ContextBias: Controlled Evaluation of Bias Persistence Under Context Shift in Text-to-Image Models

Hugging Face Daily Papers Papers

Summary

This paper introduces ContextBias and ContextBench to evaluate bias persistence in text-to-image models, finding that bias increases in semantically unrelated contexts.

Text-to-image models learn associations between concepts - in the case of this paper, people's professions, which we refer to as roles - and visual attributes. These associations can underpin many observed forms of stereotypical bias. A key open question in this area is whether these associations are stable or change when visual representations of people in professional roles are placed in different prompted contexts. We introduce ContextBias, a controlled evaluation framework, and ContextBench, a benchmark spanning 92 roles and 1,656 semantically controlled prompts, designed to isolate the effect of contextual variation on role-linked visual representations. Evaluating four state-of-the-art models on 66,240 generated images, we find that placing a role in a semantically unrelated context does not suppress role-linked attributes; instead, cross-role attribute concentration increases (pooled BI +0.047). Demographic cues, characteristic garments, and role-specific tools remain highly prevalent across context-free, related, and unrelated conditions, and are robust to semantic prompt reformulation. Scene composition and camera framing show the greatest context-sensitivity. These findings reveal a form of stereotypical persistence that remains largely invisible to context-free evaluations, highlighting the need for controlled contextual variation in bias benchmarking. Code and dataset: https://huggingface.co/datasets/shaghayegh/ContextBias , https://github.com/Sina-Emami/ContextBias
Original Article
View Cached Full Text

Cached at: 09/01/26, 03:40 PM

Paper page - ContextBias: Controlled Evaluation of Bias Persistence Under Context Shift in Text-to-Image Models

Source: https://huggingface.co/papers/2608.29847 Published on Aug 30

·

Submitted byhttps://huggingface.co/shaghayegh

kollion Sep 1

Abstract

Text-to-image models exhibit persistent role-linked visual stereotypes across varied contexts, with cross-role attribute concentration increasing rather than diminishing under unrelated prompts.

Text-to-image modelslearn associations between concepts - in the case of this paper, people’s professions, which we refer to as roles - and visual attributes. These associations can underpin many observed forms ofstereotypical bias. A key open question in this area is whether these associations are stable or change when visual representations of people in professional roles are placed in different prompted contexts. We introduceContextBias, a controlled evaluation framework, andContextBench, a benchmark spanning 92 roles and 1,656 semantically controlled prompts, designed to isolate the effect of contextual variation on role-linked visual representations. Evaluating four state-of-the-art models on 66,240 generated images, we find that placing a role in a semantically unrelated context does not suppress role-linked attributes; instead,cross-role attribute concentrationincreases (pooled BI +0.047).Demographic cues, characteristic garments, and role-specific tools remain highly prevalent across context-free, related, and unrelated conditions, and are robust to semantic prompt reformulation.Scene compositionandcamera framingshow the greatest context-sensitivity. These findings reveal a form of stereotypical persistence that remains largely invisible to context-free evaluations, highlighting the need for controlled contextual variation in bias benchmarking. Code and dataset: https://huggingface.co/datasets/shaghayegh/ContextBias, https://github.com/Sina-Emami/ContextBias

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2608\.29847

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.29847 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.29847 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.29847 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

The Cost of Context: Mitigating Textual Bias in Multimodal Retrieval-Augmented Generation

arXiv cs.CL

This paper identifies and formalizes 'recorruption' in multimodal RAG, where adding accurate context causes models to abandon correct predictions due to attentional collapse (visual blindness and positional bias). The authors propose BAIR, a parameter-free inference-time framework that restores visual saliency and penalizes textual distractors, improving reliability across medical, fairness, and geospatial benchmarks.