In this webcast, Dr. Luisa Mayer of the New England Eye Institute at Perkins talks about the development of tests of visual acuity in babies through the means of Preferential Looking. Preferential Looking is something that a baby does when they see something interesting, next to something that is not interesting.
The field of vision science and the study of the development of visual perception began in the 1950s. Dr. Mayer takes the viewer through the advances in the study of how babies see, from FPL (Forced Choice Preferential Looking) using stripes of different widths for very young babies, to OPL (Operant Preferential Looking) to hold the interest of older babies and children.
Read full transcript »
Presented by Luisa Mayer, Ph.D.
Length of time to complete: approximately 30 minutes
CHAPTER 1: Introduction
MAYER: Preferential looking is something that a baby does when they see something that’s very interesting next to something that isn’t interesting. So they prefer to look at something that’s interesting.
I’ll give you an example. A three-month-old baby, if we show a three-month-old baby a drawing of a face where all the features are regular — the eyes are where they should be, the nose is in the middle, the mouth is below — and then you put next to that a drawing of the same features of the face, but where everything’s scrambled — the eyes are over here, the nose is there, the mouth is up here.
NARRATOR: We see the image of two faces side-by-side. The image on the left is of a normal face. And on the right the shapes depicting the eyes, nose, hairline and eyebrows are randomly placed within the same space.
MAYER: What’s the baby going to look at? The baby’s going to look at that face that’s regular and not at the scrambled face. We call that the baby’s preference to look at a face. So preferential looking has a really long history as both, just natural observation of what babies do.
They naturally look at things. Things they’re interested in capture their attention and they spend a lot of time looking at them. And that observation started a whole field of vision science, studying the development of visual perception and the development of sensory visual functions in babies, using modifications of this method now that’s called preferential looking.
The first person to do that formally was someone named Robert Fantz, who was an American developmental psychologist. And he presented two stimuli like stripes and a bull’s-eye to babies and he looked at… he measured which one they preferred by having somebody just measure their looking time to the stripes, to the bull’s-eye. He would change the position, measure their looking time, say in a two-minute interval. And cumulating their looking time over that two-minute interval would tell you that the baby preferred one or the other if the time was sufficiently different.
NARRATOR: We see the two shapes as described, on the left a bull’s-eye of three concentric black rings and on the right side, six identical wide, black vertical stripes.
MAYER: And he did that starting with newborn babies, babies shortly after birth. To do that he had to create this funny-looking box that he called a looking chamber, but actually I call it the baby box. And it’s open on the bottom.
A baby is slid in lying on their back underneath this box. Someone is peeking over the top of the box, looking through a little peephole. And inside the box are either two objects or two images that the baby’s looking at. So baby’s down below looking up at these images or these objects. And the observer looking through that peephole is measuring how much time the baby looks at one or the other object.
NARRATOR: A black and white photograph shows Fantz’s apparatus.
An infant on its back gazes up at a balloon-like object at the top of the box.
An observer stands to the left of the apparatus on a small step stool and leans over the top of the box to observe the baby through a small opening.
MAYER: When he did that with newborns, he found that newborn babies actually have visual preferences. They look at things, some things more than others. And it turns out it’s pretty complicated and pretty interesting. He also studied babies up to about six months of age, sitting them upright in a little baby chair and showing them also patterns and objects.
The bottom line of his findings with newborns was that babies actually make visual discriminations. They’re not blind, they see differences between things. It’s fairly rudimentary, but nevertheless they’re making discriminations. He considered that evidence for innate visual perception in babies, that babies are able to see forms right after birth. They don’t need experience.
Now, some would argue, well, they need some experience. Well, yes, to refine their vision certainly they need experience. He also showed that babies had preferences for certain things that changed over time. They weren’t always interested in looking at the bull’s-eye versus the stripes. Later on they seemed to be more interested in stripes than bull’s-eyes.
But at all ages he found that babies preferred to look at regular faces over scrambled faces. So Fantz started doing these studies in the ‘s. And this type of work is really called the study of visual perception in babies. And it’s been ongoing, there are all kinds of modifications of this procedure used to study complex visual perception and cognition in babies.
CHAPTER 2: Visual Perception vs. Visual Sensory Function
MAYER: I want to differentiate between what I mean by sensory versus perceptual. So sensory qualities of an object are things that make up that object, that form, such as the angles or the edges of a form, the lines, the width of the lines that create that form. Light/dark differences that define where that form is in space. Color differences and so forth.
