Exploring the gist-based modulation of learning rate in visual search
Bibliographic record
Abstract
Scene categorization is a rapid and automatic visual perceptual process, occurring with less than 100 milliseconds of exposure to a natural scene image. Why do humans have this exceptional ability to immediately, and efficiently, grasp gist information? One prominent theory (Torralba et al., 2006) suggests scene gist is rapidly perceived in order to guide exploration of our visual environment towards information-rich regions (e.g., countertops in a kitchen). Such gist-based guidance allows for efficient sampling of behaviourally relevant information contained within the visual scene. This semantic information has been shown to guide attention and visual search alongside bottom-up and top-down influences. Currently, there is little agreement about whether scene gist can be used to guide attention. Across three experiments, we test the hypothesis that rapidly available guidance signals from scene gist can be leveraged to learn new attentional strategies. All experiments were variations on the scene-preview paradigm (Castelhano & Heaven, 2010). After being presented with a preview containing a degree of information relevant to the search image, participants were instructed to find a target embedded in a naturalistic scene. Critically, target location was linked to scene gist, such that the target appeared in a consistent location determined by the scene’s conceptual category. The three experimental variants are as follows: 1) within-subject and pictorial search previews, 2) within-subject and semantic search previews, and 3) between-subjects and pictorial search previews. Across these experiments, we find evidence that activating gist with scene previews increases search efficiency. Preliminary computational analyses with a combined model of visual perception (VGG16) and category learning suggest this search benefit arises in a manner consistent with formal theories of skill acquisition. These findings are consistent with a flexible learning system that leverages scene gist information in novel ways to improve visual search efficiency.
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 | 0.001 | 0.007 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 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".