3c. 'Routines'

Topics in this section :

3c.1. Dividing reports into routines, identifying the underlying processing.

3c.2. The main properties of 'routines'.

3c.3. Additional mechanisms in 'routines'

3c.4. The main 'routines' in operation of an industrial process, and the working storage and knowledge bases they use.

Section 3c : Meeting the cognitive needs 3 : 'Routines'

Building up behavioural complexity from a cognitive processing element

Lisanne Bainbridge

If the information required to meet a cognitive need is not immediately available from the environment or the knowledge base, then some cognitive processing needs to be done to find it. The approach to cognitive processing used here assumes that cognitive needs are progressively expanded into more detailed cognitive needs, until the point at which some basic mechanism is reached, such as comparing, calculating, or using the perceptual-motor system (reasons for this approach in Section 2).

(I have not determined the principles defining this lower limit. When simulating the verbal protocol, the limit was practical rather than principled - I did not go into further detail when there was no evidence in the protocol about what the detail might be.)

Going upwards into bigger groupings, cognitive needs are grouped into 'routines' which meet a cognitive need at a higher level.

This section adds some points about the nature of these 'routines', and also outlines the main cognitive needs in the furnace power control task, as an example of a complex dynamic task.

The word 'routine' is often in inverted commas, because this grouping of cognitive needs is more flexible than a routine in a conventional computer programming language, as discussed below.

'Routines' and working storage

Working storage often does not include raw data representing the exact nature of the environment. Much of what is in working storage is the result of processing, of working through cognitive elements in a 'routine', so the content of working storage is a construct. For more on these constructs, see Section 6a.

Two ways of finding this 'constructed' information appear in complex cognitive tasks :

A. develop new constructs by new thinking/ cognitive processing :

The nature of the 'routines' for doing this cognitive processing, which have been found in process operation, are discussed more in this Section.

Example from process control : such as Tables 2.2 & 2.3 - those nuclear power station operators were inferring and predicting - inferring the state inside a process which can’t be observed directly, and predicting an event or needed action.

Examples from air-traffic control :

Bisseret 1970 used a map drawing task to study controllers using paper flight strips [the task uses different technology now], and controllers made comments such as :

'it's flying from Dijon to Rolemport, estimated time of arrival Rolemport 05, so it must be about here'.

'I've got one at level 150 which is about to pass beacon RLP and another at level 170 which is about 10 minutes behind so is about here'.

They were constructing map positions from timing information, see more on this in Bainbridge, 1975.

B. use constructs previously developed, by referring to the result of previous thinking :

This is shown in 2 ways in previous figures :

- indicated by arrows in boxes, e.g. Figure 2.3 (see more on these below).

- explicitly in full, e.g. Fig.2.7.

This referring to the results of previous thinking is discussed more in Sections 4-6.

3c.1. Identifying behaviour as being part of a 'routine'

So far the examples have only been from short sections of behaviour.

For example, Table 2.1 and Figures 2.2 and 2.3 describe the furnace power controller’s cognitive behaviour done when predicting the time of an event. That behaviour has been inferred from a section of verbal protocol, and consists of subsidiary activities, which combine together into a method for finding the time.

Tables 2.2 and 2.3, the second example in Section 2 (nuclear power station operators reacting to a fault), are not from a verbal report given at the time of doing the task. They summarise operators’ cognitive behaviour derived from interviews, and each shows a sequence of different types of activity. Each line in these tables is about a different type of activity, so these would each be identified as separate 'routines'.

Here is a longer example of thinking, from another complex process control task

Table 3c.1 is part of a verbal protocol from a gas-grid controller, collected by Umbers (1976).

The main text is the protocol.

The bold comments are an analyst's inferences about the main cognitive need being met by the prior group of phrases.

(The reasons why the operator thinks about the cognitive needs in this order are not immediately obvious, see more on this in Section 4.)

