Forced Choice Preferential Looking

It appears that neonates and infants dislike boring visual stimuli. Boring is a stimulus that is quite homogeneous and unchanging. So if you present a display to a baby half of which is quite plane and the other have has some pattern to it, the baby will tend to look at the pattern. This is the basic idea behind the forced choice preferential looking technique. This technique becomes a "forced choice" method because an observer has to decide, based on their observation of the baby's head and eye movements, where the stimulus is located. The criterion response occurs when the observer is correct 75% of the time. At the risk of being unnecessarily redundant, I would like emphasize that the baby's responses are being assessed based on the observer's perception of where the baby is looking.

At least one further consideration has to be accounted for when using this technique. It is easiest to discuss this consideration when asking the question whether a baby can tell the difference between, for example, a chromatic color and white. It is possible in such an experiment that the baby might look at the brighter color regardless of its chromaticity. To avoid this confounding of brightnesses (or luminance) the test stimuli are typically presented at a range of luminances surrounding an approximation of equal luminance of the test and reference stimuli. If the responses do not fall to chance level at any luminance level then one can assume the baby discriminated the stimulus based on its chromatic characteristics.

A Stimulus Display
Adams, Courage and Mercer (1994) published a paper in Vision Research entitled "Systematic Measurement of Human Neonatal Color Vision." I present a schematic illustration of what their stimulus display probably looked like. The graphic was drawn from the description presented in their Methods section.

The stimulus display is presented on a large card which can be randomly rotated 180o. It is the observer's job who is looking through the peep hole in the center to determine whether the test stimulus is on the left or the right; a judgement the observer makes by noting where the baby is looking. This type of stimulus display is only a prototypic example. Other forms of it have been used, including projection systems where the stimuli are presented on rear projection screens where the test and reference stimuli are randomly alternated left and right. The observer's task is always the same.

Some Data
The Adams et al. (1994) study evaluated the chromatic-achromatic discrimination capabilities of newborns and 1-month old human infants. They compared these data to those of 2-month and 3-month old infants from a previous study. They found that most (74%) newborn and 1-month old babies were able to discriminate a broadband red stimulus from all relative luminances of the achromatic background. However, most in these age groups could not do the analogous discrimination with blue, green or yellow stimuli. Adams et al. reported that the failure to discriminate occurred at relative luminances that were close to photopic luminance matches between the chromatic and achromatic stimulus. The responses improved with 2- and 3-month old infants. In fact the 3-month old infants made very few discrimination errors. Adams et al. report that "...a definitive explanation for the apparent deficits in early human color vision remains elusive."(p. 1700)

Since our primary concern in this section of the book is spectral sensitivity it is worth noting that Powers, Schneck & Teller (1981) used FPL to measure scotopic (rod) spectral sensitivity in 1- and 3-month old infants. They found that one month old babies have a relative scotopic spectral sensitivity that is very much like that of adults. The major difference is that one month old babies' scotopic vision is a little more than 10 times less sensitive than that of adults. However, as these babies grow older this differential sensitivity decreases.

Peeples & Teller (1978) measured the photopic spectral sensitivity in two 2-3 month old human infants with the forced choice preferential looking method. They used a large test field (15^o x 22.5^o) which consisted of three vertical stripes that moved either to the left or the right. These test fields were presented in an achromatic screen illuminated by white light at 100 cd/m². The spectral sensitivity functions these investigators obtained were close to that of human adult functions. However, they caution that "The present data, then, do not constitute the photopic spectral sensitivity curve for human infants, but a photopic spectral sensitivity curve, taken under a particular set of adaptation conditions."(p. 52) They are more confident of visual evoked potential data for estimates of infant photopic spectral sensitivity.

It is also of interest that Brown & Teller (1989) conducted increment detection thresholds on 3-month old infants using FPL and determined that infants at this age have a functioning red-green chromatically opponent channel. It would seem therefore that by three months of age humans have four classes of visual receptors (rods and three cone classes) and at least the red-green opponent channel is operative. We do not know with any confidence which visual mechanisms in addition to rods that humans are born with and are operative under two to three months of age.