A model was developed to account for similarities and differences between picture and word processing in a variety of semantic and episodic memory tasks. The model contains three levels of processing: low-level processing of the physical characteristics of externally presented pictures and words; an intermediate level where the low-level processor makes contact with prototypical information about the looks of objects or pictures and the sounds of words; and the deepest (propositional) level where meaning is analyzed. The relationships between pictures and their names (or between visual and acoustic images) can take place directly, between the two image stores, or indirectly, via the propositional level to which both image stores have access. One difference between pictures and their names is the greater variability of appearances that objects and pictures have. A second difference is that pictures have less ambiguity of reference than their names, resulting in fewer propositional memory nodes accessed by pictures than by words. These differences are consistent with a large body of literature on picture-word processing differences. (Author/RL)