Trait impulsivity

Conceptualizations of the neurobiological basis of impulsivity have evolved over time (see Neuhaus & Beauchaine, Reference Neuhaus, Beauchaine, Beauchaine and Hinshawin press). Hippocrates viewed impulsive tendencies as the product of excessive yellow bile (Titchener, Reference Titchener1921). In the 18th century, Franz Joseph Gall argued based on phrenology that the absence of the mental faculty of “cautiousness” resulted in impulsivity (Spurzheim, Reference Spurzheim1834). In the 19th century, John Harlow provided a detailed account of perhaps the most famous head injury of all time: that suffered by Phineas P. Gage (Harlow, Reference Harlow1848). His observations of Gage were consistent with assumptions of brain–behavior relationships that were popular at the time. In particular, specific regions of the frontal cortex were believed to underlie precautious motivations, with damage to these regions resulting in disinhibition. Many also assumed that behavior was the outcome of competing mental forces. Among healthy individuals, more primitive urges were effectively constrained by inhibitory control mechanisms. Thus, damage to brain regions responsible for inhibiting impulses created disequilibrium, so “animal propensities” prevailed (Harlow, 1948).

The notion that adaptive behavior derives from a relative equilibrium between self-gratifying and restrained motivations pervades major theories of impulsivity. For example, Eppinger and Hess (1910/1915) asserted that vagotonia, a term used to describe an imbalance in the autonomic nervous system favoring the parasympathetic over the sympathetic division, was used to explain a range psychological and medical phenomena, including neurasthenia, hysteria, nervousness, and impulsive tendencies.

Several more recent theories of impulsivity posit that individual differences in approach behavior derive from activity and reactivity of mesolimbic structures. For example, in his prominent neurobiological theory of personality, Gray (Reference Gray1987a, Reference Gray1987b) proposed a mesolimbic DA behavioral approach system as the neural substrate of appetitive motivation. Subsequently, DA theories of approach motivation were used to explain impulsive, reward-seeking behaviors often observed in ADHD, CD, and other externalizing disorders (e.g., Fowles, Reference Fowles1988; Quay, Reference Quay1993). Although these early theories correctly attributed impulsivity to mesolimbic neural dysfunction, they incorrectly assumed that excessive DA activity predisposed affected individuals to impulsive behavior. As described in detail here and elsewhere, subsequent research reveals that externalizing spectrum disorders are characterized by low rather than high mesolimbic DA activity and reactivity (see Beauchaine & Gatzke-Kopp, Reference Beauchaine and Gatzke-Kopp2012; Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Gatzke-Kopp & Beauchaine, Reference Gatzke-Kopp, Beauchaine, Coch, Dawson and Fischer2007; Zisner & Beauchaine, Reference Zisner, Beauchaine, Beauchaine and Hinshaw2015).

Contemporary accounts of impulsivity operationalize the construct in a number of ways. In its most general sense, impulsivity refers to deficiencies in self-control over behavior (see, e.g., Zisner & Beauchaine, Reference Zisner, Beauchaine, Beauchaine and Hinshaw2015). The extent to which one exhibits impulsive tendencies varies continuously across the population, and normal variation in this trait is not maladaptive. Rather, normal variation in trait impulsivity is reflected in core aspects of personality, including novelty seeking, sensation seeking, extraversion, and exuberance, all of which may foster adaptive functioning, including creative and flexible thinking (see, e.g., Degnan et al., Reference Degnan, Hane, Henderson, Moas, Reeb-Sutherland and Fox2011; Sagvolden, Johansen, Aase, & Russell, Reference Sagvolden, Johansen, Aase and Russell2005).

In contrast, excessive trait impulsivity, the focus of this paper, confers vulnerability to clinically significant psychopathology. As noted above, high trait impulsivity is expressed as a preference for immediate, smaller rewards over larger, delayed rewards, and often results in behaviors that are socially inappropriate, maladaptive, and short-sighted (see Neuhaus & Beauchaine, Reference Neuhaus, Beauchaine, Beauchaine and Hinshawin press; Zisner & Beauchaine, Reference Zisner, Beauchaine, Beauchaine and Hinshaw2015). As we discuss in detail below, such behaviors reflect an overarching problem with effectively anticipating rewards and optimizing goal-directed behavior (Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005).

In discussing trait impulsivity, it is important to note that others conceptualize the construct without a clinical aspect in mind. For example, Whiteside and Lynam (Reference Whiteside and Lynam2001) attribute four etiologically distinct personality facets, including urgency, lack of planning, lack of perseverance, and sensation seeking to impulsive behavior, and Patton, Stanford, and Barratt (Reference Patton, Stanford and Barratt1995) distinguish between motor impulsiveness (actions without forethought), nonplanning impulsiveness (overemphasis on the present), and attentional impulsiveness (difficulty maintaining focus on stimuli). In addition, impulsivity is sometimes operationalized as behavioral performance on circumscribed tasks, such as go/no-go (continuous performance), stop-signal, choice serial reaction time, delay-discounting, and cost/benefit decision-making tasks (e.g., Bezdjian, Baker, Lozano, & Raine, Reference Bezdjian, Baker, Lozano and Raine2009; Bickel, Odum, & Madden, Reference Bickel, Odum and Madden1999; Dimoska & Johnstone, Reference Dimoska and Johnstone2007). Go/no-go, stop-signal, and choice serial reaction time tasks may best capture inattentiveness and impulsive action (i.e., premature responding and/or failure to inhibit responses), whereas other behavioral tasks, such as delay-discounting and cost/benefit decision-making tasks, may best capture impulsive choice. However, common self-report and behavioral measures of impulsivity capture only partially overlapping, and in some cases more distinct, aspects of impulsivity (Reynolds, Ortengren, Richards, & de Wit, Reference Reynolds, Ortengren, Richards and de Wit2006; Robinson et al., Reference Robinson, Eagle, Economidou, Theobald, Mar and Murphy2009).

Given heterogeneity in definitions of and approaches to measuring impulsivity, a well-supported case for a particular operational definition is crucial. We prefer to operationalize impulsivity for human participants using ADHD hyperactive–impulsive scale scores over other measures for three primary reasons: (a) hyperactive–impulsive symptoms of ADHD are highly heritable (e.g., Faraone et al., Reference Faraone, Perlis, Doyle, Smoller, Goralnick and Holmgren2005), far more so than most alternative measures of impulsivity (73%–85% of the variance in ADHD symptoms is explained by heritability; Nikolas & Burt, Reference Nikolas and Burt2010; Willcutt, Reference Willcutt and Barchin press); (b) trait impulsivity among those with ADHD is associated with well-replicated physiological and neural substrates (e.g., Beauchaine, Katkin, Strassberg, & Snarr, Reference Beauchaine, Katkin, Strassberg and Snarr2001; Crowell et al., Reference Crowell, Beauchaine, Gatzke-Kopp, Sylvers, Mead and Chipman-Chacon2006; Hart, Radua, Mataix-Cols, & Rubia, Reference Hart, Radua, Mataix-Cols and Rubia2012; Plichta & Scheres, Reference Plichta and Scheres2014); and (c) ADHD predicts functional outcomes including academic and occupational underachievement, and confers vulnerability to other externalizing spectrum disorders across development (see Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Beauchaine et al., Reference Beauchaine, Hinshaw and Pang2010; Neuhaus & Beauchaine, 2013; Zisner & Beauchaine, Reference Zisner, Beauchaine, Beauchaine and Hinshaw2015).

Anhedonia

Anhedonia, a term that derives from ancient Greek (ἀν- an-, “without” + ἡδονή hēdonē, “pleasure”), was formally coined by Théodule-Armand Ribot to describe an inability to experience pleasure (Ribot, Reference Ribot1896). Even before the term emerged, diminished response to pleasure was considered to be a significant if not essential presenting feature of depression (Bevan, Reference Bevan1899; Bucknill & Tuke, Reference Bucknill and Tuke1874; Clouston, Reference Clouston1896), and it was incorporated into early influential theories of schizophrenia (Bleuler, Reference Bleuler1911; Kraepelin, Reference Kraepelin1919). Interest in the role of anhedonia in psychopathology dwindled in the first half of the 20th century; but it was revitalized in the 1960s by the psychoanalyst Sandor Rado, who posited that anhedonia was one of two key symptoms that marked genetic vulnerability to schizophrenia. Rado (Reference Rado1956, Reference Rado1962) suggested that anhedonia could be transmitted genetically, and that it interfered with normal living among individuals with schizophrenia, and among their family members who did not experience psychotic features. Paul Meehl (Reference Meehl1962, Reference Meehl1973) expanded on this work by suggesting that low hedonic capacity is a characteristic of and heritable trait vulnerability to both depression and schizophrenia.

Anhedonia became a defining feature of mood disorders following Donald Klein's conceptualization of endogenous depression. He asserted that for some, depression is a biologically derived pathology, with onset of anhedonia being the first observable sign of a depressive episode (Klein, Reference Klein1974). Klein's work contributed to inclusion of anhedonia as a criterion for major depressive disorder, beginning with DSM-III (American Psychiatric Association [APA], 1980).

Klein (Reference Klein1974), DSM-III (APA, 1980), and many others describe anhedonia as diminished interest or capacity to experience pleasure in response to stimuli that are ordinarily rewarding. However, “diminished interest” implies that hedonic capacity must change from high to low over some interval of time (e.g., during a major depressive episode). In contrast, we and others assert that for some, anhedonia is both traitlike and state dependent. We therefore define trait anhedonia as a relatively stable personality attribute, characterized by chronically low motivation, and inability to experience pleasure (e.g., Loas, Reference Loas1996; McCabe, Reference McCabe and Ritsner2014; Shankman, Nelson, Harrow, & Faull, Reference Shankman, Nelson, Harrow and Faull2010). However, we recognize that anhedonia captures partly dissociable processes that may be compromised differentially. These include anticipatory versus consummatory responses to reward, social versus physical (sensory) anhedonia, differences between interest/motivation to seek rewards versus pleasure derived from rewards, and differential responding to distal versus proximal reinforcers (see Shankman et al., Reference Shankman, Katz, DeLizza, Sarapas, Gorka, Campbell and Ritsner2014). As elaborated in later sections, the aspect of trait anhedonia that is of greatest interest here is diminished positive affect in anticipation of rewards, a deficit related directly to reduced interest in and motivation to seek pleasurable stimuli.

