Determining transformation distance in similarity: Considerations for assessing representational changesa priori

Abstract

The Representational Distortion (RD) approach to similarity (e.g., Hahn, Chater, & Richardson, 2003) proposes that similarity is computed using the transformation distance between two entities. We argue that researchers who adopt this approach need to be concerned with how representational transformations can be determined a priori. We discuss several roadblocks to using this approach. Specifically, we demonstrate the difficulties inherent in determining what transformations are psychologically salient and the importance of considering the directionality of transformations.

Psychology of knowledge representation

“…Consideration should be given to both representations of content, and the encoding and retrieval contexts (i.e., transfer appropriate processing) . However, in order to make research advances, scientists need to be careful about generating a priori predictions and not postulating a particular form or process after seeing task data …”

“…59 However, in order to make research advances, scientists need to be careful about generating a priori predictions and not postulating a particular form or process after seeing task data. 60 Future research in the field of knowledge representation is likely to benefit from advances in cognitive neuroscience and grounded/embodied cognition. While Marr 61 stressed the importance of considering different levels of explanation (i.e., computational, algorithmic, and implementational) and that each level has value, research in cognitive neuroscience may be able to constrain representational theories once the underlying brain structures and functions are better understood.…”

A Comparison of Semantic Similarity Models in Evaluating Concept Similarity

“…According to Kolmogorov complexity theory, the number of transformations is the smallest number of operations that the computer program transforms one concept into the other (Goldstone, 1999). Indeed, the calculation of the semantic similarity from the transformational model needs some comparisons (Grimm et al 2012). Transformational operations are only conducted on distinct components, while identical components are not considered.…”