Electroencephalographic indices of stages of reward memory and implications for therapeutic memory reconsolidation

Background 
While there are some evidence-based treatments for substance use disorders (SUDs), maintaining abstinence remains a largely unmet challenge. Current treatments do not tackle the maladaptive environmental factors, nor neuropsychopharmacological changes, that predispose individuals to relapse. Contemporary models of addictive behaviour place aberrant reward learning at the heart of the pathogenesis and maintenance of SUDs. Recently, attention has focussed on the process of reconsolidation in memory maintenance and suppression. Reconsolidation offers a therapeutic window in which to target (and weaken) maladaptive motivational memory traces that promote relapse susceptibility, conferring longer-lasting protection against relapse. Despite the recognised promise of ‘memory therapeutics’, their implementation is currently hampered by a lack of candidate biomarkers that index different reward memory processes. 
 
Methods
Four databases were searched: PubMed, Science Direct, Embase and MEDLINE. Search terms tapped into four key ‘stages’ of associative reward memory: (i) learning (ii) retrieval (iii) destabilization and/or reconsolidation (iv) extinction. Studies had to report EEG activity during a  relevant reward memory process and  include data from human participants who were healthy; ‘at-risk’ (with high but subclinical levels of substance use or gambling); or suffering from a substance use or gambling  disorder.  
 
Results
Of the 7604 articles screened, 46 were included with a total number of 1531 participants. Amongst reward memory retrieval studies, P300 and N400 EEG components were consistently of significance. Reward learning studies identified theta power, P300 and N500 EEG components; and prediction error studies identified delta power, P300 and N500 as being key during processing.
 
Discussion & Conclusion 
Theta oscillatory power and the P300 component could be utilized to predict how well a participant will respond to the reconsolidation intervention. These signals were key in reward learning and thus could indicate if participants had successfully learnt the task which will be used during the reconsolidation intervention. Delta oscillatory power N500 EEG component activity could prove important tools for identifying if the reconsolidation paradigm has successfully induced mismatch between expected and actual outcomes, namely prediction error. Prediction error is an essential precursor for reconsolidation, and so being able to identify if it has been signalled adequately could also provide us with information as to how successfully the prticipnt will respond to the intervention. Overall, this review has identified some key neural signatures within reward memory learning and prediction error processing. The aforementioned components are currently being validated by retrospective EEG analyses on some of our existing data. If successful, we would have identified predictors for participant responses to therapeutic memory reconsolidation.