Statistical word segmentation and target detection tasks require different mechanisms
Meili Luo, Ran Cao, Felix Hao Wang · 2024
Statistical learning is a powerful mechanism that can support a variety of learning tasks. Many theories have assumed a single mechanism for statistical learning across different tasks, where a unitary mechanism is supposed to explain results from various studies, even across different modalities. In this study, we studied auditory statistical learning by comparing two different experimental paradigms, target detection and word segmentation, and examined if different mechanisms are required to explain results from the two paradigms. Previous work using the word segmentation paradigm suggested that learning is better with sequences containing uniform-length words than with sequences containing mixed-length words. If the same mechanism supports the target detection task, the same results are predicted. However, while learning was successful in both Experiments 1 and 2 with the target detection paradigm, the effect was larger in the mixed condition than in the uniform condition. We further replicated the uniform condition advantage in the word segmentation paradigm in Experiment 3. Thus, we hypothesized that the target detection paradigm required a different mechanism from those in word segmentation. To understand these mechanisms, we proposed both theoretical analyses and a computational model to simulate results from the target detection paradigm. We found that a prediction mechanism, rather than clustering, could explain the data from target detection. Crucially, this mechanism can produce facilitation effects without performing segmentation. We discuss both the theoretical and empirical reasons why the target detection and word segmentation paradigm might engage different processes, and how these findings contribute to our understanding of statistical word segmentation.