A simplicity model of concept difficulty

Pothos, Emmanuel (2016). A simplicity model of concept difficulty. [Data Collection]. Colchester, Essex: Economic and Social Research Council. 10.5255/UKDA-SN-850037

Data description (abstract)

Feldman in Nature: One of the unsolved problems in ... concept learning concerns the factors that determine the subjective difficulty of concepts: why are some concepts psychologically simple' others 'incoherent?' (p. 633, vol. 407). The proposed research addresses this issue.
ESRC funding has enabled the development of the Simplicity model for how people spontaneously divide novel stimuli into categories. Ultimately, the aim of the model is to understand why categories like 'cats' are intuitive but a category which includes 'oranges, the moon, and chairs' is nonsensical. In this project several artificial data sets will be created. Participants will be asked to classify the objects in these data sets in different ways. The Simplicity model can provide parameter-free predictions of which categorizations will be psychologically more intuitive. These predictions will be assessed against empirical measures of category intuitiveness, such as classification variability, supervised learning difficulty, and memory for category labels.
Categorization research is dominated by models of supervised categorization, which tell us how people classify new stimuli; spontaneous classification has been under-researched. This proposal is a step towards addressing this imbalance, by further examining the Simplicity model and appreciating the ways in which category intuitiveness can be computationally characterized.

Downloads

Altmetric