162 Joint Cognitive Systems
The first thing to notice from Figure 8.1 is that there are five rather than
four control modes, because a distinction is made between a tactical attended
and a tactical unattended control mode. This is in recognition of the fact that
people often seem to relax control of what they are doing when the situation
is not demanding. Consider, for instance, the following description of the
relation between attention and performance:
We see that highly learned, automatic activities may require little or
no conscious attention to be performed. These highly automated
activities can be disrupted, and when that happens they require
conscious control, often to the detriment of their normal performance.
(Norman, 1976, p. 70)
Highly learned, automatic activities are typically carried out in recurring,
hence familiar situations. This is almost the same as saying that predictability
is high; people therefore resort to routine and very often drift into doing other
things at the same time. The classical example of that is driving to work by
the normal route. While doing that, there is time to listen to the radio, talk on
the phone or plan the day ahead. The advantage is that there is spare capacity;
the disadvantage is that the reduced attention may lead to mistakes or reduced
reliability, cf. Hollnagel (1993a) or Reason & Mycielska (1982). The
familiarity of the situation and the lack of time pressure will in most cases
reduce the thoroughness of what is being done. Because of this, superficial
cues may trigger responses that in the situation are incorrect, such as making
a turn.
As Figure 8.1 also shows, the dependence of the control mode on
predictability and available time is not simple. This is reflected by the
distinction between attended and unattended tactical control. The model
represents the fact that if predictability is high, i.e., if the situation is very
familiar and if there is plenty of time, then operators will pay less attention to
what happens, hence go into a mode of unattended tactical control. The
difference between attended and unattended tactical control is that the former
represents a meticulous and careful execution of procedures and plans, while
in the latter people know what to do but do not bother to follow it in detail,
thereby becoming more prone to fail. (Note that the curves in Figure 8.1 are
theoretical, rather than empirical, i.e., they are there to illustrate the
principles. The control modes are theoretical constructs that are useful to
describe performance, but do not make any claim to exist in reality, and
certainly not as states in the brain or mind. Figure 8.1 is also misleading in
the sense that it may give the impression that the two axes, predictability and
available time, are independent of each other. This is, however, not the case.
For instance, if predictability is low, then it may take more time to think
through a situation, which means that less time remains available for other
Time and Control 163
things. Conversely, if there is little time available, then predictions may be
harder to make.)
The main features of Figure 8.1 are summarised in Table 8.1.
Predictability
As pointed out in Chapter 7, human performance relies on a mixture of
feedforward and feedback control. Feedforward control is essential to prevent
performance from becoming purely reactive with little or no opportunity to
consider the situation as a whole, hence to predict or plan ahead. In human
behaviour, feedforward is equivalent to anticipatory control, which requires
that the operator is able correctly to anticipate what will happen and prepare
himself/herself to act accordingly. Indeed, even in so fundamental a
phenomenon as perception, anticipation plays a crucial role. It stands to
reason, that the farther – and more correctly – predictions can be made, the
better will the operator be able to maintain control of the system. The ability
correctly to anticipate future events and developments depends on a number
of things, such as knowledge and experience, quality of information
available, designed support such as procedures, the regularity of the process
and environment, etc.
Table 8.1: Control Mode Characteristics
Control mode Available time
(subjective)
Familiarity of
situation
Level of attention
Strategic Abundant Routine or novel Medium – high
Tactical (attended)
Limited, but
adequate
Routine, but not
quite – or task is
very important
Medium – high
Tactical
(unattended)
More than
adequate
Very familiar or
routine, almost
boring
Low
Opportunistic Short or
inadequate
Vaguely familiar but
not fully recognised
High
Scrambled Highly
inadequate
Situation not
recognised
Full (hyper-
attention)
Available Time
The other main influence on the control mode is the available time. Since
more will be said about this in the following, it is sufficient for now to note
that when there is too little time available, then it will be very difficult to
make predictions (since that requires time), and perhaps difficult even to
respond to what happens. A shortage of time will impede feedforward control
164 Joint Cognitive Systems
and may even degrade feedback control. Conversely, if there is ample time
then it will be possible for the operator to consider in more detail actual
events and potential developments, and plan to act accordingly.
In addition to predictability and available time, there are other conditions
that may determine the control mode. One of these is how well the situation
is understood. If it is difficult to understand what is going on, for instance
because the information presentation is less than optimal, then control may be
lost even if there are no or only a few unexpected events. For the operator it
is necessary to be able to understand what happens, i.e., the present status of
the process, as well as what has happened and what may happen. What has
happened corresponds to the short-term or long-term past of the process,
while what may happen refers to the future. Knowledge of the past is
essential in order to be able to recognise conditions, diagnose events and
identify disturbances. Assumptions about future developments are necessary
in order to plan what to do, to decide on effective means of intervention, and
to make decisions about future actions.
