Spared internal but impaired external reward prediction error signals in major depressive disorder during reinforcement learning

Abstract

Collectively, these results lend support to the assumption that MDD does not impair reward processing per se during RL. Instead, it seems to alter the processing of the emotional value of (external) reinforcers during RL, when additional intrinsic motivational processes have to be engaged.

Abnormal approach-related motivation but spared reinforcement learning in MDD: Evidence from fronto-midline Theta oscillations and frontal Alpha asymmetry

“…Participants performed a probabilistic learning task (Fig. 1) previously devised and validated by Eppinger et al (2008) and used in Bakic et al (2017Bakic et al ( , 2014. Colorful line drawings (Rossion & Pourtois, 2004) were used as visual stimuli, presented against a white homogenous background on a 17-inch computer screen.…”

“…To get acquainted with the task, they completed a short practice session of 20 trials with an extra set of stimuli. The whole experiment lasted approximately 2 hours (Bakic et al, 2017).…”

“…At the behavioral level, learning was expressed as percentage of correct responses varying as a function of time, using four consecutive bins of trials ( Bakic et al, 2017).…”

“…In this study, we had the unique chance to assess, using behavioral and EEG methods, brain mechanisms of RL (using FMT oscillatory perturbations), as well as motivation (using frontal Alpha asymmetry) concurrently in a large cohort of treatment-resistant MDD patients and compare them to age/ sex/education-matched healthy controls. To explore RL, we capitalized on a well-validated probabilistic learning task (Eppinger et al, 2008), previously used and validated in our laboratory (Bakic et al, 2017(Bakic et al, , 2014. In short, the added value of this task is that three reward probabilities are manipulated concurrently, and their effects on the learning rate and the phasic signals of enhanced cognitive control can be explored by using appropriate EEG methods (van de Vijver et al, 2014).…”

“…In healthy controls, FMT should exhibit symmetric changes between response errors and negative FB as a function of reward probability (van de Vijver et al, 2014) but might be hypoactive in MDD patients, suggesting blunted cognitive control modulation during RL. However, we predicted that these group differences should likely depend on reward probability (i.e., strength of stimulus-response association), given that MDD might interfere with RL selectively when higher efforts and enhanced motivation are required to foster learning (Bakic et al, 2017;Salamone, Correa, Nunes, Randall, & Pardo, 2012;Thomsen, 2015;Treadway et al, 2012). In particular, we expected larger group differences at the FB level when reward probability was low compared with high, because a higher motivation is presumably required in this condition for maintaining an active and sustained exploration of the FB.…”