Altered neural encoding of prediction errors in assault-related posttraumatic stress disorder

Abstract

Posttraumatic stress disorder (PTSD) is widely associated with deficits in extinguishing learned fear responses, which relies on mechanisms of reinforcement learning (e.g., updating expectations based on prediction errors). However, the degree to which PTSD is associated with impairments in general reinforcement learning (i.e., outside of the context of fear stimuli) remains poorly understood. Here, we investigate brain and behavioral differences in general reinforcement learning between adult women with and without a current diagnosis of PTSD. 29 adult females (15 PTSD with exposure to assaultive violence, 14 controls) underwent a neutral reinforcement-learning task (i.e., two arm bandit task) during fMRI. We modeled participant behavior using different adaptations of the Rescorla-Wagner (RW) model and used Independent Component Analysis to identify timecourses for large-scale a priori brain networks. We found that an anticorrelated and risk sensitive RW model best fit participant behavior, with no differences in computational parameters between groups. Women in the PTSD group demonstrated significantly less neural encoding of prediction errors in both a ventral striatum/mPFC and anterior insula network compared to healthy controls. Weakened encoding of prediction errors in the ventral striatum/mPFC and anterior insula during a general reinforcement learning task, outside of the context of fear stimuli, suggests the possibility of a broader conceptualization of learning differences in PTSD than currently proposed in current neurocircuitry models of PTSD.

Diminished positive affect and traumatic stress: A biobehavioral review and commentary on trauma affective neuroscience

“…This evidence is mixed both in regard to both presence and absence of abnormalities, as well as the specific behavioral and circuit dynamics implicated as abnormal across studies. The most consistent finding across studies and paradigms is a hyporesponsivity of the ventral striatum to monetary and “social” reward receipt ( Elman et al, 2009 ; Felmingham et al, 2014 ; Ross et al, 2018 ; Sailer et al, 2008 ), which is consistent with the motivational deficits, diminished interest (as diminished subjective “liking” of a reward will induce less motivation or interest in obtaining that reward in the future), and diminished capacity to feel positive emotions that are frequently observed to characterize individuals with the PTSD diagnosis ( Hassija et al, 2012 ). Moreover, greater severity of PTSD diminished positive affect symptoms were observed to relate to more severe deficits in ventral striatal engagement to both monetary and social reward ( Elman et al, 2009 ; Felmingham et al, 2014 ), which further highlights the plausibility of this convergence.…”

“…Utilizing a two-arm bandit task (i.e. participant has a choice between two options on each decision portion of the trial and is attempting to pick the option that provides the largest monetary payoff in the outcome phase) in women with PTSD due to assault, the investigators examined how computational estimates of stimulus value (the degree to which a particular choice stimulus during the decision phase was perceived to be associated with likelihood of a future reward) and prediction errors (the difference between the outcome expected and the outcome received) differed as a function of diagnostic status, as well as how the information components of value and prediction errors may be differentially reflected in brain circuitry ( Ross et al, 2018 ). The authors utilized the common Rescorla-Wagner computational model of reinforcement learning ( Sutton and Barto, 1998 ) to derive trial-by-trial estimates of stimulus value and prediction errors for each individual, and then examined how these component estimates related to time courses of BOLD activity (separated by independent component analysis-defined spatial networks) throughout the course of the task.…”

A computational model for learning from repeated trauma

“…Previous approaches to computational modeling of PTSD have focused on defining changes in associative learning after traumatic experience (11-15). PTSD is thus framed as a consequence of underlying mechanisms for predicting threat based on previous associations .…”

“…Identifying PTSD with threat momentum may facilitate future neurobiological and translational studies of PTSD. Extensive work has shown that patients with PTSD have different learning rate parameters during fear and extinction learning (11-15) than controls in the formation of associations. This study extends these findings by offering a model of how the sequence of threat prediction errors may generate other associative learning alterations in PTSD.…”

“…Posttraumatic stress disorder has a computational description that organizes theory and neurobiological data across Marr’s three levels - associative fear learning (Figure 1B, refs. 11-15). Learning models have been successfully applied to PTSD and underlie current conceptualizations of the disorder and treatment options (16-17).…”