Post

🧠 To reason over text and track entities, we find that language models use three types of 'pointers'! They were thought to rely only on a positional one—but when many entities appear, that system breaks down. Our new paper shows what these pointers are and how they interact 👇

A key part of in-context reasoning is the ability to bind entities for tracking and retrieval. When reading “Holly loves Michael, Jim loves Pam”, the model must bind Holly↔Michael to answer “Who loves Michael?” We show that this binding relies on three mechanisms. 1/

Prior work identified only a positional mechanism, where the model tracks entities by position: here retrieving the subject from the first clause "Holly". We show this isn’t sufficient—the positional signal is strong at the edges of context but weak and diffuse in the middle. 2/

To compensate for this, LMs use two additional mechanisms. The first is *lexical*, where the LM retrieves the subject next to "Michael". It does this by copying the lexical contents of "Holly" to "Michael", binding them together. 3/

Then we have the *reflexive* mechanism, which retrieves exactly the token "Holly". This happens through a self-referential pointer originating from the "Holly" token and pointing back to it. This pointer gets copied to the "Michael" token, binding the two entities together. 4/

We show this through extensive use of interchange interventions, evaluating over 10 binding tasks and 9 models (Gemma/Qwen/Llama 2B-72B params). Across all models, we find a remarkably consistent reliance on these three specific mechanisms and how they interact. 5/

We leverage these insights to build a causal model combining all three mechanisms, predicting next-token distributions with 95% agreement. We model the positional term as a Gaussian with shifting std, and the other two as one-hot distributions with position-based weights. 6/

Finally, we evaluate our model over more natural and increasingly long tasks, showing that the ‘lost-in-the-middle’ effect might be explained mechanistically by a weakening lexical signal alongside an increasingly noisy positional one. 7/

Overall, we show that LMs retrieve entities not through a single positional mechanism, but a mixture of three: positional, lexical, and reflexive. Understanding these mechanisms helps explain both the strengths and limits of LLMs, and how they reason in context. 8/