@inproceedings{roy-malhotra-2026-modusponens,
    title = "{M}odus{P}onens at {S}em{E}val-2026 Task 11: Breaking the Plausibility Trap in {LLM}s via Conflict-Aware Ensembling",
    author = "Roy, Soumyajit  and
      Malhotra, Manav",
    editor = "Kochmar, Ekaterina  and
      Ghosh, Debanjan  and
      North, Kai  and
      Komachi, Mamoru",
    booktitle = "Proceedings of the 20th {I}nternational {W}orkshop on {S}emantic {E}valuation (2026)",
    month = jul,
    year = "2026",
    address = "San Diego, California, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.semeval-1.10/",
    pages = "65--71",
    ISBN = "979-8-89176-414-9",
    abstract = "Large Language Models (LLMs) often struggle to disentangle formal logical validity from real-world plausibility, a phenomenon known as the ``belief bias''. This paper describes our submission to SemEval-2026 Task 11. We frame the task as a calibration problem between ``System 1'' (heuristic) and ``System 2'' (logical) thinking. Our experiments reveal that standard neuro-symbolic interventions, such as Structural Chain-of-Thought (CoT) and Nonsense Augmentation, degrade performance in low-resource regimes due to an ``abstraction penalty''. Instead, we propose a Conflict-Aware Logit Ensemble. We fine-tune two variations of Qwen-2.5-14B: a standard ``Believer'' model and a bias-hardened ``Skeptic'' model trained on oversampled conflict data. By ensembling their logits via soft-voting, we achieve a Pareto-optimal balance, reducing the Total Content Effect (TCE) to 3.21 while maintaining an overall accuracy of 94.27{\%}, resulting in a Combined Score of 39.09."
}