Biases in predictions of dynamic natural scenes: contributions of motion and scene content on the accuracy and precision of prediction
Bibliographic record
Abstract
Prediction is a fundamental part of navigating our visual world. Although there is prior evidence of prediction in memory representations of dynamic natural scenes (representational momentum), there is relatively little empirical data on an explicit prediction task in this context. Our study investigates the extent to which drivers can accurately predict non-hazardous, everyday road scenes. To this end, we created a novel stimulus set of 3D videos of real road scenes, which we plan to make publicly available, recorded using a stereoscopic dashcam setup during urban and highway driving. On each trial, we showed observers a 2s preview (video or still image) of a road scene and asked them to select the image that best represents what they think the scene will look like 2s after the end of the preview in a 5AFC task. The alternatives were frames sampled from the video at 1s intervals and always included the correct (+2s) frame. We also manipulated the presence of stereoscopic depth information using a 3D display. In a sample of 48 licensed drivers who each performed 420 trials, we found that predictions were on average 0.29s farther in time than ground truth, and such bias towards the future was larger for video compared to still image previews. Prediction proportion correct was higher for video compared to still previews and for urban roads compared to highways, suggesting an important role for motion information and environmental density in prediction. Moreover, these effects were mainly driven by increased prediction precision with relatively small changes in the magnitude of future bias. Stereoscopic depth information had negligible effects on prediction performance. Our findings suggest that drivers can make predictions about road scenes, and these predictions are subject to biases similar to those affecting memory representations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.041 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".