Drug cue reactivity involves hierarchical instrumental learning: evidence from a biconditional Pavlovian to instrumental transfer task

Abstract

RationaleDrug cue reactivity plays a crucial role in addiction, yet the underlying mechanisms are poorly understood. According to the binary associative account, drug stimuli retrieve an expectation of the drug outcome, which, in turn, elicits the associated drug-seeking response (S-O-R). By contrast, according to the hierarchical account, drug stimuli retrieve an expectation that the contingency between the drug-seeking response and the drug outcome is currently more effective, promoting performance of the drug-seeking response (S:R-O).MethodsThe current study discriminated between these two accounts using a biconditional Pavlovian-to-instrumental transfer (PIT) task with 128 alcohol drinkers. A biconditional discrimination was first trained in which two responses produced alcohol and food outcomes, respectively, and these response-outcome contingencies were reversed across two discriminative stimuli (SDs). In the PIT test, alcohol and food cues were compounded with the two SDs to examine their impact on percent alcohol choice in extinction.ResultsIt was found that alcohol and food cues selectively primed choice of the response that earned that outcome in each SD (p < .001), and this effect was associated with participants’ belief that cues signalled greater effectiveness of that response (p < .0001).ConclusionsThe alcohol stimulus could not have selectively primed the alcohol-seeking response through binary S-O-R associations because the drug outcome was equally associated with both responses. Rather, the alcohol stimulus must have retrieved an expectation that the response-alcohol contingency available in the current context was more likely to be effective (S:R-O), which primed performance of the alcohol-seeking response.

Intact goal‐directed control in treatment‐seeking drug users indexed by outcome‐devaluation and Pavlovian to instrumental transfer: critique of habit theory

“…Finally, in the transfer test, the Pavlovian stimuli are presented while the two instrumental responses are available, and it is typically found that each stimulus selectively augments choice of the response that earns the outcome signalled by the stimulus. This specific PIT effect is arguably driven by each stimulus retrieving an expectation that the corresponding response–outcome contingency has a greater probability of being effective, which drives goal‐directed choice of that response (Colwill & Rescorla, ; Rescorla, ; Seabrooke et al ., ; Hardy et al ., ; Seabrooke et al ., ), although a range of theoretical accounts of specific PIT have been offered (Cartoni et al ., ). In animals, this specific PIT effect can be abolished by chronic amphetamine exposure (Shiflett, ), and perhaps by chronic alcohol exposure (Glasner et al ., ), and alcohol paired cues exert only general excitatory effects on both responses in the specific PIT test, whereas natural reward cues exert specific PIT effects (Glasner et al ., ; Corbit et al ., ).…”

“…Finally, specific PIT studies have also produced little support for habit theory. Seven experiments (in six papers) have reported no correlation between the specific PIT effect and severity of dependence in young adult smokers (Hogarth & Chase, ; Hogarth, ; Hogarth & Chase, ; Hogarth et al ., ) or young adult drinkers (Martinovic et al ., ; Hardy et al ., ), giving a 7 : 0 ratio against habit theory. Collapsing across study types gives a 17 : 5 ratio against habit theory.…”

“…Experiment 1 used standard human outcome‐devaluation and specific PIT procedures (Hogarth & Chase, ; Hogarth, ; Hogarth & Chase, ; Hogarth et al ., , ) to test for impaired goal‐directed control in treatment‐seeking opiate and alcohol dependent individuals vs. a non‐matched student sample. Experiment 2 used novel biconditional outcome‐devaluation and specific PIT procedures (Hardy et al ., ) in which the response–outcome contingencies were reversed randomly on a trial‐by‐trial basis, signalled by two discriminative stimuli, to increase the difficulty of utilising goal‐directed knowledge at test, in the hope of demonstrating a significant difference between treatment‐seeking opiate dependent individuals vs. non‐matched adult community members. The outcome of these studies will contribute to the mixed findings as to whether impaired goal‐directed control is an important feature in human addiction.…”

Alcohol use disorder symptoms are associated with greater relative value ascribed to alcohol, but not greater discounting of costs imposed on alcohol

“…Intensity of demand for alcohol correlates with various proxies for dependence, including drinks consumed per week (MacKillop and Murphy 2007), episodes of heavy drinking per week (Murphy and MacKillop 2006) and alcohol-related problems (Murphy et al 2009). Similarly, in concurrent choice procedures, where drinkers choose between alcohol and food rewards (points or pictures), preference for the alcohol reward is associated with alcohol use disorder symptom severity in both hazardous drinkers recruited from the community (Hardy and Hogarth 2017) and student drinkers (Hardy et al 2017; Hogarth et al 2018). These demand and choice data fit with the prediction of economic theory that drinkers with greater dependence symptoms ascribe greater relative value to alcohol, which could underpin persistent drinking despite costs.…”

“…In concurrent choice procedures, participants choose between a drug reward and a concurrently available natural reward alternative across a series of trials (the two rewards may be points-based, pictures or actually consumed/administered depending on the method). The claim that percent drug choice indexes the relative value ascribed to the drug versus natural reward is supported by the finding that percent drug choice reliably increases with the severity of dependence to alcohol (Hardy and Hogarth 2017; Hardy et al 2017; Hogarth et al 2018), cocaine (Moeller et al 2013; Moeller et al 2009) and tobacco (Chase et al 2013; Hogarth and Chase 2011). Importantly, concurrent choice procedures can also index sensitivity to opportunity costs, quantified by the decrease in drug choice that occurs when the magnitude of the competing alternative reward is increased.…”