Quantifying distraction in a visual search task
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Variants of the visual search task have been used to provide key insights into how individuals search for relevant information amongst irrelevant content. Using this task, researchers have sought to answer questions about when attention is necessary for search, the role of inhibition of return during search, and whether social information is “more distracting” than other types of content. In spite of this widespread use, consideration of the semantic meaning of one’s chosen distractors has not been taken into account. Thus here we sought to quantify distraction at the level of semantic similarity, rather than the more common technique of controlling for low-level luminance information. We quantified semantic similarity through the use of word vectors. To do so, we employed vector models, which represent words as lists of numbers created by machine learning models trained on large collections of text. The more semantically and syntactically similar two words are, the closer their word vectors, allowing us to measure word similarity. We chose 5 target categories with 2 targets each, and then created 10 levels of varying similarity between the target and a main distractor. On each trial, participants were told which target item to search for amongst a display of 6 images, and were instructed to click on the image with a cursor as soon as they found the target. For half of the trials, the target and main distractor images were beside each other, and for the other half of trials they were across from each other. Accuracy declined as semantic similarity increased, and participants were faster to find the target when the main distractor was located beside as compared to across from the target. Together, our data suggest that the location and similarity between a target and distractor have an effect on attention.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| 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 it