Motion adaptation induced object position bias in macaque IT and SlowFast video recognition models
Bibliographic record
Abstract
To efficiently interact with their environment, primates excel in not just recognizing objects ('what') but also discerning their spatial attributes ('where'). This dual capacity, traditionally attributed to the functional segregation of ventral and dorsal visual processing pathways, is currently being reexamined in light of emerging evidence. Recent work of Hong et al. (2016) revealed the macaque inferior temporal (IT) cortex’s role in encoding object positions. Our study further ventures into this relatively uncharted territory with three main objectives: firstly, to extend the findings of Hong et al., assessing how scaling of neural recording sites in IT influences object-position estimates; secondly, to investigate the impact of motion adaptation (a phenomenon typically associated with dorsal stream) on these estimates; and thirdly, to evaluate whether existing ventral stream models align with our observations. We performed large-scale recordings across IT cortex of 3 monkeys (~500 sites). Monkeys passively fixated Test images (640; 1 of 8 objects, varying latent parameters, embedded in naturalistic backgrounds). Indeed, we observed highly accurate (Pearson R >0.7) IT-population based linear decodes of object positions. Next, to test whether motion-direction adaptation biases position estimates, we preceded the Test image presentation by prolonged (3000 ms) oriented gratings moving in one of four directions. Remarkably, IT-based (192 sites) position decodes showed a significant bias (p<0.0001; permutation test) in the direction opposite to the preceding motion. These biases align with perceptual reports, suggesting that the IT cortex represents perceptual rather than ground-truth positions. Interestingly, simulating the experiments in-silico on SlowFast networks (video recognition model with ResNet-50 backbone) demonstrated a similar bias (absent in vanilla ResNet-50 with scaled activation mimicking neural fatigue). Our findings introduce a framework for probing how dorsal-ventral interactions could generate adaptation after-effects and a model-based hypotheses space to guide the exploration of computational mechanisms critical for dynamic scene perception.
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 | 0.000 | 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".