A new study finds the human imagination can reliably simulate the path of a single invisible moving object but struggles to keep track of two at the same time, a result that surprises researchers and has practical implications for teaching, safety and design in Thailand. The experiments, described in Nature Communications, used short animations of bouncing balls that vanished from view and asked participants to predict where and when those objects would hit; people performed well with one disappeared ball but fell to near chance with two, supporting a serial “one-at-a-time” model of mental simulation rather than a parallel one (Nature Communications PDF). The finding suggests that while our eyes and attention can monitor a handful of visible moving objects, the mind’s eye has a much narrower working capacity when it must continue motion after objects drop out of view (Harvard Gazette report).
Why this matters to Thai readers is simple: many everyday tasks — from supervising children at a busy temple fair to estimating multiple vehicles’ trajectories in Bangkok traffic, to lifeguarding on crowded beaches — depend on our ability to mentally project where moving things will be when we cannot see them. Until now, decades of research into visual attention and multiple-object tracking focused on how many items people can keep visually indexed (typically three to four), but far less was known about the imagination’s own capacity when objects are no longer visible (review of multiple object tracking literature). The new research drills into that blind spot and shows a striking bottleneck: the mind’s internal simulation appears to advance only one object at a time.
The experiments were straightforward yet revealing. Volunteers recruited online viewed short videos of animated balls bouncing in a confined space. In some trials a single ball disappeared and participants judged when it would have hit the floor. In other trials two balls vanished after different motion histories, and participants were asked to predict impact times for both. The team contrasted participant behavior with two computational models: a parallel model that simulates all objects simultaneously, and a serial model that advances one object’s trajectory at a time. Human choices tracked the serial model: participants handled one invisible object robustly but performance dropped sharply with two, even when the researchers offered monetary incentives to encourage effort (Nature Communications PDF).
The authors behind the study said the finding surprised them. An associate professor who studies intuitive physics and mental simulation noted that while people often feel they can imagine complex scenes — “If I close my eyes right now, I can see a tower of blocks falling down” — subjective vividness does not equal capacity for concurrent mental simulation. The research lead described repeated experiments that yielded the same limit and even produced amused reactions from participants when they discovered how poorly they managed two invisible trajectories at once (Harvard Gazette report).
This new constraint sits beside a substantial body of work on multiple-object tracking (MOT) in perception. Classic MOT tasks show that observers can attend to and visually track several moving items, typically up to three or four, depending on speed, spacing and task demands. Those limits arise while objects remain visible and available to visual attention and indexing mechanisms. The imagination task used by the new study removes ongoing visual input and forces the brain to continue dynamics internally; under those conditions, the brain appears to abandon parallel tracking and switch to a conservative, serial strategy (MOT review and mechanisms). In other words, seeing and mentally simulating are constrained by different processing architectures.
Several follow-up manipulations in the study help sharpen interpretation. Participants were slightly better when two balls moved in tandem before disappearing, suggesting the brain can collapse similar trajectories into a single simulation when motion is coherent. Participants performed much better when the two balls remained visible until impact, confirming that the deficit specifically concerns continued mental projection after perceptual input ceases. The authors also tested whether poor performance reflected effort conservation: they offered subjects extra pay to incentivize accurate answers, but this did not restore multi-object simulation performance, pointing away from laziness and toward a genuine cognitive capacity limit (Nature Communications PDF; Harvard Gazette report).
The methodological strength of the paper lies in triangulating behavioral data, model comparison and replication across variations. By formalizing competing hypotheses as computational models — parallel versus serial simulation — the authors tested which algorithm best matched human responses. The serial model’s superior fit suggests that the brain’s strategy for simulating unseen motion is to allocate internal “simulation time” to each object sequentially. That decision is sensible from a resource-allocation perspective: running multiple complex simulations in parallel would demand more working memory and processing bandwidth than the brain typically affords for moment-to-moment cognition.
What does this mean for Thai educators, safety professionals and designers? First, teachers demonstrating dynamic physical processes in classrooms should prioritize concrete demonstrations over asking students to mentally run multiple simultaneous simulations. When explaining collisions, free-fall, or conservation laws, live or video demonstrations that keep objects visible will better support learning than relying on students’ unaided imagination. Physics instructors at Thai universities and vocational institutes could pair brief animations with synchronized measurement tasks so students can probe trajectories while they are still visible, rather than asking them to imagine multiple hidden paths.