THE MODELLING OF TIME
Time refers to the fact that we are dealing with systems and processes that
develop and change. The primary application area of CSE is situations, where
the rate of change of the process or target system is appreciable relative to the
duration of the activities under consideration, in particular situations that
change so quickly that it is a problem for the JCS to keep pace. (Second to
that, but much less studied, are situations where the rate of change is so slow
that it, almost paradoxically, is difficult for the JCS to sustain attention.) The
rate of change in turn depends on how the system boundaries have been
defined, and there are therefore no permanent criteria for when a situation is
dynamic or not. This means two things: first, that there is limited time
available to evaluate events, to plan what to do, and to do it. Second, that the
information that is used needs to be updated and verified regularly because
the world is changing. This is one reason why it is unrealistic to describe
decision making as a step-by-step process unless the decision steps are
minuscule relative to the speed of the process.
If unexpected events occur occasionally, there may be time and resources
to cope with them without disrupting the ongoing activities, i.e., without
adversely affecting the ability to maintain control. But if the unexpected
events are numerous and if they cannot be ignored, they will interfere with
the ongoing activities, hence lead to a loss of control. This is a result of the
simple fact that it takes time to deal with unexpected events. Since time is a
limited resource, the net effect will be that there is less time, hence less
capacity remaining to do the main tasks. Such situations may occur for all
Time and Control 165
kinds of JCSs, ranging from single individuals interacting with simple
machines to groups engaged in complex collaborative undertakings. Indeed,
it soon becomes evident that, regardless of domains, a number of common
conditions characterise how well they perform, and when and how they lose
control.
Time has generally been treated as a Cinderella in both human-computer
interaction and human-machine interaction (Decortis & De Keyser, 1988).
This is also the case for cognitive engineering and cognitive science, despite
the obvious importance of time in actual work, i.e., in activities that go on
outside the controlled confines of the laboratory. The proximal reason for that
is probably the legacy from behaviourism, carried on by human information
processing psychology, which focused on how an organism responded to a
stimulus or event, rather than on how an organism or system behaved over
time. The distal reason is the fundamental characteristic of the experimental
approach to scientific investigation, whether in the behavioural or natural
sciences, which is to expose a system to an influence and take note of the
consequences or reactions (Hammond, 1993). It is thereby assumed that the
behaviour being studied can be decomposed into discrete chunks without
affecting its functional characteristics.
While it has been known since the days of Donders (1969, org. 1868-69)
that mental processes take time, the speed of actions is more important than
the speed of mental processes. In other words, the interesting phenomenon is
the time it takes to do something, such as recognising a situation or decide
about what to do, rather than the time of the component mental processes,
particularly since these cannot be assumed to be additive except for specially
created experimental tasks (Collins & Quillian, 1969). One simple reason is
that it cannot be assumed that the duration of an action – to the extent that
one can talk about this in a meaningful way at all – can be derived by
considering the duration of the elementary or component processes. Even if
the internal workings of the mind were sequential, in a step-by-step fashion,
the combination or aggregation need not be linear. Human action is
furthermore not the execution of a single sequence of steps, but rather a set of
concurrent activities that address goals or objectives with different time
frames and changing priorities. For example, in order to make decisions, a
process plant operator needs to be able to reason about temporal information
and changes, to predict the effects of his actions and of the changes he
observes, continuously to make reference to what has happened, is happening
and might possibly happen, and to co-ordinate on the temporal axis the
actions of several users (Volta, 1986).
To illustrate the vast differences in temporal demands that any serious
study of JCS has to consider, Table 8.2 shows the profile of a number of
different domains and tasks (adapted from Alty et al., 1985). The domains
range from the daily and almost trivial (cycling and driving a car) to highly
166 Joint Cognitive Systems
specialised and demanding activities (flying an aeroplane or controlling a
power station). The entries are arranged according to the maximum number
of process variables. As Table 8.2 shows, this ranges from the very few to the
exceedingly many; other domains, such as grid control, may run even higher.
Two other important aspects are how often it is necessary for the operator to
do something and how much time is available to do it. The frequency of
operator operations almost falls into two separate groups, one where actions
are required very frequently if not continuously, and one where actions are
few and far between. The time allowed category is also very interesting, since
one domain stands out from the others, namely nuclear power generating
stations. The 30-minute respite to act found here is a deliberate solution to a
serious problem, as discussed in Chapter 4.
Table 8.2: Temporal Characteristics of Different Domains
Process
type/domain
Number of
process
variables
Frequency of operator
actions
Time allowed
for operator
actions
Cycling 2 (speed,
direction)
1/second (manoeuvring)
1/minute (coasting)
Direct
Car driving < 10 1/second (heavy traffic)
1/minute (light traffic)
Direct
Steel rolling
mills
< 100 1/second Direct
Aviation 100 - 300 1/minute (landing)
2-3/hour (cruising)
1/second (manual flight)
Direct
Electronic
trading
~500 – 5,000 1-4/minute << 1 minute
Process
industries
2,000 – 10,000 5-6/hour (sometimes
clustered)
< 1 minute
Nuclear power
generating
stations
10,000 – 20,000
1/hour (usually clustered) 1-30 minutes
Representation of Time in COCOM
Effective control requires that the operator – and more generally, the JCS
controlling the process – is able to make sense of the available information
and in particular possible unexpected events, as well as able to find, choose
or generate appropriate actions or responses. The model shown in Figure 8.2
comprises several sets of time that affect the ability to remain in control. One
set comprises the time needed for various parts of an action, i.e., the
durations of the components of an action. These are the time needed to
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.