Relation-Based Categories are Easier to Learn than Feature-Based Categories

Relational reasoning is often viewed as the pinnacle of human intelligence. Accordingly, one common viewpoint is that learning categories defined by relational regularities is more difficult than learning categories defined by featural regularities. This view is supported by developmental trends in learning. Studies comparing featural and relational category learning in adults also find a feature advantage, but these studies do not ground featural and relational information in a common perceptual substrate. The current study offers an appropriate comparison between featureand relation-based category learning. Contrary to previous studies, we show how relational learning can be easier. The advantage is attributable to the flexibility of online relational comparisons between a stimulus and a memory representation of a category. Alternative explanations based on difficulties in processing absolute vs. relative stimulus information are ruled out.