A computational model for learning from repeated trauma

Abstract

Traumatic events can lead to lifelong inflexible adaptations in threat 9 perception and behavior which characterize posttraumatic stress disorder (PTSD). This 10 process involves associations between sensory cues and internal states of threat and 11 then generalization of the threat responses to previously neutral cues. However, most 12 formulations neglect adaptations to threat that are not specific to those associations. In 13 order to incorporate non-associative responses to threat, we propose a computational 14 theory of PTSD based on adaptation to the frequency of traumatic events using a 15 reinforcement learning momentum model. Recent threat prediction errors generate 16 momentum that influences subsequent threat perception in novel contexts. This model 17 fits data acquired from a mouse model of PTSD, in which unpredictable footshocks in 18 one context accelerate threat learning in a novel context. The theory is also consistent 19 with epidemiological data showing that PTSD incidence increases with the number of 20 traumatic events, as well as the disproportionate impact of early life trauma. Since the 21 theory proposes that PTSD relates to the average of recent threat prediction errors 22

Scalar-on-function regression for predicting distal outcomes from intensively gathered longitudinal data: Interpretability for applied scientists

“…Other possibilities also exist. [35] provides a computational model for simulating the effect of early and recent events on later behavior. Causal modeling, whether frequentist or Bayesian, may also be useful [see, e.g.…”

Predictive processing in mental illness: Hierarchical circuitry for perception and trauma.

“…Reinforcement learning models have recently encompassed a simpler but related notion that recent prediction errors in one context generate a momentum that can additively adjust predictions across contexts (Eldar, Rutledge, Dolan, & Niv, 2016; Rutledge, Skandali, Dayan, & Dolan, 2014; Figure 1D). We recently applied this model to PTSD (Kaye, Kwan, Ressler, & Krystal, 2019), positing that the experience of repeated threat of violence (trauma) leads to a shared threat estimation across all potential environments (a universal estimate of threat). These reinforcement learning momentum models make explicit the relationship between previous prediction errors and the change in future predictions across states.…”