The effects of statistical learning and congruency on the development of multi-modal objects

Abstract

The effects of statistical learning and congruency on multi-modal binding were examined. As pattern acquisition is stronger for within-object than for between-object associations, extending bias from within-object to within-modality was tested, and the statistical learning effect on between-modality learning assessed. Dyson and Ishfaq‟s (2008) paradigm was adapted, with frequency of within- and between-modality associations manipulated (Experiment 1), and frequency and congruency manipulated (Experiment 2). Each experiment comprised baseline (no predictive value), intra-modal (intramodal predictive value), and inter-modal (intermodal predictive value) conditions. Experiment 1 showed faster performance for within-object judgments, and fewer errors on within-object judgments, excluding the inter-modal condition. Experiment 2 replicated this, with cross-experimental analyses showing weak congruency effects. Data showed probability manipulations led mostly to interference on same-modality trials rather than facilitation on different-modality trials, suggesting while frequency of differentmodality associations did not facilitate superior performance, perhaps expectancies of frequent different-modality associations weakened sensitivity to the within-modality bias.