Experiment Duration_Pilot (final)

 

BASIC DESIGN.

 

Subjects will hear tones lasting between 600 milliseconds and 2.9 seconds, and attempt to judge their duration (they will be told to estimate durations in milliseconds, using the scale from 500 to 3000). (Note: About six subjects were run, none of which showed a significant improvement in their task.)

 

Instructions.

 

In this study you are going to be trained in a rather peculiar task that you may find interesting, and which may possibly even be useful to you some day. You are going to be estimating the duration of brief intervals. You will hear tones lasting for a randomly chosen period of time ranging between 500 msec and 3 seconds (varying in units of 1/10 of a second), and then you will try to estimate how long the tone lasted. To do that, you will use the mouse to click how long you think the interval was.

There will be buttons arranged along the bottom of the screen: push the left-most buttons for shorter durations, and push the right-most buttons when you hear longer durations. You will receive feedback after every response telling you the correct answer (an asterisk will appear above the correct button). The experiment will be divided into 12 blocks of trials, each lasting just a few minutes. Most people find this task rather challenging and at least somewhat interesting, and we hope you do as well.

 

Design:  12 blocks X 36 trials per block.

Experimental Variables: No independent variables.  Duration is randomly selected on each trial in increments of 100 msec.

 

Procedure for an individual trial:

 

1. Randomly select an integer x ranging between 6 and 29 (inclusive).

 

2. Present plus sign as warning for ½ second, then take that away, then play a 800 hz tone lasting for  x*100 msec.  (using 26 different .wav files)

 

3. Subject guesses the duration by clicking on one of 26 unlabeled buttons arranged from lowest to highest duration

  

4. Feedback is provided, giving the person 1.6 seconds to "digest" it.  Then there is a 1.5 second pause, and we go on to next trial.

 

DATA STORAGE

 

On each trial, please write out to the disk

<Block #; Trial #; Actual Duration; Subject's Estimate of Duration>

 

Also please print out a summary data file with 12 rows, containing the following columns

 

1. Block #

 

2. Average of the absolute values of the difference between the actual duration and the subject’s estimate of the duration across trials within the block. (so if one trial the subject overestimates by 100, and on another trial the subject underestimates by 100, the errors do not cancel out, and the average difference = 100.)

 

3. Average difference between the actual duration and the subject's estimate of the duration.  (so if one trial the subject overestimates by 100, and on another trial the subject underestimates by 100, the errors cancel out and the average difference = zero.)

 

4. Average squared difference between the actual duration and the subject's estimate of the duration.  (so if one trial the subject overestimates by 100, and on another trial the subject underestimates by 100, the average squared difference = 10,000.)

 

[these are 3 widely used measures of precision.]

 

(Note: There was a bug when this experiment was run where the summary data file contained the sums of each of the above, and not their averages! However, this problem was easily fixed by dividing each sum by the number of trials in each block.)