Brightness Matching

This first applet is designed to produce a set of meaningful data, while illustrating an important principle of visual perception. The applet allows you to manipulate the brightness of a gray circle centered within a larger circle of a lighter or darker color. Your task is to adjust the center of the circle on the right to be the same shade of gray as the center of the circle on the left.

As you move the slider to the right, the center of the right circle will lighten. When you think that you have a match, click on the button labeled “Record Data.” At this point another set of circles will appear and you will repeat the process. When you have made nine settings, the applet will present your data, showing you how accurate you were. (Write these data down or print them out, because you can’t retrieve them once you move on.)

A general principle of human visual perception is that a dark background will cause a spot in the center to appear lighter than it actually is. Thus in the example shown above, we would expect you to err by setting the center spot at the right lighter than it really should be. This means that the difference in brightness between the two center dots will be positive. This would apply to trials 1, 4, and 8. The reverse should happen on trials 2, 5, and 6, where the background on the left is lighter than the one on the right—here the differences should be negative. Finally, trials 3, 7, and 9 were control conditions, where the two backgrounds were the same, and we would expect most accurate settings, and relatively small (positive or negative) differences.

For your own data, calculate the mean and median differences under each of the three conditions described above. Create a table similar to the one shown below, which was created from my data.:

Results of 9 trials of color matching

Left Background	Trials	Differences	Mean	Median
Lighter Darker Equal	2, 5, 6 1, 4, 8 3, 7, 9	-.30, -.17, -.14 .07, .13, .09 -.02, -.10, -.03	-.21 .10 -.05	-.18 .09 -.03

• What do your data show with respect to the hypothesis outlined above?
• Would you have a preference for the mean over the median as the important statistic here?
• Why would the mode not be a useful measure?
• Why do you suppose that the three differences within any one line are not all the same? (This will be a very important point later when we refer to this variability of scores obtained under similar conditions as “random error.”)

Comments to: Gary.McClelland@Colorado.edu

  Return to index