Mental imagery, reasoning, and blindness
Quarterly Journal of Experimental Psychology
Although reasoning seems to be inextricably linked to seeing in the "mind's eye", the evidence is equivocal. In three experiments, sighted, blindfolded sighted, and congenitally totally blind persons solved deductive inferences based on three sorts of relations: (1) visuospatial relations that are easy to envisage either visually or spatially, (2) visual relations that are easy to envisage visually but hard to envisage spatially, and (3) control relations that are hard to envisage both visually
... isage both visually and spatially. In absolute terms, congenitally totally blind persons performed less accurately and more slowly than the sighted on all such tasks. In relative terms, however, the visual relations in comparison with control relations impeded the reasoning of sighted and blindfolded participants, whereas congenitally totally blind participants performed the same with the different sorts of relations. We conclude that mental images containing visual details that are irrelevant to an inference can even impede the process of reasoning. Persons who are blind from birth-and thus do not tend to construct visual mental images-are immune to this visual-impedance-effect. Mental Ima gery, Reasoning, and Blindness 3 Introduction One of the most fundamental and frequent ly asked questions within cognitive psychology is "What type of mental representation is used when people think, solve problems, or make decisions". If you ask people with no education in psychology how they reason, many of them say that they rely on visual mental images. For instance, with a problem such as: Adam is taller than Brenda. Brenda is taller than Cathy. Does it follow that Adam is taller than Cathy? they say that they form a mental picture in their "mind's eye" and then look at this picture to see that Adam is taller than Cathy (Knauff, Strube, Jola, Rauh, & Schlieder, 2004) . Not only non-psychologists, but also many cognitive psychologists have claimed that reasoning is inextricably linked to seeing in the "mind's eye ' ' (e.g., DeSoto, London, & Handel, 1965; Kosslyn, 1994) . The aim of the present paper is to explore the issue of mental representation in reasoning with a special focus on visual mental images. We re-examine the hypothesis that visual representations underlie reasoning, reject it, and propose an alternative view. We show that previous reasoning studies have often overlooked a possible confounding between materials that invoke visual imagery and materials that invoke spatial representations. In addition, the article is concerned with the connection between reasoning and blindness. On the one hand, the visual nature of reasoning might suggest that congenitally totally blind individuals-that do not experience visual mental images-should be impaired in reasoning with highly visual materials (e.g., Fraiberg, 1980) . On the other hand, there are several studies showing that persons who are blind from birth differ from sighted people in their use of visual images, but that they are as good as the sighted in the construction of spatial representations (e.g. Kerr, 1983). The paper begins with a brief summary of previous findings on imagery and reasoning. We focus on deductive reason-Mental Ima gery, Reasoning, and Blindness 4 ing, in which the truth of the premises ensures the truth of the conclusion. We then explain why we believe that previous reasoning studies have often overseen a possible confounding between visual imagery and spatial representations. Based on these thoughts, we outline our hypothesis that reasoning relies on spatial representations whereas visual mental imagery is not necessary in reasoning. We report three experiments with sighted, congenitally totally blind, and blindfolded sighted participants that test this hypothesis. Finally, we draw some general conclusions about visual imagery, spatial representations, and reasoning. A first study on imagery and deductive reasoning was carried out by DeSoto, London, and Handel (1965) . They investigated three-term series problems as in the preceding example and argued that reasoners represent the three individuals in a visual image, and then "read off" the answer by inspecting the image "in the minds eye". Huttenlocher (1968) also claimed that reasoners construct a visual mental image of the individuals in the problem that is analogous to seeing the individuals. Shaver, Pierson, and Lang (1975) claimed that performance on three-term series problems depends on the ease of creating an image of the given materials, the instruction to form images, and the participants' ability to form images. Clement and Falmagne (1986) found that materials rated as easy to imagine led to fe wer errors in reasoning. Pearson, Logie, and Gilhooly (1999) reported that a visual secondary task has a disrupting effect on reasoningrelated mental synthesis. Other authors questioned the role of visual imagery in reasoning. Johnson-Laird, Byrne, and Tabossi (1989) examined reasoning with three transitive relations: equal in height, in the same place as, and related to (in the sense of kinship) and found no effect of imageability on reasoning accuracy. Clement and Falmagne studied conditional reasoning tasks such as If a man walks his golden retriever, then he gets upset about his insect bite, which were assumed to be highly visual, and problems based on statements such as If the man takes an economic perspective, then he Mental Ima gery, Reasoning, and Blindness 5 uses the new memory technique, which were assumed to be difficult to visualize. A second factor was the availability of pertinent knowledge. This factor interacted with imageability, but in the direct comparison there was no difference between problems based on statements that were easy to imagine visually and those that were difficult to visualize. Newstead, Pollard, and Griggs (1986) reported a similar result and Sternberg (1980) found no difference in the accuracy of solving problems that were easy or hard to visualize and no reliable correlation between scores on the imageability items of IQ tests and reasoning ability. Richardson (1987) reported that reasoning with visually concrete problems was no better than reasoning with abstract problems. One possible resolution of the inconsistencies in the previous findings is that these studies have overlooked the distinction between visual and spatial representations. This distinction has been drawn by many researchers in cognitive psychology (e.g. Hitch (1974), visual and spatial information was maintained and processed in one uniform system-the visuo-spatial scatch pad. Recent studies, however, have shown that this part of working memory consists of two specialized temporary memory sys tems-one visual and one spatial. The visual component is responsible for retaining visual features such as shape, texture, color, and metrical distance. The spatial component holds locations and movement information, i.e., it represents what things are where. Studies using the dual-task interference paradigm showed that the maintenance of visual information such as shape or color interferes with a visual perceptual input (2004) showed a direct double dissociation. In their studies, a visual short-term memory task was more strongly disrupted by visual than spatial interference, and a spatial memory task was simultaneously more strongly disrupted by spatial than visual interference (Klauer & Thao, 2004). Brain imaging studies also provide strong evidence for a dissociation of the two systems and that different cortical areas contribute in different ways to the processing of visual and spatial information. These studies fall into two groups. The first group was primarily concerned with the re-). The second group of studies investigated the role of the ventral and dorsal pathways for visual and spatial working memory. The main motivation of these studies was to explore whether the brain areas that are specialized for the perception of objects, features, and locations are also involved if such information has to be maintained in working memory. Therefore, in a study by Smith, Jonides, Koeppe, Awh, Schuhmacher, and Minoshima (1995), partic ipants had to perform either spatial memory tasks (remembering positions of objects) or visual memory tasks (reme mbering the identity of objects). The results manifested a double dissociation, since the spatial tasks activated occipital and parietal (right-hemispherical) regions, i.e., regions of the dorsal "where" pathway, while the visual tasks resulted in activation in (left-hemispherical) inferotemporal regions, i.e., areas of the ventral "what" pathway. These results were replicated in a study by Courtney, Ungerleider, Keil, and Haxby (1996) , who found activation in the ventromedial temporal lobes, i.e., parts of the "what" pathway, when participants had to maintain faces in working memory, while the superior and inferior parietal cortex, i.e., parts of the "where" path-Mental Ima gery, Reasoning, and Blindness Kerr, N.H. (1983). The role of vision in "visua l imagery" experiments: Evidence from the congenitally blind.