Research map / Vision-language and multimodal / Human and embodied modeling
Only Brains Align with Brains: Cross-Region Patterns Expose Limits of Normative Models
Larissa Höfling, Matthias Tangemann, Lotta Piefke, Susanne Keller, Matthias Bethge, Katrin Franke
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 |
|---|---|---|---|
| evaluate diverse vision models with standard RSA/LP alignment scores | 47 pretrained vision models · BOLD Moments, visual ROIs | model-brain alignment | No clear effect |
| use linear predictivity instead of RSA as the alignment measure | best-ranked models · BOLD Moments, visual ROIs | alignment relative to brain-brain reference | Mixed |
| add alignment pattern similarity requirement to the alignment score criterion | model-ROI candidate selection · BOLD Moments, visual ROIs | number of candidate brain-aligned models | Improved |
| compare model-derived alignment patterns with brain-derived patterns | all evaluated models · BOLD Moments, visual ROIs | alignment pattern similarity | Hurt |
| compare equivalently aligned models by their cross-region alignment patterns | top RSA models · BOLD Moments, visual ROIs | alignment pattern | Improved |
| compare fMRI-derived alignment patterns across subjects | brain-brain alignment patterns · BOLD Moments, visual ROIs | alignment pattern similarity across subjects | Improved |
| compare fMRI-derived patterns to structural connectivity patterns | fMRI-derived alignment patterns · HCP structural connectivity vs BOLD Moments | alignment pattern similarity | Mixed |
| normalize scores to a lower estimate of intersubject consistency (ablation) | alignment scores · BOLD Moments, visual ROIs | discriminative power | Improved |
| train on varied Taskonomy tasks instead of recognition or video objectives | ResNet-50 Taskonomy models · BOLD Moments, visual ROIs | RSA alignment: 0.06 vs 0.22 (V1, class-object vs V-JEPA 2 giant-384) | Hurt |
Closest papers by meaning
- Bridging Critical Gaps in Convergent Learning: How Representational Alignment Evolves Across Layers, Training, and Distribution Shifts (NeurIPS 2025 · similarity 0.89)
- Scaling Laws for Task-Optimized Models of the Primate Visual Ventral Stream (ICML 2025 · similarity 0.89)
- Sparse components distinguish visual pathways & their alignment to neural networks (ICLR 2025 · similarity 0.89)
- Model–Behavior Alignment under Flexible Evaluation: When the Best-Fitting Model Isn’t the Right One (NeurIPS 2025 · similarity 0.88)
- Attention Alignment Between Humans and Vision-Language Models (NeurIPS 2026 · similarity 0.88)
- MindAligner: Explicit Brain Functional Alignment for Cross-Subject Visual Decoding from Limited fMRI Data (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.
