disease process rather than its pivotal cause. He does not say amyloid is irrelevant; he says that targeting it alone has not resolved the disease.

He points to a shift in research attention. Perlmutter says that about a decade earlier roughly a third of Alzheimer’s research was focused on beta-amyloid, compared with about 14 percent at the time of the interview, with more attention going to metabolism and the brain’s immune system. He also cites a UK trial testing metformin, a diabetes drug, and atomoxetine, an ADHD drug, in people with symptomatic Alzheimer’s. He interprets the inclusion of a metabolic drug as evidence that metabolism is receiving greater attention. The trial is testing these treatments; its results are not yet part of the discussion.

Perlmutter says he does not favor amyloid-targeting drugs, describing them as mostly ineffective and significantly dangerous. He cites a Cochrane analysis of 17 studies involving 20,342 people, each followed for 18 months, and says it found no meaningful effect in slowing progression or producing improvement. That is his characterization of the analysis. The discussion does not identify the individual drugs, provide the analysis’s outcome measures or give adverse-event figures, so it does not offer a fuller account of the review’s findings.

He also discusses a different experimental direction: 40-hertz light and sound stimulation. Perlmutter describes research led by Li-Huei Tsai at MIT, in which 40-hertz stimulation restored gamma rhythms and shifted microglia toward a supportive state in animal models. He says early human work has also been conducted. He keeps a light device on his desk but acknowledges that he cannot prove it is helping him.

A chart shown during the discussion depicts an animal study in which mice exposed to one hour of 40-hertz flickering light per day for seven days had lower levels of amyloid plaques and abnormal tau than mice kept in the dark. Bartlett also cites animal findings about plaque and tangle reductions. These are animal results, not evidence in the discussion that the same effects occur in people. An on-screen community note adds that no trial has directly compared 40-hertz stimulation with anti-amyloid drugs, and that published human studies are small and early-stage. Perlmutter’s enthusiasm for the approach is clear, but he describes evidence that is still developing; his claim that the results are better than amyloid drugs is not supported by a direct comparison in the material shown.

When Steven Bartlett asks Perlmutter to rank lifestyle contributors, Perlmutter puts lack of agency first: if people do not believe their choices matter, he says, the rest of the list will not matter either. He places sedentary behavior second, then names lack of social connection, non-restorative sleep and poor diet. After the first two, he says the remaining factors are equally ranked.

Exercise is the intervention he returns to most often. He calls movement a “multi-tool” because it affects several pathways at once: insulin function, inflammation, mitochondrial activity and blood flow to the brain. He also describes muscle as an endocrine organ. When muscle is active, it releases signalling molecules called myokines. Perlmutter says these can influence the brain and help stimulate BDNF, a growth factor involved in supporting neurons and their connections. An on-screen note says some myokines can cross the blood–brain barrier in animal studies.

Perlmutter argues for both aerobic exercise and resistance training. He refers to a meta-analysis concluding that both matter, and says building muscle may increase the supply of muscle-derived signals that benefit the brain. He cites a study using Framingham Heart Study data: among middle-aged people, the most physically active had about a 45 percent lower risk of Alzheimer’s than the least active. For older participants who carried APOE4, he says, the most active group had about a 67 percent lower risk than the least active. These are comparisons between activity groups in the study as he describes it, not a guarantee that exercise will produce a particular reduction for an individual.

To explain why activity might matter, Perlmutter returns to muscle signalling, blood sugar and inflammation. He says that active muscle can improve insulin function and that an A1C blood test, which reflects average blood sugar, can offer a signal about risk when elevated. He also argues that exercise can reduce inflammation and improve blood flow. His practical emphasis is broad: he says all forms of exercise are useful, and that people can combine aerobic activity with resistance training rather than treating them as competing choices.

He also presents movement as an alternative to relying on a drug to produce some of the same metabolic effects. Bartlett asks whether a pair of running shoes could be better than medicine; Perlmutter says exercise does more than metformin, which is being studied in an Alzheimer’s trial. That comparison is his argument about the breadth of exercise’s effects, not evidence in the discussion that exercise and metformin have been tested head to head as Alzheimer’s treatments.

