Understanding the neural and cognitive mechanisms of decision making in alcohol dependence

Abstract

Alcohol dependence is a debilitating disorder associated with many underlying neurobiological and cognitive influences. Although several studies have shown decision making deficits in alcohol dependence, few have specified the underlying cognitive and neural mechanisms responsible for the deficit. This study investigated decision making in alcohol dependence using the framework of a neural network model of decision making (Frank & Claus, 2006). This model proposes that probabilistic learning from positive and negative feedback relies on dopamine D1 and D2-type receptors in the basal ganglia (BG) in combination with phasic dopamine (DA) bursts. Complementing the BG/DA system is a system that learns about outcome magnitudes in the orbitofrontal cortex. I examined decision making in alcohol dependence and focused on three aspects of decision making behavior: learning from probabilistic feedback, learning from the relative magnitude of outcomes, and intertemporal choice. Non-dependent heavy drinkers and treatment seeking alcohol dependent (TxAD) participants completed the following tasks: probabilistic selection, reversal learning, and Stroop. In addition, participants completed a gambling and a delay discounting task while being scanned in fMRI. Results indicated that TxAD participants were relatively unimpaired at learning from probabilistic feedback, as they exhibited performance equivalent to heavy drinkers across the probabilistic selection task and the first stage of the reversal learning task. However, alcohol dependent participants exhibited differential ability to learn from large infrequent punishments on the gambling task, showing greater selection from a deck in which the frequency of punishments was low, but the magnitude of punishments was very high. In addition, TxAD participants failed to activate lateral orbitofrontal cortex in response to punishments to the same degree as heavy drinkers. In the delay discounting task, TxAD participants showed equivalent discounting rates to heavy drinkers, but brain activation patterns differed between the two groups. Whereas heavy drinkers activated a rostral region of prefrontal cortex (Brodmann area 46), treatment seekers strongly activated a more caudal region of PFC (BA 8/6), suggesting that alcohol dependent individuals may engage different types of control mechanisms when making decisions, and this may be a core deficit associated with the disorder.