A sufficiently strong prior can colour the perception of new evidence enough that every experience confirms it, so the belief cannot be updated by exposure.
The model is that perception is already a blend of prior and evidence, so a dominant prior alters what is experienced before any updating occurs. The phobic who meets a harmless dog experiences a terrifying encounter, and updates toward dogs being terrifying. It offers a mechanism for why some beliefs survive exactly the evidence that should destroy them.
In practice
- A manager convinced an employee is careless reads their next careful week as a fluke.
- A team that believes a department obstructs them experiences every delay as obstruction, including the ones that were not.
- Possible explanations
- The explanation is that perception already blends prior and evidence, so a dominant prior changes what is experienced before any updating occurs, which means the experience cannot correct the belief.
- In its favour
- The account fits phobia, where repeated exposure to a harmless stimulus fails to reduce fear, and exposure therapy's design, which works by controlling the conditions of the experience, is consistent with it.
- Against it
- It is a coinage of the rationalist community, argued from examples rather than tested, and its Bayesian framing has been questioned on whether the formalism does the work claimed for it.
- Standing
- A coinage of the rationalist community, argued from examples rather than tested. Presented in a Bayesian framing whose formal adequacy is disputed. The clinical phenomenon it draws on, that phobic exposure without structure can reinforce fear, is documented.
- Subject
- Psychology & Behavior
- Essay
- Theories of the mind, from the lab to the bookshop