TEMPER: Testing Emotional Perturbation in Quantitative Reasoning

• URL: http://arxiv.org/abs/2604.07801v1
• Date: Thu, 09 Apr 2026 04:52:18 GMT
• Title: TEMPER: Testing Emotional Perturbation in Quantitative Reasoning
• Abstract summary: We show that emotional framing reduces accuracy by 2-10 percentage points even though all numerical content is preserved.<n>We also show that neutralizing emotional variants recovers most of the lost performance.<n>Non-emotional paraphrases cause no such degradation, implicating emotional content rather than surface-level changes.
• Score: 5.171335396838995
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models are trained and evaluated on quantitative reasoning tasks written in clean, emotionally neutral language. However, real-world queries are often wrapped in frustration, urgency or enthusiasm. Does emotional framing alone degrade reasoning when all numerical content is preserved? To investigate this, a controlled emotion translation framework is developed that rewrites problems into emotional variants while preserving all quantities and relationships. Using this framework, Temper-5400 (5,400 semantically verified emotion--neutral pairs) is constructed across GSM8K, MultiArith, and ARC-Challenge, and evaluated on eighteen models (1B to frontier scale). Two core results emerge: First, emotional framing reduces accuracy by 2-10 percentage points even though all numerical content is preserved. Second, neutralizing emotional variants recovers most of the lost performance, showing both that the degradation is tied to emotional style rather than content corruption and that neutralization can serve as a lightweight inference-time mitigation. Non-emotional paraphrases cause no such degradation, implicating emotional content rather than surface-level changes. Beyond emotion specifically, the benchmark construction procedure provides a general framework for controlled stylistic translation and robustness evaluation.

Related papers

• Emotion Entanglement and Bayesian Inference for Multi-Dimensional Emotion Understanding [40.766332191425]
  EmoScene is a theory-grounded benchmark of 4,731 context-rich scenarios annotated with an 8-dimensional emotion vector.<n>We propose an entanglement-aware Bayesian inference framework that incorporates emotion co-occurrence statistics to perform joint posterior inference over the emotion vector.
  arXiv  Detail & Related papers  (2026-04-01T12:27:04Z)
• FACE-net: Factual Calibration and Emotion Augmentation for Retrieval-enhanced Emotional Video Captioning [81.33341786837974]
  Emotional Video Captioning (EVC) is an emerging task, which aims to describe factual content with the intrinsic emotions expressed in videos.<n>We propose a retrieval-enhanced framework with FActual and Emotion augmentation (FACE-net)<n>FACE-net collaboratively mines factual-emotional semantics and provides adaptive and accurate guidance for generation.
  arXiv  Detail & Related papers  (2026-03-18T07:53:15Z)
• Emotion is Not Just a Label: Latent Emotional Factors in LLM Processing [4.733799057798266]
  We study emotion as a latent factor that shapes how models attend to and reason over text.<n>We analyze how emotional tone systematically alters attention geometry in transformer models.<n>We propose an emotional regularization framework that constrains emotion-conditioned representational drift during training.
  arXiv  Detail & Related papers  (2026-03-10T05:23:18Z)
• Do LLMs "Feel"? Emotion Circuits Discovery and Control [54.57583855608979]
  We study the internal mechanisms that give rise to emotional expression and in controlling emotions in generated text.<n>This is the first systematic study to uncover and validate emotion circuits in large language models.
  arXiv  Detail & Related papers  (2025-10-13T12:24:24Z)
• KEVER^2: Knowledge-Enhanced Visual Emotion Reasoning and Retrieval [35.77379981826482]
  We propose textbfK-EVERtextsuperscript2, a knowledge-enhanced framework for emotion reasoning and retrieval.<n>Our approach introduces a semantically structured formulation of visual emotion cues and integrates external affective knowledge through multimodal alignment.<n>We validate our framework on three representative benchmarks, Emotion6, EmoSet, and M-Disaster, covering social media imagery, human-centric scenes, and disaster contexts.
  arXiv  Detail & Related papers  (2025-05-30T08:33:32Z)
• UDDETTS: Unifying Discrete and Dimensional Emotions for Controllable Emotional Text-to-Speech [61.989360995528905]
  We propose UDDETTS, a universal framework unifying discrete and dimensional emotions for controllable emotional TTS.<n>This model introduces the interpretable Arousal-Dominance-Valence (ADV) space for dimensional emotion description and supports emotion control driven by either discrete emotion labels or nonlinearly quantified ADV values.<n>Experiments show that UDDETTS achieves linear emotion control along three interpretable dimensions, and exhibits superior end-to-end emotional speech synthesis capabilities.
  arXiv  Detail & Related papers  (2025-05-15T12:57:19Z)
• Fearful Falcons and Angry Llamas: Emotion Category Annotations of Arguments by Humans and LLMs [9.088303226909277]
  We crowdsource subjective annotations of emotion categories in a German argument corpus and evaluate automatic labeling methods.<n>We find that emotion categories enhance the prediction of emotionality in arguments.<n>Across all prompt settings and models, automatic predictions show a high recall but low precision for predicting anger and fear.
  arXiv  Detail & Related papers  (2024-12-20T15:41:47Z)
• Emotion Rendering for Conversational Speech Synthesis with Heterogeneous Graph-Based Context Modeling [50.99252242917458]
  Conversational Speech Synthesis (CSS) aims to accurately express an utterance with the appropriate prosody and emotional inflection within a conversational setting. To address the issue of data scarcity, we meticulously create emotional labels in terms of category and intensity. Our model outperforms the baseline models in understanding and rendering emotions.
  arXiv  Detail & Related papers  (2023-12-19T08:47:50Z)
• Emotion and Sentiment Guided Paraphrasing [3.5027291542274366]
  We introduce a new task of fine-grained emotional paraphrasing along emotion gradients. We reconstruct several widely used paraphrasing datasets by augmenting the input and target texts with their fine-grained emotion labels. We propose a framework for emotion and sentiment guided paraphrasing by leveraging pre-trained language models for conditioned text generation.
  arXiv  Detail & Related papers  (2023-06-08T20:59:40Z)
• Emotion Intensity and its Control for Emotional Voice Conversion [77.05097999561298]
  Emotional voice conversion (EVC) seeks to convert the emotional state of an utterance while preserving the linguistic content and speaker identity. In this paper, we aim to explicitly characterize and control the intensity of emotion. We propose to disentangle the speaker style from linguistic content and encode the speaker style into a style embedding in a continuous space that forms the prototype of emotion embedding.
  arXiv  Detail & Related papers  (2022-01-10T02:11:25Z)
• Emotion-aware Chat Machine: Automatic Emotional Response Generation for Human-like Emotional Interaction [55.47134146639492]
  This article proposes a unifed end-to-end neural architecture, which is capable of simultaneously encoding the semantics and the emotions in a post. Experiments on real-world data demonstrate that the proposed method outperforms the state-of-the-art methods in terms of both content coherence and emotion appropriateness.
  arXiv  Detail & Related papers  (2021-06-06T06:26:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.