Research map / Vision-language and multimodal / Human and embodied modeling
Characterizing the Ventral Visual Stream with Response-Optimized Neural Encoding Models
Meenakshi Khosla, Keith Jamison, Amy Kuceyeski, Mert R. Sabuncu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Share weights across filter orientations (E(2)-steerable) vs standard CNN (ablation) | Response-optimized model · NSD ROI prediction | Pearson r | No clear effect |
| Response-optimized model vs task-optimized and ideal observer baselines in early visual areas | Response-optimized model · NSD early visual ROIs | prediction accuracy | Improved |
| Response-optimized vs ideal observer in high-level areas | Response-optimized model · NSD LO and VO | prediction accuracy | Improved |
| Response-optimized vs task-optimized models on synthetic stimuli | Response-optimized model · NSD-synthetic OOD | prediction accuracy | Mixed |
| Fit readout for novel subjects with only 5% of training data | Response-optimized model · Novel NSD subjects | prediction accuracy: ~500 samples | Improved |
| Use response-optimized model readout vs CORnet-S for retinotopy with few samples | Response-optimized model · pRF agreement | agreement | Improved |
| Train on brain response instead of random initialization (ablation) | Response-optimized model · ImageNet-16 transfer | classification accuracy | Improved |
| Model higher visual areas (V4, LO, VO) instead of early areas | Response-optimized model · ImageNet-16 transfer | classification accuracy | Improved |
| Optimize on brain responses rather than COCO object labels (ablation) | Same architecture · Cue-conflict dataset | shape bias: 56.7-69.4% | Improved |
| Move from early to higher visual ROI target | Response-optimized model · NSD perception and semantics RSA | alignment with human perception | Improved |
| Use predicted instead of measured responses | Response-optimized model · NSD perception | RSA agreement with human similarity | Improved |
Closest papers by meaning
- A Convolutional Neural Network Interpretable Framework for Human Ventral Visual Pathway Representation (AAAI 2024 · similarity 0.91)
- Sparse components distinguish visual pathways & their alignment to neural networks (ICLR 2025 · similarity 0.90)
- Vision CNNs trained to estimate spatial latents learned similar ventral-stream-aligned representations (ICLR 2025 · similarity 0.90)
- In Silico Mapping of Visual Categorical Selectivity Across the Whole Brain (NeurIPS 2025 · similarity 0.89)
- Flexible Context-Driven Sensory Processing in Dynamical Vision Models (NeurIPS 2024 · similarity 0.89)
- Scaling Laws for Task-Optimized Models of the Primate Visual Ventral Stream (ICML 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
