Computational and neural signatures of interoceptive learning in anorexia nervosa

PROJECT SUMMARY
Anorexia nervosa (AN) is a highly impairing, chronic, and often fatal disorder, however its etiology remains poorly
understood. Aberrant aversive learning, particularly in relation to internal bodily signals (i.e., aversive
interoceptive learning), may be a critical feature of eating disorder pathology, as interoceptive domains are linked
to greater body image disturbance, distorted hunger/satiety cues, and dysregulated affective processing in AN.
Aversive interoceptive learning is driven by discrepancies between anticipated and observed sensory states (i.e.,
prediction errors), brain-based computations associated with networks consisting of the insula, striatum,
prefrontal cortex, and amygdala. Individuals with AN demonstrate difficulties distinguishing between expected
and experienced sensations, suggesting their ability to successfully learn from body sensations is compromised,
which may maintain disordered eating. Despite this, aversive interoceptive learning is considerably understudied
in eating disorders. This is the first study to examine 1) how individuals with AN learn from aversive interoceptive
outcomes, 2) whether neuroanatomical regions supporting aversive interoceptive learning display altered
functional connectivity in AN, and 3) how behavioral and neural signatures of aversive interoceptive learning are
linked. Thirty-two adult women diagnosed with AN and 32 demographically matched healthy controls will
complete an associative learning paradigm utilizing aversive breathing restrictions and will undergo resting-state
functional magnetic resonance imaging. Interoceptive learning will be operationalized using computational
models that track trial-by-trial prediction errors (PE) and stimulus value estimates. Aim 1 will examine model-
generated latent behavioral differences in aversive interoceptive learning (e.g., learning rates) between AN
participants and healthy controls, as well as associations with clinical eating disorder measures. Aim 2 will assess
group differences in insula functional connectivity with regions linked to aversive learning and interoceptive
processing (i.e., amygdala, striatum, prefrontal cortex). Aim 3 will explore associations between insular
connectivity and learning rates. Uncovering behavioral and neural signatures of aversive interoceptive learning
will not only inform etiological models of risk and maintenance in AN, but will also signify an imperative next step
in the development of novel treatments that target both cognitive and sensory processes contributing to eating
disorder pathology. Moreover, this project will provide invaluable training in computational and neuroimaging
methodology, skills critically needed to enhance eating disorder research and treatment development.

Grant Number: 5F31MH133362-02
NIH Institute/Center: NIH
Principal Investigator: Carina Brown