View Project

Norwegian AI Directory

The embodied cognitive bases for classifier systems


Description:

This project looks at the possibility that abstract grammatical categories as they are expressed in classifier systems of nouns are instances of embodied conceptual knowledge. Such categories typically describe the noun's essence (an ANIMAL dog), its phy sical properties (a LONG pencil, a FLEXIBLE rope), or its function (a VEHICLE car). "Embodied" means that concepts are stored and reactivated in modal brain areas corresponding to the senses that were active in the process of learning: through vision, tou ch, taste, physical interaction or emotion, or a combination of these. Modal knowledge is processed locally in disparate brain areas, e.g. visual knowledge quite distant from areas for moving one's body parts. Knowing this, the first challenge regards how co-occurring modes of learning are merged in the brain to form unitary concepts. Is there a central abstraction area in the brain that after learning stores concepts irrespective of the manner in which they were learnt? If so, what function does reactiva tion of neurons in modal areas have for processing concepts? is it redundant or crucial for understanding? The project also looks at the contribution of shape, motion and emotion in knowing about living vs. non-living things. Evidence from grammars and br ain-damaged patients have independently shown that this notion is gradient (human > animal, etc). Neuroimaging evidence on these categories will reveal what brain areas support their knowledge and increase understanding of meaning-impairment. Likewise, th e importance of understanding a central vs. distributed representation of knowledge will affect how we think about treatments for knowledge- and language disabled people, since we will understand better the importance of manner of learning for subsequent storage. It will also prove or disprove embodied cognition as a theory applying equally to abstract and concrete concepts. Finally, interactional knowledge systems are relevant to programming in robotics.


Project leader: Marit Lobben

Started: 2013

Ends: 2017

Category: Universiteter

Sector: UoH-sektor

Budget: 3583000

Institution: Psykologisk institutt

Address: Oslo