So those are the components that go up to make a form, and perception is seeing the form as a form, seeing all those components put together, and that creates an object that we can discriminate. Now, the sensory visual functions that I’m talking about that have been studied over many years since shortly after Fantz’s work are visual acuity, detail vision, contrast sensitivity — the ability to see different contrasts between objects — color vision, temporal vision, motion sensitivity, stereo, vernier acuity and a number of other really basic visual functions. The research started on these things really with babies in the ‘s, so shortly after mid-century.
Actually, there are a number of different modifications of preferential looking that were used to test visual sensory function. I’m going to focus on one in particular because it’s what I know. It’s from the person I worked with, it’s where I did my graduate studies and so you’ll hear all about that. But there were a number of other researchers who used modifications of preferential looking to study behavioral visual function in babies. And I’m just going to name a couple of them because they’re prominent people, they’re people in our community and they’re good friends as well.
At MIT, Massachusetts Institute of Technology, some of the first studies were done by Richard Held and Jane Gwiazda on babies’ vision. And they were also very interested in clinical aspects — so astigmatism, an abnormality of the eye that causes visual acuity to be reduced. And also Eileen Birch who was working in their lab as a post-doc, and then now is at the Retina Foundation in the Southwest and does a lot of really good research on clinical aspects of visual function in babies.
And then there’s a pair of people — a couple — from England who have been working as long as everyone else: Jen Atkinson and Oliver Braddock from the United Kingdom. These are just some people. I can’t name all of them. There’s a list of maybe or people doing research in this area. But the person I’m going to talk the most about, who for me has the greatest influence — but of course I’m biased — is Davida Teller, who was a scientist of vision in adults, who was at the University of Washington in Seattle and had an insight about how you might approach testing visual sensory function in babies using a modification of Fantz’s preferential looking technique.
CHAPTER 3: Early Testing — Forced Choice Preferential Looking
MAYER: Now if you remember, Fantz showed two stimuli and measured whether… how much looking time the baby spent looking at each of those stimuli. Teller realized that you really can’t test sensory function as a perceptual task, at least not in this kind of way, and that what you needed to do was to create a stimulus setting where there’s just one stimulus.
That is, that’s the stimulus that’s going to measure your sensory function. And I’m going to give as an example a technique, stimulus that I’ve been intimately involved in and was the first that was used, and that’s test of visual acuity using stripes, black and white stripes in a patch. It refers to a circular patch. So that stimulus was going to be the test of visual acuity, varying the size of the width of the stripes from very coarse, very wide to very fine, and then at some point finding out what’s the threshold, or what is the acuity of the child for that stripe stimulus.
Well, okay, you’ve got a stripe and now what do you do with it? Well, what she designed was to put the stripe either on the right or the left of center, and in the other part there would be a stimulus that was a circle just like the stripes were in a circle, but it was gray and matched in average brightness to the stripes when you couldn’t see them. So you take a measuring instrument and you measure what’s the average brightness of that patch and you match it to this gray. Now, that’s because if this were brighter, the baby might be just looking at it because it’s brighter, not because they saw the stripes.
NARRATOR: We see an example of how Teller’s stripe stimuli would be presented to a test subject.
Two circles have been cut out of a large gray field. They are on opposite sides of and equidistant from the midline of the large field. In the center there is a small peephole, which allows the observer to see the test subject.
In this example, the circle on the left displays a pattern of vertical stripes, and the circle on the right displays a solid gray color that matches the large field.
MAYER: And she also put the stripes in the gray in a big surround that was the same gray approximately as the gray patch, as the blank, we call it. And that was to reduce features that might be distracting to the baby. So instead of looking at stripes, they might look at something off on the side.
Now they have only one thing to look at: stripes versus gray. So that was probably the, you know, the first… I want to say innovation she created. Other people did the same thing, but her second innovation is even more striking and a little bit more difficult to describe. And that is, there’s a person observing, just as in Fantz’s technique, a person observing a baby looking at these things.
And that person’s job she changed. She changed the observer’s job so there’s no longer measuring looking time but rather where are the stripes based upon what the baby does. So are the stripes on the right or the left? And the observer, looking at the baby, doesn’t know where the stripes are and they have to make this judgment just on the basis of what the baby’s doing.
NARRATOR: A photograph depicts an observer behind the testing apparatus. The woman is looking through the peephole to observe the test subject.
