Article Text

Download PDFPDF

Game theory
The consultation game
  1. G Elwyn
  1. Correspondence to:
 Professor G Elwyn
 Primary Care Group, University of Wales Swansea, Swansea SA2 8PP, UK;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Exploring the use of game theory to address quality

Game theory and clinical practice seem an unlikely marriage and it may not be obvious how such a combination could improve the quality of clinical consultations. But this area of applied mathematics seems set to reduce some of our quintessential human foibles to a predictable set of behaviours based on what you win and what you lose (forget about how you play the game—at last it’s officially all about winning). Game theory could be a sharp new tool for dissecting the mass of behaviours at play in the medical consultation, a historically paternalistic human interaction that stretches back as far as Hippocrates.

Medicine is a service delivered by a mix of episodic and repeated interactions between humans, medicated by the use of technologies such as tests, drugs and procedures. There is clear evidence that there is communication failure in consultations and that adherence to advice and treatment is 40–50% less than optimum, yet explanations and successful interventions to address this problem are rare. It is also clear from many empirical studies that the quality of the consultations and communication processes is far from optimum when examined from patient perspectives, and results in inefficiencies, errors and costs, not least the medicolegal expense of patient claims.


The central principle in game theory is that “players think about what others are likely to do, and do so with some degree of thought”.1 Although there are many types of games used to explore this “thinking”, all have a structure that involves an interaction leading to a reward or a loss between people, in pairs or, more recently, between and within groups. The “laboratory rat” of game theory is called the Prisoner’s Dilemma where two players independently face a decision whether to collaborate or defect (box 1).

Box 1 The Prisoners’ Dilemma

Two prisoners independently face a decision whether to collaborate or defect where their individual decisions, when considered together, lead to differing payoffs. The possibilities are mutual cooperation, mutual defection, or a situation where one has decided to cooperate and the other to defect. One-sided defection offers the largest reward to the defector and the least reward to the collaborator. Mutual cooperation offers equal rewards to both. Mutual defection also offers equal rewards to both but at a marginally lower level and, according to the theory, this is the rational strategy—the so-called Nash equilibrium, after the mathematician John Nash.

The Prisoner’s Dilemma predicts a rational strategy for players, one of mutual defection (the Nash equilibrium) which gives a safe but low return over many interactions, rather than cooperation which offers a greater reward but at a greater risk. What is surprising is that humans spurn the rational strategy more often than not, preferring instead to cooperate. This outcome is one demonstrated repeatedly, not just in the Prisoner’s Dilemma but in more complex games such as the Centipede.

How this irrational behaviour persists in the “survival of the fittest” paradigm is ironically best explained by those evolutionary theorists who have looked to game theory themselves, and suggests that our oversized cerebral cortex with its unique consciousness has arisen not from a “dog eat dog” strategy of survival but from a Machiavellian drive to forge allegiances and beat the competition using guile rather than brute force.2

The problem of cooperation is an area of active research where interdisciplinary debates between game theoreticians (mathematicians generally), economists, psychologists, and social scientists are raging. Those disciplines accommodating broader perspectives, such as psychology, argue that human interaction is complex. Decisions about payoffs, they contend, are influenced as much by personal and contextual variables as they are by simplistic mathematical type assessment of gain or loss by interacting with others. Names such as “behavioural”1 and “psychological” game theory3 signal the ongoing discussions.


Game theory could help us to understand and explain the extent of cooperation and defection in medical interactions. As a profession, we have been slow to look beyond the boundaries of medicine to other disciplines and, because of this conservatism, research into medical processes has neglected game theory as a potential theoretical framework for analytical work.

Other disciplines have been quicker to see the potential of game theory and have used it extensively to examine many areas of human life. Axelrod in particular has applied the theory in innovative ways to consider sociological issues and to examine policy implications.4 Gutek examined the differences between commercial service encounters (one-off consultations) and service relationships (repeated consultations) in many walks of life such as hairdressing, financial advice, and family physicians, and found disadvantages to both consumer and provider in each design.5 Medicine could usefully extend this work to examine where continuity adds (or detracts) value to professional-patient interactions.

Existing research on the medical consultation is devoid of a theoretical basis. While there are plenty of prescriptive models such as the “patient centred method”,6 the “Cambridge Calgary” approach7 and so on, this field does not seem to have provided a theory on which to examine the empirical findings. Descriptive work such as that by Byrne and Long,8 assessment of videotapes in the Royal College of General Practitioners’ examination,9 and the recent analyses by discourse analysts reveal that practitioners do not come anywhere near performing according to the prescribed “idealised” models.10

So where does the fault lie? With the practitioners or with the models? The models are based on professionally constructed idealised communication methods and have not been based on adequate theories of how humans interact, given the differing payoffs for patients and practitioners. For health professionals, keeping to time and limiting their exposure to large volumes of patients is a way of managing their workload, maximising income, and keeping sane. At the same time they need to retain the cooperation of patients by building trust, obtaining information, and proving to be good at solving and dealing with problems. The payoffs for doctors of engaging in advocated methods of communicating—exploring ideas, concerns and expectations, for instance—are not so clear. Payoffs for patients are different in different situations. Fast provision of a sought-for medication is all some people are looking for; others are looking for supportive ongoing relationships where chronic illnesses are explained and mastered in partnerships, while others are looking for a screening investigation but are unaware of the possible harms involved, and so on. It may well be time to pause for thought and to examine the motivations for doctors and patients in a different way.

When combined with developments in neuroscience, particularly real time MRI scans, games such as the Prisoner’s Dilemma show that cooperation leads to activation of brain areas associated with reward processing and learning, reinforcing the behaviour over and above the payoff itself.12 So it appears that humans are programmed to be more than calculating machines, interested only in maximising utility, as pure game theory would have it. Instead, we like to work together, to some extent at least, and it is a matter for further research to explain how. Game theory offers medicine a potential method, yet the complexity and the layered nature of the consultation cannot be easily modelled by one or other of the existing games in the game theory playroom. In their paper in this issue of QSHC, Tarrant and colleagues11 concede that this will be a considerable challenge. If it is possible, perhaps we can then think about improving the quality of the consultation process and create gains for practitioners, patients, and health service economists.


The author thanks Jenny Kowalczuk for contributing to the drafting of this commentary.

Exploring the use of game theory to address quality


Linked Articles