Picture Theory, Tacit Knowledge or Vividness-Core? Three Hypotheses on the Mind’s Eye and Its Elusive Size

In this study, we compared hypotheses derived from our interpretation of three imagery theories – picture theory, tacit knowledge and vividness-core. Participants were asked to generate “small” (1.2), “medium” (11 or 16), or “large” (91) images of concrete, everyday objects. Image size varied between subjects in Experiment 1, and within subjects in Experiment 2. Vividness ratings and image latency were measured. According to picture theory, vividness should increase directly with latency, and both should increase continuously with size, in both Experiments. According to tacit knowledge theory, such a continuous increase will occur only in Experiment 2 when the full range of sizes is known to the subjects. According to vividness-core theory, latency and vividness should be inversely related in both experiments, and latency should increase with size in Experiment 1 but not in Experiment 2. Results support vividness-core. Images, we conclude, are primarily derived from memories whose latent activation is reflected in reported vividness, as specified by vividness-core theory. In the recently revived “imagery debate”, one contention has been that tacit knowledge can explain the classic findings on the effects of manipulating the size of mental images without the need of postulating visual mental imagery or pictorial representations (Pylyshyn, 2002). In the classic study, Kosslyn (1975) verbally cued participants to imagine an animal (e.g., tiger) so that the entire visual image would fill one of four randomly presented squares of different areas. Latencies were longer for generating images that filled larger squares, suggesting to Kosslyn that more time was consumed to “fill out” the larger images with the imaged object parts, and, hence, with more details. According to Kosslyn's (1994) picture theory, images are depictive representations formed in a structure (visual buffer) that has space limits, which constrain image resolution. If the object is imagined so small (or so large) that one cannot appropriately represent a part in which a given detail belongs, then the detail will not be incorporated. As size increases within an optimal range, it will offer progressively more locations for representing details of the imaged object, thereby requiring continuously more time to be completely fleshed out. However, this direct relationship between size and generation latency of mental images may also be explained by tacit knowledge (Pylyshyn, 2002); if one is asked to generate a small mental image, one may generate it with few visible details because one knows from daily experience that real-world objects that are smaller are less detailed when viewed from far off. One could have generated any image at any size with any level of detail, but one did not, because of one’s (tacit) knowledge of how objects look. This is similar to the notion of ‘demand characteristics’, in that the participant is trying to understand the implications of the imagery task (and does so by relating it to actual visual experience), even though there is no explicit demand from the experimenter. Paraphrasing Pylyshyn's argument, if we are asked to generate a small image, we are likely to generate it as having fewer visible details than if we are asked to generate it as looming large directly in front of us. If the task of generating a small image, as opposed to generating a large one, entails having fewer visible details, then we can predict the result expected by the experimenter, without the need of assuming that we are actually basing our responses on a real scale of small and large images. The results of indirectly increasing or decreasing image detail by manipulating image size will be obvious and will be as expected by the subject and the experimenter (Pylyshyn, 2002, p. 163). Pylyshyn’s and Kosslyn’s claims have implications for the data we report, which concern both the time required to generate an image or image latency, and the amount of detail (detailedness) and clarity of the mental images, that is, their vividness (Marks, 1973). Although Kosslyn and Pylyshyn did not explicitly use the construct or term of vividness in their theories, other researchers have shown that introspectively available properties, such as the ones included in the present definition of vividness, can reflect the resolution of the visual buffer (e.g., Dean & Morris, 2003). In particular, vividness ratings can provide a reasonably good estimation of image detailedness (D’Angiulli, 2001). Thus, it seems that, together with image latency, vividness could be used to derive new interpretations and hypotheses from picture theory, tacit knowledge or other imagery accounts, providing also a common test bench for competing hypotheses. Indeed, a third alternative account (vividness-core) proposes that size manipulations can be confounded with changes in vividness, and that changes in image latency are really related to vividness, not to size. In a study that addressed specifically the relationship between vividness, size and image latency, D’Angiulli and Reeves (2002) found that vividness, not size, had a major effect on image latency, with the most vivid