This is another example (as in Figure 2.1) of how the analyst’s natural language understanding and knowledge of the task can be used to divide the protocol into groups of phrases on the same topic, and to infer what that topic might be.

- - -

Table 3c.1 : A section of a gas-grid controller's protocol. Adapted from Umbers' study of gas-grid control, Time 22.07. (1976, p. 321) (Bainbridge, 1992, Table 3)

I'll give you a quick resume of what the forecast was this morning,

that is forecast estimate,

and the estimate for this morning was 75 of MG [manufactured gas] and the NG [natural gas] send of 150 m.

predict state

the stock aim on both MG is 62.6 and NG 180.7

review goals

we have 3 plants at work, the 3b at C/hill, the 1c at Tipton, and No. 3 at WH,

this gives us a total cum make per hour. of 4.2 approx without B/ air

review present state

B/ air will give us something in the region of -

sorry, a correction to the total available without B/ air would be approx 4.4

and there's 0.8/hour. of B/ air,

so that we can in fact make 5.2 with full NG,

This is something in the region of 129-130 for the day.

review action availability and effect

we don't contemplate using this much

evaluate

but there is the facility for taking NG off the reforming and putting Naphtha in,

which will give us another increase of something in the region 0.30-0.35/hour.

review action availability and effect

so that we're not too badly off

evaluate

if the weather forecast which they give for tomorrow has been mostly sunny and warm,

with a max of 68 and a mean of 58 F.

predict goals

we shall cope with the send on MG with what we've got quite admirably.

evaluate

- - -

There are two features of this to emphasise :

- the controller is reviewing his plans for the day, what he needs to achieve and how to do it. As in Table 2.3 from the nuclear power station example, this controller is using anticipatory control, not feedback.

- it is the change from one topic/ type of cognitive activity to another that defines the 'routines’ for the analyst, but there is no information about why the topics are talked about in this order. The controller does not mention in his verbal report the reason why he changes from cognitive activity to another. This has to be identified by the analyst (see Section 4).

However, it turns out that what the operator does next is determined by the results of previous thinking. These issues are discussed more in Sections 4-6.

How has the underlying 'routine' been identified

The 'routines' in the cognitive processing are identified by the analyst’s natural language understanding and knowledge of the task being done. Those enable the analyst to group phrases together as being about the same topic, they are part of one purpose/ aim/ need.

In the next level of analysis, all the examples identified as being the same type of cognitive behaviour are combined as evidence for the cognitive processing used in this aspect of behaviour.

Returning to the furnace power control task : Section 2 contains a small detailed example of how the processing, underlying two sections of verbal report meeting the same cognitive need, was identified.

Figures 2.1-3 include two examples of the same behaviour. Both sections consist of a sequence of items all working towards the same aim. All lines of this figure are considered to be part of the same 'routine'.

The protocol phrases are in Figure 2.1.

The possible underlying processing is in Figure 2.2.

And a 'box element' representation of this processing is in Figure 2.3.

This type of analysis was done in detail for the entire verbal report from one experienced furnace power supply operator who controlled [a simulation of] the allocation of electric power to steel melting furnaces over a period of several hours.

(Several other experienced and inexperienced operators took part in the study, and their verbal reports were studied in less detail.)

So the examples of cognitive processes in the rest of this paper are based on more complete data than the examples in Section 2.

In developing the paper-and-pencil simulation of this protocol, the 'routines' were identified from repeated sections of the protocol. I divided the protocol phrases into groups in which the phrases appeared to be concerned with a common topic. Then I compared groups of phrases which appeared to be concerned with the same topic, and which were similar in content. The 'routines' were then developed to describe, or generate, these groups of phrases. Bainbridge (1972) contains a full account and all the box element diagrams, Section 2, part 2.2 and Bainbridge (1974, 1990) contain a simple example of the analytic method.

Above the lower limit, any grouping of cognitive needs which (when executed) meets a higher level of cognitive need has been called a 'routine', and is represented in the figures by cognitive elements linked by a dashed line, as in Figure 2.3.