Like trait impulsivity, anhedonia is operationalized in a number of ways. Common measures include the original and revised Chapman Physical and Social Anhedonia Scales (Chapman, Chapman, & Raulin, Reference Chapman, Chapman and Raulin1976), the Snaith–Hamilton Pleasure Scale (Snaith et al., Reference Snaith, Hamilton, Morley, Humayan, Hargreaves and Trigwell1995), and the Fawcett–Clark Pleasure Scale (Fawcett, Clark, Scheftner, & Gibbons, Reference Fawcett, Clark, Scheftner and Gibbons1983), which provide some convergent psychometric properties (Leventhal, Chasson, Tapia, Miller, & Pettit, Reference Leventhal, Chasson, Tapia, Miller and Pettit2006). Anticipatory and consummatory aspects of anhedonia are distinguished from one another using the Temporal Experience of Pleasure Scale (Gard, Gard, Kring, & John, Reference Gard, Gard, Kring and John2006). Examples of common laboratory-based indices of anhedonia include self-report, psychophysiological, and hemodynamic responses to tasting sucrose and chocolate (e.g., Berlin, Givry-Steiner, Lecrubier, & Puech, Reference Berlin, Givry-Steiner, Lecrubier and Puech1998; Chentsova-Dutton & Hanley, Reference Chentsova-Dutton and Hanley2010), and to hearing pleasant musical stimuli (e.g., Keller et al., Reference Keller, Young, Kelley, Prater, Levitin and Menon2013), as well as behavioral responses, such as developing a response bias toward more frequently rewarding stimuli (Pizzagalli, Jahn, & O'Shea, Reference Pizzagalli, Jahn and O'Shea2005).

As a general rule across self-report and laboratory-based measures, high anhedonia scores capture low positive affectivity, a collection of emotions related to low levels of joy, energy, excitement, and confidence (Watson, Reference Watson, Snyder and Lopez2002). This contrasts with the personality characteristic of negativity affectivity, which manifests as a general tendency to experience distress, nervousness, and hostility, which may or may not co-occur with low positive affectivity (Watson & Tellegen, Reference Watson and Tellegen1985).Footnote ⁵ Here, we operationalize trait anhedonia as deficits in incentive processing prior to the experience of reward (i.e., low interest and motivation to expend energy to receive rewards; diminished anticipatory responses to rewards). However, we recognize that the term anhedonia conventionally encapsulates deficiencies relating to both obtaining and directly experiencing rewarding stimuli, and that one or both of these functions can be affected within individuals.

DA neural circuitry

DA and reward processing

Reward processing encompasses both endogenous and exogenous responses to reinforcers, including neural responding, and how reward contingencies are used to predict, assess, and guide goal-directed behavior. The role of the mesolimbic DA system in reward processing was first studied in the mid-1950s by Olds and Milner, who demonstrated that rats implanted with stimulating electrodes in regions of this system, including the VTA and striatum, habitually self-stimulate, sometimes to the point of death (Olds & Milner, Reference Olds and Milner1954; see Milner, Reference Milner1991). Later animal studies, some of which we describe below, expanded this work, and identified midbrain DA function as fundamental in processing incentives, and in guiding future, goal-directed behaviors. These findings and others that followed provide the basis for conceptualizing the mesolimbic DA system as the brain's primary reward circuit and provide a compelling neurobiological explanation for the relationship between reward and motivation.

Incentive processing includes both anticipatory and consummatory phases of reward responding. During reward anticipation, an organism is motivated to approach potential reinforcers. During reward consummation, the organism experiences some degree of pleasure/satisfaction from receiving reinforcement.Footnote ⁶ To maximize access to reinforcers, organisms must learn to anticipate them accurately based on potentially dynamic and ambiguous reward contingencies, a process known as reward learning. The distinction between reward anticipation and receipt has received considerable support from animal and neuroimaging work (discussed below; see Berridge & Kringelbach, Reference Berridge and Kringelbach2015; Berridge & Robinson, Reference Berridge and Robinson1998; O'Doherty, Reference O'Doherty2004), and emerges in multiples lines of research. For example, separating reward anticipation from receipt parallels the concepts of “wanting” (motivation to pursue rewards) versus “liking” (satisfaction from receiving rewards) described by Robinson and Berridge in their highly influential incentive salience theory of addiction (Reference Robinson and Berridge1993, Reference Robinson and Berridge2008; Berridge & Kringelbach, Reference Berridge and Kringelbach2015). According to incentive salience theory, wanting (anticipating) is mediated by the NAcc core among other regions, whereas liking is mediated by neural processes in the NAcc shell (e.g., Berridge & Kringelbach, Reference Berridge and Kringelbach2015; Berridge & Robinson, Reference Berridge and Robinson2003; Saddoris, Cacciapaglia, Wightman, & Carelli, Reference Saddoris, Cacciapaglia, Wightman and Carelli2015).

The computational task of accurately anticipating reinforcers is accomplished in part through appropriate tonic (low-amplitude, baseline) and phasic (high-amplitude, event-related) activity of midbrain DA neurons (Schultz, Reference Schultz1998). For example, unexpected rewards elicit burst firing of DA neurons that project to the NAcc in the VS (Tremblay, Hollerman, & Schultz, Reference Tremblay, Hollerman and Schultz1998). The magnitude and time course of this spike in DA provides a temporal window for an organism to learn which behaviors increase the likelihood of receiving rewards, resulting in reinforcement of these behaviors (i.e., motivated responses; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Once the organism can predict, based on cues, the probability and magnitude of an impending reward, phasic DA reactivity shifts forward from occurring during consummation of the reinforcer to the temporal interval between cue and reward (Bressan & Crippa, Reference Bressan and Crippa2005). However, phasic DA responding continues to occur during receipt of rewards that are unexpected, or are larger in magnitude than expected. In contrast, phasic DA responding dampens following receipt of rewards that are worse than expected, or expected but absent (Bressan & Crippa, Reference Bressan and Crippa2005; Schultz, Reference Schultz1998). During reward anticipation, midbrain DA neurons encode relative value of expected rewards (magnitude of the reward weighted by probability and delay), such that large, immediate, and highly probable gains elicit the most robust neural firing (Kobayashi & Schultz, Reference Kobayashi and Schultz2008; Tobler, Fiorillo, & Schultz, Reference Tobler, Fiorillo and Schultz2005). The VS, which receives direct input from midbrain DA neurons, and from other structures implicated in reward processing, including the amygdala, insula, and orbitofrontal cortex (OFC), also indexes the relative value of predicted rewards (Knutson, Adams, Fong, & Hommer, Reference Knutson, Adams, Fong and Hommer2001), whereas reduced activation signals reward devaluation (Gottfried, O'Doherty, & Dolan, Reference Gottfried, O'Doherty and Dolan2003). Thus, DA-mediated activity of the VS and other structures is integral to associative learning, and associative learning is reflected in the ability to anticipate future rewards.

Phasic changes in mesolimbic neural responding have been assessed directly in animals (e.g., Tremblay et al., Reference Tremblay, Hollerman and Schultz1998) and indirectly through functional neuroimaging with humans. One type of functional magnetic resonance imaging task commonly used with humans is the monetary incentive delay task, adapted from tasks used in primate research to study reward processing (Knutson et al., Reference Knutson, Adams, Fong and Hommer2001; Schultz, Dayan, & Montague, Reference Schultz, Dayan and Montague1997). Parameters of the monetary incentive delay task vary across studies, but the basic framework is as follows: for each trial, participants are presented with an incentive cue that signals the possibility of winning money, avoiding losing money, or no change during that trial. Participants are then challenged to respond to a target quickly, and their performance determines the monetary outcome for that trial. This paradigm is advantageous because it separates reward anticipation from reward receipt, and has been used widely to investigate neural correlates of reward processing in clinical and nonclinical samples (e.g., Forbes et al., Reference Forbes, May, Siegle, Ladouceur, Ryan and Carter2006; Knutson, Bhanji, Cooney, Atlas, & Gotlib, Reference Knutson, Bhanji, Cooney, Atlas and Gotlib2008; Lutz & Widmer, Reference Lutz and Widmer2014; Sauder, Derbidge, & Beauchaine, Reference Sauder, Derbidge and Beauchaine2016).

Although phasic DA responding has received considerable attention in research on reward processing, tonic levels of DA in the mesolimbic system also play a role in evaluating and responding to incentives (Goto, Otani, & Grace, Reference Goto, Otani and Grace2007; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Tonic levels of DA, as assessed through DA concentrations in extrasynaptic space, are normally insufficient to stimulate postsynaptic DA receptors (Grace, Reference Grace, Solanto, Arnsten and Castellanos2001). However, tonic DA modulates phasic DA activity (Bressan & Crippa, Reference Bressan and Crippa2005). For example, tonic DA can bind to DA terminal autoreceptors and other receptor sites in the extrasynaptic space, inhibiting phasic release of DA (Grace & Bunney, Reference Grace, Bunney, Borroni and Kupfer1995). Through such processes, excessively high or low tonic DA can compromise reward learning by altering the relative strength of phasic signals to reinforcers (see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005; Schultz, Reference Schultz1998).

As alluded to above, incentive processing is related closely to motivation, defined as the degree to which an organism will exert energy in goal-directed behaviors. Like reward processing, motivation derives in part from tonic and phasic DA in the VS, and communication among structures in the mesolimbic DA circuit, including the VS (see Niv, Reference Niv2007). For example, chronically elevated extracellular DA is associated with enhanced motivation (Cagniard, Balsam, Brunner, & Zhuang, Reference Cagniard, Balsam, Brunner and Zhuang2006), whereas DA depletion and DA receptor blockage in the VS are associated with reduced motivation (Salamone, Cousins, & Bucher, Reference Salamone, Cousins and Bucher1994). Thus, intact ability to neurally encode the relative value of rewards and anticipate future rewards is important for stimulating motivated behaviors in order to obtain rewards that are “worth” the effort.

Several key takeaways from research on incentive processing are worth noting before we move forward in our discussion: (a) effective reward processing is contingent on tonic levels of DA, and on phasic modulation of DA in mesolimbic structures; (b) reward anticipation and reward consummation are dissociable, and reward anticipation may be compromised in organisms that possess intact consummatory responses; and (c) the ability to appraise reward value and anticipate future rewards influences motivation to engage in goal-directed behaviors.

Animal studies

Converging evidence from comparative studies suggests that impulsive behavior is associated with deficiencies in mesolimbic DA function (see Gatzke-Kopp, Reference Gatzke-Kopp2011; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Selective lesioning of the NAcc core, a DA-rich area of the VS, increases impulsive choice in rats (Cardinal, Pennicott, Lakmali, Robbins, & Everitt, Reference Cardinal, Pennicott, Lakmali, Robbins and Everitt2001). Similarly, overexpression of the DA transporter in the NAcc, which results in reduced levels of extracellular DA in this region, is associated with both increased impulsive choice (greater preference for immediate, smaller rewards over larger, delayed rewards), and risk proneness (greater preference for large, uncertain rewards over small but likely rewards; Adriani et al., Reference Adriani, Boyer, Gioiosa, Macri, Dreyer and Laviola2009).

