Purpose: The purpose of this paper is to explain and develop a classification of cognitive structures (or typologies of thought), previously designated as spoke, chain and network thinking by Kinchin "et al." Design/methodology/approach: The paper shows how concept mapping can be used to reveal these conceptual typologies and endeavours to place the concept-mapping method in the broader context of learning styles and learning theory. Findings: The findings suggest that spoke structures are indicative of a na?ve epistemology, or of "learning-readiness"; chain structures are indicators of "goal-orientation" and networks are indicators of expertise. Furthermore, change that comprises simple elaboration of existing spokes or chains is likely to be the result of surface learning styles and the emergence of networks indicative of deep learning. The utility of these different cognitive approaches is discussed. Research limitations/implications: The work is limited by the general lack of empirical testing, but the approach is presented as an important source of hypotheses for future research. Practical implications: The practical implications of the research are considerable. First, concept mapping provides a framework for documenting and assessing understanding at "novice" and "expert" levels. Second, where definitive criteria can be developed from the learning styles literature, cognitive change in the course of learning can be evaluated to distinguish between deep versus surface or holist versus serialist approaches, for example. Originality/value: The paper is original and comprises a synthetic approach to the study of learning style and learning theory through the use of the concept-mapping method. It has both practical and theoretical value because it suggests a new approach and is an important source of testable hypotheses. (Contains 8 figures.)