Research map / Label-efficient and robust learning / Recognition and long-tail
Probing Brain Activation Patterns by Dissociating Semantics and Syntax in Sentences
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| add paraphrase loss (PLoss) (ablation) | VGVAE · STS benchmark test (sem var.) | Pearson correlation (%): 65.7 -> 72.5 | Helped |
| add paraphrase loss (PLoss) (ablation) | VGVAE · Constituent parsing (syn var.) | F1: 26.7 -> 29.0 | Mixed |
| add word position loss (WPLoss) (ablation) | VGVAE · Constituent parsing (syn var.) | F1: 26.7 -> 35.7 | Helped |
| add jabberwocky loss (JLoss) (ablation) | VGVAE · POS tagging (syn var.) | accuracy (%): 22.6 -> 32.3 | Helped |
| add jabberwocky loss (JLoss) (ablation) | VGVAE · STS benchmark test (sem var.) | Pearson correlation (%): 65.7 -> 55.2 | Hurt |
| add jabberwocky loss to VGVAE+PLoss+WPLoss (full DFRM) (ablation) | VGVAE+PLoss+WPLoss · Constituent parsing (syn var.) | F1: 34.6 -> 40.0 | Helped |
| add word position loss to VGVAE+PLoss+JLoss (full DFRM) (ablation) | VGVAE+PLoss+JLoss · POS tagging (syn var.) | accuracy (%): 34.4 -> 38.6 | Helped |
| add all three losses (DFRM) vs baseline VGVAE | VGVAE · STS benchmark test (sem var.) | Pearson correlation (%): 65.7 -> 73.0 | Helped |
| replace direct distance-minimizing constraint with jabberwocky/paraphrase generation losses (ablation) | DFRM · semantic/syntactic similarity | performance | Helped |
Closest papers by meaning
- Can fMRI reveal the representation of syntactic structure in the brain? (NeurIPS 2021 · similarity 0.92)
- Probing Word Syntactic Representations in the Brain by a Feature Elimination Method (AAAI 2022 · similarity 0.92)
- Far from the Shallow: Brain-Predictive Reasoning Embedding through Residual Disentanglement (NeurIPS 2025 · similarity 0.91)
- Hierarchical Frequency Tagging Probe (HFTP): A Unified Approach to Investigate Syntactic Structure Representations in Large Language Models and the Human Brain (NeurIPS 2025 · similarity 0.89)
- Improving Semantic Understanding in Speech Language Models via Brain-tuning (ICLR 2025 · similarity 0.88)
- A Polar coordinate system represents syntax in large language models (NeurIPS 2024 · similarity 0.87)
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.