Second, in public safety settings — lifeguarding on Phuket beaches, traffic policing in Bangkok, or supervising children at Songkran festivals and temple fairs — professionals should be cautious about expecting accurate mental tracking of multiple hidden trajectories. Training protocols can emphasize scanning strategies and environmental cues that reduce the need for internal simulation. For example, lifeguard training might focus on sectoring the observation area and using visual anchors to maintain continuous visibility of swimmers, rather than teaching rescuers to mentally track several submerged swimmers’ predicted surfacing points simultaneously. Road-safety campaigns and driver training can stress the limits of predicting multiple hidden hazards and promote spacing, speed reduction and anticipatory scanning to keep trajectories visible longer.
Third, designers of human-machine interfaces and AI systems deployed in Thai settings should respect this human limitation. Assistive systems that predict and highlight the future positions of external objects (for example, in advanced driver assistance systems or industrial safety monitors) can present aggregated cues rather than expecting operators to mentally simulate several individual trajectories. Interfaces that visualize future paths — even simplified overlays — may compensate for human inability to run more than one unseen simulation at a time, reducing operator load and improving safety.
Beyond practical uses, the study deepens theoretical understanding of how the brain builds internal models of the world. Cognitive scientists have long proposed that the brain operates as a probabilistic simulator, running forward models to predict outcomes. The new work refines that notion by revealing a computational architecture with constrained parallelism: the brain may maintain detailed dynamic simulation for a single object while representing others with cruder heuristics or with memory traces that require serial reactivation. This has implications for computational neuroscience and artificial intelligence. AI systems that emulate human-style prediction may choose between running many low-fidelity simulations in parallel or a few high-fidelity ones serially; matching human strengths and limits could improve human-AI teaming where predictable, explainable behavior is valuable.
Critically, the authors and commentators caution against overgeneralizing the one-object limit. The study focused on short-duration predictions in relatively simple physical contexts (bouncing balls). More complex scenes, longer delays, object similarity, and prior knowledge about dynamics may allow the brain to adopt different strategies. For instance, when multiple objects move together coherently (like a flock or a vehicle convoy), the brain can compress information and simulate a group dynamic rather than distinct items, improving performance. Future research can explore whether expertise — such as athletes, air-traffic controllers, or experienced video-game players — expands imaginative capacity with extensive training, or whether the serial bottleneck reflects a hard computational constraint.
Thai researchers and institutions can play an active role in this next phase. Cognitive labs at Thai universities could replicate the task with local samples to test cultural, developmental and expertise-related moderators. Classroom studies could measure how different instructional methods interact with the imagination’s limits to affect physics learning in Thai students. Public-health and safety agencies might partner with cognitive scientists to design training that compensates for serial simulation, potentially lowering accident risk in crowded urban and coastal environments.
The findings also resonate with Thai cultural values around mindfulness and attention. Buddhist-informed practices that cultivate focused, moment-to-moment awareness emphasize single-task attention rather than multitasking. The new evidence that the mind’s eye may naturally favor serial processing offers a scientific mirror to cultural practices that prize concentrated focus. Educational and workplace policies promoting focused attention and structured scanning could align with both cognitive limits and cultural strengths, potentially improving safety and learning outcomes.
Looking ahead, several research threads are ripe for exploration. Longitudinal training studies could test whether the imagination’s one-object ceiling is malleable across weeks or months of practice. Neuroimaging experiments could reveal whether serial simulation recruits working memory and executive-control circuits in ways distinct from parallel visual tracking. Cross-cultural studies might ask if languages or play practices that emphasize dynamic narratives shape children’s propensity to simulate multiple trajectories. Finally, applied trials could test whether interface designs that externalize future trajectories lower error rates in real-world tasks, from driving to industrial control.
For Thai policymakers and practitioners the takeaway is both pragmatic and empowering. Recognize the imagination’s surprising narrowness when tasks require keeping invisible motion in mind. Reorient training, instruction and safety design to keep critical motion in view, simplify trajectories into coherent groups, and use external visualizations where possible. Schools can adapt lesson plans to favor demonstrations and real-time measurements; hospitals and rehabilitation centers can design motor-relearning exercises that scaffold perceptual input rather than relying on internal simulation; urban planners and transport authorities can prioritize designs that maintain line-of-sight and reduce the cognitive load on road users.
The research opens an accessible frontier: simple, low-cost changes in teaching and safety practice that acknowledge how people naturally think can yield outsized benefits. The authors themselves call for more work to map the tricks the mind uses when the world disappears from view and to chart where imagination is robust and where it fails (Nature Communications PDF; Harvard Gazette report). For Thai communities juggling dense social life, crowded transport and rich educational ambitions, translating these scientific insights into practices and policies could make everyday environments safer and learning more effective.