“Addiction states as dynamic changes in valuation?”

There is now compelling evidence for a “domain general” neural system in the computation of subjective value—the basis of idiosyncratic preferences and choice. Perturbations in this valuation system in both humans and other animals successfully capture many features of addiction, a psychiatric disorder characterized by maladaptive choice behavior. However, existing work has generally considered addiction as a static entity, ignoring addiction’s most elusive (and perhaps most defining) feature—its stereotyped, cyclic nature at the level of the individual, characterized by alternating periods of abstinence and drug use. I will discuss recent work in which we aim to better understand two dynamic processes at the transition between abstinence and drug use in human subjects: (1) a relatively slower process (in the order of weeks) related to uncertainty preferences that tracks drug use vulnerability which we have modeled with repeated assessments of economic choice behavior through the first months of treatment for opioid addiction and that appears to be supported by enhanced neural encoding of uncertainty and weaker connectivity within the valuation system; and (2) a faster process (in the order of minutes or hours) related to the subjective state craving that tracks immediate vulnerability and which we have modeled as a selective, gain-control like shift in the value of the object of craving. Understanding these addiction states as dynamic changes in valuation, we hope, can help identify when additional therapeutic intervention is needed on a timescale that is clinically useful as well as motivate the development of new decision- and valuation-based interventions for breaking the cycle of addiction.