Purchase this article with an account.
Zhen Li, Hiroaki Shigemasu; Generalized representation of shapes from different cues in parts of IPS areas. Journal of Vision 2019;19(10):177b. https://doi.org/10.1167/19.10.177b.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In this study, we adopted fMRI to provide evidence whether ROIs are involved in representation of convex-concave 3D shapes from binocular disparity or perspective by assessing classification accuracy of distinguishing 3D shapes using multi-voxel pattern analysis, and further investigated relationships of these representations: whether generalized representation of 3D shapes from different cues is involved in each ROI. The ROIs were defined by standard retinotopic mapping in early visual areas and localizers for higher areas. Stimuli depicted convex-concave 3D shapes formed by two slanted planes. Three types of stimuli were used: (I) Shapes from disparity using random dot stereogram. (II) Shapes from disparity using black-white dotted lines. (III) Shapes from perspective cue using black-white dotted lines. Two different types of stimuli of disparity were used in order to check whether shapes from disparity using different elements shares generalized representation with shape from perspective cue. Linear support vector machine was used for classification. Two types of classification were performed to investigate 3D shape representation: (1) Same-type stimuli classification which trained and tested classifier on the shapes from the same type of cue. (2) Transfer classification on shapes from different types of cues which includes: training on type I and testing on type III and vice versa; training on type II and testing on type III and vice versa. Results showed that for same-type stimuli classification, classification accuracy had a tendency to be higher in early visual areas than in intraparietal sulcus (IPS) areas, while for transfer classification, classification accuracy had a tendency to be higher in parts of IPS areas than early visual areas. These results indicate that parts of IPS areas may be related to more generalized representation of 3D shapes formed by slanted planes irrespective of cue types while early areas are related to low level process of cue information.
This PDF is available to Subscribers Only