The next photo we see shows a view of a baby’s face as seen through the peephole. In this example, the face is turned to the baby’s right, suggesting that the baby is responding to a stripe stimulus on that side.
MAYER: And she called it a forced choice judgment, which is really quite important. The observer had no other choice. Where are the stripes: right or left?
On every single trial, discrete trials, showing a whole range of stripes from very tiny to very large. And with that procedure, she reconceived preferential looking and she called it forced choice preferential looking.
Pretty obvious, but pretty significant change. And I’m going to abbreviate that as FPL. So I’m going to say FPL and just remember I’m talking about that technique.
CHAPTER 4: Using the Forced Choice Method
MAYER: Created out of this series of presentations of stripes of varying widths is this, what we call a psychometric function. And “psycho” is psychology, response, and “metric” is a metric — that is, some measure of sensory function. And in fact, psychometric functions are, I would say, the building blocks or the grounding for a field of visual psychophysics.
So there’s another fancy word. What is psychophysics? It is the psychological responses to the physics of the stimulus, to qualities of the physical stimulus. So we have stripe width, that’s the physical parameter. It varies over a certain range. We take the psychological responses of the observer to… in the paradigm that we’re talking about.
Where is the stimulus? You can also say, what interval is the stimulus in? Or you can say, yes or no, does the baby see them or not? That’s another way of formulating an experiment. Any sensory function can be measured with psychophysics. So it doesn’t have to be just vision, but it can be hearing, smell, touch, taste. And lots of interesting experiments are done using psychophysics for testing all kinds of sensory functions.
NARRATOR: A young girl is shown wearing a large pair of headphones while taking a hearing test. She gazes up at a woman in a white lab coat who is administering the test.
MAYER: How do you measure acuity doing that? I mean, you know, I’ve said stripes and a range of widths and so forth.
The way the technique developed was the observer’s correct and incorrect judgments on each stripe was cumulated, and a percent correct for each stripe width was calculated. So what you now have is you have a function that shows the percent correct performance of the observer as a function of stripe width.
NARRATOR: An example of a psychometric function plot is displayed. Along the X or horizontal axis are six examples of the stripe stimuli presented. The stripe width varies progressively from wide to very fine.
The Y or vertical axis represents the observer’s correct judgments as a percentage. The five points that are plotted show a diminishing percentage of correct judgments from % to less than %, which correlate to the diminishing widths of the stripe stimuli presented.
MAYER: When the stripes are very large, the baby looks, makes a really strong look at them. The observer can tell easily where the stripes are. When they get smaller and smaller, it may be a little more subtle, and the baby may look like this, may just glance a little bit at the stripes. But the observer can still tell most of the time where the stripes are. But when they get so small the baby can’t see them, they’re just two blanks for the baby, the observer is just guessing and most likely is just saying, “Oh, I think it’s there, I think it’s there,” and is wrong half the time, correct half the time.
NARRATOR: We see an example of a stimuli of very small stripes being presented, along with a solid gray circle. In the photo, the sensory qualities of the stripes are too subtle to be distinguished. A slight difference in the color of the circle to the left of the midline is the only cue. The next photo taken through the peephole shows a baby looking down slightly, exhibiting no preference for either side of the display.
CHAPTER 5: Early Testing — Operant Preferential Looking
MAYER: Who are the babies that can be tested with forced choice preferential looking, FPL? In Davida’s lab the babies that were being tested were generally about two to three months of age. Well, my colleague Jane Allen came along and the idea was well, let’s see if we can test really young babies just like Fantz did with his preferential and see how old… you know, when does acuity develop and how does it develop normally? And the person who did that was Jane Allen, who in ‘ finished her study of FPL acuity tested in babies between two weeks and six months of age. And she showed that acuity matured over that time, it got better. But it didn’t get near adult levels. So it wasn’t over.
Well, what happened after six months? Babies get bored, basically, sitting for all the numbers of trials that you have to present to do full psychometric functions. At about six months they’re no longer willing to do that. And they need some kind of reinforcement to keep looking, at least we realized that in those days. And I was in Teller’s lab at that time, and in fact, working on Jane Allen’s study.
There was another person in the lab who was a research associate. Her name was Velma Dobson, a very important person in the field of infant vision. And she and I conceived of a method to test babies, or we wanted to try this method using a reinforcement procedure to see if we could test older ages and how far we could push this technique.
