Extraction Dissonance: Not All Ensembles are Created Equal
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
Our visual system is extremely proficient at extracting statistical information, such as the average size and density of objects. But what are the mechanisms supporting this? Here, we present a new methodology for exploring this issue, and use it to show that not all ensembles are treated alike by the visual system. To investigate the perceptual organization of ensemble structures, we began by examining the perception of Pearson correlation r in scatterplots. For plots with single data populations, discrimination performance is described simply: just noticeable differences (JNDs) are proportional to the distance from r=1. Our study expanded this to consider the case where each scatterplot not only contained a "target" population of black dots, but also an irrelevant "distractor" population of red dots (Fig. 1). Observers viewed two such scatterplots side-by-side (each containing both target and distractor populations), and were asked to identify the plot with the higher target correlation. There were 100 target dots in every condition. Interference from several distractor correlation values was determined by measuring JNDs for plots with various distractor dot numerosities: 100, 50, and 25 dots. Results showed a surprising effect: for target correlations of .3 with 100 and 50 distractor dots, discrimination declined considerably when the distractor correlation changed from .9 to .999. Meanwhile, for distractors of 25 dots, JNDs were low for both .9 and .999 distractor correlations (Fig. 2). This extraction dissonance, where discrimination is considerably different for ensembles with highly similar correlations, suggests that denser .999 populations may be represented similarly to a holistic unit, rather than an ensemble. More generally, our methodology provides exciting potential for exploring the level at which set of items can be perceived as a population, versus a single visual object. Meeting abstract presented at VSS 2016
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 | 0.001 | 0.002 |
| 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) | 0.000 | 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