An integrated model of pattern and symbolic information for the recognition of an object in context

Major issues in an integrated model is the representation of pattern information and symbolic information, and the interaction between these two. One of the domains where this interaction plays an important role is the recognition of an object. The primary function of the recognition is the transformation of pattern information into symbolic information such as a name of an object. In addition to this bottom-up information processing, the recognition is also influenced by the top-down information such as context and knowledge. An overall architecture of the integrated model including connectionist part, symbolic part and context is proposed. This is followed by a presentation of a prototype integrated model which is focused on a core part of the overall integrated model. The well-known bowl/cup problem in cognitive psychology is chosen as an example of a vision task to clarify the interaction, especially the effect of context on the recognition