3c.2. The properties of a 'routine'

A listing of the special properties which may be found in 'routines' is below.

As an example of the need for special mechanisms in the representation : the items in a 'routine’ often do have to be done in a specific order, but not always.

repeat of Figure 2.2 : Combining information from 2 samples of the same activity, from Figure 2.1 phrases 5-14, to complete an account of the processing underlying the verbal reports (Bainbridge 1974, 1985).

For example, in Figure 2.2 repeated above, it is not possible to calculate Time to go until the person knows the stage length and the time so far.

However, the time so far could be calculated, and the stage length could be read, in any order. Sometimes the person does such alternatives in any order, sometimes only in one order, and those might be represented in different ways in a description of the processes underlying the behaviour (see more below in 3c.1).

That is a simple example of a 'routine'. There were over a dozen needed to describe the furnace operator’s thinking (Bainbridge 1972), see Table 3c.2 below. It wouldn’t be all that interesting to include them all here in detail. However there are some general properties which it is important to mention.

The full range of 'routines' need to include at least five special properties.

1. As just noted, the order in which the cognitive needs making up a 'routine' are met is not rigid. The cognitive needs are only met in a certain sequence if meeting some need requires the results from finding another need, otherwise the protocol evidence suggests that the thinking can be done in any order. This implies some sort of underlying parallel representation of the 'routine', which appears in sequential form only when it is actually carried out. That is, sequential processing emerges as a property of behaviour when a working method is actually used, rather than being inherent in the underlying representation which generates the behaviour. The simplification into a serial form has also just been used to produce a simple diagram. This flexibility is indicated by using a dashed line to link the elements in a 'routine'. Or by a form of brackets, see Figure 3c.1 below.

2. The protocol also suggests a second way in which a 'routine' is not a rigid sequence. The person may be able to interrupt the 'routine' at any point defined by the completion of a cognitive need, go and do something else, and then return to this point. As there were few interruptions in the furnace power control data, there is not much direct evidence on whether this interruption can occur in any level of 'routine', or only at 'higher' levels. Or on how well the interruptions are dealt with. This is discussed further with an example in Section 6.

3. The three types of link between the processing elements in a 'routine' (directly between its elements, to a compatibly structured knowledge base, and the cross-references to working storage, all discussed in Section 2) make three interconnected and interdependent networks - of the cognitive needs in a routine, of the links to and within the knowledge base, and of the cross references in working storage. These three networks are mutually reinforcing, and support the structure of a 'routine' as an entity.

4. The items seem to have an integral part-whole relation to each other, in making up the 'whole' of meeting the higher level need.

There seems to be an inherent stopping rule in many aspects of human cognition, as pointed out by the Gestalt psychologists. There is some in-built recognition that items together make a whole which is greater than the sum of the parts. In some complexity theories, forms of larger organisation appear as emergent properties when large numbers of simple items are active together. It is not possible to say, from the evidence available, whether larger forms of organisation would emerge from these box elements in action, via the properties outlined above, or whether their organisation into 'routines' can only be accounted for by adding another mechanism, such as one for recognising that given elements are sufficient to meet the need : 'it works'.

5. This will not be discussed fully until Section 7, but it is suggested a group of elements making up a working method which meets a cognitive need (i.e. a 'routine') can have meta-knowledge associated with it about the general properties of this working method, such as how long it takes, how much effort it involves, and what general type of result can be obtained. This meta-knowledge may be used in choosing between alternative methods for meeting a cognitive need, depending on the context. This meta-knowledge, and the independence of the cognitive need and the 'routine' for meeting it (the link represented by the stepped arrow) are powerful factors which will be discussed more fully in Section 7.

These points apply to well established 'routines' in cognitive skill. The element described in this paper does not provide a full mechanism for learning, but it does lead to some interesting points about learning which will be discussed more in Section 8.

