sfwork logo

Systems Action in Organisations


by Mark McKergow


This paper addresses the differences between first-level learning (change within a framework) and second-level learning (change to the framework itself) in organisations. Examples of both types are given. The use of systems thinking is considered in relation to organisational learning, and the impact of system complexity on the ways we examine and use systems concepts is discussed. Complex systems behaviour is non-deterministic, and therefore obtaining feedback from the system in action is the most feasible way of working with it. Guidelines for second level learning based on acting systemically, rather than thinking systemically, are given.


This paper is part of the fourth Learning Company conference. "Organisational learning" is one of the key management topics of the 1990s. The idea of an organisation which learns from its experience, its environment and its people is an attractive one. And yet, if we were to stumble across an actual bone fide learning organisation, how would we know? What general types of activities would we expect to see in action?

Mike Pedler, John Burgoyne and Tom Boydell, in their work on the Learning Company, have given glimpses of their own ideas of the sort of things such a company would do. They have defined a learning company (Pedler et al 1991) as;

"a company which creates learning opportunities for all its members and is able to transform itself as a whole."

Peter Senge, in his book "The Fifth Discipline" (Senge 1990), has perhaps gone further in terms of describing what a learning organisation would do. He describes a learning organisation as;

"a place where people are continuously discovering how they create their reality. And how they can change it."

Senge's "Fifth Discipline" is systemic thinking; viewing the organisation and its environment from a systems viewpoint. However, he stops short of actually telling us how we can use systemic thinking in changing and building our organisations. In this paper, I seek to explore in a little more detail the ways in which systemic principles can be used in organisational learning, and to start to produce some principles for putting the systems thinking into systems action.


Central to the ideas presented here is the distinction between first-level and second-level learning (McKergow, 1994), or more generally first-order and second-order change. The idea of first and second order change is not new. Gregory Bateson and co-workers (Bateson et al, 1956, &Bateson, 1971) pointed out the fundamental differences based on logical type difference. Watzlawick et al (1974) describe the differences in some detail, with many excellent examples. On the organisational front, Argyris and Schon (1978) described the same principles as "single loop" and "double loop" learning respectively. In all these cases, the key difference is that between change within a framework (first order/singleloop) and change to the framework itself (second order/double loop). The framework may be seen as the set of governing variables which are invariant during first order change.

Let us illustrate the difference with a sporting analogy; soccer. Imagine a football team, whose express aim is to win their matches. After an unsuccessful match, when they lost and scored no goals, they decide to learn from their experience. They might train harder, get fitter, practice taking free kicks, or change their formation. All of these may well improve their performance in future games. They are "learning to play the game better". This form of learning or change is known as "first-order", as it is carried out within a certain framework of rules, in this case the rules of soccer.

Let us contrast this with the kind of learning or change which results from a change to the rules, rather than simply a change within the rules. In our soccer example, the players might experiment with picking the ball up and running with it, and thereby play a new game. (This actually happened at Rugby School in the 19th Century, of course). This is "second-order" learning or change. Having established a new set of rules, the players would then probably spend time doing more first-level learning, to perfect their new game.

Note that second-level learning or change happens far more in real life than might be supposed. An example from the soccer scene is the recent change in rules which makes illegal back-passes by the defenders to the goalkeeper (a second order change). In the early weeks of the 1993/4 season there were some spectacular mix-ups and confusions as the players got accustomed to the rule (first level learning), together with calls from some that the old rule should be reinstated. However, within a few months all were accustomed to the new rule, and everyone was playing a (slightly) different game with no problems.

So, we now have two levels of learning which can affect what happens on the field of play; first-level, which is "learning to play the game better", and second-level, which is learning to change the rules which govern what happens at the other levels". This is illustrated in Figure 1.


How are the ideas of first and second level learning relevant to a learning organisation? A story may help illustrate:

"I was walking home one evening when I came across a drunk crawling around near a lamp post. He said that he had dropped his keys, and asked me to help him look for them. We searched fruitlessly for about ten minutes, whereupon I asked him about the circumstances under which the keys were lost. The drunk replied that he had dropped the keys under the hedge some yards away. "So why are you looking here?", I asked. "We'll never find them there - there's no light. Here, at least we can see what we're doing!".

