Reading 6.4: Building shared models
In the early 1950s, Alex Bavelas (Bavelas, 1951) undertook some simple experiments which contain a number of important lessons. In particular he showed the way in which the shape of the system structure of communication affects the evolution of learning. Hierarchical communication structures, such as the typical organisation chart, are invariably faster at solving problems in very simple cases where no new learning is required, but once the problem situation becomes complex and there is a need to develop new learning, peer-based communication is invariably more effective. This seems an invaluable lesson in the world today which faces a multiplicity of complex problems.
The experiment consisted of one group of five people arranged in a star configuration, and a second in a ring configuration (see Figure 6.4). They were provided with the means with which to communicate with each other in writing only. They were arranged so that they could write notes and pass them to the adjacent person or persons along the configuration lines, but could not communicate in any other fashion. They were not visible to each other. No communication could take place in the star configuration except through the central person, and no communication could take place across the ring, only around the line of the ring loop.
Each person in each group was given five marbles of different colours, such that only one colour was held in common by all the group members. The task for the group was to identify the colour held in common. Perhaps not surprisingly in all the trials the group arranged in the star configuration succeeded first, the central position being pivotal in that all the information could be gathered there and the problem solved at that position. Then the problem was changed in that marbles with ‘mottled, milky’ colours were substituted for the previous clear marbles with distinct colouration. This time the group arranged in the ring configuration succeeded first. Even more interestingly, it was found in repeated experiments, that in this situation the person occupying the central position of the star always gave up trying to understand the information being passed to him/her and started simply to guess the result. In the ring configuration in repeated experiments, the problem solving time reduced to that of the first experiment with clear marbles.
How do we explain these results? In the second case, it is clear that the problem becomes much more complex, the problem in either case cannot be solved without having a shared definition (shared models) of the colours. In the first case the uncertainty between the people in the groups about the definition of the colours is low, and there is sufficient shared understanding for it not to affect the problem solving. In the second case there is little shared understanding and it is difficult to achieve a common model – this can only be achieved with a great deal of communication. The volume of communication simply overwhelms the person at the centre of the star, and since they can no longer deal with it they give up and use an alternative strategy. In the ring group, the model is developed, and once developed, the problem solving time falls to the same as in the first case with the clear marbles.
Some further work on this scenario was carried out by Walter Lee and Paul Pangaro (Lee and Pangaro, 1993) from the point of view of information theory. They showed that, surprisingly, in information terms, the person in the centre of the star is operating sensibly from their point of view. They draw attention to the position of the CEO (Chief Executive Officer) who is in the centre of a much larger system configured as a star. Unsurprisingly, the CEO is not inclined to listen to the people around. The hierarchies of organisations are famous for communications only travelling in a downwards direction and nothing travelling up. Here you can see in a simple experiment the problem illustrated – to achieve communication you need to establish common models, common understandings, and to minimise the amount of communication necessary. The price you pay if this is not done is that of managers going their own way with little understanding of what is actually going on, to the detriment of the organisation.
Shared experience improves the relevance of messages simply because the sender understands (because he’s been there) the recipient’s concerns, values, prejudices and problems. The recipient, on the other hand, will perceive the sender’s message as trustworthy because he knows (by virtue of previous collaboration) that the sender possesses that understanding. The Bavelas experiments indicated that the circle members trusted each other because each node knew that his partners (the adjacent nodes) were undergoing exactly the same experience. In the star configuration, by contrast, trust broke down precisely because everyone knew that the central node had a position and experience history that was qualitatively different from the peripheral nodes. (Of course, the peripheral nodes couldn’t even trust each other because they couldn’t communicate with each other.) The relevance of the messages, particularly of those from the periphery to the centre, was low simply because the peripheral nodes did not and could not appreciate the problems, pressures, and constraints facing the central node. Thus, messages sent during the noiseless case, that suggested improving the message-sending process by specifying the sixth colour that each node didn’t have, were irrelevant to the situation faced by the central node, and would only make his job more difficult.
One way for a chief executive to establish common understanding is to surround themselves with people who share their view of the world. This is exceedingly common and does avoid the problem, but has consequences of its own.
So, because a message is formulated within a mindset, and framed within models contained within that mindset, achieving communication is not straightforward. Clearly one of the models that is important in this process is the model the source has of the destination mindset.
The message will be decoded, and the complexity unravelled, within the destination mindset using the models that are actually present. If these models are somewhat different the decoded message, then, may be very different from that intended. This raises the immediate question as to how common models develop, and here lies the importance of feedback, the return communication which creates a loop which will be seen to be a fundamental part of our modelling. In the ring group, the feedback loops between the participants functioned well so that gradually a shared model was formed. In the star group, because of the communication overload at the centre, this could not happen.