Optimization vs. Heuristic: Is less more?

Below I share the video of a talk by Gerd Gigerenzer on how people actually make decisions in an uncertain world. He underlines the important distinction between decision making under "uncertainty" (when individuals do not know the probability distributions that underlie outcomes of interest) and decision making under "risk" (where the process generating the outcomes is not deterministic but people know the random process governing these outcomes).  

The bottom line of the talk is that the best strategy for a person deciding under risk (for example when you gamble in the casino where you know the odds of being dealt a given combination of playing cards) is not necessarily the best strategy in a world of uncertainty (for example when you make a portfolio allocation). He does not criticize the assumption that individuals act rationally. Instead he makes the point that in an uncertain world rational decision making often times utilizes heuristic (or intuitive) strategies guided by fewer variables that are directly observable instead of complicated models with many variables which are not directly observable. Having observed an individual's behavior, economists tend to interpret this behavior AS IF it is the result of  optimization by an individual with full (or bounded) rationality who is guided by a complicated decision model subject to certain limitations on her mental capacity and imperfect information about the parameters of that model. He argues that heuristic models are what rational agents do and should follow given informational constraints and time limitations. Therefore, as Gigerenzer argues, the predictions we make about future behavior and outcomes can be quite sensitive to which of the two models (heuristic decision or optimization) we assume people follow.

While these points are  probably (or rather hopefully) well-taken by many economists, we currently lack a satisfactory framework to model heuristic decision making processes. The issue is further complicated by the fact that such a modelling approach needs to allow for the circumstantial nature of heuristic thinking. In other words, we need to think about the decision maker as having a toolbox of models, rather than a single model, to choose from depending on the nature of the problem at hand and the feasibility constraints she faces. In this sense it will surely be harder to discipline our formal analyses of decision making so as to agree with other economists on the basic modelling principles. It may take a while for economists to switch from the current modelling approaches as they are convenient and more importantly widely accepted.

Many times we don't care how the decision has actually reasoned when he made a decision with the excuse that we don't get to witness her thought-process anyway. However, we increasingly have greater access to behavioral and psychological clues that are suggestive of the intermediate steps she has followed to reach a given conclusion. Evidence on these intermediate steps should be checked against what outright optimization and different heuristics would suggest.