Developing a non-human primate model to dissect neural mechanisms of human facial expression processing
Bibliographic record
Abstract
Understanding how the human brain processes facial expressions requires quantitative models that bridge neural mechanisms with behavior. While many qualitative descriptions exist, the field lacks tight coupling between neural hypotheses and behavioral measurements. To address this, we developed a comprehensive approach combining behavioral measurements, large-scale neural recordings in macaques, and computational modeling to investigate the computations underlying facial emotion discrimination. We first established a rigorous behavioral paradigm, using a binary human facial expression discrimination task across six emotions (360 images), comparing facial emotion recognition between humans and macaques. To probe the neural mechanisms, we conducted chronic multi-electrode recordings in the macaque inferior temporal (IT) cortex during passive viewing of emotional faces. Using 205 logistic regression decoders, we tested how different transformations of the IT population activity predicted behavioral error patterns. We evaluated a suite of artificial neural networks (ANNs) to identify computational models that mirror these neural processes. Significant image-by-image correlations (r=0.69, p<0.001) validated macaques as a suitable model for studying the neural basis of facial emotion processing. We found that macaque IT activity significantly predicted image-level behavioral responses in both humans (70-170 ms, R=0.49, p<0.001) and monkeys (70-110 ms, R=0.69, p<0.001). Additionally, we observed that traditional action-unit models for facial expression analysis are significantly less aligned with human behavior than other ANNs (e.g., ImageNet-trained models, CLIP, simCLR). In addition, ANN-IT representations most closely matched monkey IT, as assessed by representational similarity analyses, also predicted human behavior more accurately. These findings provide critical insights into the neural computations underlying facial emotion discrimination and establish macaques as a robust model for studying these processes. By integrating neural, behavioral, and computational insights, this work provides a critical step toward developing biologically plausible models of facial expression recognition.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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 source (direct Gemma or distilled Codex), 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".