PEBL Manual Dexterity

Jump to: navigation, search
Screen shot of PEBL's Manual Dexterity Task


[edit] About

This test is a simple task which evaluates how the participant is able to move the mouse to a precise location on the screen under different levels of jitter-noise. This documents the version to be released in the PEBL Test Battery Version 0.7, PEBL Version 0.12. Earlier versions used slightly different data output records.

Note that this test is original to the PEBL battery, although it resembles several tracking and compensatory tracking tasks that have been used in previous computer test batteries.

[edit] Notes

By default, the task runs for 80 trials, ten trials each at eight different noise conditions. On each trial, the target position is chosen randomly and uniformly from within the target donut.

User's should note that the performance on this test will be highly sensitive to mouse driver parameters, as well as screen resolution and thus whether the task is run fullscreen or not).

[edit] Options

Edit dexterity.pbl in battery/dexterity to change options

    gVideoWidth <- 800
    gVideoHeight <- 600

    gTolerance <- 5      #How close you have to get to the target.
    gCursorSize <- 5     #Size of the cursor
    gNoise <- 12         #noise on each sample.
    gDrift <- .3         #drift back toward center
    gSleepEasy <- 1

    gTimeOut <-  10000  #maximum time allowed to hit target
    gMinDist <- 100                    # Minimum distance to target
    gMaxDist <- (gVideoHeight-20)/2    # Maximum distance to target

    noiseCond <- [1,3,5,7,9,12,15,20]  ##noise conditions (in pixels); blocked
    condsize <- 10         ##trials per condition

[edit] Data Output

Data are saved in a file called dext-X.csv. The column will look like:

subnum, trial, startx, starty, noise, targx, targy, distance, complete, time, moves
321, 1, 400, 300, 5, 635, 151, 287.717, 1, 717, 22
321, 2, 400, 300, 10, 559, 370, 217.076, 1, 1029, 26
321, 3, 400, 300, 10, 642, 259, 107.201, 1, 1559, 44
321, 4, 400, 300, 1, 572, 506, 116.017, 1, 1095, 29
321, 5, 400, 300, 5, 303, 81, 216.021, 1, 1656, 41

The first column is the subject code, the second is the trial index, the third and fourth are the x,y position of the start (these will always be the same), the fifth is the 'noise' condition: the size, in pixels, of the uniform jitter that is added to the cursor location on each Draw() cycle. The next columns are the target X,Y location, followed by the measured distance between start and end, followed by whether they completed the trial (i.e., it didn't time out after 10 seconds), followed by the time needed to reach target and a 'move' count, which is the number of times the mouse moved to a new position, and should be highly correlated with time. Prime DV of interest is probably time, and noise and distance are IVs that should be examined.

In addition, PEBL computes some summary statistics and saves them to a file dexterity-summary-X.txt, and reports them in a final dialog. It looks like:

       PEBL Manual Mouse Dexterity Test              
 By Shane T. Mueller
System:       LINUX
Sat Feb 12 00:59:20 2011

Summarized results:

Noise   N      Median Mean    sd
1       3       28      35      4.96655 
5       3       21      29      8.04156 
10      3       23      30.3333 8.99383 

Computing basic statistics for each level of noise used.

[edit] References

[edit] See Also

Personal tools