No doubt the routines which a person knows are very much a function of their task knowledge. One person may have several independent collections of things they can do. For example I can sew a french seam without needing to look up the instructions, but doubt whether I could bake a good eclair even if I did look up the instructions.

3c.3. Additional mechanisms in 'routines'

The 'routines' described so far consist of linked cognitive elements. More mechanisms can be involved in 'routines'. They were not needed to account for the simple fragments of steel works and nuclear operator behaviour discussed in Section 2. But they were needed in the whole collection of routines underlying the list in Table 3c.2 below.

1. Alternative sequences.

2. Conditionals.

3. Iteration.

4. 'Two arrow’ box.

5. Double-line arrow

1. Alternative sequences. As mentioned above, in many routines it is necessary to do a particular group of behaviours to meet a cognitive need, but not to do these in any set order. For example in Figure 2.2, before calculating the time when a furnace stage will end, it is necessary to find out the time when the stage started, and the length of time the stage will take. But it does not matter in what order these two needs are met, and the protocol mentions them in different orders in different examples of this behaviour.

In the box element diagrams a bracketing arrow may be used to indicate this parallel storage and freedom of sequence, as in Figure 3c.1. Here it is assumed that the choice of the order in which to do the items is random. There are more complex and more structured sequencing decisions at a higher level, as discussed in Section 4.

Figure 3c.1 : A possible way of representing the freedom of sequence in which processes are done. Compare with Figure 2.3 in which this processing is described as a set sequence.

(I don’t think this way of representing this sequence flexibility is ideal, as it could be taken to mean that the two lines of processing are done in parallel, which is not the intention.)

There are also several symbols in Figure 3c.1 which are combined from earlier figures and need to be emphasised.

a. The representation of cross references in working storage has been simplified.

A box containing a thin arrow indicates that its contents are found by reference to working storage earlier within this routine, i.e. to local working storage.

A box containing a thick arrow indicates cross reference outside this routine, to data available elsewhere in the processing (for more on this see below, also Sections 4 and 5).

b. a doubled box indicates the main cognitive need found by this routine.

c. The underlined words (italic in some figures), e.g. read, describe basic actions.

d. The parts of the routine written in UPPER CASE are explicitly mentioned in the verbal protocol. The parts written in lower case have been added by the analyst, fromthe inference that these processes must have been done, or the operator would not have been able to say the things which he did say.

2. Conditionals. A choice between alternative activities is represented by a conditional version of the box element. There is a simple example in Figure 3c.2 : the conditional question is indicated by an oval shape (as was often done at the time). The cognitive need, tested in the condition, is met either by reference to working storage, represented by thick arrows in Figure 3c.2, or by carrying out a 'routine'. Once the data has been obtained, processing continues, depending on the result of the conditional test. (For more on conditional elements, see Section 4.)

Figure 3c.2 : An example of a conditional element (Bainbridge, 1972, Figure 7.2.6).

In this type of conditional, the information needed on which the decision is based, may be found by cross reference to data already available from some other part of the thinking, as shown by the thick arrow.

Or it may be found when needed by doing some processing, examples in the next figure and see the 'two arrow box' below in 3c.4.

If the data is already available, there usually isn’t any evidence from the verbal report that the decision represented by the conditional has been made, except from what happens next. It appears to be made automatically, without conscious attention.

3. Iteration. Sometimes the protocol includes processing which is carried out on several items. This iteration does not appear to require any unusual principles. It can be represented simply by a multiple box, as in Figure 3c.4. (For more on multidimensional working storage, see Section 6.)

Figure 3c.3 : An example of iteration (Bainbridge, 1972, Figure 7.2.5). The furnace operator is checking which of the furnaces in the melting stage have already had an action made on them.

