You are in:
Home > Doing Experiments
DOING EXPERIMENTS
Because you (the student users) are not likely to have experiences in doing psychophysical experiments, an elaboration on the submenu, Experiment and Data Analysis, is made below. Clicking this link allows you to design experiments for whichever topic you have chosen from the Menu. The data analysis part of the program analyzes the data collected from an experiment that you will design and perform. Unlike the tutorial, there are no quiz items to answer. Data are displayed in both tables and graphs, and an explanation of the analysis, as given in the tutorial, is available.
Figure 1. The webpage for the Experiment and Data Analysis option for the Method of Limits.
Below are comments on the different variables that can be manipulated in the "Experiment and Data Analysis."
Measuring Precision and Accuracy. For each of the methods, the opening screen shown in Figure 1, or a similar one, will appear and you can design an experiment to determine the effects of changing one or more of the following experimental variables.
Standard stimulus: The length of the standard stimulus was set for 100 screen units in the tutorial. You can vary the standard between 30 and 380 screen units.
Comparison stimuli: For the methods of limits and constant stimuli, you can vary the range of the comparison stimulus values and the value of the midpoint of the range. The available ranges are 10, 20, 30, 40, 50, 60, 70, and 80 screen units, the available midpoints are 21 to 416 screen units. In the tutorial, the range was 20 units and the midpoint was 100 units. Since there are 21 equally spaced comparison stimuli in the method of limits and 11 in the method of constant stimuli, the range of 20 units provided a 1-unit difference between the two "nearest" comparison stimuli for the method of limits and a 2-unit difference for the method of constant stimuli. If you choose a different range, the unit difference will automatically change proportionately, e.g., the range of 40 units would produce a difference of 2 units for the method of limits and a difference of 4 units for the method of constant stimuli. The range you choose must be larger than the IU and the difference between the two "nearest" comparison stimuli must be smaller than the JND. You will need a small range for a small standard stimulus and a larger one for a larger standard stimulus. (The reason for this requirement should become apparent when you study Weber's law.) The midpoint will automatically change with the change in Point of Objective Equality (POE), but you may want to have the midpoint of a range different from the POE when you think that the PSE will differ from the POE. In the method of adjustment, the range is set automatically to 20% of the standard stimulus and the midpoint is set to the POE. For example, if the standard stimulus is 100 units, the range is set for 90 to 110 units.
Top line: You can set the top line to be either the standard stimulus or the comparison stimulus. It is the standard stimulus in the tutorial.
Horizontal separation: While the standard stimulus and the comparison stimulus in the tutorials come on the screen at random locations (vis-a-vis their horizontal separation from each other), you can choose to present the lines in fixed horizontal locations in your experiment. The standard and comparison stimuli are set to appear at random screen locations, but if you want them to appear in fixed locations, click the "fixed" button. Then, choose the desired separation of the fixed locations. Possible separations vary between 0 and 270 screen units, depending on the standard stimulus length. A separation of 0 means that the standard will be directly over or under the comparison.
Vertical separation: You can change the vertical separation between the standard and the comparison to be between 20 and 340 screen units.
Number of trials: You can vary the number of trials between 5 and 50 for the method of limits, between 40 and 150 for the method of constant stimuli, and between 10 and 100 for the method of adjustment.
Presentation order or Starting position: You can choose to have descending trials only, ascending trials only, or counterbalanced between ascending and descending trials. This option is not available in the method of constant stimuli.
Precision or Accuracy. The experiments with Weber's law and the Mueller-Lyer illusion in this website use the method of adjustment. Hence, the opening screen that appears is like that of the method of adjustment and the computer chooses the range and the midpoint of the comparison stimuli. You can manipulate the same variables as you did in the method of adjustment: top line, horizontal separation, number of trials, and starting position. Unlike the experiments with the method of adjustment, you can vary the number of stimuli between 2 and 7. For the Mueller-Lyer illusion experiment, a menu very much like that for the method of adjustment will appear but the horizontal separation may be varied between 0 and 200 screen units and the vertical separation can be varied between 20 and 340.
Possible Experiments. Below are some experiments you might try. You should try to create your own experiments as well, since the creative act of designing an experiment can be as valuable as actually executing the design.
By choosing different values for the variables you can learn more about the topics. You can vary the number of trials in different sections to learn about the stability of an obtained JND, PSE, Weber fraction, or the extent of the Mueller-Lyer illusion as a function of the number of trials. Also, you can use a small range of comparison stimulus values with a large standard stimulus in the methods of limits and constant stimuli, and learn about these methods. You will find that the JND cannot be measured when the set of comparison stimuli is not appropriate for the standard stimuli. Such an experience would give you an idea of what you must take into consideration in choosing the comparison stimuli when you want to use these methods to measure the JND and the PSE for stimuli other than the lengths of lines, such as the amount of light or weight.
Besides learning more about each topic, you can gain experience in designing your own experiments. An experiment might involve systematically varying the horizontal or vertical separation of the standard stimulus and comparison stimulus to discover changes in the JND, Weber fraction, or the extent of the Mueller-Lyer illusion. Eliminating the random horizontal placement of the comparison stimulus or placing the standard stimulus and comparison stimulus closer together should allow subjects to develop more effective judgment strategies, which in turn, should reduce the size of the JND or the extent of the illusion. Weber's law is said to fail under this condition (Weber, 1849) and you can design an experiment to find this out for yourself.
Experiments need not be limited to manipulating the variables in the menu. You can compare the three psychophysical methods with each other. (See Experiment I in SAMPLE EXPERIMENTS near the bottom of the menu.) You may want to examine the effect of different lengths of lines on the Mueller-Lyer illusion. (See Experiment II in SAMPLE EXPERIMENTS.)
The "Experiment and Data Analysis" can be used in conjunction with some external manipulation. You might hypothesize that observers with glasses (or contact lenses) would produce smaller JNDs than the same subjects not wearing glasses (or contact lenses). Observers could be asked to run the experiment twice, once with their glasses (or contact lenses) and once without. Don’t forget to counterbalance the order of testing across observers. You can also measure the extent of the Ponzo illusion by attaching masking tape on the computer screen. (See Experiment III in SAMPLE EXPERIMENTS.) How about measuring the vertical-horizontal illusion by attaching a sheet of paper with vertical and horizontal lines for standard stimuli?