Three levels of cognition in sport: the Kalén 2021 meta-analysis

Kalén et al. (2021) pooled hundreds of studies and sorted every cognitive measure used in sport into three levels: basic cognitive functions, higher cognitive functions, and sport-specific cognitive decision-making skills. Only the third level strongly discriminates between athletes of different standards. That result, published in Psychological Bulletin, is the theoretical anchor for sport-specific testing and the reason Insporta builds tests from the sport rather than from abstract stimuli.

The three levels

Level	What it measures	Ability to discriminate between athletes
1. Basic cognitive functions	Reaction time, processing speed, simple attention	Weak
2. Higher cognitive functions	Executive functions (working memory, inhibition, cognitive flexibility), anticipation on generic stimuli	Moderate
3. Cognitive decision-making skills	Sport-specific decisions on representative stimuli (match video, 360-degree scenes)	Strong, the best discriminator

The ordering is not accidental. It follows the two competing traditions in sport cognition. The cognitive-component approach (Nougier et al., 1991) predicts that athletes should excel at levels 1 and 2, because sport is supposed to train the general machinery of the mind. The expert-performance approach (Ericsson and Smith, 1991) predicts that the advantage should appear mainly at level 3, because expertise is knowledge and skill organised around a domain. The meta-analysis finds the expert-performance pattern.

Why the third level discriminates

Three mechanisms are visible across the pooled studies.

Domain-specific stimuli engage domain-specific memory. A footballer looking at a match scene recognises formations, passing lanes and pressing triggers that a non-player does not see. A footballer looking at an arrow on a screen has no such advantage. The more the test material resembles the sport, the more of the athlete’s accumulated experience it can draw on.

Generic functions are shared with everyone. Reaction time and working memory are distributed across the population and shaped by age, education and health. Athletes may sit slightly above average, which is why level 2 shows a moderate effect, but the overlap between an academy player and a schoolmate is large. Beavan et al. (2020) later showed that the trajectories of these functions in elite players follow the general population over years of training, which fits the meta-analytic picture.

Response format amplifies the effect. Travassos et al. (2013), pooling 111 effect sizes, found that expertise differences grow as the required response moves from verbal report to button press to micro-movement to in-situ action. Representative stimuli and representative responses reinforce each other, and both belong to level 3.

What the hierarchy does not say

The meta-analysis does not say generic cognition is irrelevant to sport. Level 2 has a moderate effect, and Reinhard, Mann and Höner (2025) found a generic composite explained 13% of variance in playing level, which is not nothing. What it says is that if you want to tell athletes apart, or find the ones who will progress, sport-specific measures are the better instrument, and generic ones are a poor substitute rather than a shortcut.

Nor does it say that every sport-specific test is a good one. A video decision-making test built without reliability data, or a knowledge test full of questions that reward reading speed rather than understanding, is still sport-specific and still weak. Level 3 is where the signal lives; extracting it is a separate job, covered in how to validate a sports knowledge test.

Where Insporta sits

Insporta’s tests are sport-specific by construction: the content is the rules, tactics, opponent patterns and history of one sport, every question carries a topic tag, and the wrong options are built from that sport’s own plausible misconceptions. This places Insporta at the third level of Kalén’s decomposition.

Two honest qualifications follow.

First, Insporta measures knowledge, not video-based decision-making. The strongest numbers in the literature, including the 34% in Reinhard et al. (2025) and the r = .78 reliability in Höner et al. (2023), belong to 360-degree video tests. Knowledge is the substrate of decision-making, and the applied questions in Insporta reach toward it, but the two are different constructs and the numbers are not transferable.

Second, a level-3 instrument can still be contaminated by level-1 noise. A question with a stem twice as long as its neighbours measures reading speed as much as knowledge. That is why the methodology holds stems and options inside fixed length corridors and treats response time as one dimension, never as the score itself.

Reading on

• Sport-specific knowledge, not generic brain training applies the hierarchy to the 2025 study.
• Executive functions: what a knowledge test does not measure covers level 2 in detail.