The role of DA neurocircuitry in impulsivity is especially evident in animal studies that select and compare groups based on (a) differences in inherent VS responding and (b) baseline differences in impulsive behaviors. Both impulsive choice and impulsive action (failure to inhibit behavior when it is advantageous to do so) are associated with abnormally low DA release in the VS in response to electrical stimulation (Diergaarde et al., Reference Diergaarde, Pattij, Poortvliet, Hogenboom, de Vries and Schoffelmeer2008). Among laboratory animals that are highly impulsive, as assessed by poor behavioral inhibition on the stop-signal task, DA agonism shortens reaction time to inhibit action and thereby improves impulse control (Eagle, Tufft, Goodchild, & Robbins, Reference Eagle, Tufft, Goodchild and Robbins2007; Feola, de Wit, & Richards, Reference Feola, de Wit and Richards2000). In contrast, among animals with fast reaction times at baseline (high impulse control), studies by Eagle et al. and Feola et al. demonstrate that DA agonism either exhibits no effect on impulse control or lengthens reaction time.

Impulsive animals also exhibit an increased proclivity to engage in drug use. For example, inherently low D2/3 receptor availability in the NAcc is associated prospectively with higher rates of cocaine self-administration (Dalley et al., Reference Dalley, Fryer, Brichard, Robinson, Theobald and Lääne2007). Combined, such findings suggest low tonic and phasic DA release are associated with impulse control problems.

It is important to note that relations between DA and impulsivity may be obscured entirely when inherent individual differences in mesolimbic function are not considered, because these individual differences modulate effects of DA-acting drugs. Studies that do not consider individual differences in baseline impulsivity overwhelmingly associate DA agonism with increases in impulsive action, which is in direct contrast with well-replicated findings linking low DA to impulsivity among highly impulsive animals (see D'Amour-Horvat & Leyron, Reference D'Amour-Horvat and Leyton2014). In addition to individual differences in DA neurocircuitry function, other factors relating to dosing and drug administration schedules should be considered when conducting and interpreting studies concerning pharmacological manipulation of DA in animals.Footnote ⁷

It is important to note that there is still uncertainty regarding the precise nature and causes of DA deficiency associated with trait impulsivity. A number of possible mechanisms exist, including abnormalities attributed to DA release, density, and function of the DA transporter and/or DA receptors, and functional connectivity between regions among and outside central DA systems, as well as effects of particular stimulus conditions. In all likelihood, more than one such factor drives behavioral effects, and factors underlying impulsive tendencies may differ among individuals (see Beauchaine et al., in press; Dalley, Mar, Economidou, & Robbins, Reference Dalley, Mar, Economidou and Robbins2008). Molecular genetic accounts suggest that trait impulsivity and its various behavioral manifestations across development are inherited multifactorially. Because many genes are involved, which interact with one another and with environment to affect behavior, no single genetic locus is necessary or sufficient to confer clinically expressed impulsivity (e.g., Gizer, Otto, & Ellingson, Reference Gizer, Otto, Ellingson, Beauchaine and Hinshaw2016).

Neuroimaging research with humans

Findings from neuroimaging studies among humans with ADHD and other externalizing spectrum disorders who exhibit clinically significant impulsivity parallel and in some cases clarify findings from animal studies. Single photon emission computed tomography, positron emission tomography (PET), and magnetic resonanc imaging studies implicate compromised mesolimbic and/or mesocortical DA function in ADHD (e.g., Ludolph et al., Reference Ludolph, Kassubek, Schmeck, Glaser, Wunderlich and Buck2008; Vles, Feron, & Hendrikson, Reference Vles, Feron and Hendriksen2003; Volkow et al., Reference Volkow, Wang, Newcorn, Fowler, Telang and Solanto2007, Reference Volkow, Wang, Kollins, Wigal, Newcorn and Telang2009), CD (e.g., Rubia, Smith, et al., Reference Rubia, Smith, Halari, Matukura, Mohammad and Taylor2009), SUDs (e.g., Martin-Soelch et al., Reference Martin-Soelch, Leenders, Chevalley, Missimer, Kunig and Magyar2001; Volkow, Fowler, & Wang, Reference Volkow, Fowler and Wang2004), and antisocial traits (e.g., Oberlin et al., Reference Oberlin, Dzemidzic, Bragulat, Lehigh, Talavage and O'Connor2012). Children with ADHD exhibit smaller VS volumes, which correspond with severity of hyperactive/impulsive symptoms (Carmona et al., Reference Carmona, Proal, Hoekzema, Gispert, Picado and Moreno2009; Sauder et al., Reference Sauder, Beauchaine, Gatzke-Kopp, Shannon and Aylward2012). Individuals with ADHD also demonstrate reduced striatal responding to smaller, immediate rewards and larger, delayed rewards in functional magnetic resonance imaging (fMRI) studies (e.g., Plichta et al., Reference Plichta, Vasic, Wolf, Lesch, Brummer and Jacob2009).Footnote ⁸ Furthermore, those with ADHD (e.g., Carmona et al., Reference Carmona, Hoekzema, Ramos-Quiroga, Richarte, Canals and Bosch2012; Furukawa et al., Reference Furukawa, Bado, Tripp, Mattos, Wickens and Bramati2014; Hoogman et al., Reference Hoogman, Aarts, Zwiers, Slaats-Willemse, Naber and Onnink2011; Scheres et al., Reference Scheres, Milham, Knutson and Castellanos2007; Ströhle et al., Reference Ströhle, Stoy, Wrase, Schwarzer, Schlagenhauf and Huss2008), early-onset criminal offending (Cohn et al., Reference Cohn, Veltman, Pape, van Lith, Vermeiren and van den Brink2015), and SUDs (Beck et al., Reference Beck, Schlagenhauf, Wustenberg, Hein, Kienast and Kahnt2009; Wrase et al., Reference Wrase, Schlagenhauf, Kienast, Wüstenberg, Bermpohl and Kahnt2007) all exhibit VS hypoactivation during reward anticipation and/or reward outcome compared with controls. These findings are consistent with hypofunctionality of DA circuitry observed in impulsive animals.

No clear evidence has emerged from animal research to describe the relationship between phasic DA responses to immediate rewards and impulsivity (Luman, Tripp, & Scheres, Reference Luman, Tripp and Scheres2010), and competing theories suggest that such responses are suppressed (e.g., Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005) or unaffected (e.g., Tripp & Wickens, Reference Tripp and Wickens2008). In some neuroimaging research, ADHD is associated with similar or perhaps even greater VS responses to reward outcomes (as opposed to reward anticipation), compared with controls (Furukawa et al., Reference Furukawa, Bado, Tripp, Mattos, Wickens and Bramati2014; Scheres et al., Reference Scheres, Milham, Knutson and Castellanos2007; Wilbertz et al., Reference Wilbertz, van Elst, Delgado, Maier, Feige and Philipsen2012). This is consistent with theories that link timing and duration over magnitude of DA responding to ADHD (see also Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005).

Some of the strongest neuroimaging evidence for deficient mesolimbic responding as a neural substrate of impulsivity derives from studies that assess neurobiological consequences of DA agonists among those with ADHD. DA agonists, including methylphenidate, increase extracellular DA levels in the striatal pathway (e.g., Volkow, Fowler, Wang, Ding, & Gatley, Reference Volkow, Fowler, Wang, Ding and Gatley2002). Among both children and adults with ADHD, methylphenidate normalizes (a) attenuated striatal responses to reward (Vles et al., Reference Vles, Feron and Hendriksen2003; Volkow et al., Reference Volkow, Fowler, Wang, Ding and Gatley2002), (b) frontocingulate underreactivity during error processing (Rubia, Halari, Mohammad, Taylor, & Brammer, Reference Rubia, Halari, Mohammad, Taylor and Brammer2011), and (c) functional connectivity deficits between mesolimbic and mesocortical brain regions (Rubia, Halari, Cubillo, Mohammad, & Taylor, Reference Rubia, Halari, Cubillo, Mohammad and Taylor2009). Thus, DA agonists, which increase availability of DA in the striatum, decrease impulsive behaviors by increasing DA availability to levels experienced by individuals without these symptoms (e.g., Hinshaw, Henker, Whalen, Erhardt, & Dunnington, Reference Hinshaw, Henker, Whalen, Erhardt and Dunnington1989; MTA Cooperative Group, 1999; see Beauchaine & McNulty, Reference Beauchaine and McNulty2013).

Translational models of externalizing behavior

As described above, the strength of a given reinforcer to motivate future behaviors is dependent on several factors, including the magnitude and predictability of the reinforcer and the interval between rewarded behavior and receipt of the reinforcer (see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). These factors are coded by phasic neural firing in the VS and interconnected brain regions, which signal relative values of expected reinforcers. Associative learning is affected adversely when tonic DA firing is suppressed (Tripp & Wickens, Reference Tripp and Wickens2008). Under such conditions, an organism is less able to predict and assign value to reinforcers, because the temporal window within which reward learning can occur is narrowed. As a result, larger and more frequent incentives are required to form behavior–reward associations, and to predict future reward outcomes (Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). It should be noted that more frequent trials of nonreward may also be required for extinction of learned behavior–reward associations.

It follows from patterns of neural activity and reactivity described above that impulsivity emerges as a preference for immediate over delayed incentives given a general failure to associate behaviors with rewards that are temporally distant from one another (see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Thus, immediate but smaller rewards are appealing compared to larger delayed rewards, because immediate choices are more easily associated with reward and therefore reinforced. Impulsive action may reflect slower learning of which behaviors to select in order to progress toward one's goals (i.e., maximize reward yields). For example, the go/no-go task requires the organism to respond to presentation of “go” cues and withhold responses following “no-go” cues. Failure to inhibit “no-go” responses may occur when responses for “go” cues are sufficiently reinforced, but extinction processes for erroneous, ineffective, or insufficient behavior are impaired. This leads to less differential responding between cue types. Consequently, an organism may exhibit generalized responding to cues regardless of type, and fail to withhold responses appropriately (see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). This is consistent with neuroimaging research demonstrating a failure to migrate neural activity from mesolimbic structures to the anterior cingulate when previously rewarded behaviors are extinguished among boys with ADHD and/or CD (Gatzke-Kopp et al., Reference Gatzke-Kopp, Beauchaine, Shannon, Chipman-Chacon, Fleming and Crowell2009).

Contemporary ontogenic process models assert that heritable trait impulsivity is a transdiagnostic vulnerability to all externalizing spectrum disorders (Beauchaine et al., Reference Beauchaine, Hinshaw and Pang2010, in press; Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Beauchaine, Shader, & Hinshaw, Reference Beauchaine, Shader, Hinshaw, Beauchaine and Hinshaw2016). As noted above, children, adolescents, and adults with ADHD, which often begins a developmental trajectory to more severe externalizing conduct, exhibit diminished DA reactivity to incentives (see Plichta & Scheres, Reference Plichta and Scheres2014; Rubia, Reference Rubia2011). This hypofunctionality may emerge at least in part through low tonic DA (e.g., Vles et al., Reference Vles, Feron and Hendriksen2003; Volkow et al., Reference Volkow, Wang, Newcorn, Fowler, Telang and Solanto2007; see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Although impulsive, reward-seeking behaviors upregulate DA levels temporarily, with associated increases in positive affect, they do so only transiently, and do not address underlying DA deficiency (Gatzke-Kopp & Beauchaine, Reference Gatzke-Kopp, Beauchaine, Coch, Dawson and Fischer2007; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Thus, low tonic DA, combined with low phasic responding to reward cues, may lead to a disproportionately high preference for immediate over delayed rewards and diminished ability to act with foresight.

