Search papers, labs, and topics across Lattice.
This study investigates whether LLMs can infer latent sequential structure beyond simple action frequencies in controlled two-player and one-player tasks, crucial for accurate behavioral simulation. Using Rock--Paper--Scissors and n-gram continuation, the research separates distribution matching from conditional rule following to assess strategy identification and Markov rule adherence. Findings reveal that extended context does not enhance structure identification, correct recognition doesn't guarantee faithful simulation, and higher-order dependencies significantly impair rule recovery, suggesting apparent fidelity can mask flawed generative processes.
LLMs' apparent behavioral fidelity in sequential tasks can mask fundamentally incorrect generative mechanisms, failing to recover latent structure even with extended context.
Large language models (LLMs) are increasingly used as interactive agents and simulators, yet it remains unclear whether they can recover latent sequential structure beyond surface action frequencies. This distinction is critical for behavioral simulation, where actions are often shaped by prior context rather than marginal frequencies alone. We study this question using controlled two-player Rock--Paper--Scissors interactions and a one-player stochastic n-gram continuation task. Across these experiments, we test whether LLMs can identify latent strategies, follow simple Markov rules, and sustain higher-order conditional dependencies. Our framework separates distribution matching from conditional rule following. Results show that longer context does not improve identification, correct recognition does not ensure faithful simulation, and higher-order dependencies substantially degrade rule recovery. Apparent behavioral fidelity can therefore mask incorrect generative mechanisms.
