@inproceedings{nwadike-etal-2025-library,
    title = "Library-Like Behavior In Language Models is Enhanced by Self-Referencing Causal Cycles",
    author = "Nwadike, Munachiso S  and
      Iklassov, Zangir  and
      Aremu, Toluwani  and
      Hiraoka, Tatsuya  and
      Heinzerling, Benjamin  and
      Bojkovic, Velibor  and
      AlQuabeh, Hilal  and
      Tak{\'a}{\v{c}}, Martin  and
      Inui, Kentaro",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.acl-long.1232/",
    pages = "25365--25377",
    ISBN = "979-8-89176-251-0",
    abstract = "We introduce the concept of the $\textit{self-referencing causal cycle}$ (abbreviated $\textit{ReCall}$ ){---}a mechanism that enables large language models (LLMs) to bypass the limitations of unidirectional causality, which underlies a phenomenon known as the $\textit{reversal curse}$. When an LLM is prompted with sequential data, it often fails to recall preceding context. For example, when we ask an LLM to recall the line preceding ``O say does that star-spangled banner yet wave'' in the U.S. National Anthem, it often fails to correctly return ``Gave proof through the night that our flag was still there''{---}this is due to the reversal curse. It occurs because language models such as ChatGPT and Llama generate text based on preceding tokens, requiring facts to be learned and reproduced in a consistent token order. While the reversal curse is often viewed as a limitation, we offer evidence of an alternative view: it is not always an obstacle in practice. We find that $\textit{ReCall}$ is driven by what we designate as $\textit{cycle tokens}${---}sequences that connect different parts of the training data, enabling recall of preceding tokens from succeeding ones. Through rigorous probabilistic formalization and controlled experiments, we demonstrate how the cycles they induce influence a model{'}s ability to reproduce information. To facilitate reproducibility, we provide our code and experimental details at https://anonymous.4open.science/r/remember-B0B8/."
}