aiqbench analysis
This study investigates how assigning personas to a multimodal language model alters its internal representational geometry versus its actual behavioral outputs, revealing a significant dissociation between the two. The research demonstrates that while prompt framing strongly reshapes internal representations in later layers, persona content has a more direct impact on final judgments, challenging the assumption that representational distance directly predicts functional relevance. These findings provide critical insights into the architecture of artificial cognitive systems and how simulated perspectives are structurally encoded within large models.
The Geometry of Simulated Perspectives: Dissociating Representation and Behaviour in a Multimodal Language Model
arXiv:2611.08262v1 Announce Type: new Abstract: Multimodal language models can simulate different observers, yet it remains unclear how assigned perspectives reshape their internal representations and whether these changes govern their judgements. Here, we investigate the representational geometry of persona-conditioned visual interestingness in Qwen3-VL-8B-Instruct. Across 100 images and 53 prompting conditions, we systematically vary persona content and the linguistic framing used to introduce it, while recording behavioural ratings and layer-wise hidden states. We observe a systematic reorganisation of representational geometry across processing depth: early layers primarily encode image identity, whereas persona-specific structures emerge in later layers. Strikingly, prompt framing reshapes internal representations more strongly than persona identity, whereas persona content has a greater impact on behavioural judgements. This dissociation is further illustrated by minimal identity statements that alter representations even without changing responses. These findings demonstrate that representational distance does not necessarily reflect behavioural relevance. More broadly, they reveal how simulated perspectives emerge as distinct structures within the representational space of artificial cognitive systems, highlighting a fundamental distinction between the geometry of internal representations and their functional consequences.
Abstract:Multimodal language models can simulate different observers, yet it remains unclear how assigned perspectives reshape their internal representations and whether these changes govern their judgements. Here, we investigate the representational geometry of persona-conditioned visual interestingness in Qwen3-VL-8B-Instruct. Across 100 images and 53 prompting conditions, we systematically vary persona content and the linguistic framing used to introduce it, while recording behavioural ratings and layer-wise hidden states. We observe a systematic reorganisation of representational geometry across processing depth: early layers primarily encode image identity, whereas persona-specific structures emerge in later layers. Strikingly, prompt framing reshapes internal representations more strongly than persona identity, whereas persona content has a greater impact on behavioural judgements. This dissociation is further illustrated by minimal identity statements that alter representations even without changing responses. These findings demonstrate that representational distance does not necessarily reflect behavioural relevance. More broadly, they reveal how simulated perspectives emerge as distinct structures within the representational space of artificial cognitive systems, highlighting a fundamental distinction between the geometry of internal representations and their functional consequences.
30 pages, 10 figures, 2 tables. Code: this https URL data: this https URL
Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2611.08262 [cs.AI]
(or arXiv:2611.08262v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2611.08262 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 289 words
Related coverage
Lens: Bringing the Right Semantic Perspective into Focus for Training-Free Multimodal Representation Learning
arXiv · Sep 18, 2026
Order-Invariant Answers, Order-Sensitive Representations in Mathematical Reasoning
arXiv · Sep 24, 2026
Mitigating Bias in Large Vision-Language Models via Counterfactual Ensemble Decoding
arXiv · Aug 25, 2026
Complementary Supervised and Self-Supervised Representations for Out-of-Distribution Graph Learning
arXiv · Oct 7, 2026