Sleep is another lever, in part because of the brain’s glymphatic system, which Perlmutter describes as a waste-clearance network. He agrees with Bartlett’s “street cleaners” metaphor for the system’s activity during sleep, but emphasizes that the cleaners are active during deep sleep. Restorative sleep also supports memory consolidation and, in Perlmutter’s account, helps reduce inflammation and improve insulin function. He says even one poor night can impair insulin response, blood-sugar control and decision-making.

Bartlett asks how sleep affects decisions. Perlmutter contrasts the prefrontal cortex, which he calls the “adult in the room,” with the amygdala, which he describes as making fast, impulsive decisions. The prefrontal cortex helps people weigh consequences and plan; inflammation, he says, can weaken its ability to regulate the amygdala. He connects this to diet and the broader spread of ultra-processed foods, arguing that chronic inflammation can impair the brain systems involved in deliberation.

Social connection has a place in Perlmutter’s model for both biological and practical reasons. He refers to Blue Zones, areas associated in the discussion with long life and lower dementia rates, and notes that their residents tend to be integrated into community. He proposes oxytocin as one mechanism: trust and connection can increase oxytocin, which he says helps keep microglia in a supportive state.

An on-screen community note describes microglia as resident immune cells that make up about 5 to 10 percent of brain cells and help defend and clear waste from the central nervous system. Perlmutter’s broader point is that social connection is not merely a matter of mood or lifestyle preference in his framework: he sees it as one of several inputs that may affect the brain’s immune environment. He returns to connection at the end as both an individual practice and a way of relating to the wider world.

Perlmutter’s food advice is less a single diet prescription than a focus on metabolic signals. He identifies blood sugar, blood pressure, body fat and blood lipids as important measures of metabolism. In a demonstration, he pours teaspoons of sugar to show the amount he says is present in a 12-ounce glass of orange juice: eight teaspoons. Bartlett recalls having thought orange juice counted as a healthy serving of fruit. Perlmutter’s point is that vitamins and fiber do not cancel out the sugar load.

Bartlett describes using a continuous glucose monitor to observe how foods affect his own blood sugar. Perlmutter says he was surprised by the response to cashews, and also saw a temporary rise during exercise. He presents the monitor as a way to learn about individual responses, not as something everyone must wear continuously. An on-screen note explains that intense workouts can temporarily raise blood sugar because stress hormones signal the liver to release glucose, even though working muscles also draw glucose from the blood. Perlmutter’s example is a case for measuring how a change affects an individual, not for treating every temporary rise as evidence that exercise is harmful.

The discussion includes a disagreement about gluten sensitivity. Bartlett cites estimates suggesting that celiac disease affects around 1 percent of people and that only a fraction of people who identify as gluten-sensitive react specifically to gluten in blinded food challenges. Perlmutter replies that non-celiac gluten sensitivity is more common than celiac disease and recounts a patient whose headaches disappeared on a gluten-free diet, returned when she resumed gluten, and improved again after avoiding it. An on-screen note describes fructans, found in foods such as wheat, onions and garlic, as a possible source of symptoms that can resemble gluten-related problems. Bartlett raises that possibility, along with other explanations for why people feel better when changing their diet; Perlmutter maintains that some people do have non-celiac gluten sensitivity. The exchange leaves the two positions in view: Perlmutter’s patient example supports his clinical claim, while Bartlett emphasizes that self-reported improvement does not establish that gluten itself was the cause.

Perlmutter adds a concern about glyphosate, the herbicide he says is used on wheat in the United States to dry crops before harvest. He argues that glyphosate can affect gut bacteria and mitochondria. His broader point is that the gut microbiome may affect the brain through substances produced by microbes, changes in gene expression and inflammatory signals that can travel from the gut to the brain. These are claims he advances as part of his metabolic account; the discussion does not provide a study or quantify the effects of glyphosate exposure on dementia risk.

He recommends polyphenol-rich foods, including berries, onions, dark chocolate, coffee and extra-virgin olive oil. Perlmutter describes polyphenols as compounds that can activate the NRF2 pathway, which he says supports antioxidant production and reduces inflammation. He also says these compounds interact with gut bacteria. An on-screen chart ranks dark chocolate, flaxseed, black olives, blueberries, hazelnuts, plums, cherries, strawberries, coffee and red onions by polyphenol content per 100 grams. Perlmutter’s practical emphasis is on including colorful, relatively unprocessed plant foods.

