---
title: "Researchers Crack Open GPT-OSS-20b to Find Hidden Symbol Systems Inside"
subtitle: "An 8-year interpretability project shows that LLM representations can be closely approximated by symbolic role-filler structures, enabling precise behavioral edits."
url: "https://alphasignal.ai/news/researchers-crack-open-gpt-oss-20b-to-find-hidden-symbol-systems-inside"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-09-03T07:03:04.222Z"
updated: "2026-09-03T07:05:44.178Z"
type: "Paper"
topics:
  - "Llms"
  - "Reasoning"
subtopics:
  - "Hallucinations"
  - "Structured Output"
upvotes: 11
source_url: "https://arxiv.org/abs/2608.29530"
comments: 0
truncated: true
---

# Researchers Crack Open GPT-OSS-20b to Find Hidden Symbol Systems Inside

> An 8-year interpretability project shows that LLM representations can be closely approximated by symbolic role-filler structures, enabling precise behavioral edits.

Modern language models operate on continuous vectors yet excel at tasks that traditionally required discrete symbol manipulation: arithmetic, logic, code execution, syntax. A new paper from R. Thomas McCoy, Paul Soulos, Tal Linzen, and Paul Smolensky offers a concrete resolution to that apparent paradox. Inside those messy vectors, neural networks are quietly building something that looks like a symbol system, and you can swap the real vectors out for a closed-form symbolic equation without breaking the model.

The [paper](https://arxiv.org/abs/2608.29530) is the culmination of an eight-year project. Its central claim: the vector representations of a wide range of neural networks can be closely approximated by symbolic structures, meaning you can replace a network's entire representation-generating process with a closed-form equation instantiating a symbolic structure while the network's behavior remains largely unchanged. That holds not just for toy recurrent nets but for seven modern LLMs including Gemma-3-27b, GPT-OSS-20b, Qwen3-14b, OLMo-2-13B, and Llama-3.1-8b.

## The binding problem vectors were never supposed to solve

Symbolic AI has a long history of representing sentences as trees, math as expressions, and logic as formulas. Neural networks encode everything as points in high-dimensional space instead. Addition is order-insensitive, so if you just sum embeddings, `cats chase dogs` and `dogs chase cats` look identical. Cognitive scientists call this the binding problem, and it's why the Linear Representation Hypothesis, the idea that concepts are directions in vector space, can't fully explain how models handle structure.

The authors' hypothesis is that networks solve this using Tensor Product Representations (TPRs), a formalism introduced by Smolensky in 1990\. In a TPR, a structure is decomposed into fillers (the content, like the word "cats") paired with roles (the position, like "subject"). Each filler-role pair is combined via tensor product and the pairs are summed, producing a single vector that preserves who-goes-where.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [The Emergent Symbolic Structure of Artificial Neural Networks](https://arxiv.org/abs/2608.29530)
- [DISCOVER GitHub repository](https://github.com/tommccoy1/discover/)
- [HTML version of paper on arXiv](https://arxiv.org/html/2608.29530)