nucleation

QUESTION-325: Is persistent avoidance in anxiety maintained by an inflated uncertainty estimate about the avoided states, as Brown, Price and Dombrovski propose, or by a pessimistic evaluation of one's own future avoidance, as Zorowitz, Momennejad and Daw propose, rather than by an inflated threat value?

2 min read

Properties4
StatusOpen
Tagspsychopathology-and-treatment, learning-and-conditioning
Linewill-organization
Cited as “bears on” byTHEORY-1190 — In a Markov decision process, an agent that evaluates actions as if it will fail to avoid danger at later steps, while in fact choosing optimally at each step, spreads threat value to states far from the danger and avoids early and widely, which reproduces qualitatively exaggerated threat appraisal, fear generalization and several laboratory findings associated with anxiety, THEORY-910 — Brown, Price and Dombrovski propose, without a model run or data, that in aversive environments avoiding uncertain states keeps their uncertainty high and so perpetuates avoidance, that anxiety strengthens this loop by misreading noise as change, and that exposure works by reducing the uncertainty rather than an excess of learned threat, THEORY-915 — In Zorowitz, Momennejad and Daw's simulations of a Markov decision process, an agent that evaluates actions as if it will fail to avoid danger at later steps, while in fact choosing optimally at each step, spreads threat value to states far from the danger and avoids early and widely, which reproduces qualitatively exaggerated threat appraisal, fear generalization and several laboratory findings associated with anxiety

Graph View

Why it is a question

THEORY-910 holds what Brown, Price and Dombrovski propose, without a model run or data; THEORY-915 holds what Zorowitz, Momennejad and Daw’s simulations showed, with no data fitted. Whether either mechanism operates in people, and which, is untested.

What would answer it

The former promote_when of each, unchanged.

THEORY-910. Two results. First, a simulation: an uncertainty-sensitive temporal-difference agent in an aversive Markov decision process, with noise misattributed to volatility or estimation uncertainty, shows avoidance that persists because the avoided state’s uncertainty is never reduced, and stops when that uncertainty is reduced by forced sampling with the threat estimate held fixed. Second, people: in an instrumental avoidance task, measured uncertainty-learning parameters predict how much people avoid, and a manipulation that lowers uncertainty about the avoided option without changing its expected threat reduces avoidance. It is refuted if avoidance in such tasks tracks the expected threat and not the uncertainty, or if reducing uncertainty with threat fixed leaves avoidance unchanged. Consistency with clinical observations about exposure cannot settle it.

THEORY-915. Data, not more simulation: in a sequential avoidance task with known transition structure, people’s choices are fit better by a model with a pessimistic weight on their own future actions than by an optimal agent with inflated threat values or altered transition beliefs, and that weight relates to anxiety; or the model’s own prediction confirmed, a free-choice premium reduced or reversed in anxious people. It is refuted if anxious avoidance in such tasks is explained as well by altered threat values with an optimal policy, or if the bias appears where no future avoidance is at stake. A qualitative match of new simulations to further published effects cannot settle it.

Backlinks