To Copy or Not to Copy: Copying Is Easier to Induce Than Recall

Paper in proceeding, 2026

Language models used in retrieval-augmented settings choose between parametric knowledge stored in their weights and contextual information in the prompt. This work presents a mechanistic study of that choice: using an approach based on steering vectors, we carry out interventions at selected layers and token spans, and we are able to toggle the model’s behavior in two directions: Copy→Recall (suppressing context use) and Recall→Copy (inducing the model to copy any token from the context). Experiments on three architectures (decoder-only and encoder/decoder) and two open-domain QA benchmarks show consistent behavior shifts under moderate scaling while monitoring accuracy and fluency. Mechanistic analyses of attention routing, MLP contributions, and layer-wise probability trajectories reveal an asymmetry: inducing copying is an easy "reactivation" process that can be triggered at different locations in the input, while restoring recall is a "suppression" process that is more fragile and strongly tied to object-token interventions.