My dissertation project tested current theories of the picture superiority effect (better memory for pictures than words). A popular theory is dual coding (Paivio, 1991), which states that pictures are better remembered because they are encoded with two specific codes (pictorial and verbal) while words elicit just a single code (verbal). The extra code associated with pictures give them an advantage during retrieval. A more recent theory (transfer appropriate processing) discussed by Weldon and Roediger (1987) also claims that pictures and words elicit different types of information at encoding, but further states that the type of processing engaged by the retrieval task is also important. According to the theory, if the retrieval task requires conceptual processing, any items that are encoded with more conceptual processing will be more likely retrieved than items encoded with little or no conceptual processing. It is believed that pictures are encoded with more conceptual processing and this is why they are better recalled on conceptual memory tests. For tasks that require perceptual processing (e.g., traditional implicit memory tasks) a match in the type of percept encoded during study will be important. If words are used as cues on perceptual tests, studied words will be better recalled. If pictures are the cues on the perceptual test, studied pictures will be better recalled. Weldon and Roediger (1987) presented data which supported these predictions of the theory.
More recently, however, Weldon and Coyote (1996) reported results from implicit and explicit conceptual tasks that did not fully support the transfer appropriate processing explanation of picture superiority. Picture superiority was found for explicit conceptual tasks, but not for implicit conceptual tasks. Since implicit and explicit task performance can both be influenced by conscious memory (associated with conceptual processing) and automatic memory (associated with perceptual processing), memory task performance may not be the best measure of a specific type of processing. Therefore, my dissertation included four experiments that compared conscious (conceptual) and automatic (perceptual) estimates of memory to compare the amount of each type of processing required for retrieval of studied pictures and words. The first three experiments involved memory tasks that are believed to differ in the amount of conceptual and perceptual processing required for retrieval. The tasks included stem completion (mixed processing), category production (mostly conceptual), picture fragment identification (mostly perceptual for pictures) and were given in a process dissociation procedure (Jacoby, 1991). Multinomial process tree models were used to estimate the two types of memory processes. As predicted by transfer appropriate processing theory, pictures showed greater conceptual estimates for all tasks. Perceptual estimates depended on the type of cues given for the task. Perceptual estimates were higher for words on the word stem competion and category production (given with a one-letter cue) tasks, but were higher for pictures on the picture fragment identification task. These results supported transfer appropriate processing as an explantation of the picture superiority effect.
Implicit memory for pictures has not been studied as often as memory for words. The important differences between encoding and retrieval of pictures and words may have implications for the use of pictures in the process dissociation procedure. In addition, different models describing the relationship between conscious and automatic memory may need to be developed for the use of picture stimuli.