What is necessary (and not) for rule learning?

Geambasu, A. 1, 2 , Spierings, M. J. 3, 2 , Levelt, C. C. 1, 2 & ten Cate, C. 3, 2

1 Leiden University Centre for Linguistics
2 Leiden Institute for Brain and Cognition
3 Institute of Biology Leiden

In order to understand the conditions under which adults are able to extract XXY/XYX-type patterns and generalize them to novel input, we conducted an artificial grammar-learning task in the style of Marcus et al. (1999). Three participant groups were exposed via reinforced training and tested in a two-alternative forced choice task. Another eight groups were exposed via passive familiarization and tested in a yes/no paradigm. In each group we systematically varied the number of familiarization items, test items and experimental instructions. Our results indicate that adults do not readily learn a rule or generalize it without feedback, unless they are explicitly instructed to what they should attend. We also find that amount of variety during familiarization has surprisingly little effect on the ability to generalize. We discuss the implications of our results for understanding the biases adults have when they are faced with the task of implicitly abstracting rules from a speech stream.