Classic 1980s arcade game validated as a window into executive functions

In the 1980s, cognitive scientists Mané and Donchin built a deceptively simple arcade game to study how people acquire complex skills. Players pilot a frictionless spaceship, bombard a space fortress, identify and destroy mines, and capture bonus symbols, all at the same time. Four decades later, that game, known as the Space Fortress task, has finally received the rigorous psychometric scrutiny it long lacked. A new pre-registered study of 145 healthy adults shows that the task produces normally distributed scores with good-to-excellent reliability, and that performance is meaningfully tied to executive functions, the suite of cognitive abilities that governs how humans plan, adapt, and control their behavior in demanding situations.

The Space Fortress task was originally conceived as a laboratory paradigm for complex skill acquisition, and its designers were explicit about why. Perceptual, cognitive, and motor skills are all required of the player, along with specific knowledge of the rules and game strategy. Unlike traditional laboratory measures such as the Stroop task or dual N-back, the game offers a dynamic, semi-ecological environment that mimics real-world cognitive demands more closely, and its video game format promotes the motivation and engagement known to facilitate cognitive training outcomes. Crucially, its four subtasks are interdependent rather than independent: missiles earned in the bonus task are needed to destroy mines and the fortress, and careless flying accumulates negative points that drag down the total score. This interactivity mirrors operational settings like aviation, where adjusting an aircraft’s vertical speed also alters its airspeed and forces a throttle correction.

To characterize the cognitive machinery underlying performance, the researchers anchored their investigation in the influential model of Miyake and Friedman, which decomposes executive functions into three correlated but separable components: mental set shifting, information updating in working memory, and inhibition of prepotent responses. This three-factor structure has replicated robustly in healthy adult populations and comes with a well-established battery of nine cognitive tasks. The theoretical case for linking the game to all three components is compelling. The bonus task resembles a 1-back paradigm, demanding continuous updating of symbol sequences in working memory. The fortress subtask functions like an inverted go/no-go paradigm, since shots must be spaced more than 250 milliseconds apart or the fortress’s vulnerability resets, yet a rapid double shot under 250 milliseconds is required to destroy it. The mine subtask demands withholding the prepotent urge to shoot every mine until proper identification is achieved, and the multitask structure creates strong incentives for shifting between subtasks.

The study recruited 180 participants in Toulouse, France, between February 2022 and November 2023, ultimately analyzing 145 adults aged 18 to 39 with a mean age of 24.6 years. A priori power analysis, based on an expected correlation of 0.30 between executive functions and game performance with 90 percent power and a 0.01 significance level, called for 157 participants. During the first laboratory session, participants familiarized themselves with the game for roughly 12 minutes using variable training, practicing each subtask separately before combining them. They then completed ten three-minute runs, alternating between monotask runs, involving only flying and fortress destruction, and multitask runs engaging all four subtasks. The best multitask score served as the performance measure, with points allocated by fixed rules: destroying the fortress earned 250 points, type-1 and type-2 mines earned 50 and 60 points respectively, while being hit, crossing game borders, or failing to destroy a mine before it disappeared cost 35 to 100 points.

The second session, held within a week, administered the nine executive function tasks in a fixed order: antisaccade, letter-memory, color-shape, number-letter, Stroop, keep-track, dual 2-back, category-switch, and stop-signal. Three tasks measured inhibition, three measured updating, and three measured shifting, yielding composite scores for each component plus an overall executive function index. The results on psychometrics were encouraging. Raw scores ranged from minus 1330 to 5160 points, with a mean of 1492.6, and the distribution showed only a slight leftward skew of minus 0.21 and near-zero kurtosis of minus 0.04. A Kolmogorov-Smirnov test revealed no significant deviation from normality. Reliability, assessed with intraclass correlation coefficients across the five multitask runs in a single session, reached 0.785 for single measures and 0.948 for the average of five runs, both statistically significant, placing the task in the good-to-excellent range by conventional benchmarks.

Concurrent validity proved equally solid. The correlation between the best game score and the composite executive function score was 0.50, meaning that executive functioning accounted for 25 percent of the variance in performance. Notably, the composite score correlated more strongly with the game than any individual component, suggesting that the interaction of shifting, updating, and inhibition matters more for this task than any single ability alone, an observation that echoes debates about a global factor overseeing individual executive components. Among the subcomponents, inhibition showed the strongest correlation at 0.42, followed by updating at 0.38 and shifting at 0.25. The strength of the inhibition link was unexpected, since the researchers had predicted updating would dominate. Exploratory regression models reinforced the pattern: a model including the composite executive function score and four covariates explained 46.7 percent of the variance, while a model separating the three components explained 52.5 percent, with inhibition and updating remaining significant predictors and shifting dropping to null once the others were accounted for.

