Cognitive Daily

It’s pretty obvious to most people that the two shoes in this picture are the same size:

i-43084841c05b9591eff0e62d471ce383-granrud1.jpg

ResearchBlogging.orgBut if you actually took a ruler to your computer display it would show that the image of my right shoe is nearly twice as wide as the image of my left shoe. Young children will mistakenly say that the real left shoe is smaller than the right shoe; it’s only gradually that we learn to take the context of the photo into account and conclude that the shoes are the same size: the shoes appear to be a matched set; the left shoe is a bit out of focus, the railing to the left shows that the left shoe is farther away than the right shoe; we know more distant objects tend to appear smaller, and so we determine that they are the same size. Of course, it’s also relatively easy to fool the perceptual system:

In an Ames room, the height of the walls, the pattern on the floor, and the size of the doors is varied so that our usual visual cues don’t work. We’re fooled into thinking the person on the left is a giant, while the person on the right seems tiny (in fact she’s simply farther away than she appears).

If we can be so easily fooled, it might make you wonder: What, exactly, are we learning about the sizes of distant objects as we grow older? The prevailing notion for decades has been that we gradually learn cues to the size of objects and how far away they are. By the time they are ten, most kids are about as good at judging the size of distant objects as adults — they are said to have achieved “size constancy.” But recently Carl Granrud has begun to challenge that notion. Even in normal perceptual circumstances (in an open field, say, rather than an Ames room), adults make systematic errors, overestimating the size of distant objects. Granrud believes that the perceptual system simply doesn’t give us enough information to accurately judge the size of distant objects, and so we develop a system of compensation. While young kids undercompensate, adults often overcompensate.

Granrud asked kids ranging from age 5 to 10 to judge the size of white disks mounted on sticks in a field, either 6.1 meters away or 61 meters away. The children stood next to nine reference disks, and after each disk was shown to them, they pointed to the reference disk that matched its size. As expected, all the kids were accurate judging the size of the nearby disks, and older kids were better than younger kids judging the size of the faraway disks.

But Granrud didn’t stop there. He also tested each child on her or his knowledge of basic perceptual rules for judging size and distance of objects. For example, they might be shown a photo like the picture of my shoes at the start of this post, and asked whether the shoes were the same size or different, or like this picture of Nora “holding up” the leaning tower of Pisa:

i-f032a086bb6d87f00dda2ff13059ee66-pisa1.jpg

Kids who recognized that pictures like this are just tricks got credit for a correct response. The responses were divided into high- and low-scoring groups. This graph shows how high-scoring kids compared to low-scoring kids in each age group:

i-b995a27cf65365bc63288bf10e6f25b2-granrud2.gif

As you can see, when kids did better on the test of visual rules, they were also more accurate judging the size of the distant disks, regardless of their age. It appears that these rules can be learned by children as young as five. Granrud says this supports his contention that size constancy is a matter of compensating for the apparent distance of an object. When we see a far-away object, we make an estimate of its distance, and based on that knowledge, we make an estimate of its size.

This also makes it very clear why the Ames room is so effective: if we can’t make a reasonable estimate of the distance to a person, it’s nearly impossible to judge her size, and so we assume that we’re seeing a giant!

Granrud, C. (2009). Development of size constancy in children: A test of the metacognitive theory Attention, Perception & Psychophysics, 71 (3), 644-654 DOI: 10.3758/app.71.3.644

Comments

  1. #1 KeithB
    June 4, 2009

    I think this also applies to the apparent larger size of the moon near the horizon. I wonder if a young child would see it as the same size?

  2. #2 Dave Munger
    June 4, 2009

    Good point, Keith — a quick search reveals the moon illusion is actually *stronger* for young children. So judging sizes at extremely long distances (approaching infinity) might be a very different process.

  3. #3 albedo
    June 4, 2009

    The apparent change of size of the moon has to do with the lack of visual clues helping to judge the size or distance when looking up in the sky. At the horizon you have all kinds objects that help you to estimate the “correct” size: trees, houses, haze. If you only have a dark background in the zenith then you see the moon “as big as it really is”. The result is not “spoiled” by treacherous clues.
    Perceptually the sky is not like a hemisphere around you but rather like an oblate spheroid cut in half. Imagine half an m&m. The sky above your head feels “nearer” than the horizon.

  4. #4 Janne
    June 4, 2009

    An interesting point about the Ames room-illusion is that it shows just how weak our object experience is for judging size. It’s normally assumed that direct experience with an object is a reliable cue to its size – as adults we know how big a shoe is, so we see both shoes as about the same size. But in fact, the Ames room shows that bottom-up cues for distance easily overrides our own personal experience with intimately familiar objects (close relatives and friends). If knowledge about particular objects really are a distance cue, it’s not a very strong one.

  5. #5 Markk
    June 4, 2009

    I had a strange experience walking up to a large Wind Turbine (over 300 feet high). It looked smaller, just not that big. Then I went inside and looked up the tower. Whoah … it looked liked miles. When I walked outside it was like my brain clicked. Now it looked tall. Very funny feeling. I guess I just couldn’t realize how big it was since it was not something I had been close to. I had seen them for years along this highway at a distance.

  6. #6 JBM
    June 5, 2009

    Interesting! Linked to another phenomenon in depth perception (and to the pisa tower picture): The leaning tower of pisa, an illusion of perspective, where the right pisa is seemingly more tilted than the left.
    see http://scienceblogs.com/sciencepunk/2009/03/how_to_straighten_the_leaning.php

  7. #7 Lobster
    June 5, 2009

    Is there something wrong with me? The left shoe DOES look smaller than the right show to me… Yet I feel like if the right shoe were a little higher up or the picture a little wider (to show more of the railing on the side), they would look the same. I need more of a visual cue than the carpet.

    It’s not that I’m particularly young or that I’m new at this whole “seeing” thing, but I have been effectively blind in one eye all of my life (alternating exotropia), and so obviously never developed binocular vision.

  8. #8 albedo
    June 5, 2009

    I remember a small river that ran through the town I lived. It looked so huge and wide to me – also the bridge over it – when I was five or six years old.

    Eight years later I visited my old hometown again and couldn’t believe that this small river and bridge were those I remembered from my childhood.

    That’s when I understood that children see the world very differently from adults.

  9. #9 alice
    June 10, 2009

    I was born with amblyopia, commonly called a lazy eye. I learned late in life that I did not have the same depth perception which people who have two normally operating eyes have. Although I was prone to running into things as a child, as an adult, I no longer find my face in contact with the edge of a door. I do have trouble parking or catching a ball, for instance, because don’t have binocular vision.

    But my brain, for the most part has learned to judge distance by environmental clues such as the size of an object which is bigger is probably nearer. I think that’s how it works.