# Do VLMs Have Bad Eyes? Diagnosing Compositional Failures via Mechanistic Interpretability

Source: [Do VLMs Have Bad Eyes? Diagnosing Compositional Failures via Mechanistic Interpretability](<https://www.sota2.com/research/paper/68fa7add1e209131cea85949>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-27T07:00:40.261Z

Authors: Ashwath Vaithinathan Aravindan, Abha Jha, Mihir Kulkarni

Published: 2025-08-20T01:15:28.000Z

## Abstract

Vision-Language Models (VLMs) have shown remarkable performance in integrating visual and textual information for tasks such as image captioning and visual question answering. However, these models struggle with compositional generalization and object binding, which limit their ability to handle novel combinations of objects and their attributes. Our work explores the root causes of these failures using mechanistic interpretability techniques. We show evidence that individual neurons in the MLP layers of CLIP's vision encoder represent multiple features, and this "superposition" directly hinders its compositional feature representation which consequently affects compositional reasoning and object binding capabilities. We hope this study will serve as an initial step toward uncovering the mechanistic roots of compositional failures in VLMs. The code and supporting results can be found https://github.com/Mystic-Slice/Do-VLMs-Have-Bad-Eyes.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2508.16652>)

[PDF](<https://arxiv.org/pdf/2508.16652.pdf>)
