Understanding person acquisition using an interactive activation and competition network

Face perception is one of the most developed visual skills that humans display, and recent work has attempted to examine the mechanisms involved in face perception through noting how neural networks achieve the same performance. The purpose of the present paper is to extend this approach to look not just at human face recognition, but also at human face acquisition. Experiment 1 presents empirical data to describe the acquisition over time of appropriate representations for newly encountered faces. These results are compared with those of Simulation 1, in which a modified IAC network capable of modelling the acquisition process is generated. Experiment 2 and Simulation 2 explore the mechanisms of learning further, and it is demonstrated that the acquisition of a set of associated new facts is easier than the acquisition of individual facts in isolation of one another. This is explained in terms of the advantage gained from additional inputs and mutual reinforcement of developing links within an interactive neural network system.