The rapid assessment of infant visual acuity has been a topic of considerable interest in recent years. One of the most successful methods has been preferential looking (PL). The PL technique takes advantage of the fact that, given a choice between a high contrast, low spatial frequency grating and a blank field, most infants will preferentially fixate the grating. We have published evidence demonstrating that some visible gratings are not preferentially fixated but are, in fact, avoided (Held et al., 1979). We showed infant psychometric functions that were non-monotonic. As spatial frequency increased, preferential looking fell from near lOOo/, preference for the grating to significantly less than 50% before rising to levels not significantly different from 50% at high spatial frequencies. Since any significant deviation from 50% indicates an ability to see the stimulus and since these “dips” had not been previously reported in the PL literature, we suggested, on the basis of published data, that PL acuity might have been underestimated when only those points significantly above 50% were considered. We have also reported on a fast PL method that gains in reliability when these “dips” occur (Gwiazda ef al., 1980). Recently, two articles have appeared in this journal reporting failures to find “dips” below 50% in PL experiments (Teller et al., 1982: Banks et al., 1982). Both papers also question the usefulness of the fast method proposed in Gwiazda ef al. (1980). The papers raise interesting and important issues pertaining to the measurement of PL acuity. In this letter, we will address three questions raised by these articles: (I) why don’t Banks et ul. and Teller er al. find dips below 50% in their PL data? (2) In light of these results, what is the significance of the dip? (3) How do the shapes of the psychometric function and the constraints on infant testing limit the choice of psychophysical methods? Teller ef ul. (1982) report that they do not find negative preference dips in PL data from Allen (1979). In our original paper (Held ef al., 1979) we argued that their version of PL would not produce dips. Teller et al. (1982) agreed. In our version, infants are seated in a dark room facing two bright stimuli. One is a grating. At low frequencies, infants prefer the grating over the other stimulus which is a bright but blank region. If the grating is aversive, however, the infant looks at the other bright field and preferential looking for the grating drops below 50%. In the Teller version, the grating stimulus is presented on either one side or the other of a large gray field. The entire field is of the same average luminance as the grating. There is no second bright stimulus, as such. Thus, if the grating is aversive, the infant will look about at random since there is no second choice stimulus. No dip below SOY/, would be expected and none is seen. Teller et al. argue that they get preference above 50% where we get preference below 50% (see their Fig. 4). They suggest that their observers, given feedback about the actual position of the stimulus, would learn to interpret the infant’s aversion as a clue to the side containing the stimulus. We do not give the observer feedback. Their suggestion is certainly possible though Banks et al. argue that it is unlikely that an observer could change criteria rapidly enough as the stimulus changed from one spatial frequency to another. As a cause for the dip, Teller et al. suggest that “neural non-linearities” might produce a dimming of the gratings that are fixated less than 5096 of the time. Rather than an aversion for these gratings, they suggest that the infant’s behavior may reflect a simple preference for the brighter stimulus, the blank field. Banks et al. note that their method should have revealed any such effect. As it did not, they consider the explanation to be implausible. Another possible explanation for their failure to find negative values in the range of our negative values is the shallowness of the psychometric functions cited by Teller er al. Our psychometric functions are characteristically much steeper. Of the 16 infants whose functions are shown in Fig. I of Held ef al. (1979), I5 drop from better than 80’~ preference for the grating to near or below a 50%
[1]
Davida Y. Teller,et al.
Do preferential looking techniques underestimate infant visual acuity?
,
1982,
Vision Research.
[2]
J M Wolfe,et al.
Quick Assessment of Preferential Looking Acuity in Infants
,
1980,
American journal of optometry and physiological optics.
[3]
R. Held,et al.
Stereoacuity of human infants.
,
1980,
Proceedings of the National Academy of Sciences of the United States of America.
[4]
Martin S. Banks,et al.
A failure to observe negative preference in infant acuity testing
,
1982,
Vision Research.
[5]
R. Held,et al.
Infant visual acuity is underestimated because near threshold gratings are not preferentially fixated
,
1979,
Vision Research.
[6]
D. Mayer,et al.
Visual acuity development in infants and young children, as assessed by operant preferential looking
,
1982,
Vision Research.
[7]
R Held,et al.
The Development of Vergence Does Not Account for the Onset of Stereopsis
,
1983,
Perception.
[8]
D. Teller,et al.
Visual acuity for vertical and diagonal gratings in human infants.
,
1974,
Vision research.