Geon (psychology)

Geons (geometric icons) are simple 3-dimensional forms such as spheres, cubes, cylinders, cones or wedges. One often-cited [Kirkpatrick, K. Object Recognition. Department of Psychology, University of York. [http://www.pigeon.psy.tufts.edu/avc/kirkpatrick/default.htm] ] theory of object recognition, Biederman's "Recognition by Components Theory" (RBC) [Biederman, I. (1987) Recognition-by-components: a theory of human image understanding. Psychol Rev. 1987 Apr;94(2):115-47. [http://content.apa.org/journals/rev/94/2/115] ] proposes that visual input is matched against structural representations of objects in the brain. These structural representations consist of geons and their interrelations (e.g., an ice cream cone could be broken down into a sphere located above a cone). Geons can be used to represent a large number of possible objects with very few components; e.g., 24 geons can be recombined to create over 10 million different two-geon objects.

Properties of Geons

There are 2 essential properties of geons:
# Viewpoint-invariance: they can potentially be distinguished from one another from almost any perspective (one exception being that from an end-on view, a cylinder can look like a sphere).
# Stability: recognition of geons is often robust to occlusion and degradation by visual noise.

Criticism of Geons

Some more recent studies [Tarr, M. J.; Williams, P.; Hayward, W. G.; Gauthier, I. (1998) Three-dimensional object recognition is viewpoint dependent. Nature Neuroscience; 1:275-277. [http://www.nature.com/neuro/journal/v1/n4/full/nn0898_275.html#B2] ] suggest that three-dimensional object recognition can be viewpoint "dependent", rather than "independent", in contrast to the predictions of recognition-by-components theory and geons. This argument follows in part from the observation that different regions of the brain respond to different viewpoints of a common object, say, the human face. Since each region responds differently to different viewpoints of the same human face, the brain must not use a common underlying geon-based schema for face recognition, instead relying on a viewpoint-dependent encoding schema where matches are determined by similarity.

Notes