However hard we search for the keys under the lamp post, we're very unlikely to find them there. No matter what we do, how long and how carefully we try, how much time, effort and resource we put in, we're not going to find the keys as long as we keep looking in the wrong place. Now, can we sometimes keep trying and keep trying to achieve something in our organisations, and end up thinking we've tried "everything" and not succeeded? When in fact, we were looking under the wrong lamppost. Unless we can know how to move away from the lamp post we're at (ie, do second level learning), we may well end up like the drunk - stuck.

In a learning organisation, we might therefore expect to see things happening at all three levels, as shown in Figure 1.

* The "doing" level, relating producing our products, carrying out our purpose and satisfying and delighting our customers.

* The "first learning" level, relating to improving what we do at the "doing" level.

* The "second learning" level, where we change the rules which govern what happens at the other two levels by defining the "game".

The rest of this paper considers what these "first learning" and "second learning" activities might be.


A lot of what is currently classed as "organisational learning" happens at the first level. The basis of TQM and quality improvement programmes often focuses on learning to "play the game better", by improving what already happens. Processes like the Deming cycle of"plan - do - check - act" and David Kolb's experiential learning cycle (experience - reflect - generalise - plan) are excellent templates for first level learning. In a recent review of "building a learning organisation", David Garvin (1993) points out the benefits of following such examples, and concludes that managing the learning process and measuring what's done are the key points. He cites organisations such as Boeing, Xerox and General Electric, and processes such as benchmarking, system and process audits, information exchange and strategic reviews as being those which a learning organisation would undertake. These are all examples, in my view, of excellent first-level learning.

First-level learning can be characterised by working on a "logical, rational and linear" basis. It takes the classical Western scientific paradigm (that all objects and events can be understood in terms of their ultimate elements, analysis and reductionism is the key to understanding, and that cause is necessary and sufficient to define effect), and uses those assumptions to focus on improvement. And very good results have been found. If you're not doing first level learning, please start. If you're already doing it, please carry on. There are lots of good improvements to be had from these methods. However, you may have tried some of them and not had the success you wanted, or that your consultants promised. Or you might want more change or learning than you've been able to get so far. Keep reading!


Earlier in this paper, we defined second-level learning as changing the rules within which the doing and the first-level learning takes place. So what are these rules anyway? They are the things which remain inviolate during the activity we are considering, and so can take many forms. They are usually not formal rules in the sense of being written down, but the players act as if these rules were in place.

The rules can be (for example):

* the patterns of behaviour exhibited by some or all players at certain times or events.

* the relationships between interlinking parameters or phenomena.

* the framework of understanding and belief by which individual players make sense of the world around them, or

* the framework of collective understanding built up between various players during past events and exchanges.

Because we all make sense of the world through our "mental models" (as described by Argyris &Schon and Senge amongst others), we have to move our minds to a higher level to start to appreciate the differences our models make to the ways we think and act, and to be in with a chance of changing our own models and those of other people. This shift of mind takes us above our own perceptions and into a world of systemic phenomena.


The theme of this conference is "dialogue". As I understand it, there are links between the shift of mind of becoming aware of one's own models, and the practice of dialogue. If dialogue is a process of building a shared understanding by means of two-way communication, then it is also about changing our models as other people change theirs, so that a shared understanding results. Suspending our assumptions about the subject and people concerned seems to be a key part of establishing dialogue. Dialogue can therefore be a means to second level learning (if practised well and carefully).


The systemic paradigm is a rather different one to the classical Western scientific paradigm I mentioned above. It came to prominence in the 1950s via cyberneticists Norbert Wiener and Ross Ashby, and has gathered pace in the last 20 years with developments such as chaos theory and complexity in the physical sciences.

