Linear Separability in Categorisation and Inference: A Test of the Johnson-Laird Falsity Model
Abstract
Johnson-Laird suggests that difficulties in problem solving can be explained by the mental models theory. This study tests linear separability effects in categorisation and inference as an alternate explanation, hypothesising that categorisation and inference would be easier for linearly separable (LS) functions than nonlinearly separable (NLS). Thirty two participants were tested on one LS and one NLS functions over repeated trials. Results indicated that categorisation and inference were significantly more difficult for NLS functions, but only for the highest performing participants on some trials. Among poorer performing participants there were no significant differences between response rates and response times. The most likely explanations for these findings are the complexity and duration of the experiment, rather than lack of support for the linear separability hypothesis. Implications for the military and research communities and suggestions for future research are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2014
- Accession Number
- ADA606957
Entities
People
- Armando Vozzo
- George Galanis
- Susannah J. Whitney
Organizations
- Defence Science and Technology Group