5.1 Introduction

In recent years, there has been tremendous support for working toward the RDoC goals of identifying the neurobiological mechanisms that cut across or are common to multiple psychiatric disorders. Identifying the pathophysiological mechanisms underlying transdiagnostic symptoms will improve the validity of disease classifications by grouping individuals based on multiple dimensions of behavior and biology. This could potentially account for heterogeneity and comorbidity observed among DSM diagnostic categories. One of the constructs that provides an excellent opportunity for transdiagnostic research is reward processing, as reward-related dysfunction is reported across several psychiatric disorders including schizophrenia and addiction, but especially mood disorders. Current research suggests that abnormal reward sensitivity is involved across the entire mood disorder spectrum, with blunted reward-related functioning observed as a risk for major depression, whereas abnormally elevated reward-related functioning reported as a risk for bipolar disorders (BDs). However, the results are mixed and there is considerable debate regarding the degree to which shared or distinct profiles of reward system dysfunction contribute to the motivational deficits observed across unipolar and bipolar mood disorders. This is due to the fact that reward processing is not a unitary construct, but multifaceted, and comprises subcomponents with distinct and overlapping neurobiological underpinnings. For example, different components of reward processing are suggested to include reward anticipation, reward consumption, and reward learning. In addition, both unipolar depression and BDs are highly heterogeneous. Consequently, considerable research has focused on understanding if these discrete components of reward processing could explain the heterogeneity observed across mood disorders and lead to finer parcellation of symptom clusters. This will further lead to identification of homogeneous subgroups based on symptomatology, a prerequisite process to the development of targeted prevention and treatment strategies.

The primary focus of this chapter is to review the neural mechanisms underlying different reward processing components, including reward anticipation, reward consumption, and reward learning, in major depressive disorder (MDD) and BD. The chapter will also review evidence for these neural mechanisms acting as state-like versus trait-like markers of mood pathology. In addition, we will comment on future analytical approaches that are needed to understand the complex reward-related disruptions observed across mood disorders.

5.2 Neurobiology of Reward Processing and Its Components

The components of reward processing mentioned here – reward anticipation, reward consumption, and learning map onto the core reward components (reward responsiveness, reward valuation, and reward learning) of reward processing outlined in the Positive Valence Systems Matrix of the National Institute of Mental Health’s (NIMH’s) Research Domain Criteria (1–3). Reward responsiveness (or anticipation) is associated with the ability to anticipate or represent future incentives. In contrast, reward valuation refers to the processes by which the value of a reinforcer is computed as a function of its magnitude, predictability, time to expected delivery, and the effort required to obtain it. Finally, reward learning incorporates the process by which an organism acquires information about reward-predictive cues and by which novel reward outcomes subsequently shape behavior.

Evidence suggests that these processes map onto partially distinct neural circuitry (for a review, see 4). Briefly, the reward system has been linked to the frontal–striatal circuit that responds to stimuli involving the anticipation and receipt of rewards (e.g., 5). The ventral striatum is thought to play a central role in reward anticipation and is involved in processing both primary (e.g., food) and secondary (e.g., monetary) rewards. In contrast, studies examining reward consumption implicate “hedonic hotspots” in the ventral pallidum and nucleus accumbens that are the site of endogenous opioid receptors (6), as well as the encoding of reward value in the orbitofrontal cortex (OFC; 7). In the past decade there has been a burgeoning interest in applying computational algorithms to dissect reward learning in healthy and psychiatric populations. Using these models, individual differences can be estimated by measuring trial-by-trial variability in learning. Learning occurs when there is a deviation between the expected and actual outcome, quantified as a prediction error which is then used to update value estimates that support better prediction of future rewards. Nonhuman primate findings have shown that phasic firing of dopamine (DA) neurons in the ventral tegmental area (VTA) encodes reward prediction error (RPE). These midbrain DA RPE signals are then transmitted to the striatum and cortex and used to update stimulus-action values and guide goal-directed behavior (8, 9). Consistent with this, human fMRI studies have described RPE signals in cortico-striatal circuits including the striatum, midbrain, and prefrontal cortex (10, 11), and these signals have been found to be disrupted under conditions that affect phasic DA signaling (e.g., DA antagonists) (12–14).

5.3 Major Depressive Disorder

5.4 Is Reward Dysfunction a Potential Endophenotype of MDD?

According to Gottesman and Gould (35), endophenotypes (or intermediate phenotypes) are measurable components (e.g., neurophysiological, biochemical, neuroanatomical, cognitive, or neuropsychological) that exist between the behavioral symptoms of a disease and genes. To be considered an endophenotype, a reward dysfunction should meet the following criteria (35): (a) specificity (i.e., reward dysfunction is strongly associated to a given condition); (b) heritability; (c) state independent (i.e., reward dysfunction is stable over time and independent from clinical state and treatment); (d) cosegregation (i.e., reward dysfunction occurs more frequently in affected compared to non-affected relatives of an ill individual); (e) familial association (i.e., reward dysfunction is more frequent in relatives of ill individuals than the general population); and (f) biological and clinical plausibility (36). Because endophenotypes are presumed to be more proximal to genes than clinical diagnoses, they may help us identify genetic variants and associated genes with small samples. Identification of genetic variants and associated genes using small samples. Endophenotypes are thought to be to be disorder-specific. Blunted striatal response to both anticipation and consumption has been suggested as an endophenotype of MDD, as blunted striatal response to rewards has been reported in individuals with current depression, remitted MDD, and non-affected family members of individuals with MDD. Next, endophenotypes must be heritable and although no studies have investigated the heritability of striatal reward responses, behavioral response to rewards and reward-related ERPs is reported to be heritable. In addition, blunted striatal response to reward response has been observed within never-depressed offspring of depressed versus healthy parents (37, 38). This blunted striatal response to both anticipation (39) and consumption (40, 41) prospectively predicted worsening of depressive symptoms over two years among adolescents, even after controlling for baseline symptoms. Furthermore, reduced FRN amplitude – a deflection in the ERP thought to originate from RPE activity in the dorsal ACC and striatum, predicted first onset of MDD in a two-year follow-up among never-depressed adolescent girls (26). Finally, the reliability of striatal response to rewards has been poor to moderate (17). Nevertheless, taken together, these studies suggest that neural markers of altered reward processing are a promising endophenotype for MDD risk. However, future studies will need to replicate these findings over a broad developmental range and to investigate the specificity and reliability of such findings.

5.5 Bipolar Disorder

5.6 Heritability of Reward Dysfunction in BD

BD is a highly heritable disorder, with an estimated heritability of 60–80% (47). The offspring of patients are reported to have a tenfold risk of developing BD relative to the general population. Therefore, identifying endophenotypes (or intermediate phenotypes) based on biological constructs that are associated with BD and reported in individuals at risk of BD could elucidate the pathophysiology of BD. Consistent with this, abnormalities in reward-related neural responses have been reported in healthy offspring of parents with BD. For example, similar to BD patients, healthy offspring of parents with BD also had elevated vlPFC activation during reward consumption (47, 58). Similarly, they also show elevated vlPFC–striatal functional connectivity during reward processing that was specific to offspring of parents with BD and not observed in healthy controls or offspring of parents with non-BD psychopathology (59), underscoring the specificity of these findings to BD and a potential “intermediate phenotype” for risk of BD.