U. of Utah, David Eccles School of Business, United States
When entrepreneurs envision counterfactual futures, learning how to reach them means learning how to make the presently untrue true. We propose that confidence in succeeding with such attempts should follow two intertwined learn- ing processes: calibrating forward looking confidence by creating and revising a causal logic of how to create a contrarian future, something we will call confidence type 2, and calibrating backward looking confidence or confidence type 1 by draw- ing inferences from data. Using formal modeling and logic, we argue that if the probability of a proposition (that is, confidence type 1 in this proposition) falls beyond a threshold, traditional Bayesian learning gives the entrepreneur no option but to abandon the causal logic altogether and persistence become irrational–an expression of overconfidence. However, we formulate the conditions under which the entrepreneur can rewrite the causal logic, and rationally revise confidence type 2, and rationally persist. Finally, we show that confidence type 2 defines how available probabilities should be mapped to the prospects of success when pursuing a contrarian state of the world. Our results have implications for un- derstanding the linkage between rationality and innovation and the human/ AI interface in crafting strategy.
