Hold your finger down on a key for exactly three seconds. Now draw a line exactly seven centimeters long. Both tasks ask you to turn a number into a produced quantity, and both are things people do badly at first and better with practice. Whether the brain treats them as the same kind of problem has been surprisingly hard to settle.
Jiaxuan Teng, Arne D. Ekstrom, and Eve A. Isham at The University of Arizona set out to compare the two directly. In a preprint posted to bioRxiv on July 29, 2026, they describe a study in which participants produced temporal intervals and line lengths during a training phase, with some receiving feedback on how close they came and others receiving none. Everyone was then tested without feedback. Finally, the researchers introduced magnitudes nobody had trained on: new durations, new line lengths. That last phase was the real question. Does whatever people learn during training stay glued to the specific quantities they practiced, or does it generalize?
The first result is a similarity. Time and space followed learning curves of the same shape: quick gains at the start, then a plateau. People got most of the improvement they were going to get early, and after that the curve flattened. Theoretical accounts of how we learn to judge durations have disagreed about whether time estimation should behave like other magnitude judgments at all, so a shared trajectory is worth noting on its own.
The second result is a difference, and it runs consistently in one direction. Spatial production reached higher accuracy and higher precision than temporal production, got there faster, and held the advantage more steadily. Accuracy here means landing near the target; precision means landing in roughly the same place each time, whether or not that place is correct. You can be reliably wrong. Lines beat intervals on both counts.
What feedback did
Feedback helped in both domains, but not everywhere at once. The authors report that its effect on precision showed up early, during training itself: people who were told how they were doing became more consistent while still practicing. The accuracy benefit arrived later, becoming clear during the test phase after the feedback had been withdrawn. That split is the sort of detail that is easy to miss with a single summary score, and it suggests feedback does at least two separable things rather than one general thing.
Those gains did not evaporate when the practiced magnitudes did. Participants who had trained with feedback carried the benefit into the transfer phase, on durations and lengths they had never seen. The effect was stronger for spatial production, which fits the broader pattern of space outperforming time throughout the study. So what people acquired was not a memorized set of specific answers. Something more portable came along with it.
The transfer data also revealed a familiar distortion. Estimates drifted toward the middle of the range: shorter magnitudes were overproduced, longer ones underproduced. Researchers call this a central tendency effect, and it is a well-documented signature of how people handle magnitude generally. Ask someone to reproduce a very short interval and they will tend to overshoot. Ask for a long one and they will fall short. Learning improved performance without erasing this pull toward the average.
Why it matters
The practical reading is straightforward. If you are teaching a skill that depends on producing the right duration, expect it to be harder work than teaching the equivalent spatial skill, and expect the results to be shakier. Musicians, surgeons, athletes, and anyone learning to time an action all operate in this territory. The finding that feedback benefits generalize to untrained magnitudes is the more encouraging piece: training on a handful of examples appears to build something that extends past those examples, at least within the range tested.
The theoretical reading is more careful. Time and space have long been argued to share a common magnitude system in the brain, with debate over how deep that sharing goes. This study lands in the middle, and the authors present it that way. The learning dynamics overlap in shape but come apart in quality and durability. Both explanations that treat time as just another magnitude and explanations that treat it as entirely separate have trouble accommodating that combination.
Some caution is warranted. This is a preprint, meaning it has not yet completed peer review. The abstract available here does not specify how many people took part, the exact durations and lengths used, or the size of the differences, and those numbers matter for judging how large and how reliable the effects are. What the work establishes is a direction and a shape: two magnitude skills that learn alike and perform unalike, with feedback pushing both forward and carrying further in space than in time.