Computer Methods and Programs in Biomedicine, 32 (1990) 147.

-149 147
Elsevier

COMMET 01094

Section II. Systems and programs

Computerized Mackworth vigilance clock test

R o l f W e i t k u n a t a n d Michael Bestler

Institut fiir Psychologie, University of Munich, Munich, F.R.G.

A computerized vigilance test, the Mackworth clock test, is described. The program runs on IBM-compatibleXT and
AT computers with a graphics adapter.

Vigilance testing; Mackworth vigilanceclock test; IBM personaJ computer

1. Introduction double-jumps occurring at inconstant time inter-

vals. Whereas building appropriate, hardly porta-
Vigilance testing is often necessary in experimen- ble hardware is both expensive and time-consum-
tal designs of general, developmental, clinical, and ing, the test is well suited for adaption on com-
other branches of psychology. In addition, testing puters where it can be executed without the active
subjects' ability to concentrate on a certain task is involvement of an experimenter. The procedure is
increasingly relevant in ergonomic and medical not language-dependent and can be used on dif-
sciences. For example, it becomes more and more ferent machines, in different applications, and re-
important to examine (side)effects of drugs on peatedly as it is rather insensitive to learning
vigilance, as compounds are further refined in effects. Finally, transferring the procedure to a
their (intended or unintended) operation upon the computer offers the advantage that computations
central nervous system. While there is a variety of on the results can be performed right away. Since
paper-and,pencil tests to do vigilance testing with, the Mackworth clock test can be described as a
apparative methods are increasingly being utilized. task of sustained signal detection, evaluation along
One reason is that apparative procedures usually the lines of signal detection theo%, is adequate
have a higher ecologic validity than rather abstract under certain conditions: in addition to the signal
and sometimes language-dependent paper-and- event (any irregular jump of the hand), the non-
pencil measures. signal condition must be defined. Since the event
One of the most widely applied apparative can theoretically happen during every unit of time
vigilance tests is the Mackworth clock test [1.1. that t h e h a n d is moving (usually every second),
Originally, subjects had to monitor a hand on a every regular jump would imply a non-signal event.
clock without a dial over an extended period of Therefore it is recommended that subjects are
time, usually 2 h. The task was to detect irregular informed about the frequency of signal events as
well as about the length of the testing period in
advance. In this way, the ratio of signal and
non-signal events becomes obvious and can be
Correspondence: Dr. R. Weitkunat, Institut fiir Psychologie,
Universit~it Miinchen, Leopoldstr. 13, 8000 Miinchen 40, used for the computation of signal detection scores
F.R.G. also. For example, if 30 signal events are to occur

0169-2607/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

148

during a 60 rain testing period, noise trials can be system can be found. Otherwise, a clock is dis-
set at 30 as well, since a signal event can statisti- played (function dra.ctock) and, for 1 h, the
cally be expected to occur every other minute. subject has to monitor the clock ticks in order to
respond to any irregular tick, that is either faster
or slower than the usual 1 s ticks. This modifica-
tion of the original Mackworth procedure was
2. Hard- and software
introduced to render habituation, learning, and
constitution of perceptual a n d / o r response sets.
During the development of a program to be used Every time some irregular clock tick occurs, the
widely, possibly on different computers, certain
subject has to press a key on the keyboard within
aspects need to be considered. Two important
5 s after the event. Fast and slow ticks occur
points are: (1) the programming language, and (2)
pseudo-randomly, 15 each. The sequence of fast
the type of computer to be used as the basis
and slow signal events is determined in the array
system for development. Usually, a good answer
s Loufast. At the end of the test, the results of the
to the first question is just C. Aside from im-
vigilance task are displayed on the screen (func-
portant structural advantages, the computing lan-
tion r e s u l t s ) , giving correct responses, false
guage C is now widely used by professional pro-
alarms, misses, and correct rejections, along with
grammers and it is highly portable, compared, for
the signal-detection parameters d ' and r , com-
example, to different Basic dialects. The answer to
puted as described by Brophy [2]. Computations
the second question is more difficult. One accepta-
of the latter two parameters are performed within
ble and practical solution is to uti]iTe an IBM-
functions dprime and catc_y_and_z. The pro-
compatible system. These machines are exten-
gram is well documented in order to facilitate
sively used in laboratories all over the world, they
adjustments of the above-described parameters.
are rather inexpensive and they certainly suffice to
Test length was set to 60 min, during which 30
do the present task. However, the fact has to be
irregular jumps take place. Since this information
taken into consideration that in cases where
is available to the subject too, the number of noise
graphic output as well as other machine- or oper-
trials can be set to 30 as well. The times array
ation system-specific language extensions are used,
initialized with pseudo-random numbers contains
portability is restricted to IBM microcomputer-
the times of signal-event occurrences. The num-
compatible machines, usually housing either CGA,
bers indicate the event-times in seconds from the
EGA, VGA or Hercules graphic adapters. Since
beginning of the test. The minimal latency be-
Borland's implementation of C (Turbo C) supplies
tween two events was set at 60 s. Function
a convenient graphics interface as well as drivers
for all of the above graphic adapters along with
key_check, which is called from the main func-
some convenient extensions to ANSI C, the pro- tion every second, checks for the correct acquisi-
gram was written with version 2.0 of this compiler. tion of hit and false alarm responses. Since only
one false alarm is accepted between two signal
events or between the beginning of the te~t and
the first irregular tick, a maximum of 30 hi'.s and
3. 3ystem description false alarms, respectively, can occur. Timing is
done without using assembly language routines
In order to perform the vigilance test, the subject (Biihrer et al. [3]) fully in Turbo C: the function
is seate .A. . .; n. . . f.r.~. ,...*.
Of the ,-,~mr~.*o~
v . . . . screen and
asked to read the instructions displayed by the increment the volatile variable bi o:s_counter ev-
program (function i n s t r u c t i ons). Recognition ery millisecond or - - at the end of the program
of the graphic adapter (function i ni t i a l i z e ) as - - to return to the original frequency. Function
well as adjustment of horizontal and vertical dis- de lay is called from the mai n function in order to
tances (function modify) is performed automati- realize exact timing of 1 s, 1.5 s, as well as 0.5 s
cally and the program stops if no or no adequate tick intervals.
 149

Since the function de tay is called after all US $. Additionally, you may send along a for-
graphical operations, after checking the keyboard, ~l.tp
matted .,4 IBM-AT disk (1.2 Mbyte).
and after evaluation of eventual responses, exact
timing of clock ticks is accomplished without any
interrupt-handling overhead. The program was References
tested to run on IBM-compatible XT and AT
microcomputers with Hercules, EGA, and V G A [1] N.H. Mackworth, Researches on the Measurements of Hu-
graphic adapters. man Performance. Special Report Series No. 268 (Medical
Research Council, London, 1950).
[2] A.L. Brophy, Alternatives to a table of criterion values in
4. Availability signal detection theory, Behav. Res. Methods Instrum.
Comput. 18 (1986) 285-286.
[3] M. Biihrer, B. Sparrer and R. Weitkunat, Interval timing
In order to receive a printed copy of the program, routines for the IBM PC/XT/AT microcomputer family,
please send a self-addressed envelope along with 5 Behav. Res. Methods Instrum. Comput. 19 (1987) 327-334.