Middle spatial frequencies are needed for face recognition only when learned faces are unfiltered: More evidence from spatial frequency thresholds for matching
Bibliographic record
Abstract
A number of studies (Gold, Bennett, & Sekuler, 1999; Nasanen, 1999; see Parker & Costen, 2001 for review) have suggested that middle spatial frequencies (SFs) are optimal for face recognition. A few recent studies (Liu et al., 2000; Collin et al., 2003; Kornowski & Petersik, 2003) have cast doubt on this, suggesting that perhaps spatial frequency overlap is the more important factor in determining how well spatially filtered faces are recognized. The latter studies predict that if learned faces are filtered in the same way as the tested faces, little or no advantage of middle SFs for face recognition will be found. At VSS 2004 (Collin & Martin, 2004), we presented data on SF thresholds for face recognition when comparison faces were unfiltered vs. when they were filtered in the same way as the test face. Those data showed that middle SF are needed for face recognition only when the comparison faces are unfiltered. However, the thresholds were gathered by the method of adjustment, a method thought to be vulnerable to observer criterion shifts. This opened up a potential alternative explanation for our results. Here we present similar data, but gathered by the method of constant stimuli. Observers performed a 4AFC task where they were asked to match a test face to one of four comparison faces. The test face was spatially filtered to a range of low-pass and high-pass cut-offs. In one condition, the four comparison faces were unfiltered. In the other condition, the comparison faces were filtered in the same way as the test face. Our data show that more central SFs are sought out when comparison faces are unfiltered than when they are filtered. These data are in accordance with our previous study, suggesting that those results were not due to criterion shifts. This suggests that the high efficacy of middle SFs in face recognition is task-dependent and may arise due to interference from non-middle SFs in unfiltered learned images.
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 | 0.001 |
| 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 | 0.000 |
| Scholarly communication | 0.000 | 0.002 |
| Open science | 0.001 | 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".