At eighteen, the human brain processes information faster than it ever will again. At sixty-seven, the same brain has acquired a vocabulary that will not begin to decline for another decade. The peer-reviewed evidence from a 48,537-person study has overturned the popular assumption that the human mind has a single peak.

The standard scientific view, for most of the twentieth century, was that human intelligence peaked in late adolescence or early adulthood and then began a long, gradual decline. The intuition behind the view was that the brain, like the rest of the body, reached its biological peak in the years immediately following maturation and that everything afterwards was a slow process of deterioration. Most popular discussions of cognitive ageing still rest on the same intuition.

The peer-reviewed evidence does not support it.

What the evidence supports instead is a much more interesting picture. Human intelligence is not a single thing with a single peak. It is at least six different cognitive abilities, each operating on its own developmental trajectory, each peaking at a different age, and each declining at a different rate. Some of these abilities peak in the late teens. Others peak in the late sixties. The gap between the earliest and the latest peak is approximately forty-nine years, which is longer than most working careers.

What the study found

In 2015, two postdoctoral researchers, Joshua Hartshorne at the Massachusetts Institute of Technology and Laura Germine at Harvard and the Massachusetts General Hospital, published a paper in Psychological Science that brought together the most comprehensive cognitive-ageing data available at the time. They had been running large-scale online cognitive testing experiments through two web platforms, gameswithwords.org and testmybrain.org, which allowed people of any age to participate in short cognitive tasks from any computer. Over several years, they accumulated test data from a much broader cross-section of the adult population than traditional laboratory studies had been able to reach.

The barrier to studying cognitive ageing in the past had been a simple practical one. Laboratory cognitive studies are typically conducted on college students (because universities make them available) and retirees (because they have the time). The thirty-five years in the middle, between age twenty and age fifty-five, had been chronically under-sampled because adults in that age range have jobs and families and do not have the time to come to a psychology laboratory. Hartshorne and Germine’s web-based approach solved the sampling problem by making the tests take only a few minutes and accessible from any device.

The study analysed cognitive test results from 48,537 web participants. To verify that the web-based findings were not an artefact of the online platform, the researchers also analysed the standardised normative data from the Wechsler Adult Intelligence Scale and the Wechsler Memory Scale, which are the most widely used clinical IQ measurement instruments in the world. Both data sources produced the same pattern.

The pattern was that different cognitive abilities peak at radically different ages.

On a lightly related note, we came across this video that explains how our quirks, across all ages, might actually signal high intelligence. If you’ve ever wondered whether you’re “weird”, this video is for you. 

The six peaks

Processing speed, which measures how quickly a person can take in new information and respond to it, peaks at approximately age eighteen or nineteen. After that, it declines steadily. A forty-year-old does not process information as quickly as an eighteen-year-old, and this is one of the most consistent findings in cognitive ageing research.

Short-term memory, measured by tests of how many items a person can hold in working memory at once, peaks at approximately age twenty-five. The peak is followed by a gentle plateau through the late twenties and a slow decline that becomes noticeable in the late thirties.

Face recognition, which had been studied separately by Germine and colleagues in 2011 using more than 44,000 participants and published in the journal Cognition, peaks at approximately age thirty-two. This finding, when it was first published, was genuinely surprising. Face recognition had been assumed to be a perceptual skill that matured early, alongside basic visual processing in childhood. The data showed that the ability to learn new faces and retain them continues to develop for more than a decade after most other perceptual skills have peaked.

The ability to concentrate, focus on novel information, and ignore distractions peaks at approximately age forty-three. This is the cognitive ability most closely associated with executive function, the capacity to plan, to inhibit impulses, and to keep multiple competing demands in mind without losing track of any of them. It is the cognitive ability that working adults rely on most heavily, and it does not reach its lifetime maximum until well into middle age.

Emotional understanding, measured by tests of how accurately a person can identify the emotional states of other people from facial expressions, body language, and tone of voice, peaks at approximately age forty-eight. This finding has perhaps the strongest implication for everyday life, because it suggests that the cognitive capacity for empathy, social judgment, and reading other people is one of the latest-maturing abilities in the human mind. It also suggests that the conventional wisdom that older adults are better at understanding other people may have a measurable biological basis rather than being a function of accumulated life experience alone.

Vocabulary, and more generally what cognitive psychologists call crystallised intelligence, continues to rise until approximately age sixty-seven. Crystallised intelligence is the accumulated store of factual knowledge, learned skills, and verbal ability that a person has built up over their lifetime. By the strongest peer-reviewed evidence, it does not peak until almost five decades after processing speed has begun to decline, and even after the peak, vocabulary tends to remain stable into the seventies before any noticeable decline begins.

What this overturns

The traditional model of cognitive ageing, sometimes called the single-peak model, held that there was one cognitive trajectory for the human mind, rising through childhood, peaking somewhere between the late teens and the late twenties, and then declining at a roughly steady rate through the rest of the lifespan. The model was useful for understanding broad averages, but it produced a misleading picture of what was actually happening at the level of specific cognitive abilities.

