Statistical context matters when it comes to expectations, study finds
Should you sell your stock or buy more of it? What is the best meal to order at your favorite restaurant? Such decisions are often made by considering expected outcomes — but where do those expectations come from?
New research from neuroscientists at Brown University’s Carney Institute for Brain Science found that when it comes to revising expectations, statistical context matters. The study was published in August in eLife, a peer-reviewed open access scientific journal for the biomedical and life sciences.
“People form expectations based on experience and revise these expectations when confronted with contradictory information,” said Matthew Nasser, co-author of the study and assistant professor of neuroscience. “But not all information drives people to revise expectations to the same degree. For example, information that indicates a fundamental change, such as a new chef at your favorite restaurant, can facilitate rapid belief revisions. But this might not always be the case, particularly if your favorite chef takes a lot of sick days.”
Nassar and Michael Frank, Edgar L. Marston Professor of Psychology, measured brain signals with electroencephalography while human participants played a computer game that forced them to rapidly revise beliefs in response to new information. They found that a specific brain signal, the P300, predicted how much people would learn in a changing environment, but did so differently in different contexts.
In a context like the one where the chef was expected to be hired and fired rapidly, the larger the P300, the more the participants revised expectations. But in the case where the chef is apt to call in sick, the larger the P300, the less the participants revised expectations. The researchers interpret their results as suggesting brain signals such as the P300 might not provide a volume knob on learning, but rather facilitate the assignment of an unexpected outcome—such as a meal—to an alternative causal process—like the chef.