Modeling How Naming Experiences Bias Sorting Performance

Abstract

Previous studies assume that language specificity might affect naming patterns, whereas perceptual similarity based sorting might be universal. Study 1 was aimed to examine the hypothesis, but it was found that the correlations of sorting based on perceptual similarity between Chinese and Dutch (or French) are surprisingly lower than previous findings. We argue that perceptual similarity might be also affected by naming experiences. In study 2, we adopted backpropagation neural networks to model the sorting patterns observed in Study 1. During training, the inputs of models were composed of the numeric vectors of names and semantic features. Nevertheless, only semantic features were fed to networks for simulating the sorting results. After training, the simulation outcome of networks showed high correlation with the experimental results. These studies suggest that we should reconsider the involvement of naming experiences for the sorting processes. Keywords: naming, sorting, perceptual similarity, backpropagation neural networks


Back to Table of Contents