arXiv:2609.34091v2 cs.LG

From Perturbation Sensitivity to Sparse Autoencoder Feature Interpretation: An EEG Foundation-Model Case Study

Yucong Cao, Chenqi Li, Tingting Zhu

제출 2026-11-04

원문 초록

Sparse autoencoders (SAEs) split a model's internal activity into many sparse units, or latents, and a latent is often named after a concept when removing that concept from the input changes its activity. We identify a consequential confound in this inference: if removing the concept changes the input more than the comparison edit does, a larger response may reflect a larger edit rather than the concept. Brain-signal (electroencephalography, EEG) models provide a controlled test, because how much an edit changes the EEG input can be measured directly. We measure the SAE response as the change in how often each latent is active, averaged over all latents. Across 27 combinations of model, dataset, and layer, removing the 8--12 Hz alpha band changes this response 7.33 times as much as removing an equally wide 30--34 Hz control band, but it also changes the signal far more. Matching the two edits per trial on squared time-domain distortion lowers the mean ratio to 1.42, with 19 of 27 combinations above one. Matching thus substantially reduces, but does not eliminate, the alpha--control contrast. The residual supports relative sensitivity under this single control, not general spectral specificity or an individual alpha-representing latent. We use an evidence-to-claim map to separate what this comparison supports from stronger claims of specificity and feature semantics.

제출 이력

확인된 버전만 표시합니다. 전체 제출 이력은 arXiv 원문에서 확인하세요.

  1. arXiv:2609.34091v2제출 시각 미확인

함께 살펴볼 논문

공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.

공통 태그로 연결된 공개 논문이 아직 없습니다.

토론

이 논문과 연결된 포럼 글이 아직 없습니다.