The role of mental rotation in letter processing by children and adults.

Children and adults identified or discriminated the version (normal or backwards) of letters presented in 10 different orientations between o and 180 degrees. Reaction time to discriminate version increased linearly with orientation for both children and adults, but reaction time to identify was not strongly influenced by orientation for either children or adults. This suggests that both children and adults mentally rotate a representation of the letter to discriminate version, but that neither children nor adults rotate letters to identify them. Mental rotation is an important skill that children and adults can use to deal with misoriented stimuli (Cooper & Shepard, 1973; Marmor, 1975, 1977)For example, in a version task in which subjects must judge whether or not a character is presented in a normal (e.g., R) or backwards (e.g., fl) version, subjects apparently rotate misoriented characters (e.g., y ) to the upright position before making the judgment (Cooper & Shepard, 1973). Mental rotation is characterized by a linear relation between reaction time and the angle through which the mental representation is rotated, indicating a continuous transformation that is executed at a constant rate (Shepard & Metzler, 1971). It is apparent from the literature that rotation rate increases with age from childhood to adulthood (see Table 11); our experiment focuses on two other issues on which children and adults might differ. First, it is apparent that adults do not need to rotate alphanumeric characters to identify them (Corballis & Nagourney, 1978; Corballis, Zbrodoff, Shetzer & Butler, 1978), despite claims that some knowledge about orientation must be available before identity can be established (Rock, 1974). The evidence against rotation is that the effect of orientation in the identification task is substantially smaller than that observed in a version task believed to require rotation (Cooper & Shepard, 1973). This finding may reflect the extensive practice adults have had with alphanumeric characters, and may not replicate with children. That is, mental rotation may be a more general strategy in children than in adults, and in particular, children might be expected to rotate letters (which are relatively unfamiliar) to identify them. Second, even in the version task, which has provided much of the evidence for mental rotation, it is apparent that adults do not rotate stimuli in all orientations. The function relating reaction time to orientation typically has a quadratic component, reflecting a range from o to about 60 degrees in which reaction time is independent of orientation, followed by a range from 80 to 180 degrees in which reaction time increases linearly with orientation (Cooper & Shepard, 1973; Hock & Tromley, 1978). The range from o to 60 degrees reflects a range of 'perceptual uprightness' which adults have presumably developed from extensive experience with alphanumeric characters. Because children have had less experience, they might be expected to have "This work was supported by a Canadian Commonwealth scholarship to the first author and by Grant 410-77-0797 from the Humanities and Social Sciences Research Council of Canada to the second author. Requests for reprints should be sent to Jennifer M. Young, Department of Applied Psychology, OISE, 252 Bloor St West, Toronto, Ontario, M5S1V6. Canad. J. Psychol./Rev. canad. Psychol., 1980,34 (3) 265 a narrower range of perceptual uprightness than adults. The existing literature provides some evidence on this issue: Marmor (1975, 1977) found no evidence of a quadratic component in children from 4.5 to 8.6 years old, whereas Childs and Polich (1979) found such trends in children g.25 and 11.08 years old. However, Marmor used pictures of bears and cones instead of the alphanumeric stimuli used by Childs and Polich, Hock and Tromley, and Cooper and Shepard, and it is possible that her pictures were not sufficiently familiar to produce the perceptual uprightness effect; indeed she found no such effect in adults with these pictures. Thus, the existing literature may overestimate the age at which the perceptual uprightness effect first appears. To test these notions, we administered version and identification tasks to children and adults using alphanumeric stimuli for both tasks. Since the version task is generally believed to involve rotation, it provides a standard against which performance on the identification task can be evaluated; if identification requires rotation, the slope of the orientation function must be at least as steep as the one observed in the version task. Moreover, the alphanumeric stimuli allow a clearer assessment of the perceptual uprightness effect than do Marmor's stimuli. The critical questions are first, whether children's letter identification involves mental rotation; and second, whether the range of perceptual uprightness in children is as broad as that for adults.