In addition to providing a neurobiological explanation for impulse control problems, the DA deficiency model of impulsivity may also account for mechanisms that underlie responses to certain interventions. For example, raising intrinsically low levels of tonic DA via administration of DA agonists improves associative learning among those with ADHD by extending the interval during which behavior–reward contingencies are encoded (see Gatzke-Kopp & Beauchaine, Reference Gatzke-Kopp, Beauchaine, Coch, Dawson and Fischer2007). DA agonists may also reduce novelty- and sensation-seeking behaviors by diminishing an organism's dependence on immediately available rewarding stimuli to increase striatal activation (Arnsten, Reference Arnsten2006). Other neural processes, such as error detection, may also improve when reinforcement contingencies are presented sufficiently close together to accommodate a shorter window for associative learning (see Luman, Oosterlan, & Sergeant, Reference Luman, Oosterlaan and Sergeant2005). For example, providing immediate, performance-based rewards is associated with improved task accuracy and increased neural responding to error and task-relevant stimuli among children with ADHD (Rosch & Hawk, Reference Rosch and Hawk2013). It is also a core component of effective treatments for more severe conduct problems (e.g., Ialongo, Poduska, Werthamer, & Kellam, Reference Ialongo, Poduska, Werthamer and Kellam2001).

Interim conclusions

Considerable evidence from animal research and neuroimaging studies with humans suggests that for many individuals, trait impulsivity derives from chronically low tonic DA and diminished DA reactivity to incentive cues in the VS and other mesolimbic structures (see Gatzke-Kopp, Reference Gatzke-Kopp2011). As a result, trait impulsive individuals exhibit deficits in associative learning and a preference for immediate but small rewards over larger delayed rewards. This preference for immediacy also manifests in reduced motivation to engage in tasks that require sustained effort, and tasks that are not intrinsically motivating (see Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Raising DA to normal levels improves impulse control by facilitating associative reward learning, increasing positive affect (see Ashby, Isen, & Turken, Reference Ashby, Isen and Turken1999), and reducing reliance on external reinforcers that are immediately available.

Animal studies

Because animals cannot self-report state or trait anhedonia, it must be inferred from behavior under carefully controlled experimental conditions. For example, anhedonia is sometimes operationalized as aberrantly low likelihood of seeking pleasurable stimuli despite intact motor function. Animal models suggest that DA is neither necessary nor sufficient for experiencing pleasure from consuming appetitive stimuli (i.e., liking), but may be involved critically in motivational aspects of reward processing (i.e., wanting). Manipulating mesolimbic DA through pharmacological agonism/antagonism, electrical stimulation, and lesions induced by neurotoxins, such as 6-hydroxydopamine, alters an animal's propensity to expend energy to gain rewards, but does not influence hedonic effects of rewards, such as food palatability (see Berridge & Robinson, Reference Berridge and Robinson1998). For example, mice that are engineered genetically to have synaptic DA levels that are 70% higher than normal expend more energy to obtain sweet rewards, but are no more likely to derive pleasure from sweets themselves, as assessed through orofacial reactions (Peciña, Cagniard, Berridge, Aldridge, & Zhuang, Reference Peciña, Cagniard, Berridge, Aldridge and Zhuang2003). Similarly, mice that are engineered genetically to be incapable of producing DA exhibit a preference for sucrose and saccharin over water, just as control mice, but initiate licking behavior for the sweet-tasting substances less frequently (Cannon & Palmiter, Reference Cannon and Palmiter2003). The authors suggest that DA is not required to develop a taste preference for sweets, but may be involved in goal-directed behavior to obtain sweets. This parallels findings among depressed humans, who experience hedonic effects of sweet tastes similarly to healthy controls (Dichter, Smoski, Kampov-Polevoy, Gallop, & Garbutt, Reference Dichter, Smoski, Kampov-Polevoy, Gallop and Garbutt2010).

Animals with deficient mesolimbic DA levels also exhibit a preference for low-cost/low rewards when high-cost/high rewards are available, which is sometimes considered to be a behavioral indicator of anhedonia (see Salamone, Correa, Farrar, & Mingote, Reference Salamone, Correa, Farrar and Mingote2007). This preference is maintained when reward delays are made equivalent across high- and low-cost reward trials (Floresco, Maric, & Ghods-Sharif, Reference Floresco, Maric and Ghods-Sharifi2008), and disappears when low and high rewards require equal energy expenditure (Denk et al., Reference Denk, Walton, Jennings, Sharp, Rushworth and Bannerman2005). Thus, animals understand reward contingencies yet choose options that require less effort. Correa, Carlson, Wisniecki, and Salamone (Reference Correa, Carlson, Wisniecki and Salamone2002) also reported that animals with selective DA lesions in the NAcc are just as likely as controls to exert effort for rewards that require one lever press, but respond less frequently when rewards require five lever presses. Such findings implicate low DA in diminished motivation, or reduced goal-directed behaviors to receive reward.

Neuroimaging research with humans

Anhedonia is often viewed as a core component of depression and schizophrenia, and like impulsivity, is expressed along a continuum in the population (Keller et al., Reference Keller, Young, Kelley, Prater, Levitin and Menon2013). Nonpathological anhedonia involves both reduced ability to anticipate and adapt one's behavior to obtain incentives and reduced motivation to work for incentives. For example, Pizzagalli et al. (Reference Pizzagalli, Jahn and O'Shea2005) reported that undergraduates with high self-reported anhedonia failed to show increases in response bias for rewards, whereas participants with low self-reported anhedonia exhibited this response bias. Thus, the high anhedonia group failed to adjust their behavior based on prior rewarded experiences, suggesting deficient ability to anticipate rewards to maximize gains. Pizzagalli et al. propose that this behavioral pattern may reflect the anhedonic phenotype of depression. Undergraduate students with higher self-reported anhedonia are also less willing to expend effort to receive rewards in laboratory-based tasks that are designed to evaluate reward motivation (Treadway, Buckholtz, Schwartzman, Lambert, & Zald, Reference Treadway, Buckholtz, Schwartzman, Lambert and Zald2009).

To our knowledge, only two studies have assessed the relationship between anhedonia and neural responding to incentives in nonclinical, non–high-risk samples. Both studies used versions of the monetary incentive delay task. The first study, conducted by Wacker, Dillon, and Pizzagalli (Reference Wacker, Dillon and Pizzagalli2009), included a predominately nonclinical sample (N = 28). Reduced NAcc volumes and diminished activation in the NAcc to monetary incentives were both associated with anhedonia (as assessed by the anhedonic depression subscale of the Mood and Anxiety Symptom Questionnaire; Watson et al., Reference Watson, Weber, Assenheimer, Clark, Strauss and McCormick1995), but not with other symptoms of depression and anxiety. The authors failed to find compromised activation to reward cues, which is more consistent with the anhedonia literature. However, their paradigm allowed for monetary gains on 50% of reward trials. This reward schedule is lower than that used in similar tasks from other studies (e.g., ~66% of trials were rewarded in Knutson et al., Reference Knutson, Adams, Fong and Hommer2001). Thus, participants were less able to predict reward outcomes, which, as described above, results in elicitation of greater phasic NAcc firing during reward receipt, but not during presentation of reward cues. Wacker et al. (Reference Wacker, Dillon and Pizzagalli2009) acknowledge that this difference in methods complicates interpretation of their findings. It is therefore unclear whether anhedonia involves dysfunctional neural responding to anticipatory, consummatory, or both phases of reward processing.

The second study was performed by Eisenberger et al. (Reference Eisenberger, Berkman, Inagaki, Rameson, Mashal and Irwin2010), who, unlike Wacker et al. (Reference Wacker, Dillon and Pizzagalli2009), excluded participants with a current DSM-IV Axis I diagnosis. Participants (N = 39) were randomly assigned to placebo or a low-dose injection of the endotoxin Escherichia coli. Endotoxin was administered to induce an inflammatory response, because inflammation is implicated in onset and maintenance of depression (Miller, Maletic, & Raison, Reference Miller, Maletic and Raison2009). Compared with placebo, endotoxin exposure was associated with both blunted VS activation to reward cues (but not reward receipt) and increased self-reported and observer-assessed depressed mood.Footnote ⁹ Group differences in VS activation mediated the relationship between endotoxin exposure and increases in observer-rated depressed mood. Although the purpose of this study was to investigate the role of inflammation in depression, findings support a functional relationship between blunted VS responses to reward cues and depressive symptoms.

Two other neuroimaging studies have investigated anhedonia among healthy adults without past or current mental illness, but neither included functional tasks that assessed reward responding to incentives. The first study, conducted by Harvey, Pruessner, Czechowska, and Lepage (Reference Harvey, Pruessner, Czechowska and Lepage2007), required participants to view images of positive, negative, and neutral valences. Among other findings, higher anhedonia scores, as measured by the Chapman Revised Physical Anhedonia Scale, were associated positively with activation in the right ventromedial prefrontal cortex (vmPFC), and negatively with anterior caudate volumes, which extended to the VS in a less stringent follow-up analysis. Although the researchers' hypotheses of enhanced vmPFC activation and reduced striatal volumes were supported by their findings, they did not identify reduced VS activation to pleasant stimuli as predicted.

The second study, conducted by Keller et al. (Reference Keller, Young, Kelley, Prater, Levitin and Menon2013), investigated the relationship between trait anhedonia (measured using the positive affect factor of the Mood and Anxiety Symptom Questionnaire—Short Form) and neural responding to musical stimuli. Trait anhedonia was associated with reduced pleasantness ratings of stimuli, and with hypoactivation of the right NAcc, basal forebrain, bilateral hypothalamus, and cortical regions, including orbitofrontal, anterior, and posterior cingulate, and ventromedial prefrontal cortices. Trait anhedonia correlated negatively with effective connectivity between mesolimbic and paralimbic regions, including the bilateral insula and the OFC.

Anhedonia in MDD

As noted above, anhedonia characterizes several psychiatric disorders, including major depression, schizophrenia, and Parkinson disease (Winograd-Gurvich, Fitzgerald, Georgiou-Karistianis, Bradshaw, & White, Reference Winograd-Gurvich, Fitzgerald, Georgiou-Karistianis, Bradshaw and White2006). Although anhedonia is not required for a DSM-5 diagnosis of major depression, anhedonia and depressed mood are the two core diagnostic features of the disorder (APA, 2013). Reported estimates of clinically significant anhedonia among adults with major depression vary widely from 37% to 92% (Buckner, Joiner, Pettit, Lewinsohn, & Schmidt, Reference Buckner, Joiner, Pettit, Lewinsohn and Schmidt2008; Lemke, Puhl, Koethe, & Winkler, Reference Lemke, Puhl, Koethe and Winkler1999; Pelizza & Ferrari, Reference Pelizza and Ferrari2009). Similarly, estimates of anhedonia among depressed children and adolescents range from 10% to 57% (Luby, Mrakotsky, Heffelfinger, Brown, & Spitznagel, Reference Luby, Mrakotsky, Heffelfinger, Brown and Spitznagel2004; Yorbik, Birmaher, Axelson, Williamson, & Ryan, Reference Yorbik, Birmaher, Axelson, Williamson and Ryan2004). Such large discrepancies may be attributable to use of different measures of anhedonia, different thresholds for determining clinical significance, and differences in symptom presentation and severity among study samples. Regardless, these findings indicate that anhedonia is a common but not ubiquitous feature of depression.

