What Does “Smarter” Mean When We Talk About Exercise?
This series examines whether repeated exercise changes cognitive performance in adults who are generally healthy or sedentary but do not have diagnosed major cognitive impairment. Its specific focus is the match between exercise modality and cognitive domain: whether aerobic, resistance, or multicomponent programs affect broad cognition, executive control, working memory, processing speed, or episodic memory in the same way. That focus matters because “smarter” sounds like one thing, while the studies measure several different abilities.
The question is not trivial. Sedentary behavior is associated with concerns about brain health, and exercise is attractive as a modifiable, scalable intervention. But a useful answer has to be more precise than “exercise is good for your brain.” A person might become faster at a timed task, better at holding information in mind, or more capable of switching between rules without showing a parallel improvement in delayed recall. Those are meaningful changes, but they are not interchangeable—and none is automatically an increase in general intelligence.
The evidence considered here comes from two complementary reviews. One 2026 systematic review and network meta-analysis examined 17 randomized controlled trials involving 2,187 sedentary adults and compared aerobic exercise, resistance exercise, and multicomponent exercise. A second 2026 review focused specifically on repeated resistance exercise in healthy adults; it included 17 studies and extracted data from 1,270 participants. The reviews overlap in subject matter, not necessarily in the trials they analyzed. Their value lies in allowing two views of the same question: a broad comparison of exercise types and a closer look at what resistance training may do across cognitive domains.
The title asks for more than the studies measured
A study can measure whether scores change on a cognitive test. It cannot, by that fact alone, answer whether a person has become more intelligent in the broad sense used in everyday speech. Intelligence includes many capacities and is influenced by education, health, environment, sleep, mood, and opportunity. The two reviews do not test all of those influences. They ask a narrower intervention question: after a program of repeated exercise, do participants perform better on selected measures than people in comparison conditions?
That narrower question is still important. It converts a vague promise into something that can be compared across studies. The exercise categories are also practical families rather than single activities. Aerobic exercise includes rhythmic, continuous activities such as walking, jogging, cycling, or swimming. Resistance exercise works against an external resistance to build strength or endurance. Multicomponent exercise combines two or more modes, such as aerobic and resistance work, sometimes with balance or mind–body elements. A finding about one family should not be silently transferred to every program that happens to be called exercise.
The population boundary matters just as much. The sedentary-adult review defined low activity in several operational ways, including doing less than 150 minutes of moderate-intensity activity per week, doing less than 60 minutes of moderate-to-vigorous activity, failing to meet activity guidelines for several months, exercising fewer than twice a week over a longer period, or scoring low on a validated activity questionnaire. “Sedentary” is therefore a study definition, not a permanent identity or a judgment about a person’s motivation. The review also allowed generally healthy adults as well as people with cardiometabolic risk factors such as overweight or prediabetes, provided they had no major cognitive impairment.
The resistance-focused review considered adults who were cognitively healthy or did not have a diagnosed neurocognitive disorder. These populations are useful for prevention-oriented questions, but “healthy” does not mean identical baseline health, and neither label makes the results transferable to people with dementia, Parkinson’s disease, severe psychiatric illness, or another condition that could change safety and response. The reviews excluded such conditions for good methodological and practical reasons. The exclusion also sets a limit on the headline: these findings describe the possibility of supporting cognition before major diagnosed impairment, not a treatment for established disease.
The studies also concern multi-session programs, not the claim that one workout permanently sharpens the mind. That distinction keeps the time scale honest. A single session may produce a temporary change in arousal or attention, but the reviews assembled evidence about programs repeated over weeks or months. The question is therefore whether regular exposure is associated with measurable adaptation, not whether exercise creates an immediate feeling of mental clarity. It also means that adherence is part of the real intervention. A program that produces an effect under supervision but is rarely completed in ordinary life has a different practical value from one that a person can sustain, even if the studies do not give us a simple formula for converting adherence into cognitive benefit.
The comparison condition matters as well. In the broad review, the control group was not supposed to receive another intervention that could affect cognition. In the resistance review, comparison conditions ranged from non-exercise or usual activity to health education, balance work, or another exercise condition. A passive control and an active control do not ask exactly the same question: the first contrasts training with little structured activity, while the second asks what extra benefit one program adds beyond another engaging activity. This is one reason a positive group difference should be read as evidence about a particular comparison, not as a universal distance between exercise and all alternatives.