4. 'Two arrow' box. As mentioned above, for some cognitive needs the data needed was obtained sometimes by carrying out a 'routine', sometimes by cross reference to data obtained recently by other processing. This is represented in the diagrams by a 'two arrow' box, i.e. a box which both contains an arrow indicating cross reference, and has a stepped arrow for calling on a 'routine'. This representation means that working storage is referred to if the data is available, otherwise the 'routine' is carried out. For an example, see Figure 3c.3 above (melt list) - this list may already been have found during a previous use of this 'routine', or it can be obtained by reading the display.

This 'two arrow’ mechanism is the basis for several important aspects of behaviour discussed later. In particular it means that either serial or parallel processing can emerge from the same mechanism, depending on circumstances.

Figure 3c.4. Generalised examples of "two-arrow box".

A visual reminder that the information meeting a need can be obtained in several different ways, depending on what has been done previously.

This type of box may appear alone (A) or within a conditional (B).

The observed cognitive behaviour requires there are underlying mechanisms for several ways that the current value of a variable can get into working storage (which in these diagrams is represented by a box), and these are represented by different symbols in the diagrams :

- it is already 'in the box' as the result of previous processing at this point.

- it can be obtained by cross reference to the result of previous processing elsewhere, which is stored in another 'box’. The cross reference is represented by an arrow in this box.

- it can be obtained by doing some cognitive processing. The kinked arrow represents a possible goal-means link to the method used for finding it.

The right (output) box in a box element is sometimes empty, sometimes contains an arrow, which indicates that the information needed has been found in some previous processing.

There are two versions of this arrow :

A single-line/ thin line arrow in a box indicates reference to data available locally (in the same diagram/ 'routine').

A double line/ thick line arrow links to data found elsewhere in the processing.

In Figures 3c.1, 2 and 4 this mechanism is shown by a thick line arrow, but in many other figures it is shown by a double-line arrow.

This data elsewhere, and so the thick arrows, indicate a key part of the 'overview', the stored information available to most of the processing and providing the context for deciding what to do and how to do it. See particularly Sections 5- 6.

Figure 3c.5 : Double-line arrow in box.

These symbols are not intended to imply the actual way this done by neurones. Or even the way these mechanisms might be realised in a particular computer programming language. This symbol just indicates the type of mechanism which is needed to realise the observed cognitive processing. This symbol is not intended to imply there’s a distinct (group of) neurone(s) which store the current value, cross referenced by two different types of dendrites which link to other groups of neurones. We don’t know enough about how the brain does the amazing things it can do to make any sort of inference.

Figure 3c.3 above represents a small 'routine' which uses all four of the additional mechanisms and symbols. The aim of this 'routine' is to make a list of the furnaces which are currently both melting (shown by a red light) and using less than 100% of the optimum electric power. If a furnace is charging, and so using no power (0%), the operator notes important features of this stage. A flat serial diagram is very inadequate as a presentation of the dynamic and adaptive nature of the actual cognitive processing in action.

Section 3c.4. The main 'routines' in process operation.

The Working Storage and knowledge bases used by the 'routines'

As noted above, I used language evidence to divide the furnace operator’s verbal report into segments each serving a different cognitive function.

I then brought together all the instances of a given function, and identified the processing underlying this behaviour (analysis as described in Section 2).

In the furnace power supply control task, I found more than a dozen main 'routines' (see Table 3c.2 below). These main cognitive needs (predict, evaluate, etc,) can be found in many other complex dynamic tasks. Though of course in different tasks the details of how these needs are met are not in exactly the same form - compare the bold activity labels in Tables 3c.1 and 3c.2.

The following Table 3c.2 extracts, from the box element diagrams describing each type of cognitive activity (Bainbridge 1972), the main characteristics of each type of activity.

Each part of the Table has 4 sections summarising the main features of this aspect of the task thinking - the working storage and knowledge bases involved.

Note that these items have not been extracted from raw data on the behaviour. They are extracted from the outcome of a sequence of analyses :

- divide protocol into segments.

- bring together the segments about the same behaviour

- combine data from the different instances to extract the underlying processing (e.g. Figure 2.2).

