ENS, Jaurès, 29 rue d'Ulm, 75005 Paris
We make many decisions throughout the day. Some of these decisions are the sort familiar to economists studying decision under risk. The economic decision maker chooses between lotteries specified in terms of the probabilities and values of various possible outcomes: "a 50% chance of 100 Euros else nothing" or "40 Euros for sure".
But most everyday decisions involve selection of actions or perceptual decisions where we have imperfect knowledge of probabilties. Should I attempt to cross the street right now or wait? What is the probability I will trip and fall? What is the probability that the approaching taxi driver is distracted?
I'll describe how to devise motor tasks potentially equivalent to economic decision under risk and describe experiments evaluating human performance in such tasks. I will describe how we can use the tasks to study how we represent and use information about our own motor uncertainty and uncertainty about events in the environment around us. How accurate are our estimates? Are they biased? Do we know how good or bad our estimates are, a form of metacognition?
Last we will consider a metacognitive task in which the decision maker initially has little information about the probabilities in a motor lottery task but has the opportunity to "buy information" about probability. Each unit of information provides additional information about task probabilities. The key question is, does the decision maker -- buying information -- know when to stop?