The authors propose several explanations for inhibition’s prominence. Prepotent response inhibition, the ability to suppress dominant automatic responses, correlates strongly with resistance to distractor interference, both of which the game demands when players must refrain from firing too quickly at the fortress or shooting mines before identifying them. Inhibition may also tap a common executive factor reflecting the ability to actively maintain goals and bias ongoing processing, for instance stopping fire when missiles run out and reorienting toward a bonus to reload. A third possibility involves processing speed, which shares substantial variance with inhibition; faster processing of the game interface, from missile stock to fortress status, would plausibly drive better adaptation. As for updating, participants who verbalized their shot count or kept track of their missile stock in working memory freed themselves from constantly checking the interface, strategies that likely involve the phonological loop and support situational awareness, the capacity to understand the current and future state of the game that has been shown to correlate strongly with working memory.

Shifting’s weak showing may reflect a mismatch between what the standard battery measures and what the game actually requires. The laboratory shifting tasks involve reactive transitions between two predefined tasks cued by the experiment, whereas the game likely depends on self-initiated, context-driven shifting across four interconnected subtasks, a distinction between proactive and reactive control discussed extensively in the cognitive control literature. The covariate analyses added further nuance. Video game experience correlated positively with performance at 0.22, but this relationship vanished in the regression models, possibly because the internally developed questionnaire did not distinguish between fast-paced competitive titles like StarCraft or Fortnite and slower games like Animal Crossing. Age showed a weak negative correlation that did not reach significance, likely limited by a sample in which only 17.2 percent of participants were 30 or older, and education level showed no relationship in a highly educated sample averaging 16 years of schooling.

Sex produced the largest covariate effect by far. Men outperformed women with a large effect size of Cohen’s d equal to 1.28, and sex remained a strong predictor even after accounting for executive function scores, video game experience, age, and education. The authors note that this replicates earlier findings attributed to men being more likely to discover the optimal strategy of circling the fortress, and they propose additional contributing factors: social pressure to compete, which is visible in France where men account for roughly 94 percent of esports participants; men’s tendency to play more competitive action games whose keyboard controls overlap with the game’s; and an academic imbalance in the sample, since most male participants studied aeronautical engineering and were likely more familiar with inertia, momentum, and thrust. The researchers caution that anyone using the task as a dependent measure should account for this difference, as it may reflect a psychometric limitation.

The study also flagged a potential floor effect: some participants struggled with ship control and accumulated negative scores, and these low performers also showed lower composite executive function scores, suggesting the game may be less suitable for individuals with weaker executive functioning and warranting caution in clinical populations. Nevertheless, the findings open four promising directions: semi-ecological executive function assessment, since the task can be understood in about fifteen minutes and assessed with a single three-minute run compared with the roughly 90 minutes required by the standard battery; complex skill acquisition research, the game’s original purpose; executive function training, where its adjustable parameters suit adaptive difficulty protocols, though past training studies have yielded inconsistent transfer results, including one early report that pilot trainees trained on the game were 30 percent more likely to pass their flight exam while later replications found limited or no benefit; and mental workload investigation, where participant feedback indicated perceived workload differed substantially between monotask and multitask conditions. Future work should now test long-term test-retest reliability across sessions separated by weeks or months, and extend validation to older adults and clinical groups where floor effects may emerge.

Subject of Research: Psychometric validation of the Space Fortress task and its relationship with executive functions in healthy adults

Article Title: The space fortress task: Psychometrics evidence and relationship with executive functions

Article References: Chenot, Q., & Scannella, S. (2026). The space fortress task: Psychometrics evidence and relationship with executive functions. Heliyon, 12(16), Article e45593. https://doi.org/10.1016/j.heliyon.2026.e45593

Image Credits: AI Generated

DOI: 10.1016/j.heliyon.2026.e45593

Keywords: Space Fortress, executive functions, psychometrics, cognitive training, working memory, inhibition, cognitive assessment, video game experience, sex differences, skill acquisition, space, fortress