Face space is not linear: Empirical evidence of curvature and compression
Bibliographic record
Abstract
Face perception studies and models often assume that face space is linear, that is, that changes in gender or emotion produces linear changes in facial features. In the current study, we test the linear face space assumption by measuring additivity (i.e. whether the combination of gender and emotion is predicted by the sum of its components), proportionality (i.e. whether the ratio of internal-to-physical changes is constant, or whether there are compression or expansion effects at higher gender or emotion intensities), and directionality (i.e. whether increasing gender or emotion intensity introduces qualitative changes). Participants were told to produce faces that corresponded to various intensities along male/female and happy/sad continuums, including combinations of the two dimensions. They produced these faces using Poirier and Faubert (VSS 2010; in revision)’s technique, using sliders to adjust 53 components of facial expression including the shape and position of eyes, eyebrows, mouth, nose, and head. Data from 7 participants show that (1) variability along gender and emotion was captured by 3 dimensions: gender, emotion, and curvature (see below), accounting for 70.7% of variability in features (F(61, 500)=1.8, p=.0002), (2) gender and emotion were linearly additive within the range tested (F(12,560)=1.2, p=.30), (3) there was evidence of compression at high intensities for both gender (F(5, 30)=3.6, p=.012) and emotion (F(5,30)=5.0, p=0.002), (4) maximum expansion occurs at 0.5x female and 0.5x happy, and (5) the face space is significantly curved along both gender and emotion (R2s=49.1% & 57.4%, Fs(1,10)=9.7 & 13.5, ps=.011 & .0043 respectively) meaning that intensity changes introduced qualitative changes that cannot be captured as linear feature changes. The presence of deviations from linearity (e.g. compression, expansion, and curvature) implies that linear morphs as commonly used in experiments and virtual reality introduce both quantitative and qualitative systematic distortions, at least for variations involving gender and/or happy/sad. Meeting abstract presented at VSS 2012
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 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 | 0.000 |
| 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 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".