a conceptual and computational model
Quan Wang, Pramod Chandrashekharia, Gabriele Spina · 2011
Abstract-We propose a new theory explaining the Camiliarityto-novelty shift in inCant habituation. In our account, inCants' interest in a stimulus is related to their learning progress, i.e. the improvement oC an internal model oC the stimulus. Specifically, we propose inCants preCer the stimulus Cor which its current learning progress is maximal. We also propose a new algorithm called Selective Learning SelC Organizing Map (SL-SOM), a biologically inspired modification to SOM, exhibiting Camiliarity-to-novelty shiCt. Using this algorithm we present experiments on a robotic platCorm. I. INTRODUCTION Familiarity-to-novelty shift is a commonly observed phenomenon in infant habituation experiments. Habituation and its derived paradigms are important tools to understand cognition of preverbal infants. In infant habituation experiments, participants undergo a training phase of habituation in which the images or dynamic displays of objects are shown repeatedly. Then they are tested on familiar or novel stimuli. According to novelty preference theory