Whereas the classical paradigm has that cause is necessary and sufficient to define effect, and that complete understanding is possible with complete data and analysis, the systemic view is based in the idea of feedback loops. With feedback, things can influence many other things, including themselves. When we start to examine the world, we find many examples of feedback in action. A simple example is that of a central heating boiler with a thermostat. The boiler runs until the room is heated up, whereupon the feedback mechanism turns off the boiler. The heat in the room "caused" the boiler to go off. Or was it the boiler running in the first place? So the boiler "caused" itself to go off? The simple "cause and effect" model is insufficient to handle looping phenomena.

There are three basic kinds of systemic loop:

Negative feedback (or balancing) loop: The system is stable, and small perturbations are damped out. Example; thermostat.

Positive feedback (or snowballing) loop: The system tends to accelerate away from the starting point. Example; stock market bubble.

Feedforward loop: This is less widely recognised than the feedback loops described above. Feed forward is where the expectation of what will happen influences what actually happens. Example; Expectation of a petrol shortage can lead to panic buying, and a "genuine" petrol shortage, even though the supplies of petrol are uninterrupted.

The concept of leverage is also important in thinking systemically. In our customary, linear thinking a small change usually has a small effect. But given the ability of systems to amplify or damp out change, we can find that a small change in one part of the system leads to large changes in other parts. The "butterfly effect", where the flap of the wings of a butterfly over the Indian Ocean can set in motion winds which build up, over the course of several months, into a hurricane threatening the Florida coast, is an example of leverage.

The final point to bear in mind when thinking systemically is that the normal "cause and effect" thinking is insufficient. We seem to grow up considering the world "out there", which we can "do things to" and sometimes "does things to us". Systemically, we have to consider the ways in which we ourselves are part of a set of systemic loops and links, and how our own thoughts and actions are linked to other things, which may in turn link back to our actions again. The example of an office dispute (some of you may be familiar with this!) is instructive:

Supervisor: "I have to keep an eye on him, to make sure he doesn't get into difficulties and to make sure he's doing the right thing."

Worker: "I'm fed up with the boss keeping tabs on me. He obviously doesn't trust me to do the job. I'm not going to do my best till he leaves me alone."

Supervisor: "He's not doing very well. I've got to keep a closer eye on him."

Worker: "There he goes again. If he won't let me get on with it, it his own fault!"

Supervisor: "His performance is getting worse. I'll have to keep showing him how to do it."

and so on... Who caused this situation? Somebody might have started it once upon a time, but now the looping impasse and downward spiral is being held in place by both parties. So, asking who caused it is not helpful. What is more helpful is asking how would be better.


There are a number of ways of using systemic concepts in looking at change in organisations. One school of thought is to use system dynamics to build models of the organisation, to establish the ways in which the various aspects of the organisation influence each other, and therefore themselves. The building of the system models helps the managers concerned to focus on how their organisation is working, what "rules" are currently being seen in the operation, what the consequences of those rules are and what changes they might make. (See for example Wolstenholme and Stevenson, 1994)

System dynamics are useful for studying and understanding many processes, for example production lines. However, a developing school of thought says that, in reality, living people are far too complex to be modelled by what is essentially a mathematical process, and favours a different approach which owes quite a lot to the ways in which physicists are beginning to handle very complex systems. And these "complex" systems can look deceptively simple.


Imagine a water drop falling onto a highly polished baking tray (Waldrop, 1993). Droplets form. There are two counteracting forces at work. Gravity tries to spread the water as thinly as possible, to a position of stability (negative feedback). Surface tension, in the other hand, tries to keep the water in large droplets, the larger the better (positive feedback). The result is a complex pattern of droplets, which is never the same twice. Perhaps the most surprising thing is that it is actually impossible to calculate precisely what the pattern of droplets will be for definite, even with all the computing power and infinite amount of time. We could calculate probabilities of certain patterns, but we could not predict what was going to happen. It's not just difficult, it is provenly impossible. The only way to find out the pattern is to do it!. By making the experiment, we can find out, for this drop and this tray, what the pattern is. And, of course, for the next drop, the pattern will be different again.

If this is a complex system, then what about an actual living person? If a man hits a billiard ball with a cue, in a certain direction, with a certain force, we can just about calculate what happens to the ball. But if the man hits a colleague with exactly the same force and direction, can we calculate what happens to the man? No chance. In this world of complexity, anything that involves a living being is in the "complex" and uncalculatable league.


