Imagery

According to Cicero, it was the Greek poet Simonides who first recognized the utility of mental imagery for memory. During a brief absence from a banquet at which the poet recited the entertainment, the roof of the building caved in, mangling the guests so badly that recognition of the bodies was impossible. Simonides (who likely already had developed a fairly good memory in the context of his job) realized that he could remember the faces and clothing of the guests in various locations around the room; thus he was able to identify the corpses for their families. Secondarily, perhaps, the art of memory was born (see Yates, 1966, for full histories).

Image and imagery generally are used to refer to those concrete, perceptual, and usually visual modes of thought that appear to represent the physical world relatively directly. These are clearly distinguishable from verbal thought processes, which are arbitrary in the linguistic sense of there being no necessary relationship (other than social agreement) between words and their referents. Mental images of things experienced thus often appear to be fundamentally related to their meanings. It is difficult to think of a pizza without some olfactory or gustatory image or to think of one's mother without experiencing a visual image of her face.

Consistent with subjective impressions, a variety of studies have indicated that visual imagery and visual perception have some similar qualities. Several studies by Shepard and his colleagues, for example, have examined mental rotation (Cooper and Shepard, 1973). This paradigm typically involves asking people to judge whether two visually presented stimuli (e.g., letters or three-dimensional shapes) are identical or mirror reflections of each other. Shepard's results have consistently indicated that, in order to decide on the possible identity of the two stimuli, subjects first mentally rotate one of them to the same orientation as the other. This is evidenced by the fact that reaction times to make the identity judgments increase regularly as the angular difference between the two stimuli increases from 0 to 180 degrees and then decreases to 360 degrees (as the rotation can be made "backward").

Kosslyn and his colleagues have demonstrated similar findings in a series of studies on mental scanning (Kosslyn, 1973). In this paradigm, subjects memorize several landmarks on a simple map and then are asked to form an image of it. Consistently, the time required for subjects to scan from one location to another on their images is a direct function of the linear distance between the two locations on the original map, thus reflecting the analogue character of mental images. Further studies by Kosslyn, Finke, and others have shown that people can mentally construct twoand three-dimensional figures that have emergent properties not predictable from their component parts. Such findings indicate a perceptionlike quality of mental imagery that has been supported by some neurophysiological evidence (Farah, 1984).

Although the psychological processes underlying these phenomena may not be entirely clear, the utility of mental imagery and pictures to facilitate memory has been recognized for centuries. Simonides's method of loci, for example, in which successive items are imaged at specific locations along a familiar route, has long been used for learning ordered lists of unrelated items or remembering a series of points to be made during a speech. Imagery also can be helpful in learning foreign vocabulary, as in forming an image of a cadaver falling from a table for the Italian cadere (to fall) or someone harvesting cogs in a field for cogliere (to pick or gather). In addition, imagery seems central to the development of thinking and memory in children, and a progression from motor processes to visual images to verbal processes is assumed by most major theories of cognitive development.

In scientific studies of memory, there are several typical findings that indicate an important role of imagery: Words rated as referring to more concrete, highly imageable things are better remembered than words referring to abstract, nonimageable things—the concreteness effect; the use of imagery in learning (either by instruction or spontaneously) leads to better memory than nonimaginal strategies—the imagery effect; and pictures are better remembered than words—the picture superiority effect.

This apparent centrality of mental imagery in thinking and memory has led to several theoretical frameworks for studying human memory and cognition that give images an essential role. The most completely articulated of these, the dual coding model of Paivio (1971, 1986), is depicted in Figure 1(a). The model includes a verbal system specialized for dealing with sequential, especially linguistic information and an imagery system specialized for dealing with non-verbal, holistic information. It explicitly entails that concrete experiences generate perceptual memory traces (images) that preserve the structural attributes of the input. Those generated memory representations are assumed to be functionally equivalent to perceptual representations "in the sense that an image of an object generated to its name has the same mnemonic properties as the image evoked by the object itself" (Paivio, 1986, p. 144). Among other consequences, the storage of information-rich perceptual images is assumed to account for the findings that mental comparisons of things on physical dimensions (such as size) yield similar patterns of response times regardless of whether the objects, their pictures, or their printed names are presented.

The hypothesized existence of separate but interconnected verbal and imaginal systems in the dual coding model explains the concreteness effect in terms of coding redundancy: Concrete words usually result in dual verbal and imaginal memory codes (via the intersystem, referential crossover), whereas abstract words usually result in only a single, verbal code. Such redundancy provides two alternative "routes" to the traces for concrete words during recall and provides a "backup" if one code is forgotten. In learning of concrete word pairs (i.e., in paired-associate learning), imagery is assumed to provide a means of integrating the meanings of the two words into a single, imaginal unit that can be retrieved later in the context of one of the words as a cue. Integration of this sort is assumed to be unavailable for abstract word pairs, which must be stored as two-unit verbal strings. Finally, pictures are assumed by the dual code model to be better remembered than words in part because images may be inherently more memorable than verbal traces but primarily because pictures are likely to elicit verbal naming (and hence dual memory codes) with a somewhat higher probability than words elicit imagery.