The model that the Hartshorne and Germine evidence supports is sometimes called the multiple-trajectories model, or the differentiated cognitive ageing model. In this picture, there is no single human intelligence that peaks at a single age. There are multiple cognitive systems, each with its own developmental schedule, each governed by different neural mechanisms, and each responding to different aspects of life experience. Some of these systems are calibrated by biology, mature early, and decline in the predictable way the single-peak model describes. Other systems are calibrated by experience, mature late, and continue to develop for much longer than the single-peak model allowed.

The differentiation matters because it changes the question. The question is no longer “When does the human mind peak?” but rather “Which part of the human mind peaks when?” and the answer turns out to be different for every cognitive ability that has been studied.

Why this happens

The biological explanation, on the current evidence, is that different cognitive abilities depend on different brain systems, and the different brain systems mature and age on different schedules. Processing speed depends heavily on the integrity of white-matter tracts, the insulated cables that connect different brain regions, and white-matter integrity peaks in the late teens before beginning a slow decline that continues for decades. Working memory depends on prefrontal cortex function, which matures somewhat later. Face recognition appears to depend on specialised regions of the temporal lobe that continue to refine their function into the early thirties. Emotional understanding involves a distributed network including the prefrontal cortex, the temporal lobes, and the limbic system, and the integration of this network appears to continue maturing into middle age. Vocabulary depends on the accumulated network of semantic associations stored in the temporal cortex, which can continue to grow throughout life as long as new words and concepts continue to be encountered.

Cognitive psychologists distinguish between fluid intelligence, which is the ability to think on one’s feet and solve novel problems, and crystallised intelligence, which is the accumulated store of learned knowledge. Fluid intelligence is more dependent on biological hardware that peaks early. Crystallised intelligence is more dependent on accumulated experience that builds up over decades. The traditional model collapsed the two and described an average that fit neither.

The honest limitations

Several methodological caveats apply to the Hartshorne and Germine study and to the broader literature on which it is built.

The web-based testing platform recruits participants who choose to take cognitive tests online. These participants are not a random sample of the general population. They tend to be more educated, more motivated by intellectual curiosity, and more comfortable with computers than the population average. The researchers attempted to control for this by also analysing the standardised Wechsler normative data, which is drawn from more representative samples. The two data sources produced consistent findings, which is reassuring but not conclusive.

The study is cross-sectional, not longitudinal. It compares people of different ages at the same point in time rather than tracking the same individuals as they age. This means that the apparent age-related changes could in principle be partly cohort effects rather than true ageing effects. The researchers argued, plausibly, that the consistency of the findings across multiple cognitive abilities and across both data sources makes a pure cohort explanation unlikely, but the cross-sectional design is a real limitation.

The cognitive abilities tested are necessarily a subset of the full range of human cognition. The study measured the cognitive functions that can be reliably tested in short online tasks, which excludes many of the more complex and ecologically meaningful cognitive abilities, including practical problem-solving, creative reasoning, social intelligence in real-world contexts, and the various forms of accumulated expertise that develop over a working career.

What it means for general readers

Several things follow from the differentiated cognitive ageing picture that are worth saying clearly.

The first is that there is no single age at which the human mind is at its best. The cognitive abilities that matter for any given task depend on what the task requires. A task that depends on processing speed favours people in their late teens and early twenties. A task that depends on emotional judgment favours people in their late forties. A task that depends on accumulated vocabulary and knowledge favours people in their sixties. Most real-world tasks depend on a mix of multiple abilities and therefore do not have a single optimal age.

The second is that the cognitive decline that begins in the late teens for processing speed is not the decline of human intelligence as a whole. It is the decline of one specific cognitive ability while several other abilities are still developing rapidly. The phenomenon that people sometimes describe as feeling slower at age forty than they did at age twenty is real, but it coexists with the fact that the same people are typically much better at concentration, emotional understanding, and vocabulary at forty than they were at twenty.

The third is that the cognitive abilities that develop latest are also the abilities most heavily dependent on experience rather than on biological hardware. The implication is that the practices that build experience, including reading widely, engaging with complex social situations, learning new skills, and exposing oneself to novel cognitive challenges, are also the practices that continue to develop the latest-peaking cognitive abilities for the longest.

Most cognitive abilities, on the strongest current evidence, are not at their peak in their owner’s twenties. Most are still developing. The forty-eight-year peak for emotional understanding sits almost three decades after the eighteen-year peak for processing speed. The sixty-seven-year peak for vocabulary sits almost five decades after it.

The human mind, in other words, takes a lifetime to reach its full capacity, and different parts of it reach that capacity at very different points in the lifetime. The question of whether someone is past their cognitive prime cannot be answered without specifying which cognitive ability is being asked about.

On the strongest current reading of the evidence, the answer for most adults is that several of their most important cognitive abilities are still rising.