Because anhedonia is so common to major depression, disentangling state-dependent effects of mental illness from trait anhedonia in psychiatric populations remains a challenge. Nevertheless, mesolimbic DA dysfunction in depression is reported widely (e.g., Lambert, Johansson, Ågren, & Friberg, Reference Lambert, Johansson, Ågren and Friberg2000; Meyer et al., Reference Meyer, Krüger, Wilson, Christensen, Goulding and Schaffer2001). Striatal hypoactivation to positively valenced stimuli is a well-replicated finding among depressed adolescent and adult samples (e.g., see Forbes & Dahl, Reference Forbes and Dahl2011; Forbes et al., Reference Forbes, May, Siegle, Ladouceur, Ryan and Carter2006, Reference Forbes, Hariri, Martin, Silk, Moyles and Fisher2009; Groenewold, Opmeer, de Jonge, Aleman, & Costafreda, Reference Groenewold, Opmeer, de Jonge, Aleman and Costafreda2013; Morgan, Olino, McMakin, Ryan, & Forbes, Reference Morgan, Olino, McMakin, Ryan and Forbes2013). For responses to incentives more specifically, current depression is associated with failure to develop a response bias toward rewarding stimuli (e.g., Henriques, Glowacki, & Davidson, Reference Henriques, Glowacki and Davidson1994; Pizzagalli, Iosifescu, Hallett, Ratner, & Fava, Reference Pizzagalli, Iosifescu, Hallett, Ratner and Fava2008), impaired reward-based reversal learning (Robinson, Cools, Carlisi, Sahakian, & Drevets, Reference Robinson, Cools, Carlisi, Sahakian and Drevets2012), and blunted VS responses to anticipation and receipt of monetary incentives (e.g., Forbes et al., Reference Forbes, Hariri, Martin, Silk, Moyles and Fisher2009; Pizzagalli et al., Reference Pizzagalli, Holmes, Dillon, Goetz, Birk and Bogdan2009; Smoski et al., Reference Smoski, Felder, Bizzell, Green, Ernst and Lynch2009). Pizzagalli et al. (Reference Pizzagalli, Holmes, Dillon, Goetz, Birk and Bogdan2009) also demonstrated that anhedonic symptoms and depression severity correlate negatively with smaller caudate volumes. The one exception to this body of literature is a study by Knutson et al. (Reference Knutson, Bhanji, Cooney, Atlas and Gotlib2008), who found that NAcc activation did not differ between depressed participants and controls during reward anticipation or receipt. The authors suggest that their sample of depressed adults may not have exhibited VS hyporeactivity during reward anticipation as they predicted specifically because anhedonic features were not sufficiently prominent.

The null finding by Knutson et al. (Reference Knutson, Bhanji, Cooney, Atlas and Gotlib2008) should not be considered anomalous. As presented above, anhedonia in depression is common, but not universal. A large body of research indicates that depression emerges from a multitude of equifinal pathways involving genetic-, neural-, neuroendocrine-, and trait-level vulnerabilities that interact with environmental risk factors, including adversity in childhood, ongoing stressors, and social instability (see Charney & Manji, Reference Charney and Manji2004; Kendler, Gardner, & Prescott, Reference Kendler, Gardner and Prescott2002; Kim, Reference Kim2008; Klein, Kotov, & Bufferd, Reference Klein, Kotov and Bufferd2011). Trait anhedonia may be a characteristic feature of a subpopulation of individuals with major depression, which may account for why only a subset of depressed individuals exhibit mesolimbic DA dysfunction when interindividual variability in reward responding is considered (Foti, Carlson, Sauder, & Proudfit, Reference Foti, Carlson, Sauder and Proudfit2014). Accordingly, greater attention to the extent to which individual differences in reward processing predict alternative phenotypic presentations of illness and treatment response may be especially valuable (see Forbes et al., Reference Forbes, Hariri, Martin, Silk, Moyles and Fisher2009; e.g., Morgan et al., Reference Morgan, Olino, McMakin, Ryan and Forbes2013). The importance of identifying clinical subgroups on the basis of etiological vulnerabilities and not collections of behavioral symptoms is revisited multiple times in this article.

Anhedonia in populations at high risk for major depression

Anhedonia may be present prior to and/or after major depressive episodes, and thus may be best characterized as a traitlike vulnerability (McCabe, Reference McCabe and Ritsner2014). Self-reports of anhedonia predict the onset of major depression 1 year later (Dryman & Eaton, Reference Dryman and Eaton1991), and undergraduate students and individuals with histories of depression report individual differences in hedonic capacity that are stable across long periods of time (Clark, Fawcett, Salazar-Grueso, & Fawcett, Reference Clark, Fawcett, Salazar-Grueso and Fawcett1984; Shankman et al., Reference Shankman, Nelson, Harrow and Faull2010; Watson & Walker, Reference Watson and Walker1996). Anhedonia is also experienced both during and between major depressive episodes (Schrader, Reference Schrader1997). Despite such findings, and a long-established theoretical argument for anhedonia as a trait vulnerability to depression, very few laboratory-based studies have investigated prospective links between anhedonia and depression, and studies of anhedonia in cases of remitted depression cannot rule out scarring effects.

Anhedonia, self-inflicted injury (SII), and borderline personality disorder

As summarized above, both anhedonia, assessed via self-report, and blunted VS responding to incentives, assessed through neuroimaging, are exhibited by vulnerable and currently depressed samples. Another subpopulation of interest for the present discussion is individuals with a history of SII, including those who have attempted suicide and/or engage in nonsuicidal self-harm (see Crowell et al., Reference Crowell, Beauchaine, McCauley, Smith, Vasilev and Stevens2008; Kaufman, Crowell, & Stepp, Reference Kaufman, Crowell, Stepp, Beauchaine and Hinshaw2016; Nock, Reference Nock2010). Individuals who engage in these behaviors are both trait impulsive and anhedonic, and lie at the nexus of internalizing and externalizing psychopathology (see Beauchaine, Klein, Crowell, Derbridge, & Gatzke-Kopp, Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Crowell, Derbidge, & Beauchaine, Reference Crowell, Derbidge, Beauchaine and Nock2014; Crowell, Kaufman, & Beauchaine, Reference Crowell, Kaufman, Beauchaine, Tackett and Sharp2014). Anhedonia predicts suicidal ideation (Winer et al., Reference Winer, Nadorff, Ellis, Allen, Herrera and Salem2014) and violence toward self and others (Fawcett et al., Reference Fawcett, Scheftner, Fogg, Clark, Young and Hedeker1990; Nordström, Schalling, & Asberg, Reference Nordström, Schalling and Asberg1995; Sadeh, Javdani, Finy, & Verona, Reference Sadeh, Javdani, Finy and Verona2011). Furthermore, as with other groups characterized by anhedonia, those who have attempted suicide exhibit deficits in reward processing, including impaired reward/punishment-based reversal learning (Dombrovski et al., Reference Dombrovski, Clark, Siegle, Butters, Ichikawa and Sahakian2010), a high preference for immediate over delayed rewards, and difficulties anticipating positive future events (see van Heeringen, Bijttebier, & Godfrin, Reference van Heeringen, Bijttebier and Godfrin2011).

Research on the neural substrates of reward responding in SII is still in its infancy, but structural brain alterations in mesolimbic circuitry are implicated in major depression among individuals with a history of suicide attempts (Jia et al., Reference Jia, Huang, Wu, Zhang, Lui and Zhang2010), and those at high risk for suicide based on family history (Wagner et al., Reference Wagner, Koch, Schachtzabel, Schultz, Sauer and Schlösser2011). A history of self-harm is also associated with reduced functional connectivity in striatal circuitry, both during depressive episodes and in remitted depression (Marchand, Lee, Johnson, Thatcher, & Gale, Reference Marchand, Lee, Johnson, Thatcher and Gale2013; Marchand et al., Reference Marchand, Lee, Johnson, Thatcher, Gale and Wood2012). In the only study to date that has specifically evaluated neural correlates of reward responding among self-injurers, adolescents who engaged in SII exhibited less striatal activation in anticipation of incentives compared with controls (Sauder et al., Reference Sauder, Derbidge and Beauchaine2016).

Although depressed adolescents who engage in SII share many similarities with other depressed groups (e.g., high negative affectivity; low positive affectivity), they are more likely to follow a developmental course toward borderline personality disorder (BPD; see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Crowell, Beauchaine, & Linehan, Reference Crowell, Beauchaine and Linehan2009; Crowell, Kaufman, & Lenzenweger, Reference Crowell, Kaufman, Lenzenweger, Beauchaine and Hinshaw2013), and to engage in suicidal behaviors, gestures, and threats (diagnostic criteria of the disorder; APA, 2013). Although emotion dysregulation is regarded as a core feature of BPD (see Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Linehan, Reference Linehan1993), trait impulsivity and anhedonia are also common, and correlate with severity of symptoms (Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Marissen, Arnold, & Franken, Reference Marissen, Arnold and Franken2012). In the only study to date to investigate reward responding in BPD, participants with BPD or ASPD exhibited blunted prefrontal and striatal responding to monetary incentives (Völlm et al., Reference Völlm, Richardson, McKie, Elliott, Dolan and Deakin2007), suggesting deficient neural reward processing.Footnote ¹⁰

Interim conclusions

The construct of anhedonia typically encompasses reward processing related to both anticipatory/motivational and consummatory aspects of incentives. It is likely that some anhedonic individuals have deficiencies in both domains, whereas others have deficiencies restricted to only one or the other. However, findings from behavioral neuroscience and human neuroimaging studies suggest that distinguishing between motivation and the experience of pleasure is warranted, and that DA neurotransmission may be more critically involved in the former component. From this perspective, growing evidence suggests that high trait anhedonia may reflect poor associative reward learning and diminished motivation to exert effort to obtain rewards, both of which are mediated by a hypofunctioning mesolimbic DA system. These findings parallel research on the neural bases of trait impulsivity, described above. Major depression is a heterogeneous construct, and anhedonia is not equivalent to, or a necessary component of, depression. Additional research on anhedonia in the context of clinical, high-risk, and nonclinical populations, as well as disentangling the neural substrates of different aspects of depression, including anhedonia, may help to further clarify relationships between DA neural circuitry responding and trait anhedonia.

