Generalizing Psychological Similarity Spaces to Unseen Stimuli
This work addresses a domain-specific challenge in cognitive science for researchers studying conceptual representation, but it is incremental as it builds on existing MDS and ANN methods.
The paper tackled the problem of generalizing psychological similarity spaces to unseen stimuli by proposing a mapping from raw stimuli using artificial neural networks, showing that linear regression from convolutional ANN activations to MDS-derived spaces can be successful, with results sensitive to the number of dimensions.
The cognitive framework of conceptual spaces proposes to represent concepts as regions in psychological similarity spaces. These similarity spaces are typically obtained through multidimensional scaling (MDS), which converts human dissimilarity ratings for a fixed set of stimuli into a spatial representation. One can distinguish metric MDS (which assumes that the dissimilarity ratings are interval or ratio scaled) from nonmetric MDS (which only assumes an ordinal scale). In our first study, we show that despite its additional assumptions, metric MDS does not necessarily yield better solutions than nonmetric MDS. In this chapter, we furthermore propose to learn a mapping from raw stimuli into the similarity space using artificial neural networks (ANNs) in order to generalize the similarity space to unseen inputs. In our second study, we show that a linear regression from the activation vectors of a convolutional ANN to similarity spaces obtained by MDS can be successful and that the results are sensitive to the number of dimensions of the similarity space.