Dynamic hypothesis testing: mechanisms that drive decisions under uncertainty

When assessing a situation to choose the best action, confirmation bias is a potential problem. Confirmation bias is where the final decision is pulled towards the focal decision option. The focal decision option is whichever decision option is the focus of the individual’s attention. Attention commonly focuses on the decision option that is preferred (preferred decision option), the decision option that is expected to fit the situation (expected decision option), or the decision option that is suggested by the information provided (framed decision option). A framework is developed that breaks confirmation bias down into confirmation of preference, confirmation of expectation, and confirmation of frame, based on which decision option is the focus of attention. Confirmation bias is also distinguished from four groups of confirmation processes. A new model, the dynamic hypothesis testing (DHT) model, is developed towards a unified explanation of the three types of confirmation bias and the associated confirmation processes. Methods: Four experiments were conducted with university students paid for their time. Participants were required to make an action decision based on their situation assessment of a novel business competition scenario. Participants formed situation assessments and chose actions based on information they selected from a grid of information. Action preferences were manipulated in all four experiments. The situation supported by the information was manipulated in the third and fourth experiments. Results: In Experiments 1 and 2, the framed decision option influenced information selection when participants were asked to select all the information required prior to receiving the requested information, but the framed decision option had no effect on information selection when participants received the information immediately after selecting each item of information. Dynamic Hypothesis Testing: Mechanisms that drive decisions under uncertainty Experiments 3 and 4 found that the preferred and expected decision options influenced final situation assessments and chosen actions, but not information selection. Experiments 3 and 4 also found that the preferred decision option influenced decision thresholds, which determine which action is chosen given the situation assessment. Experiment 4 demonstrated that preferred and expected decision options influenced final situation assessments via information interpretation. Both information interpretations and situation assessments were exaggerated by the preferred and expected decision options compared to a yoked control group who were not required to make a decision. When the expected and preferred decision options opposed each other, there was some indication that the expected decision option had a greater influence on information interpretation and final situation assessments than the preferred decision option in this case. Discussion: Empirical findings are used to refine the DHT model. The DHT model provides advancement towards a unified explanation of confirmation of frame, expectation, and preference. A variety of processes that have been called confirmation bias are incorporated into the model and their relationship with confirmation outcomes are outlined. The DHT model expands on parallel constraint satisfaction mechanisms in an associative network to explain confirmation bias processes and outcomes. Implications for theory, research, and practice are discussed.