There are a wide variety of methods for measuring psychophysical thresholds. This applet lets you explore five of them.

This applet is really only going to make sense if you are familiar with the methods. It is not intended as a tutorial for the methods, but as an opportunity for you to convert your book knowledge into laboratory knowledge.


Each button below corresponds to a psychophysical experiment, using the indicated method. In each experiment you make judgments about the height of two lines. On the bottom right is a target line with Muller-Lyer "wings" that may influence your percept of line height. On the top left is a test line. The height of the target line is 100 pixels. The height of the test line is an experimental variable. The experiment is interested in determining what test line height will look the same as the target line. Differences between the test and target line heights will indicate the magnitude of the Muller-Lyer illusion. In each experiment the next trial is started by pressing the space bar. This enters the user's response and starts the next trial. The response is always either a "n" or "m" keypress. The user should press "n" when the test line looks to be smaller than the target line. The user should press "m" when the test line looks to be larger than the target line.

Method of constant stimuli

In this experiment, user compares the presented test and target lines and decides if the test line is larger or smaller than the target line. Two hundred test lines of different lengths are presented in random order. After finishing all trials, a graph will appear that plots the percentage of time a test line was said to be "bigger" than the target line, as a function of the test line length (in pixels). The 50% threshold can be estimated from this psychometric function.

Method of limits

In this experiment, the user again compares the test and target bars and responds "smaller" (n) or "bigger" (m). Even without a keypress from the subject for a new trial, the computer immediately presents a new test bar that is slightly larger (or smaller, as appropriate) and the user makes the comparison again. When the user changes his judgment (from smaller to bigger or vice-versa),the program uses the test bar length as an estimate of the threshold for perceived equal length. This procedure is repeated 20 times, 10 for initial bars that are much smaller than the target and 10 for initial bars that are much larger than the target. After the experiment, the computer presents a graph that plots the frequency with which different test bar lengths were found to be the threshold. Separate curves are plotted for ascending trials (initial bar much smaller than target), descending trials (initial bar much larger than target),and both. The computer also calculates the threshold (given in the title of the window).

Method of adjustment

This experiment is much like the method of limits, except the user is free to increase or decrease the test bar's length as much as they like. When the user is satisfied that the test and target bar lengths look similar, they should press the "space bar". The computer keeps track of the test bar lengths found by the user to calculate a threshold. This is done 20 times. At the end of the experiment a graph shows the frequency of selecting different test bar lengths as perceptually equivalent to the target bar length. The computer also calculates the threshold (given in the title of the window).

Truncated staircase method

In this experiment the user interacts much as in the Method of constant stimuli, simple deciding if the test bar is larger or smaller than the target. However, the stimuli are not presented randomly, but are systematically adjusted to zero-in on the length that is perceived to be equal to the target bar length. 100 trials are presented. At the end of the experiment, a graph appears that plots the frequency with which different test lengths were presented. This gives an idea of where the method was estimating the threshold to be. The final test bar length is taken as the threshold, and this value is given in the title of the window.

Stochastic approximation

This method is much like the truncated staircase, but it uses a different algorithm for selecting the presented target stimuli. The information in the final graph is essentially the same as for the truncated staircase.

Copyright Purdue University 1997