The McGill Diverse Face Database: 92 Complex Mental States Across Socially Perceived Racial Categories
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
Theory of Mind (ToM)—the ability to infer others’ mental states—is fundamental to social cognition. Social categorization, the grouping of individuals into in-group or out-group categories, shapes these inferences. These processes co-occur during facial perception, with recent research suggesting both shared neurocognitive mechanisms and modulation of ToM by social category cues. However, existing tools for studying the impact of racial diversity on social cognition are limited: some databases prioritize racial representation but restrict stimuli to the six basic emotions, while others broaden mental state diversity but lack diversity in social appearance. Here we introduce the McGill Diverse Face Database, a validated set of 1,286 images of 14 actors from socially perceived racial categories portraying 92 complex mental states. Validation included three experiments: (1) a four-alternative forced-choice task assessing recognition accuracy, (2) a “point-and-click” task rating valence and arousal in a two-dimensional affective space, and (3) a trait-rating task evaluating perceived actor characteristics. Participants accurately identified mental states across categories (77 % of stimuli). Mean valence–arousal ratings revealed a non-linear one-dimensional manifold structure that correlated with behavioral measures. An interactive online visualization allows users to explore this “Theory of Mind manifold” (https://hctor99.github.io/TheoryofMindManifold/). By integrating social-category diversity with complex emotional expression, this database provides a new resource for studying how socially perceived group membership shapes the perception and inference of mental states.
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 | 0.004 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.009 | 0.003 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.004 | 0.009 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.048 | 0.079 |
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