Vision Rex Vol. 22, pp. 739 to 744. 1982 0042-6989/82/070739-06$03.00/O Printed in Great Britain Pergamon Press Ltd zyxwvutsrqp THE CONTRAST SENSITIVITY OF HUMAN INFANTS TO GRATINGS DIFFERING IN DUTY CYCLE MARTIN S. BANKS and BENJAMIN R. STEPHENS Department of Psychology, The University of Texas at Austin, Austin, TX 78712, U.S.A. (Received 20 July 1981; in revisedform 24 November 1981) Abstract-Human infants’ contrast thresholds for rectangular wave gratings differing in duty cycle were measured. The forced-choice preferential looking technique was used to estimate thresholds in fifteen 8- to IO-week old infants. The results indicated that contrast threshold varied systematically with duty cycle. Our findings were consistent with the predictions of a linear, multiple channel model and one version of a linear, single channel model but were inconsistent with the predictions of a contour density and another version of a single channel model. INTRODUCTION Several aspects of pattern vision change dramatically during the first year of life. Visual acuity and sensi- tivity to contrast increase (Dobson and Teller, 1978; Salapatek and Banks, 1978). Pattern preferences undergo noteworthy change as well; older infants exhibit stronger preferences for patterns composed of small elements than younger infants do (Fantz et al., 1975). Three groups of investigators have measured the contrast sensitivity function (the function relating contrast threshold to spatial frequency for sinusoidal gratings) at various ages in an attempt to characterize the manner in which pattern vision capabilities de- velop early in life (Atkinson et al., 1977; Banks and Salapatek, 1978; Pirchio et al., 1978). The emphasis in this line of work has been on characterizing pattern *The evidence that optical defects are not the primary cause of infants’ contrast sensitivity deficits is mostly in- direct yet reasonably convincing. Several types of optical defects can affect the cOntrast sensitivity of a visual sys- tem. Three of these-spherical aberration, chromatic aberration, and diffraction-have little effect on adult sensitivity at the spatial frequencies to which the infant visual system is sensitive (0-4c/deg) (Campbell and Gubisch, 1967; Westheimer, 1963). Thus, barring the un- likely possibility that one or more of these three defects is very much larger in infant eyes, they are unlikely to contribute significantly to the sensitivity deficits ob- served in infants. One optical defect which can constrain sensitivity to Mc/deg is cloudiness or the presence of significant opacities in the optic media. This defect can also be ruled out, however, because ophthalmoscopic examinations indicate that the media are relatively free of opacities or cloudiness early in life (e.g. Cook and Glasscock, 1951). Large spherical refractive error or accommodative error are other defects which can affect sensitivity to frequencies in the range of the infant CSF (Green et ul., 1980). These errors can also be ruled out though because most infants in the age range tested can and do accommodate reasonably accurately to the target distances used in infant CSF experiments (e.g. Banks, 1980). detection thresholds rather than pattern preferences. Infant contrast sensitivity functions (CSFs) reveal numerous age-related changes. The most pronounced change, however, is the steady increase in contrast sensitivity at medium and high spatial frequencies from early to later infancy (Harris et nl., 1976). To what extent can the change in various aspects of infant pattern vision be related to changes in the CSF? Linear systems theory states that the output of a linear system to any input can be predicted if one knows the system’s modulation transfer function. This approach can be used justifiably to model the optical processing of the eye since most optical systems are linear. The optical quality of the young infant’s eye exceeds the demonstrable acuity of the system as a whole, however, so the cause of the contrast sensi- tivity deficit appears to be primarily of neural rather than optical origin*. Thus, the shape of the CSF at a given age and the way in which it changes during infancy is probably mostly dependent on the charac- teristics of neural interactions rather than optical pro- cessing. Since one does not know to what extent linear systems assumptions are violated by neural mechanisms in the infant visual system, it is unclear how successful linear systems approaches might be in predicting thresholds and preferences in young infants. Banks and Salapatek (1981) showed that spatial fre- quency cut-offs (acuity thresholds) for different types of gratings could be predicted reasonably accurately from infant CSFs and linear theory. Specifically, they reanalyzed two older studies of infant acuity. In one case, they predicted the difference in cut-offs between rectangular wave and square wave gratings at three ages. In another, they predicted the difference in cut- offs between a square wave grating and a hand-drawn grating whose duty cycle and period varied. Atkinson et al. (1977a) showed that preferences for blurred faces could also be predicted from infants’ acuity cut-offs. Despite the reasonably good optics of the young eye, spatial frequency cut-offs are more likely to be 739