Preamble. What vision is NOT.

As every eighth grader knows, the story of vision begins as light passes through the lens of the eye to be focused on the retina, forming a tiny upsidedown image of the scene in front of the eye. Up to this point, the eye functions like a camera, and the retinal image is simply a faithful picture of the outside world, with colors, shapes, and motions still literally present. But as any eighth grader could figure out, the story to this point does not yet offer an explanation of vision, but only a precondition for it. After all, if you severed the optic nerve, a person would be completely blind even if his eyes were functioning perfectly. The retinal image is not sufficient for visual perception to occur. Obviously, the full explanation of how vision works depends on processing in the brain.

So, what happens to that retinal image? How does the brain transform it into an understanding of the visual and spatial world? An eighth grader might suppose that the optic nerve simply transmits the image, like cable TV, back to the brain, where the brain simply "looks" at it to see what’s there. But this is still an inadequate answer: To say that the "mind’s eye" "looks" at an "inner picture" explains nothing, without an explanation of what this inner "looking" could be. And, is there anything like an "inner picture" to be looked at? Many people believe in inner pictures, but our labs so far have offered several reasons to doubt this hypothesis:

Could we come to understand vision in this widgety way? Could we see it as a bundle of parallel feature-detecting pathways, where increasingly sophisticated widget-detectors report their own micro-specialized results? Tonight we continue our explorations of the mechanisms of vision. Once again, prepare to be surprised.

The image on the retina is flat. The eighth grader’s theory of vision imagines that picture is recreated in the brain, where it is presumably flat as well. But we understand the world with depth, the objects in it having solid, 3-dimensional form. How do we get the third dimension out of flatland? Most of us learn standard answers to these questions: the "pictorial depth cues," like perspective, and other cues to depth, like binocular disparity (exploiting the slight difference in the point of view of our two eyes). These cues do work. When we see depth in pictures, for example, we have to rely on perspective and occlusion (a nearby object blocks the object it stands in front of). It requires surprisingly little information to activate a depth cue. Even the sketchiest of pictures often specifies a cue sufficient to pop out into the third dimension.

Given the usual line-up of these cues, familiar from intro psychology courses, you may suppose that depth perception is fully explained. After all, these cues seem sufficient to explain depth perception in all the normal cases. People do respond to them. End of story.

Or is it? Perhaps you are wondering, Are there other mechanisms of depth and form perception beyond the static cues of traditional perceptual psychology? We’re glad you asked! Tonight’s lab reveals a set of cues you may not have suspected, and explores their workings.

As a warm-up discussion, we will enumerate the traditional cues. This will be the background for developing new non-natural stimuli that do not offer the eye the usual cues to depth or 3-dimensional form. If these non-natural stimuli nonetheless pop out into the third dimension, then there must be some other depth cue, some other perceptual mechanism, to explain it.

We’ll explore two main types of 3-d form perception:

The goal will be to develop experiments to specify the processes involved in each kind of perception.

This is best done with a partner.

This lab uses the "Insight 2" software package, as did last week’s. As you arrive in the lab, look on the desktop screen to see if the TAs have already copied the "Insight 2: In color" folder onto the desktop. If it is there, jump to step 3. Otherwise, follow all these steps.

1.The path:

Mac HD

Network Servers

Class software

2. Click on the folder Insight 2.

It will take a minute to copy.

3. Open the "Insight 2:In color" folder.

4. Click on the icon labelled "Insight 2" (just that, and nothing else). You may have to expand the window to find this file.

5. A main menu will open. You may see a notice about the Spatial Vision module of the program. If so, just click OK.

6. Click on Form & Motion.

After you click on Begin, you’ll be able to experiment with several moving stimuli. These are called Movies. You can vary several aspects of each movie: the shape of the stimulus, the number of dots, the "correlation" or uniformity of dot motion, the background, and others. (You use various pull-down menus to make these changes.) After you set the parameters you want, pull down the File menu, and select Generate Movie. Then pull it down again, and select Play movie. When you want to stop the movie, move your mouse to the center of the screen and double click. Note that, in general, parameter settings stay on until changed for a new movie.

In all the experiments in tonight’s lab, the visual system begins with the simple stimulus of dots in motion. When there are enough dots, when there are not too many irrelevant dots, and when the motion is "right," the perception of form results. The lab focuses on two kinds of forms: regular geometric forms and "biological forms." In your report, address these questions:

One way to think of the question: From the independent moving dots, the brain extracts certain information which it combines with other information (including, perhaps, previously learned information). What information is extracted, and how? What does the brain "look for" in the moving dots?

You can develop any experiment you want as you justify your main conclusion. However, here are four possible experiments to warm up on. Each may offer relevant clues to the processes of 3-d form perception. If you and your partner perform the experiments and discuss the questions raised for each one, you will find that you have many ideas for the main report question. Your answers for these warm-up questions need not be written out or handed in.

1. Experiment 1: Compare the effects of reducing dot correlations (1.00, 0.75, 0.35) and reducing the number of dots in the object (125, 50, 10, 2) for one of the geometric shapes (e.g. a sphere). Describe the two manipulations on your perception of the moving shape. Which degrades your perception more? Why?

2. Experiment 2: Compare your perception of the biological motion object with each of the background types (stationary, random, random motion, drifting). Which background degrades your perception of biological motion the most? The least? Why?

3. Experiment 3: Which is degraded more by adding a random motion background -- your perception of a moving geometric shape or of the biological motion? Why?

4. Experiment 4: The biological motion object is made of just 12 dots. Are 12 dots enough to generate the perception of a complete geometric shape (e.g. a sphere)? Why or why not?