From concrete examples to abstract relations
A CogSci abstract comparing feature-based and relational category learning, with single-trial EEG set against a decision model.
The idea
Relational categories group objects that share few features. Until recently, neuroimaging of category learning had mostly studied categories defined by features. This proceedings abstract compares the two in matched learning tasks and uses EEG for the timing, following an fMRI study from the same laboratory.
The approach
The analysis follows category membership trial by trial and sets those EEG dynamics against computational estimates from decision-making models. Sorting trials by their place in the learning sequence shows frontal theta and the P300 tracking feature-based and relational categorisation differently as learning proceeds.
- Feature-based and relational categories in matched tasks
- Single-trial EEG, not a single average
- Frontal theta and the P300 against a decision model
- Roberts, Goldwater, Livesey, Giron, and Davis
- Proceedings of the Cognitive Science Society, 39
The paper: escholarship.org .