When to Call an Apple Red: Humans Follow Introspective Rules, VLMs Don't

TL;DR

This paper introduces the GCA dataset to evaluate whether vision-language models and humans follow their own reasoning rules, revealing that models often violate their introspective rules while humans generally adhere to theirs.

Contribution

The study presents a new benchmark dataset, GCA, and provides empirical evidence that VLMs miscalibrate their self-knowledge, unlike humans who remain faithful to their rules.

Findings

VLMs violate their own rules in nearly 60% of cases.

Humans mostly follow their stated rules, with violations due to overestimating coverage.

World-knowledge priors reduce faithfulness in VLMs, unlike in humans.

Abstract

Understanding when Vision-Language Models (VLMs) will behave unexpectedly, whether models can reliably predict their own behavior, and if models adhere to their introspective reasoning are central challenges for trustworthy deployment. To study this, we introduce the Graded Color Attribution (GCA) dataset, a controlled benchmark designed to elicit decision rules and evaluate participant faithfulness to these rules. GCA consists of line drawings that vary pixel-level color coverage across three conditions: world-knowledge recolorings, counterfactual recolorings, and shapes with no color priors. Using GCA, both VLMs and human participants establish a threshold: the minimum percentage of pixels of a given color an object must have to receive that color label. We then compare these rules with their subsequent color attribution decisions. Our findings reveal that models systematically violate their own introspective rules. For example, GPT-5-mini violates its stated introspection rules in nearly 60\% of cases on objects with strong color priors. Human participants remain faithful to their stated rules, with any apparent violations being explained by a well-documented tendency to overestimate color coverage. In contrast, we find that VLMs are excellent estimators of color coverage, yet blatantly contradict their own reasoning in their final responses. Across all models and strategies for eliciting introspective rules, world-knowledge priors systematically degrade faithfulness in ways that do not mirror human cognition. Our findings challenge the view that VLM reasoning failures are difficulty-driven and suggest that VLM introspective self-knowledge is miscalibrated, with direct implications for high-stakes deployment.