Types of skill, and Rasmussen's SRK schema

(Note : In 'Types of Skill' the word 'skill' is used in the general sense of development of task ability from beginner to expert, not with the limited meaning of a specific type of task processing.)

I remember when Rasmussen first put forward his skill-rule-knowledge based model at a conference workshop.  There was so much disagreement between participants in the discussion, I was very surprised when he took it further.  His first publication on it was a Risø report in 1974.

There are several problems, mainly because :

- the words translate with slightly differently references in different languages, 

- different cultures have different attitudes to training and expertise,

- it is not an account of processing types which is adequate to help with complex task design.

Perhaps the word 'skill' in English translates in Danish into a word which means something more akin to the way Rasmussen uses the word.

In England since mediaeval times there has been, for practical tasks or ’trades’, a tradition of master-apprentice relations controlled by 'guilds', and people who have been through that system of training are considered 'skilled' and of higher status.  There are 4 levels of 'skill’ : unskilled, apprentice, journeyman, master.

While the 'professions’ (such as doctors and lawyers) are part trained in university/college, and part by on-the job training similar to apprenticeship. 

But people in many other countries don’t understand what we are talking about.  And in modern times anyway most people here no longer work at making things by hand, and that notion of 'skill' no longer has so much relevance.

There can be 3 general types of task processing : perceptual-motor skills, familiar cognitive skills, and problem solving skills - see the 'Types of Skill' Figure 6 in the Development of Skill paper and below.

And there are the 3 labels suggested by Rasmussen : skill based, rule based, knowledge based.

But different people associate different Rasmussen labels with different types of task processing.

For reasons to do with language and culture, take perceptual-motor skills : some people label them as 'skill based', some people label them as 'rule based' and some people as 'knowledge based'.  Similarly with familiar cognitive skills : some people label them as 'skill based', some label them as 'rule based'. . . and so on.

It’s possible to make a 3x3 table. I was delighted when I came across someone who filled the final cell in this table for me (putting forward an argument for calling perceptual-motor skills 'knowledge based').

The relation between the categories in 'Types of skill' and in Rasmussen's 'Skill-Rule-Knowledge based' schema

Lisanne Bainbridge

Department of Psychology, University College London

April 1989, unpublished note

Several people have pointed out that there are parallels between the types of skill that I have discussed (e.g. Bainbridge, 1989) and Rasmussen’s skill/rule/knowledge-based framework (e.g. Rasmussen, 1983).  The aim of this note is to acknowledge these parallels and to point out some difficulties.

There are two types of difficulty with Rasmussen’s S/R/KB formulation :

a.  the words ‘skill’, ‘rule’ and ‘knowledge’ are very rich and used with many meanings in different languages and cultures.  This has had the result that, although Rasmussen himself may use these words with consistent meaning, there is no consensus among other people about which of these categories any particular type of task behaviour might be assigned to.

b.  there are limitations to this schema as a model of human information processing.

This note will concentrate on the first of these issues, the use of terms.  

[For the limitations as a model of human information processing in complex tasks, see for example Bainbridge, 1997, and summary diagrams below.]

The usage of the terms 'skill', 'rule', and 'knowledge based' which will be mentioned have not all been used in published papers, but people have used them in discussion with me.  I have come across all the possible combinations - people who call rule-based what other people call skill, people to who call knowledge-based what other people call rule, etc.  This may occur because of differences in nuance when translating between different European languages, for example there is no simple way of translating the UK notion of ‘skill’ into French. 

There is some ambiguity about whether, by the word ’rule’, Rasmussen means behaviour which follows a standard sequence which has been developed through experience, or behaviour which follows instructions devised by someone else.  Actually ‘instruction following’ is not a unique type of cognitive behaviour but is a type of task, which may be done using any of the cognitive mechanisms, as mentioned in the last section.

Perceptual-motor Skills (LB)

These are actions using only senses and muscles which after practice become ‘automatic’, they require no conscious monitoring.

1.  Skill based :  Rasmussen apparently uses the word ‘skill’ in this limited sense, to refer to this type of behaviour.  Both in British industry and among psychologists, the word ‘skill’ has a much more general meaning, referring to the increased speed, accuracy and efficiency, even change in working method, which can develop in any type of behaviour after practice.  (I wonder whether the Danish word used to label perceptual-motor behaviour is translated into 'skill' in English, though 'skill' in English has wider implications.)

2.  Rule based :  some people say that perceptual-motor skills are examples of rule-based behaviour, on the grounds that this type of behaviour appears like, and can be modelled by, the operation of a production rule (if-stimulus-then-action).