Neural bases of trait irritability

Compromised mesolimbic DA function, including low tonic and diminished anticipatory responses to incentives, is a core neural substrate of anhedonia and chronic irritability among both clinical and nonclinical populations (e.g., Beauchaine et al., Reference Beauchaine, Gatzke-Kopp and Mead2007; Shaw et al., Reference Shaw, Stringaris, Nigg and Leibenluft2014; Yamawaki, Okada, Okamoto, & Liberzon, Reference Yamawaki, Okada, Okamoto and Liberzon2012). PET studies demonstrate that low tonic midbrain DA activity is associated with trait irritability among healthy individuals (Laakso et al., Reference Laakso, Wallius, Kajander, Bergman, Eskola and Solin2003), whereas high DA activity is associated with a pleasant mood state and high hedonic capacity (see, e.g., Ashby et al., Reference Ashby, Isen and Turken1999; Depue, Luciana, Arbisi, Collins, & Leon, Reference Depue, Luciana, Arbisi, Collins and Leon1994). Thus, tonic DA hypofunctioning affects how incentives are experienced subjectively (i.e., liking), which in turn affects behavior (Berridge & Kringelbach, Reference Berridge and Kringelbach2015; Blum et al. Reference Blum, Braverman, Holder, Lubar, Monastra and Miller2000, Reference Blum, Chen, Braverman, Comings, Chen and Arcuri2008; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Among healthy individuals, the magnitude of DA release in the VS in response to DA agonism correlates with hedonic responding (Drevets et al., Reference Drevets, Gautier, Price, Kupfer, Kinahan and Grace2001), and upregulating DA availability increases hedonic expectations for positive future life events (Sharot, Shiner, Brown, Fan, & Dolan, Reference Sharot, Shiner, Brown, Fan and Dolan2009). Such findings suggest that raising mesolimbic DA levels is associated with improved mood and hedonic capacity.

It is interesting to note that some individuals report an unpleasant subjective reaction to DA agonism, which may reflect individual differences in tonic mesolimbic DA function (Volkow et al., Reference Volkow, Wang, Fowler, Logan, Gatley and Gifford1999). Healthy individuals who report liking the effects of infusions of methylphenidate, a DA reuptake inhibitor, have fewer D2 receptors in the striatum compared to those who report disliking the experience, and more D2 receptors correlate with greater intensity of unpleasantness (Volkow et al., Reference Volkow, Wang, Fowler, Logan, Gatley and Gifford1999). Because D2 receptors are autoreceptors, a lack of this receptor type may reflect compensation for diminished DA activity in reward circuitry (Blum et al., Reference Blum, Braverman, Holder, Lubar, Monastra and Miller2000). Thus, Volkow et al. (Reference Volkow, Wang, Fowler, Logan, Gatley and Gifford1999) conclude that DA agonism among individuals with intrinsically low mesolimbic DA activity is more likely to be experienced as pleasant because it raises DA levels to within an optimal range for enjoying the experience. In contrast, those with intrinsically higher mesolimbic DA activity are more likely to find the experience aversive because DA agonism pushes activity outside an optimal range of function. In addition, depressed individuals also rate subjective effects of DA agonism as more pleasant than healthy controls, and exhibit comparatively low mesolimbic functioning following dextroamphetamine administration (Tremblay et al., Reference Tremblay, Naranjo, Graham, Herrmann, Mayberg and Hevenor2005). Low DA synthesis capacity in the striatum and midbrain is also associated with aggressiveness (Schlüter et al., Reference Schlüter, Winz, Henkel, Prinz, Rademacher and Schmaljohann2013), which, as noted above, is often observed with irritability. Collectively, these findings lend further support to the supposition that low tonic and low phasic DA contribute to trait irritability and altered responses to reinforcers.

Contribution to externalizing disorders

Several contemporary models of ADHD assert that chronic irritability motivates hyperactivity and excessive reinforcement-seeking behavior, hallmarks of the combined and hyperactive–impulsive presentations (Blum et al. Reference Blum, Braverman, Holder, Lubar, Monastra and Miller2000, Reference Blum, Chen, Braverman, Comings, Chen and Arcuri2008; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005). Sagvolden et al. (Reference Sagvolden, Johansen, Aase and Russell2005) assert that reinforcement-seeking behaviors provoke phasic DA release, which serves to temporarily reduce feelings of irritability, anhedonia, and boredom caused by low tonic DA activity. As noted above, however, any hedonic value gained from these behaviors is transient, and does not address the underlying DA deficiency. Thus, searches for more frequent and larger incentives are undertaken to alleviate the aversive mood state (see Beauchaine et al., Reference Beauchaine, Gatzke-Kopp and Mead2007; Gatzke-Kopp, Reference Gatzke-Kopp2011; Gatzke-Kopp & Beauchaine, Reference Gatzke-Kopp, Beauchaine, Coch, Dawson and Fischer2007; Sagvolden et al., Reference Sagvolden, Johansen, Aase and Russell2005).

Contribution to depression and SII

Although depression can result from interactions among a multitude of biological vulnerabilities and environmental risk factors, a core etiological factor for many with the disorder is hypofunctioning DA-mediated reward processing. Deficient neural responding to incentives is a neural substrate of low positive affectivity, anhedonia, and irritability (see Forbes & Dahl, Reference Forbes and Dahl2005, Reference Forbes and Dahl2011; Nestler & Carlezon, Reference Nestler and Carlezon2006). Lower` striatal responding in anticipation of monetary incentives is associated with lower positive affect in real-world settings (Forbes et al., Reference Forbes, Hariri, Martin, Silk, Moyles and Fisher2009), and predicts increases in depressive symptoms over time (Morgan et al., Reference Morgan, Olino, McMakin, Ryan and Forbes2013). Dampened VS activation to reward is also associated with impaired mood reactivity to positive events (Foti et al., Reference Foti, Carlson, Sauder and Proudfit2014). Like anhedonia, irritability is particularly relevant to certain subpopulations with depression, including those with notable mood/emotion dysregulation and emotional lability. Irritability is associated with histories of suicidal ideation, suicide attempts, and nonsuicidal self-injury (e.g., Conner, Meldrum, Wieczorek, Duberstein, & Welte, Reference Conner, Meldrum, Wieczorek, Duberstein and Welte2004; Ernst et al., Reference Ernst, Lalovic, Lesage, Seguin, Tousignant and Turecki2004; Herpertz, Reference Herpertz1995; Pendse, Westrin, & Engström, Reference Pendse, Westrin and Engström1999; Stålenheim, Reference Stålenheim2001). Individuals who repeatedly engage in SII are likely to experience negative/irritable mood states and use self-harm as a coping strategy to mitigate emotional distress (Brown, Comtois, & Linehan, Reference Brown, Comtois and Linehan2002; Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Nixon, Cloutier, & Aggarwal, Reference Nixon, Cloutier and Aggarwal2002; Nock, Prinstein, & Sterba, Reference Nock, Prinstein and Sterba2009; Taylor, Peterson & Fischer, Reference Taylor, Peterson and Fischer2012). Trait irritability may also help to explain the high co-occurrence of self-harm and aggression directed toward others (Leibenluft & Stoddard, Reference Leibenluft and Stoddard2013; O'Donnell, House, & Waterman, Reference O'Donnell, House and Waterman2015).

Vulnerability to externalizing disorders, depression, and heterotypic comorbidity

Trait irritability is common among individuals who are impulsive and anhedonic, and confers vulnerability to externalizing spectrum disorders and depression. Seroczynski, Bergeman, and Coccaro (Reference Seroczynski, Bergeman and Coccaro1999) reported that impulsivity and irritability share overlapping genetic influences, a relationship that is stronger than that linking impulsivity and physical aggression. Greater irritability correlates with greater severity of hyperactive–impulsive ADHD symptoms (Sobanski et al., Reference Sobanski, Banaschewski, Asherson, Buitelaar, Chen and Franke2010), and raising mesolimbic DA levels via psychostimulant administration improves irritability among children with ADHD (de la Cruz et al., Reference de la Cruz, Simonoff, McGough, Halperin, Arnold and Stringaris2015). Such findings have led some to suggest that chronic irritability is a core feature of ADHD (e.g., Downson & Blackwell, Reference Downson and Blackwell2010; Skirrow, McLoughlin, Kuntsi, & Asherson, Reference Skirrow, McLoughlin, Kuntsi and Asherson2009).

Irritability is also associated with more severe anhedonia among depressed adolescents (Gabbay et al., Reference Gabbay, Johnson, Alonso, Evans, Babb and Klein2015), and is a common experience during major depressive episodes among adults, even when excluding for bipolar spectrum features (Fava et al., Reference Fava, Hwang, Rush, Sampson, Walters and Kessler2010; Judd, Schettler, Coryell, Akiskal, & Fiedorowicz, Reference Judd, Schettler, Coryell, Akiskal and Fiedorowicz2013; Perlis et al., Reference Perlis, Fava, Trivedi, Alpert, Luther and Wisniewski2009). Trait irritability and depression share significant heritability, and irritability in childhood predicts depression in early adulthood (Savage et al., Reference Savage, Verhulst, Copeland, Althoff, Lichtenstein and Roberson-Nay2015). Of note, irritability concurrent with all major depressive episode criteria except sad mood or anhedonia is exceptionally rare, which suggests it is linked closely to core features of major depression (Fava et al., Reference Fava, Hwang, Rush, Sampson, Walters and Kessler2010; Kovess-Masfety et al., Reference Kovess-Masfety, Alonso, Angermeyer, Bromet, de Girolamo and de Jonge2013).

We discuss irritability here because it frequently co-occurs with both externalizing spectrum disorders and depression, and because it is likely an important vulnerability to heterotypic comorbidity. Irritable children and adolescents with ADHD have significantly higher rates of depressive disorders compared to those who are not irritable (Ambrosini, Bennett, & Elia, Reference Ambrosini, Bennett and Elia2013; Sobanski et al., Reference Sobanski, Banaschewski, Asherson, Buitelaar, Chen and Franke2010). Similarly, chronic but not episodic irritability in early adolescence predicts ADHD and ODD during late adolescence, as well as MDD in early adulthood (Leibenluft, Cohen, Gorrindo, Brook, & Pine, Reference Leibenluft, Cohen, Gorrindo, Brook and Pine2006). Moreover, irritability in adolescence is associated with concurrent ADHD, mood disorders, ODD, and CD, and predicts MDD and dysthymia 20 years later (Stringaris, Psych, Cohen, Pine, & Leibenluft, Reference Stringaris, Psych, Cohen, Pine and Leibenluft2009). Finally, irritability is associated with high rates of comorbidity of impulse-control disorders and depression among adults (Fava et al., Reference Fava, Hwang, Rush, Sampson, Walters and Kessler2010), and may mark more severe psychopathology (Judd et al. Reference Judd, Schettler, Coryell, Akiskal and Fiedorowicz2013; Perlis et al., Reference Perlis, Fava, Trivedi, Alpert, Luther and Wisniewski2009; Sobanski et al., Reference Sobanski, Banaschewski, Asherson, Buitelaar, Chen and Franke2010).

