The Dog the Cat Chased Stumped the Model: Measuring When Language Models Abandon Structure for Shortcuts

When language models correctly parse "The cat that the dog chased meowed," are they analyzing syntax or simply familiar with dogs chasing cats? Despite extensive benchmarking, we lack methods to distinguish structural understanding from semantic pattern matching. We introduce CenterBench, a dataset of 9,720 comprehension questions on center-embedded sentences (like "The cat [that the dog chased] meowed") where relative clauses nest recursively, creating processing demands from simple to deeply nested structures. Each sentence has a syntactically identical but semantically implausible counterpart (e.g., mailmen prescribe medicine, doctors deliver mail) and six comprehension questions testing surface understanding, syntactic dependencies, and causal reasoning. Testing six models reveals that performance gaps between plausible and implausible sentences widen systematically with complexity, with models showing median gaps up to 26.8 percentage points, quantifying when they abandon structural analysis for semantic associations. Notably, semantic plausibility harms performance on questions about resulting actions, where following causal relationships matters more than semantic coherence. Reasoning models improve accuracy but their traces show semantic shortcuts, overthinking, and answer refusal. Unlike models whose plausibility advantage systematically widens with complexity, humans shows variable semantic effects. CenterBench provides the first framework to identify when models shift from structural analysis to pattern matching.

Personality Matters: User Traits Predict LLM Preferences in Multi-Turn Collaborative Tasks

As Large Language Models (LLMs) increasingly integrate into everyday workflows, where users shape outcomes through multi-turn collaboration, a critical question emerges: do users with different personality traits systematically prefer certain LLMs over others? We conducted a study with 32 participants evenly distributed across four Keirsey personality types, evaluating their interactions with GPT-4 and Claude 3.5 across four collaborative tasks: data analysis, creative writing, information retrieval, and writing assistance. Results revealed significant personality-driven preferences: Rationals strongly preferred GPT-4, particularly for goal-oriented tasks, while idealists favored Claude 3.5, especially for creative and analytical tasks. Other personality types showed task-dependent preferences. Sentiment analysis of qualitative feedback confirmed these patterns. Notably, aggregate helpfulness ratings were similar across models, showing how personality-based analysis reveals LLM differences that traditional evaluations miss.

BriNet: Towards Bridging the Intra-class and Inter-class Gaps in One-Shot Segmentation

Few-shot segmentation focuses on the generalization of models to segment unseen object instances with limited training samples. Although tremendous improvements have been achieved, existing methods are still constrained by two factors. (1) The information interaction between query and support images is not adequate, leaving intra-class gap. (2) The object categories at the training and inference stages have no overlap, leaving the inter-class gap. Thus, we propose a framework, BriNet, to bridge these gaps. First, more information interactions are encouraged between the extracted features of the query and support images, i.e., using an Information Exchange Module to emphasize the common objects. Furthermore, to precisely localize the query objects, we design a multi-path fine-grained strategy which is able to make better use of the support feature representations. Second, a new online refinement strategy is proposed to help the trained model adapt to unseen classes, achieved by switching the roles of the query and the support images at the inference stage. The effectiveness of our framework is demonstrated by experimental results, which outperforms other competitive methods and leads to a new state-of-the-art on both PASCAL VOC and MSCOCO dataset.