Insight from a Visual Trick
The size-weight illusion may be easier to understand if we couch it in terms of a more familiar visual illusion, the Ponzo, or railroad track, illusion (b) [see “The Quirks of Constancy,” Illusions; Scientific American Mind, August/September 2006]. Two horizontal bars are shown lying between two longer converging lines. Although the bars are identical, they are not seen as such: the top bar appears longer than the bottom bar. We can explain the illusion in terms of a visual effect called size constancy; if two objects of identical physical size are at different distances from a viewer, they are correctly perceived as being the same physical size, even though the images cast by them on the retina are different sizes. Quite simply, the brain “understands” there is a trade-off between retinal image size and distance and, in effect, says, “That object’s image is small because it is far; its actual size must be much bigger.” To evaluate distance, the visual system uses various sources of information called “cues,” such as perspective, motion parallax, texture gradients and stereopsis. It then applies the appropriate correction for distance in order to judge true size.
But with the Ponzo illusion, the two horizontal bars are the same physical size on the retina. The converging lines provide a powerful trigger to read them—falsely in this case—as lying at different distances away (as though you are peering down a railroad track and see the railroad ties at increasing distance). Because your visual system “believes” the top line is farther away, it infers that the top line must really be larger than its size on the retina would indicate (relative to the other line). You therefore perceive it as being larger.
To put it differently, size constancy scaling enables you to perceive accurately the size of objects when you correctly perceive distance to those objects. In the Ponzo illusion, however, the misleading depth cue from the converging lines causes you to misapply the size constancy algorithm so that the top line is seen as being larger. Remarkably, the illusion overrides the visual signals from the retina informing the visual size-judgment centers in the brain that the two bars are exactly the same length. And because these mechanisms are all on autopilot, knowing that they are identical in size does not correct the illusion.
The situation with size and weight is analogous. (Read “actual weight signaled by muscles” for “actual retinal image size.”) Your brain says, “For the big object, I expect the muscle tension to be much greater in order to lift it.” But because the muscle tension required is much lower than expected, the object is felt as unexpectedly light. This experience overrides your judgment of actual weight signaled by your muscles.
Remember that we said the size-weight judgment system is on autopilot. So we can ask how dumb or smart it is on its own. What if we now use as test objects a disk and a ring of the same outer diameter (c), and, as with the standard size-weight illusion, we adjust each of them so that they have the identical physical weight? Of course, as before, anyone picking up the ring will expect it to weigh much less because it looks as if it has less total volume. But you (the experimenter, aware of the size-weight illusion) would predict the reverse—that the hollow ring would be felt as being much heavier than the solid disk. In fact, in collaboration with Edward M. Hubbard, now at INSERM in France, we have found that a subject will experience no size-weight illusion; she will correctly judge the objects to be the same weight. The brain seems to merely utilize the outer diameter in making the judgment, rather than
the overall volume. This experiment shows that the visual system is not sophisticated enough to understand that what is relevant is the total mass, not the outer diameter alone.