Interim conclusions

Trait irritability, which is the tendency to be annoyed, angered, and impatient, is an aversive mood state that is characterized by low tonic mesolimbic DA activity and low phasic mesolimbic DA responding to incentives. Therefore, it shares neural bases with both trait impulsivity and anhedonia. As a transdiagnostic, trait-level vulnerability to psychopathology, irritability is often experienced by individuals with ADHD and individuals with major depression and appears to be especially likely among individuals comorbid for both disorders. Among those with ADHD, irritability may serve as a motivator to engage in hyperactive and reward-seeking behaviors, which elicit mesolimbic DA release, temporarily improving a chronically aversive mood state.

Subcortical moderators of midbrain DA activity and heterotypic comorbidity

Although externalizing spectrum disorders and unipolar depression are both characterized by anhedonia, avolition, and irritability, they differ in at least one important way: those with externalizing disorders exhibit excessive approach behavior (e.g., novelty, sensation, and reward seeking), whereas those with depression and oft co-occurring anxiety exhibit passive avoidance behavior (i.e., withdrawal; see, e.g., Beauchaine, Reference Beauchaine2001; Corr & McNaughton, Reference Corr, McNaughton, Beauchaine and Hinshaw2016). At the symptom level, this difference is difficult to reconcile with the high rates of heterotypic comorbidity described above. However, the apparent discrepancy can be resolved by differentiating between neural substrates of approach and passive avoidance tendencies. Toward this end, it is important to consider that complex human behaviors are subserved by multiple neurobiological systems that interact functionally with one another (see, e.g., Beauchaine et al., Reference Beauchaine, Neuhaus, Zalewski, Crowell and Potapova2011; Zisner & Beauchaine, Reference Zisner, Beauchaine and Cicchetti2016).

Here we consider how one such network, the septohippocampal system, interacts with midbrain DA function to affect behavior. The septohippocampal system includes extensive serotonergic (5-HT) projections from the raphe nuclei, and noradrenergic projections from the locus ceruleus, to a number of brain regions, including limbic structures (e.g., amygdala and hippocampus), the anterior cingulate cortex, and more frontal regions (Gray & McNaughton, Reference Gray and McNaughton2000; McNaughton & Corr, Reference McNaughton and Corr2004). The septohippocampal system and the amygdalaFootnote ¹¹ (Gray's behavioral inhibition system) is a phylogenetically old neural system that responds to competing motivational objectives (i.e., approach-approach, approach-avoidance, and avoidance-avoidance). When such situations are encountered, septohippocampal activation promotes passive avoidance until the organism decides what action to take. Individual differences in activity and reactivity of this system give rise to corresponding individual differences in trait anxiety (Corr & McNaughton, Reference Corr, McNaughton, Beauchaine and Hinshaw2016; Gray & McNaughton, Reference Gray and McNaughton2000; McNaughton & Corr, Reference McNaughton and Corr2004). More broadly, the septohippocampal system helps us narrow attention, assign emotional significance to events, and encode them into memory (see Phan, Wager, Taylor, & Liberzon, Reference Phan, Wager, Taylor and Liberzon2002; Phelps, Reference Phelps2004). Excessive septohippocampal activity confers vulnerability to anxiety disorders and depression (Gray & McNaughton, Reference Gray and McNaughton2000), whereas deficient septohippocampal activity potentiates vulnerability to externalizing behavior (see Corr & McNaughton, Reference Corr, McNaughton, Beauchaine and Hinshaw2016). McNaughton and Corr (Reference McNaughton and Corr2004) conceptualize depression in particular as a behavioral manifestation of excessive septohippocampal system activity, which results in maladaptive responses to threat under circumstances in which the most appropriate response is to approach threat in order to obtain incentives. Instead, those who are depressed fail to approach because aversive outcomes of approach behavior are perceived as unavoidable, and outweigh any possible gains that could be obtained through action.

Functional outcomes

The research presented above suggests that heterotypically comorbid internalizing and externalizing symptoms are associated with co-occurring functional impairment in neural systems that differentiate between pure disorders. At the behavioral level of analysis, anxiety confers some protective effects among those with externalizing psychopathology. For example, those with ADHD and comorbid anxiety are less impulsive and exhibit a less severe response inhibition deficit than those with ADHD alone (see Shatz & Rostain, 2006). Furthermore, children and adolescents with CD and comorbid anxiety are less physically aggressive, regarded less negatively by peers, and experience fewer police contacts than youth with CD alone (Walker et al., Reference Walker, Lahey, Russo, Frick, Christ and McBurnett1991). Externalizing youth with comorbid anxiety are also more sensitive to punishment, and therefore more concerned with negative consequences of their behavior compared to nonanxious externalizing youth, which likely accounts for why this externalizing subgroup exhibits fewer behavior problems and better responses to behavioral treatment (e.g., Beauchaine, Neuhaus, et al., Reference Beauchaine, Neuhaus, Gatzke-Kopp, Reid, Brekke and Olliges2016; Jensen et al., Reference Jensen, Hinshaw, Kraemer, Lenora, Newcorn and Abikoff2001). Other research, however, suggests that externalizers with internalizing symptoms exhibit more severe conduct problems and greater risk of long-term social and academic impairment than noncomorbid individuals (Biederman et al., Reference Biederman, Ball, Monuteaux, Mick, Spencer and McCreary2008; Blackman, Ostrander, & Herman, Reference Blackman, Ostrander and Herman2005; Daviss, Reference Daviss2008; Ezpeleta, Domènech, & Angold, Reference Ezpeleta, Domènech and Angold2006).

Comorbid internalizing and externalizing psychopathology shape the ways in which disorders present in other ways as well. Anhedonia in BPD is related to poor impulse control, and excessive approach behaviors, but anhedonia among healthy controls is associated with the opposite pattern: withdrawal behaviors (Marissen et al., Reference Marissen, Arnold and Franken2012). Compared to depression, BPD is associated with elevated risk for suicide attempts, but comorbidity between BPD and depression is associated with more serious and more frequent suicide attempts than depression or BPD alone (Soloff, Lynch, Kelly, Malone, & Mann, Reference Soloff, Lynch, Kelly, Malone and Mann2000). High trait impulsivity is a known vulnerability to self-harm (Allely, Reference Allely2014; Crowell et al., Reference Crowell, Beauchaine and Linehan2009), and some have suggested that ASPD and BPD may reflect sex-moderated manifestations of a single core vulnerability (see Beauchaine et al., Reference Beauchaine, Klein, Crowell, Derbidge and Gatzke-Kopp2009; Crowell et al., Reference Crowell, Beauchaine and Linehan2009; Paris, Reference Paris1997; Phillipsen, Reference Philipsen2006).

Interim conclusions

Thus far, we have demonstrated that high trait impulsivity, conferred by low tonic and low phasic mesolimbic DA function, is an important vulnerability to externalizing spectrum disorders, whereas high trait impulsivity combined with high trait anxiety (subserved by excessive septohippocampal and amygdalar activity/reactivity) is implicated in comorbid externalizing and internalizing conditions. Trait anhedonia and irritability, which emerge from the same subcortical, mesolimbic substrates as trait impulsivity, are transdiagnostic vulnerabilities to both externalizing and internalizing spectrum disorders, and thus, heterotypic comorbidity.

Although chronic dysfunction in mesolimbic circuitry versus septohippocampal circuitry confer vulnerability to distinct trait-level predispositions (trait impulsivity and trait anxiety, respectively), functional interactions between neural systems affect how these vulnerabilities present (e.g., anhedonia in externalizing disorders contributes to excessive approach, whereas anhedonia in internalizing disorders contributes to passive avoidance). Investigations into neural correlates of Internalizing × Externalizing interactions are only beginning to emerge, and should shed light on how interdependent neural circuits influence one another in both clinical and healthy populations.

Externalizing psychopathology

Volitional regulation of impulsivity requires modulation of the striatum by interconnected neural systems including the anterior cingulate, orbitofrontal, and dorsolateral prefrontal cortices (see Beauchaine, Reference Beauchaine2015a; Heatherton, 2011; Heatherton & Wagner, 2011). When mesolimbic DA dysfunction is combined with deficient functioning of these cortical regions, externalizing psychopathology, beginning with ADHD and progressing to more severe externalizing disorders in high-risk environments, is especially likely (Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Beauchaine et al., Reference Beauchaine, Hinshaw and Pang2010, in press; Beauchaine, Neuhaus, et al., Reference Beauchaine, Neuhaus, Gatzke-Kopp, Reid, Brekke and Olliges2016; Zisner & Beauchaine, Reference Zisner, Beauchaine, Beauchaine and Hinshaw2015). This pattern of neural dysfunction is frequently exhibited among externalizers. For example, in addition to striatal abnormalities, which are widely reported in ADHD (see above), the disorder is also associated with ACC abnormalities, including hypofunctionality across a range of tasks (see Bush, Valera, & Seidman, Reference Bush, Valera and Seidman2005), reduced volumes (Frodl & Skokauskas, Reference Frodl and Skokauskas2012; Makris et al., Reference Makris, Biederman, Valera, Bush, Kaiser and Kennedy2007; Seidman et al., Reference Seidman, Valera, Makris, Monuteaux, Boriel and Kelkar2006), and reduced connectivity between the ACC and striatum (e.g., Shannon et al., Reference Shannon, Sauder, Beauchaine and Gatzke-Kopp2009). In addition, individuals with ADHD often display diminished error-related negativity, an electrophysiological signal generated by the ACC that reflects error monitoring (Shiels & Hawk, Reference Shiels and Hawk2010). Given the crucial role of this region in performance- and goal-related processing, disruptions in the ACC may contribute to altered ability to adjust one's behavior based on feedback, an ability that is compromised in ADHD and other externalizing disorders (e.g., Bush, Reference Bush2011; Gatzke-Kopp et al., Reference Gatzke-Kopp, Beauchaine, Shannon, Chipman-Chacon, Fleming and Crowell2009).

Higher order processing of incentives is mediated in large part by the OFC, a region that is also compromised in externalizing disorders. For example, individuals with ADHD have smaller OFC volumes (Hesslinger et al., Reference Hesslinger, Van Elst, Thiel, Haegele, Hennig and Ebert2002), but exhibit stronger functional connectivity between OFC, striatum, and ACC (Tomasi & Volkow, Reference Tomasi and Volkow2012). This is compatible with functional evidence from reward-based tasks suggesting that individuals with ADHD may also have OFCs that fail to differentiate between rewards of differing value (Wilbertz et al., Reference Wilbertz, van Elst, Delgado, Maier, Feige and Philipsen2012), and are hyperresponsive to certain types of gains (Ströhle et al., Reference Ströhle, Stoy, Wrase, Schwarzer, Schlagenhauf and Huss2008). Orbitofrontal dysfunction may contribute to increased saliency of rewards once those rewards are of sufficient magnitude to initiate mesolimbic reactivity.

