Human Intuition, Decision Analysis, and the Value of Information

Our intuitive decisions are subject to all kinds of biases and fallacies according to extensive research in the psychology of judgment. Decision analysis provides practical tools to help bring greater rationality to important decisions. Decision analysis has long been intertwined with computer science. John von Neumann, famous for the “von Neumann architecture” of the modern digital computer, also (with Oscar Morgenstern) developed expected utility theory, the logical foundation of decision analysis. After early battles about the representation of uncertainty in AI and expert systems, Bayesian probability networks and decision analysis methods are now in widespread use. I’ll give examples to illustrate how and why this has happened. A key contribution of decision analysis is the *expected value of information* (EVI), which can be a powerful guide for our decisions on collecting and analyzing data. But EVI is often misunderstood -- it works in the context of an explicit decision analysis, with identified decisions, objectives (utility) and uncertainties represented as probabilities.

Speaker: Max Henrion, Lumina