A cognitive map is more useful than a single score
The most important vocabulary distinction is between global cognition and particular cognitive domains. Global cognition is usually a broad composite or screening estimate assembled from several tasks. It can be useful as a summary of overall performance, but it may hide a mixed profile. If one ability improves while another stays unchanged, a composite can show a modest average shift without revealing which process moved. A broad score is therefore a map of the territory, not a complete description of every road through it.
The individual measures matter because different tests place different demands on a participant. A timed response task may be sensitive to speed and attention; a rule-switching task adds executive control; a delayed-recall task requires information to be retained across an interval. The reviews tried to group measures by their main cognitive construct, but no test is a perfectly pure window onto one ability. Composite batteries can smooth random noise and provide a useful overall estimate, while also concealing a divergence between subtests. Practice effects, scoring direction, and the choice of task can all affect how an apparent change should be interpreted.
Executive function concerns the control of goal-directed behavior: inhibiting a habitual response, switching between rules, solving a problem, or adapting when information conflicts. These abilities matter when a task requires more than recalling a fact. Working memory is the temporary maintenance and manipulation of information while an activity is under way—for example, keeping several instructions in mind while applying them. Processing speed is the efficiency with which relatively simple operations can be completed under time pressure. A faster response on a valid speed task is not the same as being wiser or knowing more; it is evidence about one component of performance.
Episodic memory concerns storing and later retrieving information about events or material encountered earlier. Delayed recall and recognition tasks are designed to test that longer-term process. It is easy to assume that any exercise-related brain benefit must appear as better memory, especially because discussions of exercise often mention the hippocampus. But the outcome has to be measured, and the present evidence does not show the same level of consistency across all memory results as it does for several other domains.
This distinction is also why a result can feel more modest than the word “cognition” suggests. Imagine a person who becomes better at holding a phone number in mind while dialing it, but not better at recalling a list learned yesterday. Both tasks involve memory in ordinary language, yet they draw on different processes and may respond differently to training. The reviews do not require us to choose between “the brain improved” and “nothing happened”; they allow a more informative answer in which some abilities move, some do not, and some remain too variable to call with confidence.
This map prevents a common error in interpreting the title. Suppose a program improves executive control and processing speed but produces a mixed result for episodic memory. Saying simply that “cognition improved” would be directionally fair but scientifically incomplete. Saying that participants became “smarter” would go further than the measurements allow. The more useful statement identifies the ability that changed, the size and consistency of the change, and the population in which it was observed.
Two evidence lenses, one cautious thesis
The broad review uses a network meta-analysis because many individual trials compare one exercise type with a sedentary control rather than comparing aerobic, resistance, and multicomponent exercise directly. A network can connect those fragments through direct and indirect comparisons: if aerobic exercise and resistance exercise have each been compared with a common control, the network can estimate their relative position even when the two were not tested against each other in the same trial. That is valuable when the evidence is scattered, but it also creates a boundary. The indirect path depends on the studies being sufficiently comparable in participants, interventions, outcomes, and other conditions. Differences in those features can make a neat ranking look more precise than the underlying evidence really is.
The broad review therefore gives us two different kinds of information. Direct comparisons tell us what happened in studies that actually placed the relevant groups side by side. Indirect comparisons help fill gaps and estimate relative probabilities, but they carry additional assumptions. The confidence in “which is better” can consequently be lower than the confidence that some exercise is better than remaining sedentary on a particular outcome. Keeping those claims separate will matter later, when the network’s modality rankings are put next to the resistance review’s more detailed domain estimates.
The resistance-focused review asks a different question. Instead of ranking all exercise families, it examines whether repeated resistance exercise affects several cognitive abilities in healthy adults. That focus reveals why a modality should not be dismissed because it ranks below another one in a broad network. Resistance exercise may show useful effects for some domains even if the comparative ranking is uncertain. Conversely, a positive resistance result does not prove that resistance training is the best choice for every cognitive goal.
The central thesis follows from putting these lenses together. Exercise should be understood as a set of interventions acting on a set of cognitive outcomes, with evidence of different strength for each connection. The practical judgment this series supports is therefore conditional: identify the cognitive goal, understand what the studies actually measured, compare the feasible exercise options, and keep the certainty of the evidence attached to the conclusion. The next question is how such different exercise programs could plausibly reach the brain and produce different patterns of performance—and where that mechanism story stops being proof.