One modified version of the dual coding model is depicted in Figure 1(b). In this alternative, dual verbal and imaginal processing systems are assumed to operate at a level akin to Baddeley's (1986) working memory, which has both visual and verbal components. The dual processing systems can account for most of the results concerning verbal processes and image manipulation and inspection. In contrast with the multiple, modality-specific memory codes of the original dual coding model, however, this version assumes that long-term memory involves some more generic, semantic, or propositional code common to both concrete and abstract information (Potter, 1979). Information is retrieved from this conceptual memory and images are constructed in visual working memory. Imagery is still invoked to explain the beneficial effects of imagery instructions, material concreteness, and picture presentation, but the locus of their effects on memory is placed on the distinctiveness of imaginally processed items within the encoding and retrieval contexts (Marschark et al., 1987). Interitem integration is assumed to be possible for abstract as well as concrete materials via conceptual relations.

One difficulty in explaining the effects of imageability on memory is that several other variables cooccur with it. Dimensions such as concreteness and word frequency have been shown to be positively related to rated imagery values, whereas others, such as generality of reference and associative set size, have been shown to be negatively related to it. The effect of imagery can be empirically or statistically separated from all of those other variables, however, and remains a significant predictor of memory when the others are controlled (Paivio, 1986). Exceptions begin to occur only when memory for more complex materials like concrete and abstract texts is considered or when other meaningful contexts make the importance of imaginal processing less essential (Marschark and Cornoldi, 1990).

The precise role of imagery in language comprehension and problem solving remains unclear, although it clearly plays a role in visually and spatially oriented situations, such as route learning and chess playing, which seem to involve analogue mental manipulation. Distinguishing imagery from the nature of long-term memory codes highlights the role of individual differences in imagery ability. A variety of standardized and well-documented tests that involve tasks such as mental paper-folding, mental rotation, and mental scanning have provided evidence of large differences between individuals in the vividness, speed, and frequency of image generation. The imagery abilities tapped by these tests generally are not predictive of memory ability even for concrete materials. Nonetheless, it is clear that variability in some imagery abilities can have specific and marked effects in a variety of cognitive domains, including memory. Individuals who have suffered head injuries, for example, typically exhibit deficits in visual-spatial tasks such as finding hidden figures or reconstructing previously presented pictures. They may also fail to exhibit concreteness effects in memory and be particularly slow at making imaginal comparisons (Richardson, 1990). People who score high on tests of image manipulation skill tend to be faster in comparing their images on particular dimensions, although they may be no faster in generating those images in the first place.

Among the more puzzling imagery findings yet to be explained is why people who are totally, congenitally blind still show apparent imagery and concreteness effects (DeBeni and Cornoldi, 1988). These effects appear not to be attributable to tactile knowledge, because they are just as readily obtained when the to-be-remembered items are things for which tactile experience is unlikely (e.g., a moon, a tiger, or a tower). Clearly, imagery is a multidimensional construct that has great theoretical and practical utility but no simple psychological explanation.

Bibliography

Baddeley, A. (1986). Working memory. Oxford: Clarendon Press.

Cooper, L. A., and Shepard, R. N. (1973). The time required to prepare for a rotated stimulus. Memory & Cognition 1, 246-250.

DeBeni, R., and Cornoldi, C. (1988). Imagery limitations in totally congenitally blind subjects. Journal of Experimental Psychology: Learning, Memory, and Cognition 14, 650-655.

Farah, M. J. (1984). The neurological basis of mental imagery: A componential analysis. Cognition 18, 245-272.

Kosslyn, S. M. (1973). Scanning visual images: Some structural implications. Perception and Psychophysics 14, 90-94.

Marschark, M., and Cornoldi, C. (1990). Imagery and verbal memory. In C. Cornoldi and M. A. McDaniel, eds., Imagery and cognition, pp. 133-182. New York: Springer-Verlag.

Marschark, M., Richman, C. L., Yuille, J. C., and Hunt, R. R.(1987). The role of imagery in memory: On shared and distinctive information. Psychological Bulletin 102, 28-41.

Paivio, A. (1971; reprint 1979). Imagery and verbal processes. Hillsdale, NJ: Erlbaum.

—— (1986). Mental representations: A dual coding approach. Oxford: Oxford University Press.

Potter, M. C. (1979). Mundane symbolism: The relations among objects, names, and ideas. In N. R. Smith and M. B. Franklin, eds., Symbolic functioning in childhood, pp. 41-65. Hillsdale, NJ: Erlbaum.

Richardson, J. T. E. (1990). Imagery and the brain. In C. Cornoldi and M. A. McDaniel, eds., Imagery and cognition, pp. 1-45. New York: Springer-Verlag.

Yates, F. A. (1966). The art of memory. Chicago: University of Chicago Press.

MarcMarschark