Introduction to Behavioral

6

I N T R O D U C T I O N

PT

•

I

It is important to understand that the term ‘rationality’ is used in many different 

senses, depending on the discipline of the user of the term; even within the discipline 

of economics there are different meanings.  When we refer to people acting rationally 

in the everyday sense we usually mean that they are using reason. This kind of action 

is often contrasted with people being prompted either by emotional factors or by 

unconscious instinct. However, economists have tended to regard this interpretation of 

rationality as too broad and imprecise.

Instead, they have started out from a tightly specifi ed means–end framework of 

rational decision-making, as a particular interpretation of instrumental rationality. In 

that framework, individuals are assumed to entertain preferences over a set of available 

courses of action and act such as to realize their most preferred outcome. At the heart 

of this model lie several basic assumptions about the nature of these preferences:

Completeness   Individuals entertain a preference ordering across all alternative 

courses of action that they face.

Transitivity 

Individuals make consistent choices, in the sense that if A is   

 

preferred to B, and B is preferred to C, then a rational individual  

 

will prefer A to C.

These two axioms together ensure that individuals will be able to pick at least one most 

preferred course of action out of the various alternatives they face. Both axioms may be 

relaxed in certain ways while it will still be possible to meaningfully talk of instrumentally 

rational choice. But for the most part, economists have added stronger assumptions in 

addition to these rationality axioms, either to simplify technical treatment, or sometimes 

just out of tradition. Two important additional assumptions, sometimes referred to as 

the ‘economic’ assumptions that are added to the two rationality axioms above, are that 

more of an economic good is preferred to less of it (‘monotonicity’), and that averages 

are preferred to extremes (‘convexity’).

However, this simple model of economic rationality is only applicable to decisions 

under certainty, such that outcomes are unambiguously tied to actions. As soon as one 

allows for uncertain outcomes, more complex frameworks of analysis become necessary, 

based on mathematical theories of uncertainty such as probability theory. The standard 

model for these contexts is usually augmented by the twin assumptions of expected 

utility maximization and Bayesian probability estimation. Further assumptions are 

necessary to adapt the model to decision-making stretching over a period of time into 

the future, notably assumptions regarding time preference and discounting of future 

horizons. 

But even this framework is not yet suffi ciently general for all decision contexts 

studied by economists. Uncertainty may not just be an exogenous factor, in the sense 

of being given independently of the decision taken. You may, for example, decide to 

act on a weather forecast predicting sunshine with 90% probability by leaving your 

umbrella at home. Unless you are subject to superstitious beliefs, you would not accept 

that this decision has any effect on whether it will actually rain in the end or not. 

Many economic problems are subject to yet a different kind of uncertainty that is 

endogenous to the situation studied. This is behavioral uncertainty that arises from the 

mutual dependencies involved in the strategic interaction of two or more individuals. 

Assume you are walking down a narrow lane and fi nd yourself walking towards another 

individual heading in the opposite direction. Whether or not you will brush coats with 

---

7

N AT U R E   O F   B E H A V I O R A L   E C O N O M I C S

CH

•

1

that individual will not just depend on your own actions but also on how the other 

side behaves. Economists have used a strong assumption known as the common 

knowledge assumption as a further augmentation of the standard model. This is a 

stricter assumption, whereby it is not suffi cient for each person or player to be rational, 

they must also know that all other players are rational, and that all other players know 

that all other players are rational ... ad infi nitum. 

Finally, some economists hold the view that the rationality of individual behavior 

should be judged not on the level of the individual but on the level of systemic 

outcomes. This tends to be the view of Vernon Smith, who has been particularly 

concerned with examining the predictions of economic rationality in terms of long-run 

market equilibria. Smith does not accept the norms of the standard model in terms of 

individual behavior, and believes that individuals can violate these norms and still act 

rationally according to his view of rationality. This view equates rationality with the 

end results of the decision-making process as far as market effi ciency is concerned. For 

Smith, if markets are effi cient, for example, in terms of market clearing, then this is 

evidence that individuals are rational.

On the other hand, by other defi nitions of rationality, people may act rationally 

and the predictions of the standard model may prove incorrect; this tends to be the 

view of Kahneman and Tversky, whose approach will be discussed in detail in Chapter 

5. Unlike Smith, Kahneman and Tversky do accept the norms of the standard model 

as a benchmark for judging rationality. By these standards they claim that individuals 

frequently act irrationally. However, they also argue that the systematic errors and 

biases that they fi nd in their empirical studies do not necessarily constitute irrational 

behavior. We see here a theme emerging that will run through the other chapters of this 

book, by which the standard model of economic rationality, under which a considerable 

amount of frequently observable behavior would have to be classed as irrational, gives 

way to alternative conceptions of rationality that more properly account for observed 

behavior.

At one extreme we have a view, which was perhaps fi rst formulated by Ludwig von 

Mises (1949), that any action must by defi nition be rational. This approach essentially 

defi nes rationality in terms of revealed preference. If we perform a certain act it must 

be because we have a preference for doing so; if we did not have such a preference 

then we would not perform the act. Associated with this approach is the view that 

‘a pronouncement of irrational choice might seem to imply nothing more than our 

ignorance about another’s private hedonic priorities … individual tastes are not a 

matter for dispute, nor can they be deemed rational or irrational’ (Berridge, 2001). 

The problem with such an approach is that it obscures the important factors involved in 

terms of the determination of revealed preference, and therefore, while it is a coherent 

view, it is not very useful in terms of aiding analysis and understanding since it remains 

consistent at the price of becoming a tautology.

Similar to the above view is the argument that evolution has necessarily produced 

organisms that form true beliefs and that reason rationally (Fodor, 1975; Dennett, 

1987). However, this view has been much criticized as misunderstanding the role of 

natural selection in the evolutionary process. Most evolutionary biologists agree that 

natural selection does not guarantee that rational beings will evolve, or even intelligent 

beings for that matter.

---