Predicting Typicality Effect from Individual Brain Activity Patterns in Healthy Subjects – Multi-Voxel Pattern Analysis (MVPA)

Family resemblance hypothesis (Rosch & Mervis, 1975) proposed that highly typical items (e.g., sparrow) of a semantic category (e.g., bird) share most semantic features with other members of the same category, while less typical items (e.g., penguin) share fewest common features with other members. Neuroimaging studies (e.g, Iordan et al., 2016) have observed different brain activity patterns in visual areas, such as the lateral occipital complex. Damage to other semantic regions such as the anterior temporal lobe can also cause semantic dementia. This study aims to apply multi-voxel pattern analysis (Haxby et al., 2001) to investigate which brain regions (semantic and visual) show sensitivity to classification of the neural patterns being encoded between typical and atypical exemplars in different semantic categories in healthy subjects. We will use pattern-classification algorithm (linear support vector machine) to classify individual’s brain activity patterns from an fMRI semantic feature verification task.