3.  Knowledge based :  some people say that perceptual-motor skills are examples of knowledge-based behaviour, on the grounds that they can only be done by someone who has learned a great deal about the task, so they express a person’s knowledge of the task.

Familiar Cognitive Skills (LB)

All skills (task processes) other than perceptual-motor ones are cognitive, they involve cognitive (mental) processing, and build up working storage about the state of the task.  The clumsy term ‘familiar cognitive skills’ refers particularly to the type of cognitive skill which arises in repeated task situations, when a person has developed a standard method for doing the task, after experience.

1.  Skill based :  evidently this is a type of skill, as behaviour becomes more efficient after practice.

2.  Rule based :  as I understand it, this is the type of behaviour which Rasmussen is usually referring to when he uses the term ‘rule-based’.

3.  Knowledge based :  see above under perceptual-motor skill.  Also standard methods for doing a cognitive task can make use of back-up reference knowledge, for example a mental model of the structure or function of an industrial process, or knowledge about where its components are within the building.

Prototype-using Skills (LB)

Rasmussen does not include this type of behaviour as a distinct category.

Problem-solving Skills (LB)

People can be more or less skilled (in the sense of expertise) in reacting to situations for which a standard method of working has not yet been, or cannot be, devised.  People can develop : 

- problem solving strategies, and 

- frameworks for organising knowledge in a way which makes it convenient to refer to for a particular purpose.

And learning anything new involves problem solving.

1.  Skill based :  problem solving, i.e. dealing with non-standard situations, is behaviour which can improve with experience.  In the UK, the ‘skilled’ tradesman is someone who can formulate their own approach to making or doing something.

        The simplest problem solving strategy is trial-and-error.   Even learning perceptual-motor skills by trying to copy a demo involves problem solving - learning what to direct attention to, and setting the gains and lags of how to react to that information.   The perceptual-motor parts of the task may not be conscious, but there is conscious direction of the attempts to get the task right.

2.  Rule based :  some people assign problem solving behaviour to the rule-based category, on the grounds that it makes use of knowledge which is expressed in production rule format.

3.  Knowledge based :  Rasmussen himself uses the term ‘knowledge based’ to refer to these more knowledge-intensive types of behaviour.

Note on Instruction following (which is not always 'rule-based')

The task of following instructions illustrates the difficulty of assigning a particular type of task to any one of Rasmussen’s three categories.  It is frequently assumed that giving instructions to people minimises the amount of cognitive processing they have to do, but this is in fact not necessarily the case.

1.  Skill based :  Written or spoken instructions essentially cannot be followed by using perceptual-motor skills, as understanding language involves the use of working memory.  Instructions conveyed by picture or gesture may be responded to in this way.

2.  Rule based :  well designed instructions in familiar formats might be followed by using a standard strategy for interpreting the instructions.

3.  Knowledge based : understanding unfamiliar instructions can involve a considerable amount of problem solving,  see Bainbridge, 1989.  Even these days, when companies should know better, probably most people have had the experience of trying to use new electronic equipment or software and finding the instructions incomprehensible.

Summary notes added later, more emphasis on the differences between the two schemes :

Use of the three types of cognitive behaviour is flexible.  All tasks may start with being done by problem solving, and someone may return to this in unfamiliar circumstances :

Figure 6 from Development of skill, reduction of workload.

In complex dynamic tasks, cognitive processing is not passive/ bottom up/ input driven but active/ top down/ cognition driven, not linear but cyclic and contextual.

Figure 2 from The change in concepts needed to account for human behaviour in complex dynamic tasks.

The overview may be built up to contain information about : what is happening and why, what information is needed, what to expect to happen, what these imply for the task, what best to try to achieve, and how and when to take action.  And it provides the context for choice of later physical and cognitive activities, what to do, how to do it, and when to think and act.

Top-down processing includes actively looking for the information needed in thinking, while bottom-up processing reacts to, only happens because of, input signals.

- - -


Bainbridge, L. (1989)   Development of skill, reduction of workload.   in Bainbridge & Ruiz Quintanilla (eds) Developing Skills with New Technology,  Wiley.

Bainbridge, L. (1997) The change in concepts needed to account for human behaviour in complex dynamic tasks.

IEEE Transactions on Systems, Man and Cybernetics, Part A: Systems and Humans, 27, 351-359.

Rasmussen (1983)  Skill, rules and knowledge.  IEEE Transactions on Systems, Man, and Cybernetics,  SMC-13, 257-266.

© 1989, 1997, 2022 Lisanne Bainbridge


Popular posts from this blog

Ironies of Automation

Complex Processes Review : References