Information transmission between individuals through social learning is a foundational component of cultural evolution. However, how this transmission occurs is still debated. The copying account draws parallels with biological mechanisms for genetic inheritance, arguing that learners copy what they observe as they see it. On the other hand, the reconstruction account argues that learners recreate only what is relevant and reconstruct it using pragmatic inference, environmental and contextual cues. Distinguishing these two accounts empirically using typical transmission chain studies is difficult because they generate overlapping predictions. In this study we present an innovative methodological approach that generates different predictions of these accounts by manipulating the task context between model and learner in a transmission episode. We provide an empirical proof-of-concept showing that, when a model introduces embedded signals to their actions that are not intended to be transmitted, learners’ reproductions are more consistent with a process of reconstruction than copying.