Peter Senge gives us some good advice about systems thinking. But he didn't give us a guide towards systems action - what do we actually do?

The first thing is to become very clear about what our goal is in a particular situation. Describing a goal in clear, concrete and positive terms is the first step towards achieving it. Do not focus on the "problem", or try to understand why the problem occurs. Focus instead on the solution.

Secondly, become aware of what happens, again in clear and concrete terms. What do we do in a certain situation? What do other people do? What happens? What do we expect to happen? What happens if we do something different?

Thirdly, remember that we are a part of the complex systems of loops and linkages that makes up our world, and that we have the freedom, if we choose, to act differently and therefore change what's going on around us. Remember, a small change in one part of a system can lead to large changes in other parts. If, on the other hand, we think that the problem is "out there", then that thought becomes the problem.

Fourthly, bear in mind that what happens is what happens, no more and no less. What happens is useful feedback for us about the results of our actions. Notice it and use it!

Fifthly, notice what helps in producing all or part of our goal. Sometimes things help, sometimes they don't.Notice what helps and do more of it.

Sixthly, work with what happens. Do not get concerned with what "must" happen, "ought to "happen, "should" have happened or anything else. Stick to observing what actually happens.

Finally, if your efforts havn't got you towards your goal, then you've had excellent high quality feedback that what you've done isn't working. So, do something else. If at first you don't succeed, then keep trying different things until you do. And if you've tried several different things already, maybe they weren't different enough.

The seven points above are high level, and deserve amplification. However, there is not space here to do that. I am continuing to develop an approach to management, leadership and organisational change based on systemic principles, and hope to make further material available in due course. In the meantime, try these things for yourself, and see if it makes a difference for you.


To return to the question I posed at the start of this paper, how would we know a learning organisation if we came across it? I propose that we would see:

* On the "doing" level; a considerable amount of activity aimed towards delighting customers.

* On the first learning level; activities directed towards improving what happens at the doing level, involving everyone in the organisation in some way, and

* On the second learning level; activities which revise and re-invent the framework within which the other two levels of activity take place. The guidelines for systems action are suitable for operating at this level of learning.

I hope that this paper is a useful contribution to the conference, and to the development of learning companies and organisations in the future. There are a number of workers investigating this field at present, and recent contribution in print include Paul Tosey on "Interfering with the Interference" (1993) and Tim Coldicott on "Chief Executive styles for the 1990s" (1994). I will be very happy to receive feedback on this paper as I continue developing these approaches. Thank you for your attention.



Argyris C and Schon DA, "Organisational Learning", Addison-Wesley (1978)

Bateson G; "Steps to an Ecology of Mind", Ballantine, New York (1971)

Bateson G, Jackson DD, Haley J and Weakland J; "Toward a Theory of Schizophrenia", Behavioural Science1, pp 251 - 264 (1956)

Coldicott T; "Chief Executive Styles for the 1990s", Organisations and People 1:2 pp 18 - 21(1994)

Garvin DA; "Building a Learning Organisation", Harvard Business Review 74, 4, pp78-91(1993)

McKergow MW; "What's systemic thinking got to do with anything?", Organisations and People, 1:1, pp 16 - 20 (1994)

Pedler M, Burgoyne JG and Boydell T; "The Learning Company", McGraw-Hill (1991)

Senge P; "The Fifth Discipline", Doubleday, New York (1990)

Tosey P; "Interfering with the Interference", MEAD 24, Association for Management Education and Development (1993)

Waldrop MM; "Complexity", Viking(1993)

Watzlawick P, Weakland Jand Fisch R; "Change", Norton (1974)

Wolstenholme E and Stevenson R; "Business Process Re-Engineering Using Systems Modelling", issued via the Strategic Planning Society (1994)

[Published as "Learning at a higher level: Systems Action in Organisations", in Learning Company Conference 1994: Collected Papers (eds G Welshman, T Boydell, JBurgoyne and M Pedler), pp 146 - 153 (1994)]