Neuronal correlates of reward and loss in Cluster B personality disorders: A functional magnetic resonance imaging study

Abstract

Decision making is guided by the likely consequences of behavioural choices. Neuronal correlates of financial reward have been described in a number of functional imaging studies in humans. Areas implicated in reward include ventral striatum, dopaminergic midbrain, amygdala and orbitofrontal cortex. Response to loss has not been as extensively studied but may involve prefrontal and medial temporal cortices. It has been proposed that increased sensitivity to reward and reduced sensitivity to punishment underlie some of the psychopathology in impulsive personality disordered individuals. However, few imaging studies using reinforcement tasks have been conducted in this group. In this fMRI study, we investigate the effects of positive (monetary reward) and negative (monetary loss) outcomes on BOLD responses in two target selection tasks. The experimental group comprised eight people with Cluster B (antisocial and borderline) personality disorder, whilst the control group contained fourteen healthy participants. A key finding was the absence of prefrontal responses and reduced BOLD signal in the subcortical reward system in the PD group during positive reinforcement. Impulsivity scores correlated negatively with prefrontal responses in the PD but not the control group during both, reward and loss. Our results suggest dysfunctional responses to rewarding and aversive stimuli in Cluster B personality disordered individuals but do not support the notion of hypersensitivity to reward and hyposensitivity to loss.

Introduction

Rewarding and punishing stimuli result in an increase or decrease of the probability of antecedant actions, thereby shaping behaviour. In recent years, functional neuroimaging studies in humans using primary (e.g. O'Doherty et al., 2001a, O'Doherty et al., 2002) and abstract rewards (e.g. Breiter et al., 2001, Elliott et al., 2000, Elliott et al., 2003) have advanced our understanding of the neuronal correlates of reinforcement processing and have corroborated previous findings from single-cell electrophysiological and lesion studies in animals. This research has implicated a network of interconnected brain regions mediating the behavioural and motivational effects of reward, including ventral striatum, dopaminergic midbrain, amygdala and orbitofrontal cortex (OFC; for a review see O'Doherty, 2004). Distinct functions have been attributed to these different regions. For instance, amygdala, striatum and midbrain have been found to respond to the presence of reward regardless of value (e.g. Elliott et al., 2000, Elliott et al., 2003); in contrast, a more complex pattern of responses has been identified in medial and lateral OFC suggesting a possible role for higher order processing of reinforcing stimuli, such as the integration of stimulus attributes and emotional value (e.g. Elliott et al., 2003, Kringelbach et al., 2003, O'Doherty, 2004). This in turn allows the salience of reinforcing stimuli to be updated and modulated following changes in contingencies and the subsequent use of this information in action selection.

The role of reward system components in the response to punishment or loss is less clear. Several fMRI studies in humans have suggested the striatum has an important role. Jensen et al. (2003) have identified ventral striatum responses in anticipation of sensory aversive stimuli. Other authors have shown ventral striatum activity associated with anticipation and following the presentation of both monetary rewards and punishments (e.g. Knutson et al., 2001, Delgado et al., 2003). Decreased BOLD signal in dorsal and ventral striatum has been observed following punishing feedback (Delgado et al., 2000). These findings suggest that this structure is not functionally specific to reward but may have a more general role in the processing of reinforcing stimuli. Other authors have identified BOLD signal changes in lateral OFC following punishment (e.g. O'Doherty et al., 2001b, Remijnse et al., 2005). This might reflect the involvement of this structure in response inhibition (Aron et al., 2003). In addition, anterior cingulate and thalamus (Knutson et al., 2000), right amygdala (Zalla et al., 2000), insula (O'Doherty et al., 2003) and hippocampus/parahippocampus (Elliott et al., 2000) have been associated with the experience of loss or punishment in humans.

Dysfunctional responses to reinforcing stimuli have been proposed to underlie the psychopathology in substance use and impulsivity-related personality disorders (Petry, 2002). The latter encompass two personality disorders within Cluster B of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; APA, 1994): antisocial (ASPD) and borderline (BPD) personality disorder. These two disorders share some common characteristics, particularly high levels of impulsive behaviour; some authors have argued that ASPD and BPD are manifestations of the same underlying pathology in male and female individuals respectively (e.g. Paris, 1997). There is considerable co-morbidity between these two personality disorders: in male individuals a co-occurrence of up to 50% has been identified (Zanarini et al., 1998, Chabrol and Leichsenring, 2006). Becker et al. (2005) noted that symptoms related to impulsive behaviour in BPD were not significantly more efficient in diagnosing BPD than ASPD. It therefore seems justified to consider these two disorders together as an impulsivity-related personality disorder as has previously been suggested by other authors (e. g. Goethals et al., 2005).

A number of aetiological models of impulsivity-related personality disorders have been put forward. Early accounts (Gray, 1987) postulated two distinct motivational systems: a behavioural activation system (BAS) which is sensitive to reward cues and a behavioural inhibition system (BIS) which is sensitive to punishment. In this model, impulsive–aggressive behaviour as observed in impulsive Cluster B PD is proposed to result from an imbalance of these two systems, either due to an oversensitivity of the BAS or due to hyporesponsiveness of the BIS.

More recent models of impulsive personality disorders have focused on behavioural choice in the context of reward and punishment. The reward dominance theory suggests that antisocial individuals show greater responsivity to reward and decreased sensitivity to punishment in situations where both types of stimuli are available (Scerbo et al., 1990). Impulsive individuals focus on the prospect of reward even if environmental cues indicate possible later punishment (Budhani and Blair, 2005). Preference for shorter delays in reward-choice tasks has been demonstrated in borderline (Dougherty et al., 1999) and antisocial personality disorder (Moeller et al., 2002) and in probation and parole groups (Cherek et al., 1999, Cherek et al., 1997). Other authors have shown that antisocial groups perform poorly on passive avoidance tasks by failing to inhibit punishable responses (e. g. Dikman and Allen, 2000).

The coding of stimulus reward values is crucial for generating appropriate reward-directed behavioural responses i.e. in guiding individuals' selection of advantageous over disadvantageous behaviour based on previous experience (Kringelbach and Rolls, 2004). This process has been suggested to be impaired in individuals with ventromedial prefrontal brain damage (e.g. Damasio et al., 1990, Damasio, 1994) accounting partly for their impulsive behaviour. These observations in “acquired sociopathy” (Saver and Damasio, 1991), brain imaging (Goyer et al., 1994, Raine et al., 2000, Herpertz et al., 2001, Donegan et al., 2003, Goethals et al., 2005) and neuropsychological evidence (Lapierre et al., 1995, Bazanis et al., 2002, Dolan and Anderson, 2002, Rogers, 2003) in borderline and antisocial personality disordered individuals have led to a proposal that some of their psychopathology can be explained by ventromedial and orbitofrontal dysfunction.

While a number of studies have investigated performance and neuropsychological correlates of behavioural choice in individuals with impulsive Cluster B PD no brain imaging studies have been conducted examining neuronal correlates of stimulus selection in the context of positive and negative outcomes in these patients.

In this study we used two stimulus selection tasks to investigate the impact of positive and negative outcomes (further referred to as reward and loss task) on BOLD responses in impulsivity-related personality disorders. We hypothesized that control participants would show BOLD signal increases in orbitofrontal regions during positive and negative feedback. We further hypothesized that ASPD and BPD patients (in comparison to healthy controls) would show reduced signal change in prefrontal (particularly orbitofrontal) regions in both tasks with increased BOLD response in reward and decreased BOLD response in loss related brain areas.