12/30/2012

Continuous semantic space mapped on human cerebral cortex


With recent advances in neuroimaging and data analysis methods, it has become possible to even map the neural basis of semantic concepts. There are studies showing that distinct object categories such as faces and outdoor scenes are differentially represented in the human brain, however, one of the most profound observations in early reaction time studies conducted on processing of semantic concepts (and categories) is that semantically similar words cause largest priming effects, as if semantically similar concepts would be represented close to one another in a “semantic space” so that spreading of activation would facilitate processing of concepts related to preceding ones. However, it has not been empirically shown whether such semantic space, where semantically similar concepts would be represented close to one another in a gradient/continuum, can be found on the human cortical surface.

In their recent study, Huth et al. (2012) presented healthy volunteers feature films during 3-Tesla functional magnetic resonance imaging. They then derived a large number (altogether 1705) of object and action, as well as higher-category, names based on WordNet lexicon and labeled the movies so that a time course was obtained for the presence of each concept name in the movies. These concept time courses were then regressed against the brain hemodynamic responses recorded during movie watching, and the results suggested that there indeed is a continuous semantic space on the human cortex. These results provide highly exciting novel information on how concepts are mapped in the human brain, and overall the study presents a new type of methodological approach that offers exciting possibilities for further studies on the neural basis of language, one of the most fundamental of human cognitive functions.

Reference: Huth AG, Nishimoto S, Vu AT, Gallant JL. A Continuous semantic space describes the representation of thousands of object and action categories across the human brain. Neuron (2012) 76, 1210–1224. http://dx.doi.org/10.1016/j.neuron.2012.10.014

No comments:

Post a Comment

Any thoughts on the topic of this blog? You are most welcome to comment, for example, point to additional relevant information and literature on the topic. All comments are checked prior to publication on this site.