- describe this underlying processing in a box element representation (e.g. Figure 2.3), a 'paper-and-pencil’ simulation.

- identify from this representation : the working storage and knowledge bases used/ needed/ produced by this type of behaviour.

- - -

Table 3c.2. The working storage and knowledge bases used by each of the furnace operator’s 'routines'

Key to indents in the Table :

WS : the (input) information describing the current state which is already in working storage - the previously obtained information needed by this 'routine'.

KB : the information provided by the knowledge bases.

MODULE : the name/ purpose of this type of aspect of processing (called a 'routine' elsewhere in this review).

WS : the output of the processing ; the result, plus associated future events, information needs, and actions.

Table 3c.2. The working storage input and output, and the knowledge bases used, for each 'routine' in the furnace power supply control task (inferred from the simulation).

- - -

This Table shows the main 'routines' used by the furnace power supply controller.

Several of these cognitive needs arise in nearly all process operation tasks (compare bold labels in Table 3c.2 and Table 3c.1). Though of course the details will differ greatly in different tasks. The furnace power supply controller could not immediately be a good gas grid controller nor a nuclear power station operator. Though he might well be able to recognise and understand what they are doing more easily than a general member of the public could.

The furnace power supply task also does not include all the types of thinking involved in all process operating tasks especially, for example, tasks in which multi-dimensional situations develop over time.

There can be no claim that the listing of 'routines' in Table 3c.2 is complete for all process operating tasks. And certainly the knowledge bases referred to will be different, also details of the processing.

Some of the following Sections of this review use extracts from the information in Table 3c.2.

The Knowledge Bases used are discussed more in the next Section, 3d.

The Working Storage features will be discussed more in Sections 4-6.

- - -

The sequence of 'routines'

As already described, the controller’s thinking has been divided into 'routines' by identifying (by natural language understanding and knowledge of the task) when the speaker changes from one type of cognitive activity to another.

The operator is doing a control task, so keeps repeating the same type of behaviour (so there are several instances of each behaviour, which made this detailed type of analysis possible). He needs to keep track of where he is in his thinking, and what he has thought about before (if it is still relevant).

These activities don’t always appear in exactly the same sequence, so what does determine the order they are done in ? how does the person decide what to do next ?

Section 4 is about what determines the sequence in which topics are thought about, and how this was identified. That Section also introduces the concept of the 'overview'.

Sections 5a and b describe some of the 'sequencers' found.

Section 6 discusses the overview which links them together.

- - -

Summary of main points in Section 3c

* Cognitive needs are often met by processing 'routines', themselves made up of cognitive processing elements, to some lower limit of basic processes.

* Working storage contains the result of the thinking which meets a cognitive need, so contains transformed data, not simply a representation of the outside world.

* These 'routines' have different properties from those of a conventional programming language.

* The two-arrow box symbol represents the way in which, if data is not available from working storage, it is met by carrying out a 'routine', i.e. this mechanism may produce serial or parallel processing depending on the circumstances.

* The processing element can be in a conditional form.

* The elements in a 'routine' have a part-whole relation to each other.

* The three types of link in a 'routine', between its elements, between the items in working storage, and to a compatibly structured knowledge base, make three interdependent and mutually reinforcing networks.

* The cognitive needs and the cross references structure the working storage. The contents of working storage are available in parallel, and are structured by their relation to the needs they meet and the processing used to meet them.

* The order in which the cognitive needs making up a 'routine' are met is not rigid, serial processing is an emergent property from a parallel representation.

* A complex dynamic task involves several main classes of cognitive need.

Access to other papers on this site via the Home page.

In this review, there are 3 main groups of topics.

1. Introduction.

Basic element, sources of data which meet cognitive needs.

2. The cognitive processing element.

Meeting an information need :

3a. by finding it in the environment.

3b. from a stored knowledge base.

3c. by working through a 'routine',

or referring to the result of using a routine elsewhere.

3d. more on knowledge bases.