arXiv cs.CL· Xiutian Zhao, Luqi Sun, Bj\"orn Schuller, Berrak Sisman·· 12 小时前AI 评分24
大型音频语言模型的认知障碍判断：因果影响集中，诊断信息分散
Concentrated Causal Influence, Distributed Diagnostic Information: Cognitive Impairment Decisions in Large Audio-Language Models
AI 导读
研究通过 CI-contrast 神经元（CICN）考察大型音频语言模型（LALM）判断认知障碍时的内部机制：在八个开源 LALM 和 Pitt、PROCESS-2 两个语料上，五个合格模型中有三个在消融少量 CICN 后出现大幅类别定向的预测偏移，且语音识别的掩码同样控制纯文本决策。
正文 · 原文
Abstract:Large audio-language models (LALMs) can assess cognitive impairment (CI) from speech, yet their answers reveal neither which internal components control the decision nor how much CI information is accessible in their representations. We investigate both questions through CI-contrast neurons (CICNs), selected by class-contrastive activation frequency in decoder MLPs. Across eight open LALMs and two corpora (Pitt and PROCESS-2), we causally test CICNs against size- and layer-matched random masks and activity-matched controls. In three of the five models that meet our eligibility criteria, ablation of small CICN sets causes large, class-directed prediction shifts; speech-identified masks also control transcript-only decisions, and three of four primary Pitt masks retain specificity on PROCESS-2. Position-restricted interventions locate the influence before the final answer position, most strongly after the speech or transcript evidence under letter-labeled options, largely as a decision-direction shift. Beyond causal localization, supervised linear readouts of frozen CICN activations achieve 0.92--0.97 AUC within Pitt, versus 0.67--0.84 for the models' direct answers. Matched random and all-neuron readouts reach comparable performance; top-ranked CICNs show only an exploratory advantage at very small unit budgets. These results distinguish concentrated causal control over output decisions from broadly distributed diagnostic information for CI detection in LALMs.
| Subjects: | Audio and Speech Processing (eess.AS); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.32061 [eess.AS] |
| (or arXiv:2610.32061v1 [eess.AS] for this version) | |
| https://doi.org/10.48550/arXiv.2610.32061 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xiutian Zhao [view email]
[v1]
Fri, 23 Oct 2026 01:56:25 UTC (262 KB)
来源：arXiv cs.CL · arxiv.org