Abstract for Psychonomic Society Annual Meeting, Boston 2009
How does perceptual similarity influence the learning of name-face associations? For answers, we exploited (i) realistic, synthetic faces, and (ii) monosyllabic Chinese names. The perceptual similarity space for each stimulus class was defined by multidimensional scaling. Then associative recognition was measured with stimulus sets whose similarities were manipulated parametrically. In alternating study and test blocks, subjects studied a fixed set of face-name pairs, and were tested with preserved- and rearranged-pairs. Over successive trial blocks, correct recognitions of preserved face-name pairs increased, while false recognitions of rearranged pairs decreased. Face- and name-similarity each strongly influenced associative recognition.These similarity effects were accommodated within a novel, hybrid model in which an Interactive Activation and Competition (IAC) network was integrated with NEMo, our global matching framework for recognition. Learning of face-name associations reflects both an increasingly precise representation of individual stimuli, and a sharpening of activation within the associative network.