And what we did was we stole from audiologists who use this technique to test hearing sensitivity in babies. And basically it’s a little bear, an animated bear who is playing a drum in a box that is otherwise black, but then when it gets turned on, the box lights up and the baby plays the drum. We placed a box on either side of the FPL screen and when the observer said “Baby sees the stripes,” pushed a button, stripes were on the left or the right, the bear in the box on the left lit up and the bear played his drum. And so the baby was reinforced for looking at stripes.
NARRATOR: In a photograph, an infant sits on his mother’s lap in front of an OPL display. There are dark black boxes on either side of the flat gray display field. The stripe stimulus has just been presented to the baby’s left.
The next photo depicts what happened. Because the observer judged the stripes to be on the left based on the baby’s reaction, the observer then pressed a button to record that choice. As a result of making a correct judgment, the box to the left of the display lights up and the mechanical bear plays its drum.
MAYER: When the observer was wrong, nothing happened. So it wasn’t that it happened all the time, it only happened contingent on the baby’s looking at the stripes. Well, that worked pretty well and it worked for babies up to the age of five years. We could test now children from six months — or actually earlier than that — all the way up to five years of age, with one exception. Toddlers were really tough.
And we gave up on -month-olds. -month-olds are a terror. When we measured full psychometric functions in children over this whole age range and compared it with Jane Allen’s data, we found at overlapping ages the results were the same. So that said that operant preferential looking, or OPL, wasn’t biasing the results in any way.
It didn’t make babies see better, it just made it easier to test them. So now we had techniques that would cover the whole age range, and we could say we know now that acuity is near adult levels at five years of age. It’s very poor at two weeks and it gradually increases in a steady way between two weeks and five years.
NARRATOR: This graph depicts the development of grating acuity from early infancy to preschool age. Acuities obtained by forced choice preferential looking between two weeks and six months are shown by open circles. Acuities obtained by operant preferential looking between five months and five years are shown by filled circles. Grating acuity improves from / at two weeks to about / at five to months, and to / at five years.
CHAPTER 6: FPL and OPL Apparatuses.
MAYER: So we have this technique and we have apparatuses. I call them apparatuses, this FPL apparatus and OPL apparatus. And those devices or pieces of equipment were fairly simple and actually sort of primitive. They were made out of gray cardboard, mounted on wooden frames and held together with pushpins and fishing wire and glue.
NARRATOR: In a photo, a graduate student in Davida Teller’s lab is holding a baby in front of the FPL screen with the black and white grating pattern on his right. A close examination of the photo shows the many pushpins holding the various pieces of gray cardboard together.
MAYER: Well, over the years… I mean this work was done in the ‘s, so it was… Engineers say, “You know, you do a breadboard model “and you get the proof of concept and then you go on to do more sophisticated technological things,” and so that’s what happened.
NARRATOR: Now we see an OPL device with some metal frames and glass panels in front of the right and left display fields.
A young girl reaches out and touches the panel on the left. Based on the toddler’s reaction, the observer correctly judged the striped stimuli to be on the left. As a result, a reinforcement of Cheerios cereal has been dropped from a plastic tube into a glass dish that is within the child’s reach.
MAYER: People started using equipment that was more advanced — for example, they used cathode ray tubes, and then eventually video displays and computers now. And so some very interesting, important work that was done and continues to be done using these more complicated stimulus displays. Still using preferential looking as the technique.
NARRATOR: We see in a photo a baby sitting up in front of a box containing two video monitors side by side. We can see that a display of stripes is in one of the screens.
MAYER: Something we did do in Teller’s lab and in Children’s Hospital, where I went, was we replaced the observer and the holder as separate people, but actually put that person into the same role. So the observer and the holder were now the same person — the person holding the baby, showing them the stripes and the observer. And how that worked was there was a video camera behind a peephole that was showing the image of the baby on a video monitor, that was then projected to a mirror and the observer was able to watch the babies looking and make judgments on the babies looking. Now the observer, of course… I mean the holder/observer couldn’t see where the stripes are, they shouldn’t see them either.
NARRATOR: In this photo, Velma Dobson is seen holding a baby, her three-month-old son. He is looking at a grating on the FPL screen, while Velma observes his face in the mirrored video image projected by a camera behind the central peephole in the screen. We can see a black and white image of the baby’s face in a monitor that has been placed on a shelf.
MAYER: Over the many decades — as I say, we’re now in the seventh decade of studies using preferential looking — a number of different visual functions have been tested using behavioral techniques, using preferential looking types of techniques. And that would include visual acuity is the most-studied, and over the widest age range.