His preference is for dark chocolate above 80 percent cacao, partly because higher cacao content generally means less sugar. He argues that modern diets are mismatched with what human biology evolved to expect: more ultra-processed food and added sweeteners, and less of the fiber and plant compounds he associates with traditional diets. He cites a figure of 60 percent of adult calories from ultra-processed food in the United States. An on-screen community note gives a different estimate from government data for 2021–23: 53 percent for adults and 62 percent for children and teenagers. The figures are presented differently in the discussion and note; the article does not treat them as interchangeable.

Perlmutter links type 2 diabetes to dementia risk, revising Bartlett’s suggested estimate of a 300 to 400 percent increase. He says more recent data support at least a 75 to 85 percent increase in Alzheimer’s risk, and adds that Parkinson’s risk is also higher among people with type 2 diabetes. A community note on screen gives ranges for diabetes-associated risks across Alzheimer’s, all-cause dementia and Parkinson’s, and says the association is strongest for vascular dementia. Perlmutter’s argument is that metabolic health is relevant across more than one neurodegenerative condition.

He also describes visceral fat as metabolically active rather than inert storage. It releases inflammatory chemicals called cytokines, which he says can contribute to chronic, low-level inflammation. His metaphor is that Alzheimer’s may be less like a cytokine “storm” than a “drizzle” over a long period. He stresses that inflammation itself is not always harmful: the body needs inflammatory responses, while excessive or prolonged inflammation is what he considers problematic.

On alcohol, Perlmutter says earlier studies suggested a U-shaped relationship with dementia risk, with lower risk at modest consumption than at none, but says reassessment suggests any alcohol may be harmful to the brain. He describes alcohol as directly toxic to neurons and as a potential influence on microglia. When Bartlett cites a laboratory finding that binge-level alcohol reduced microglia’s ability to clear beta-amyloid by 15 percent, Perlmutter agrees with the finding but reiterates that amyloid is not, in his view, the pivotal player. The community note identifies the result as a 2018 laboratory study in rat microglia exposed to a high concentration of alcohol; it is not a measure of the effect of a drinking episode on a person’s brain.

David Perlmutter discusses APOE as a genetic risk factor. People inherit two APOE alleles, he explains, and the combinations can include APOE2, APOE3 and APOE4. He says APOE2 is associated with lower risk, APOE3 is neutral and APOE4 with increased risk. He gives a five-fold increase for one APOE4 copy and a possible twelve-fold increase for two copies.

An on-screen community note qualifies those estimates: compared with people without APOE4, one copy is linked to roughly 2 to 4.5 times the odds of Alzheimer’s and two copies to roughly 8 to 15 times, with substantial variation by ancestry. Perlmutter’s point is not that APOE4 is harmless, but that it is a predisposition rather than a determination. He says lifestyle may offset some inherited risk, and cites the activity findings among APOE4 carriers as an example.

He also mentions that many other genetic factors contribute to Alzheimer’s risk, and that knowing one’s genetic status does not dictate an outcome. The practical implication in his argument is to use risk information to motivate attention to movement, blood sugar, sleep and other modifiable factors—not to treat a genetic result as a diagnosis. He rejects the idea that a person who learns they carry APOE4 is necessarily “on the road to Alzheimer’s.”

Perlmutter is more cautious about medication than the discussion of metformin might suggest. He says metformin shares some metabolic effects with exercise and is being studied in Alzheimer’s, but explicitly says he is not recommending that healthy people start taking it. He prefers exercise, notes that metformin may have downsides—including possible vitamin B12 deficiency—and refers to the proposed TAME trial examining metformin and longevity. Bartlett raises concerns about routine use in healthy people, including possible interference with exercise adaptations; Perlmutter agrees that potential downsides matter. The conversation leaves metformin as a subject for study and medical judgment, not a self-directed prevention recommendation.

