Statistical Learning of Novel Objects Across Differing Levels of Abstractions

Abstract

People learn something new every day, whether it is the fact that Sydney is not the capital of Australia or that among multiple levels of abstraction, statistical learning may be constrained with certain biases. Research has suggested that statistical learning can be either largely unconstrained, or constrained to certain types of patterns and regularities. This project studies whether or not statistical learning is constrained, within an environment which possesses multiple levels of abstraction. The Bloop novel object stimuli are used to test category and object-specific level learning independently in Experiments 1 and 2, and then together simultaneously in Experiment 3. Through these three experiments, evidence points to statistical learning potentially being constrained when multiple levels of abstraction are present.