Higher level object recognition is actually not that well understood. Much of our current understanding comes from cognitive neuropsychology (the study of cognitive impairment), in particular, through the study of people suffering from agnosia, an impairment of the ability to recognise visual objects. There are two main forms of agnosia: apperceptive (inability to form a representation of the object) and associative (the inability to attribute identity to the object). People with associative agnosia can perceive objects but cannot name them because their access to the necessary semantic information has been disrupted. This does not mean, however, that we cannot recognise an object if it does not have a name...but we do need to be able to successfully match perception with pre-existing representationa/semantic knowledge about the object. There is much debate about what such representations may consist of (stored templates? flexible combinations of shapes? images? words?)
Interestingly, neurons have been identified that respond selectively to many forms of information about particular object, person or concept (known sometimes as concept/grandmother cells). For example, a particular cell in your brain will 'fire' when you see/hear/read about Jennifer Aniston. Therefore, it appears that stored representations are multimodal. I'm not sure what the minimum requirements for recognition are...obviously the more you know about someone/something the easier they are to remember. A name helps but I don't think it is neccessary. Afterall, animals and infants show object recognition without the use of words, and I'm sure you have been able to recognise someone 'off the tv' without knowing their name.
Last edited by Bluth.; 11-07-2012 at 17:09.