He also mentions his own use of methylated B vitamins because of a genetic variant affecting his handling of B vitamins and an elevated homocysteine level. He takes DHA from fish oil, reporting about 1,000 milligrams per day. These are personal details, not a general supplement protocol.

Perlmutter says lifestyle changes may still matter after cognitive problems have begun. He cites a 20-week study involving 51 people with Alzheimer’s or mild cognitive impairment, led by Rudolph Tanzi and Dean Ornish. Participants changed their diet, increased exercise, reduced stress and meditated for an hour a day. Perlmutter says 70 percent stabilized or improved. He describes the diet as plant-forward and Mediterranean-style, with polyphenol-rich foods, healthy fats and social interaction.

He uses the result to support the proposition that lifestyle deserves attention even after diagnosis. The study was relatively small and the intervention lasted 20 weeks; Perlmutter does not describe the breakdown between participants who stabilized and those who improved. The discussion presents his summary of the study, not its full methods or a basis for comparing the intervention with other treatments.

He also discusses ketogenic diets as a way to target brain energy metabolism. Bartlett describes a ketogenic diet as very low in carbohydrate, perhaps below 50 grams a day, with more fat; Perlmutter emphasizes good fats such as olive oil, nuts and seeds. He says a ketogenic approach may help the brain use ketones when glucose metabolism is impaired, and refers to work in both Alzheimer’s and Parkinson’s. He acknowledges that the diet can be difficult to sustain and says he is not recommending it continuously for everyone. His stated goal is metabolic flexibility: being able to enter ketosis more readily when fasting or eating very low carbohydrate.

Bartlett cites figures for cognitive and functional improvements from ketogenic diets or ketone supplementation, including estimates of the share of brain energy supplied and changes in test scores. Perlmutter responds that targeting mitochondria is relevant to the brain’s energy needs. The figures are introduced by Bartlett; the discussion does not unpack the studies behind them. Perlmutter says the brain requires glucose, and that some parts of it depend on insulin to use energy. He names the hippocampus, hypothalamus, prefrontal cortex and olfactory bulb as insulin-sensitive areas, connecting them to memory, metabolism, decision-making and smell.

Perlmutter’s position on treatment is shaped by his view that available options are inadequate, but he does not rule out future treatments. He acknowledges that promising approaches may emerge, including a study from Havana that he says found improvement among Alzheimer’s patients receiving intranasal erythropoietin, while the placebo group declined. He describes the result as striking, without detailing the study’s size or methods. His point is that he welcomes treatment research even as he urges people not to wait for a future drug before addressing modifiable conditions.

For listeners already facing a diagnosis, his immediate recommendations are to pay attention to blood sugar and weight, exercise, and maintain social connection. He presents these as actions worth considering alongside care, not as a promise that the disease will reverse.

Perlmutter’s account of his father gives the prevention message its personal force. His father was a brilliant, compassionate brain surgeon who developed Alzheimer’s. Perlmutter says he recognized the dementia late, after his father had begun losing things, forgetting where he was and blaming others for misplaced objects. He recalls that his father initially denied the changes and accused people of moving things when he could not find them.

Perlmutter says his father’s illness lasted about five years from the point when his behavior changed to his death. Early on, his father knew something was happening, which Perlmutter describes as distressing. Later, he became unaware. As the disease progressed, Perlmutter says, its effects became less of an issue for his father and more of an issue for the people around him. By the end, his father was unable to engage meaningfully with other people or activities.

The proximity of his father’s room to Perlmutter’s neurology practice brought the family experience into his clinical work. He says that when families told him they could not understand what caring for a parent with dementia was like, he would point out that his own father was living across the parking lot. He could speak to them not only as a neurologist but as a family member living through the same disease.

That history helps explain why Perlmutter objects to passive reliance on future treatments. He frames his insistence on agency as an attempt to offer a strategy rather than reassurance. Age and inherited factors matter, he says, but habits affecting metabolism and the brain’s immune system are also worth addressing.

The closing instruction is simple: reconnect. Perlmutter means relationships with other people, and he links that social connection to the brain connections he says can be damaged by destructive microglial activity. He argues that reconnecting with others and with the planet can benefit brain health and life more broadly. It brings together his biological argument about social connection and his personal account of what dementia takes from families.