Sunday, November 24, 2013

That is so random: The PEBL Random Generation Task

Human random number generation has been studied for many years, maybe most famously by Alan Baddeley in his exploration of the central executive, and Akira Miyake in a later study of executive function.  Tasks such as this have been around a lot longer, and Towse and Neil (1998) published a lot of nice metrics of randomness that can be used to assess the quality of randomness of a sequence.

To implement a PEBL version of this, I modeled the version described in Miyake's paper, and implemented many of Towse and Niel's randomness measures to provide an automatic report. 




That Miyake paper described asking for verbal responses though, accompanying a beep presented at a rate of 1 every 800 ms.  I used this basic paradigm, except:
  • I used manual responses, with the number line at the top of the keyboard (the keypad works as well, but participants are instructed to use the number line).
  • I accompanied the beep with a big red flashing circle.  This enables visual feedback. This can be turned off by setting useFlash from 1 to 0.
  • I also added visual feedback regarding the response.  Without it, it is difficult to know whether you are actually 'playing'. 
Although the metrics are generally able to handle any number of 'tokens' most of them including letters,  there is a challenge translating from verbal to numeric responses.  We really can only key 8 responses directly because thumbs won't fit on the number line, so it would be cleaner, and generally possible, to use just 8 numbers to get responses.  I elected to stay with 10-responses, to make the measures more amenable to the ones produced by other studies.

The data files for the test are saved in the data\ folder of the randomgeneration\ task, each within a directory named by the subject code.  Two files are saved, one for the raw data, and one summary file that computes and records the different statistics.  An example is below.  The PEBL Random Generation task will be available in PEBL 0.14.


 References:
  • Baddeley, Alan. "Random generation and the executive control of working memory." The Quarterly Journal of Experimental Psychology: Section A 51.4 (1998): 819-852.
  • Towse, John N., and Derek Neil. "Analyzing human random generation behavior: A review of methods used and a computer program for describing performance." Behavior Research Methods, Instruments, & Computers 30.4 (1998): 583-591.
  • Miyake, Akira, et al. "The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis." Cognitive psychology 41.1 (2000): 49-100.

Random Generation task Assessment.  PEBL Random Number Generation Test.
 Shane T. Mueller, Ph.D.
 Based on method described by Miyake et al.
 Randomness scoring using metrics described by Towse & Niel, 1998
Sun Nov 24 08:27:42 2013

Number of responses:             15
Mean response time:              4.4
Response table:
--------------------
Number       Count
--------------------
0        2
1        3
2        1
3        1
4        1
6        1
7        2
8        2
9        2
--------------------
Coupon score:                              16
Redundancy:                                3.27936
RNG Score:                         [1-0]   0
Guttmann's Null-score Quotient:    [0-100] 66.6667
Adjacency score (Ascending):               40
Adjacency score (Descending):              6.66667
Turning point  (100 is expected):          69
Phase length:
--------------------------------------
Phase length   Count   Expected
--------------------------------------
1               3       5
2               1       2.01667
3               1       0.527778
--------------------------------------


Repetition gap Mean:                        5.33333
Repetition gap Median:                      5
Repetition gap Median:                      5
First order distance:

   First-order distance
--------------------------------------
Difference    Produced     Expected (est)
--------------------------------------
-9            0.0714286    0.00860086
-8            0            0.0213021
-7            0.0714286    0.030203
-6            0            0.0426043
-5            0            0.0485049
-4            0            0.060306
-3            0.0714286    0.070407
-2            0.0714286    0.0774077
-1            0.0714286    0.0915092
0             0            0.10011
1             0.428571     0.0887089
2             0.0714286    0.0807081
3             0            0.070107
4             0            0.060406
5             0.0714286    0.0478048
6             0            0.0369037
7             0            0.0324032
8             0            0.0214021
9             0.0714286    0.0106011
--------------------------------------


RNG2:                                    0.156803
--------------------------------------

 
Post a Comment