Research map / Label-efficient and robust learning / Recognition and long-tail
Probing Word Syntactic Representations in the Brain by a Feature Elimination Method
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 |
|---|---|---|---|
| Use MVNP rather than INLP to remove POS | BERT-base word embeddings · POS classification | accuracy: 16.55 ± 2.03 -> 21.76 ± 2.68 | Hurt |
| Use MVNP rather than INLP to remove NE | BERT-base word embeddings · NE classification | accuracy: 5.02 ± 0.89 -> 12.30 ± 6.38 | Hurt |
| Use MVNP rather than INLP to remove SR | BERT-base word embeddings · SR classification | accuracy: 25.20 ± 0.70 -> 23.40 ± 1.93 | Helped |
| Use MVNP rather than INLP to remove DEP | BERT-base word embeddings · DEP classification | accuracy: 13.39 ± 0.38 -> 13.96 ± 1.80 | Hurt |
| Use MVNP rather than INLP to remove POS | ELMo word embeddings · POS classification | accuracy: 11.12 ± 0.61 -> 21.70 ± 3.35 | Hurt |
| Use MVNP rather than INLP to remove NE | ELMo word embeddings · NE classification | accuracy: 12.09 ± 0.71 -> 13.79 ± 8.77 | Hurt |
| Use MVNP rather than INLP to remove SR | ELMo word embeddings · SR classification | accuracy: 17.21 ± 1.85 -> 22.03 ± 1.52 | Hurt |
| Use MVNP rather than INLP to remove DEP | ELMo word embeddings · DEP classification | accuracy: 17.76 ± 0.28 -> 15.98 ± 1.64 | Helped |
| Remove a syntactic feature using MVNP | BERT-base word embeddings · SCWS | Pearson correlation: 0.6993 -> 0.6870; 0.6820; 0.6775; 0.6875 | Hurt |
Closest papers by meaning
- Probing Brain Activation Patterns by Dissociating Semantics and Syntax in Sentences (AAAI 2020 · similarity 0.92)
- Hierarchical Frequency Tagging Probe (HFTP): A Unified Approach to Investigate Syntactic Structure Representations in Large Language Models and the Human Brain (NeurIPS 2025 · similarity 0.88)
- Far from the Shallow: Brain-Predictive Reasoning Embedding through Residual Disentanglement (NeurIPS 2025 · similarity 0.86)
- A Polar coordinate system represents syntax in large language models (NeurIPS 2024 · similarity 0.86)
- fMRI predictors based on language models of increasing complexity recover brain left lateralization (NeurIPS 2024 · similarity 0.84)
- TopoLM: brain-like spatio-functional organization in a topographic language model (ICLR 2025 · similarity 0.84)
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.