I attended a fascinating talk yesterday on “Behavioral Economics and the Problem of the Rational Inner Self” by Mantas Radzvilas at the Sidney M. Edelstein Center for the History and Philosophy of Science, Technology and Medicine at the Hebrew University of Jerusalem. Mantas reconstructed what he considers the main theoretical approach in current behavioral economics, which he calls the “preference reconstruction approach”. He characterizes its main principles as follows: “(1) The decision-maker’s expected utility function must capture all the motivations which are relevant for decision-maker’s evaluation of choice options, irrespective of what the nature of those motivations is. (2) The decision-maker is modelled as a rational maximizer of expected utility.”

The curious thing about this approach seems to be that it typically “relies on an implicit assumption that psychological factors may influence the decision-maker’s evaluation of choice options, but never the process of reasoning by which s/he chooses among them. Hence, this approach is always compatible with the neoclassical expected utility maximization principle.” Mantas cites as an example the inequity aversion model by Fehr and Schmidt (1999, “A theory of fairness, competition, and cooperation”), which models agents as experiencing a utility loss from non-egalitarian allocation of material payoffs.

According to Mantas, the fact that the process of rational choice is not detailed creates several problems for these approaches. Among other things, preference reconstruction models can make almost any kind of behavioral anomalies appear rational. This makes the models virtually non-falsifiable. It also deprives them of predictive and explanatory power. How do we know that agents actually act out of an inequity aversion? Because they say so? Is that good evidence? No uncontroversial answer seems to be forthcoming.

The question then is why many behavioral economists still prefer preference reconstruction approaches over alternative approaches? Mantas sees a connection to the application of results of behavioral economics in behavioral welfare economics. The latter branch of economics looks into ways to maximize the social welfare by influencing the (suboptimal) choices people actually make. It rests on knowledge about the cognitive biases humans typically fall prey to in economic choices and that are taken to compromise people’s personal welfare. It obviously assumes that, if human agents would be aware of these biases, they would prefer to overcome them and thereby maximize their personal welfare.

But, again, there is a serious problem here: The decision maker in behavioral welfare economics is presupposed as completely and purely rational but as trapped in various mechanisms that distort the choice she would make otherwise. The decision maker also has perfectly informed, internally coherent, intertemporally and intercontextually stable preferences. In short, the decision maker is the core of Kahneman’s system 2. Its decision processes are just disturbed by psychological factors. Curiously, however, no account is given of how the rational inner self works. (Note that the inner rational self model has also been criticized by Infante, Lecouteaux and Sugden 2016, “Preference purification and the inner rational agent.” However, Mantas criticism summarized here goes substantially beyond their paper).

Mantas suspects that the persistent popularity of the internal rational self rests in the fact that behavioral welfare economics would immediately become a paternalistic enterprise if it rejected the rational self. Paternalism, of course, is exactly what this branch of economics wants to resists heavily. The dilemma is that, as a consequence, behavioral welfare economists must rely on something which cannot be tested or predicted, which immediately calls into question the scientific character of this branch of economics.

I asked Mantas whether research from neuroeconomics or cognitive science may help to resolve this issue in any way for behavioral welfare economics. I was presupposing here the working hypothesis of www.insosci.eu, according to which good explanations will eventually have to be mechanistic. He answered that he had little hope in this kind of research, at least the one available today.

I agree that that the existing research is mixed on whether the brain actually contains modules that roughly correspond to system 1 and (inner rational self) system 2 (for a positive and a negative position on this question, see Trepel et. al. 2005, “Prospect theory on the brain?”, and Osam 2004, “An evaluation of dual-process theories of reasoning”). However, I do believe that neuroscientific research will eventually have something to say here. If research reveals the existence of a “rational module” in the brain that is regularly distorted but that can be isolated in principle, then behavioral welfare economics will be in a much better position to justify nudging policies. Should it turn out, however, that the brain turns out so complex that processes corresponding to system 1 and system 2 cannot be modularized and distinguished (the more realistic scenario, in my view), then this will be sufficient to deny the existence of a rational inner self. As a consequence, behavioral welfare economics will be put in the unpleasant position of having to embrace a version of economic paternalism. Suddenly, it will have to decide externally what personal welfare is, and nudging policies will become inconsistent with liberalism and individualism. The fact that a mechanistic explanation of decision making under risk and uncertainty may have these important normative implications offers a strong incentive for pursuing this research strategy far beyond the purely scientific interest. I am optimistic that the www.insosci.eu project will contribute some first steps in this direction.