Autoregressive large language models (LLMs) process text token-by-token, yet the human language system operates over multi-word units. We ask whether aggregating LLM representations at the phrase level yields a closer correspondence to human reading behavior and language cortex than the default word-level representations, and whether phrase-segmentation fine-tuning amplifies this correspondence. Using Meta-Llama-3.1-8B (base and fine-tuned), we provide three converging lines of evidence. First, phrase-level attention features predict regressive eye-saccade patterns more closely than word-level features; a partial correlation analysis with a shuffled-boundary control indicates that this is not solely an aggregation artifact and that linguistic chunk boundaries explain unique variance beyond word-level attention. Second, fMRI encoding analyses show that fine-tuning selectively improves phrase encoding in left superior temporal gyrus and inferior frontal gyrus, with no improvement for word representations. Third, representational similarity analysis confirms a phrase-specific gain in model-brain geometric alignment. These results identify phrase-level representation as a critical granularity for LLM–human correspondence and suggest that targeted training can model human-like compositional processing, linking computational representations to hierarchical theories of language.

Naturalistic language comprehension often involves interpretations that go beyond literal meaning. In continuous narratives, literal and non-literal meanings are tightly intertwined, making them difficult to distinguish computationally. Here, we combined literal sentence representations and human-annotated non-literal interpretations for model-brain alignment. Using fMRI data recorded during passive listening to the Chinese version of The Little Prince, we annotated sentences containing non-literal meaning with human-written interpretations of their implied meaning. We then derived the literal and non-literal representations from LLaMA3.1-8B and evaluated their correspondence with neural activity using whole-brain encoding models. Literal representations aligned strongly with left-lateralized frontotemporal regions, whereas non-literal interpretations showed broader right-hemisphere involvement. Combining the two further improved encoding performance in the bilateral temporal and dorsal frontal cortices, suggesting that naturalistic comprehension engages complementary levels of meaning.

Large Reasoning Models (LRMs) have achieved remarkable performance across diverse domains, yet their decision-making under conflicting objectives remains insufficiently understood. This work investigates how LRMs respond to harmful queries when confronted with two categories of conflicts: internal conflicts that pit alignment values against each other and dilemmas, which impose mutually contradictory choices, including sacrificial, duress, agent-centered, and social forms. Using over 1,300 prompts across five benchmarks, we evaluate three representative LRMs - Llama-3.1-Nemotron-8B, QwQ-32B, and DeepSeek R1 - and find that conflicts significantly increase attack success rates, even under single-round non-narrative queries without sophisticated auto-attack techniques. Our findings reveal through layerwise and neuron-level analyses that safety-related and functional representations shift and overlap under conflict, interfering with safety-aligned behavior. This study highlights the need for deeper alignment strategies to ensure the robustness and trustworthiness of next-generation reasoning models. Our code is available at https://github.com/DataArcTech/ConflictHarm. Warning: This paper contains inappropriate, offensive and harmful content.

Syntactic movement is a core concept in generative linguistics to account for word-order variation and long-distance dependencies, but its psychological and neurobiological status remains debated. Here, we test the neural reality of movement in English and Chinese by correlating brain activity during naturalistic listening with syntactic node counts, traces and word embeddings derived from X-bar style tree annotations. We find that deep structure significantly predicts neural responses in English but not in Chinese, providing partial support for movement-based accounts while revealing clear cross-linguistic differences.