Internalizing psychopathology

As discussed above, the vmPFC, OFC, and ACC are implicated in both reward processing and top-down emotion regulation. Unipolar depression is associated with dysfunctional reactions to emotion-evoking stimuli in regions related to affective processing, including the vmPFC, OFC, and ACC (e.g., see Kerestes et al., Reference Kerestes, Davey, Stephanou, Whittle and Harrison2014; Koenigs & Grafman, Reference Koenigs and Grafman2009; Murray, Wise, & Drevets, Reference Murray, Wise and Drevets2011; Stuhrmann, Suslow, & Dannlowski, Reference Stuhrmann, Suslow and Dannlowski2011). However, it is important to note that motivated/volitional regulatory control of emotions occurs through inhibition of the amygdala by lateral regions of the PFC (see Buhle et al., Reference Buhle, Silvers, Wager, Lopez, Onyemekwu and Kober2014; Clauss, Avery, & Blackford, Reference Clauss, Avery and Blackford2015; Heatherton, 2011; Heatherton & Wagner, 2011). The lateral PFC, primarily implicated in cognitive control, does not project directly to the amygdala, but rather exerts inhibitory effects though its connections with medial regions of the PFC, which have dense projections to and from the amygdala (Johnstone, van Reekum, Urry, Kalin, & Davidson, Reference Johnstone, van Reekum, Urry, Kalin and Davidson2007; Öngür & Price, Reference Öngür and Price2000; Urry et al., Reference Urry, Van Reekum, Johnstone, Kalin, Thurow and Schaefer2006). For example, cognitive reappraisal of affective stimuli is associated with activation of the lateral PFC and deactivation of the amygdala (see Buhle et al., Reference Buhle, Silvers, Wager, Lopez, Onyemekwu and Kober2014), whereas unsuccessful cognitive reappraisal (increased negative emotion) may reflect failure of the lateral PFC to suppress amygdala responding (Wagner, Davidson, Hughes, Lindquist, & Ochsner, Reference Wager, Davidson, Hughes, Lindquist and Ochsner2008).

Depression may involve failure of the lateral PFC to suppress the amygdala via the medial PFC (Johnstone et al., Reference Johnstone, van Reekum, Urry, Kalin and Davidson2007), an assertion supported by hypofunctionality of the lateral PFC and reduced functional connectivity between the dlPFC and the amygdala among depressed individuals (see Koenigs & Grafman, Reference Koenigs and Grafman2009; e.g., Dannlowski et al., Reference Dannlowski, Ohrmann, Konrad, Domschke, Bauer and Kugel2009; Johnstone et al., Reference Johnstone, van Reekum, Urry, Kalin and Davidson2007; Siegle, Thompson, Carter, Steinhauer, & Thase, Reference Siegle, Thompson, Carter, Steinhauer and Thase2007). Given the role of these lateral regions in regulating emotions, appraisals and attributions initiated by the dlPFC to dampen effects of negative, emotionally evocative stimuli among healthy individuals may not be applied effectively among those who are depressed, contributing to rumination and heightened negative affect (see Gotlib & Hamilton, Reference Gotlib and Hamilton2008; Hamilton et al., Reference Hamilton, Etkin, Furman, Lemus, Johnson and Gotlib2012).

Emotion regulation

Emotion regulation refers to processes that guide the experience and expression of emotions to facilitate adaptive behavior (Thompson, Reference Thompson and Thompson1990). Emotion regulation skills, subserved by effective cortical regulation of subcortical emotion generation circuits, protect against development of psychopathology (see Beauchaine, Reference Beauchaine2015a; Beauchaine et al., Reference Beauchaine, Gatzke-Kopp and Mead2007; Beauchaine & Thayer, Reference Beauchaine and Thayer2015). As noted above, subcortical vulnerabilities to trait impulsivity, anxiety, and anhedonia are necessary but insufficient to produce functional impairment. Rather, the combination of subcortical dysfunction and deficient top-down modulation of subcortical systems makes psychopathology most likely (Beauchaine, Reference Beauchaine2001, 2015; Beauchaine & Gatzke-Kopp, Reference Beauchaine and Gatzke-Kopp2012; Beauchaine et al., Reference Beauchaine, Gatzke-Kopp and Mead2007; Fineberg et al., Reference Fineberg, Potenza, Chamberlain, Berlin, Menzies and Bechara2010). This reflects a primary role of the prefrontal cortex: to appropriately modulate/inhibit responses in the service of effective behavior and emotion regulation (see, e.g., Dillon & Pizzagalli, Reference Dillon and Pizzagalli2007; Etkin et al., Reference Etkin, Egner and Kalisch2011; Schore, Reference Schore1996).

Developmental and environmental considerations

The brain undergoes profound structural changes during development, with cortical regions reaching maturity comparatively later than phylogenetically older, subcortical structures (Gogtay et al., Reference Gogtay, Giedd, Lusk, Hayashi, Greenstein and Vaituzis2004; Toga, Thompson, & Sowell, Reference Toga, Thompson and Sowell2006). Among typically developing individuals, peak volume of the PFC occurs around ages 10–11, and neural maturation into adulthood includes cortical thinning in the parietal and right dorsal frontal areas, and cortical thickening in language-processing areas in the frontal and temporal lobes (see Lenroot & Giedd, Reference Lenroot and Giedd2006; Toga et al., Reference Toga, Thompson and Sowell2006). Adolescence in particular is a developmental period marked by a number of neural changes, including enhanced DA innervation of the frontal cortex and increased DA concentrations in cortical and subcortical regions (Wahlstrom, Collins, White, & Luciana, Reference Wahlstrom, Collins, White and Luciana2010).

Underdevelopment of prefrontal regions that subserve self-regulatory functions, combined with low mesolimbic DA activity/reactivity, is a shared vulnerability to impulsivity and anhedonia in internalizing and externalizing disorders across the life span (see Beauchaine et al., in press). For example, in adolescence, the absence of fully intact self-regulatory regions, which continue to develop into adulthood (Lenroot & Giedd, Reference Lenroot and Giedd2006), combined with normal upregulation of DA functioning, lead to increased saliency of and motivation to seek rewards that then manifest as increased risk-taking and novelty-seeking behaviors (see Chambers, Taylor, & Potenza, Reference Chambers, Taylor and Potenza2003; Wahlstrom et al., Reference Wahlstrom, Collins, White and Luciana2010). Thus, among typically developing adolescents, impulsive tendencies are subserved by DA hyperresponsivity to reward unchecked by inhibitory mechanisms. Similarly, Forbes and Dahl (2012) suggest that adolescents with low reward responding are vulnerable to depression because of imbalances in frontostriatal reward circuitry, which are especially discrepant compared with those of typically developing individuals during adolescence. In other words, a trajectory toward depression may be more likely among individuals with inherent mesolimbic DA vulnerability, coupled with underdevelopment of higher order cortical regions involved in self-regulation and emotion regulation.

Like depression, externalizing vulnerability results at least in part from underdevelopment of frontostriatal circuitry (see, e.g., Cubillo, Halari, Smith, Taylor, & Rubia, Reference Cubillo, Halari, Smith, Taylor and Rubia2012). Although ~65% of adults who experienced childhood ADHD maintain at least some symptoms into adulthood (Faraone, Biederman, & Mick, Reference Faraone, Biederman and Mick2006), individuals with ADHD whose symptoms remit with age may gain self-regulatory skills that match their unaffected peers through eventual maturation of frontostriatal circuitry and/or recruitment of alternative circuitry to compensate for enduring frontostriatal deficits (Vaidya, Reference Vaidya, Stanford and Tannock2012). For example, the medial PFC, which has direct projections to the striatum, develops more slowly among children with ADHD, and children with more impairing and persistent ADHD-related impairment have medial PFCs that are and remain thinner into adolescence compared to those with better outcomes and controls (Shaw et al., Reference Shaw, Lerch, Greenstein, Sharp, Clasen and Evans2006, Reference Shaw, Malek, Watson, Sharp, Evans and Greenstein2012). Similarly, fronto-striato-parietal dysfunction among adults with persistent ADHD correlates negatively with the severity of ADHD symptoms (Cubillo et al., Reference Cubillo, Halari, Smith, Taylor and Rubia2012; Schneider et al., Reference Schneider, Krick, Retz, Hengesch, Retz-Junginger and Reith2010). Finally, among boys with CD, most of whom experience homotypic comorbidity with other externalizing disorders, normal maturation (gray matter pruning) of the PFC is not observed in late childhood/early adolescence (De Brito et al., Reference De Brito, Mechelli, Wilke, Laurens, Jones and Barker2009). Thus, failures of prefrontal maturation influence the presence, persistence, and severity of symptoms of both internalizing and externalizing disorders.

Given the significance of bottom-up and top-down neural function in the expression of psychopathology, it is important to address environmental influences on prefrontal and subcortical neurodevelopment. Early-life adversity, including child abuse and neglect, malnutrition, and prenatal exposure to neurotoxins, substances of abuse, and high maternal stress, are associated with epigenetic changes that can have profound effects on neural development and function (see Archer, Oscar-Berman, Blum, & Gold, 2012; Gatzke-Kopp, Reference Gatzke-Kopp2011). For instance, the structure and function of DA pathways are altered by both chronic and acute stress, possibly through physiological responses enacted by the hypothalamus–pituitary–adrenal (HPA) axis (Pani, Porcella, & Gessa, Reference Pani, Porcella and Gessa2000). Environmental insults can result in hyper- and hypofunctioning of DA neural circuitry, depending on the severity and chronicity of insult, and are associated with increased risk for psychopathologies characterized by aberrant motivation/reward-processing, including ADHD, substance abuse, and depression (see Beauchaine et al., Reference Beauchaine, Neuhaus, Zalewski, Crowell and Potapova2011; Gatzke-Kopp, Reference Gatzke-Kopp2011).

Physical abuse in childhood predicts smaller OFC volumes, which in turn are associated with social difficulties with peers and executive functioning deficits (Hanson et al., Reference Hanson, Chung, Avants, Shirtcliff, Gee and Davidson2010). In addition, adversity accrued across childhood is associated with smaller PFC volumes (Hanson et al., Reference Hanson, Chung, Avants, Rudolph, Shirtcliff and Gee2012), as is being raised in poor neighborhoods (Hanson et al., Reference Hanson, Hair, Shen, Shi, Gilmore and Wolfe2013). Moreover, neglect in childhood is associated with more diffuse frontal white matter organization (Hanson, Adluru, et al., 2013), and chronic exposure to stress results in less dendritic branching and lower neural spine densities in the PFC among rodents (e.g., Holmes & Wellman, Reference Holmes and Wellman2009).

Neural vulnerabilities to psychopathology that emerge early in development may be exacerbated by environmental stress to make psychopathology more likely. For example, among rodents, a history of chronic stress predicts functional alterations in neural, behavioral, and pharmacological responses to acute stress (e.g., Finlay, Zigmond, & Abercrombie, Reference Finlay, Zigmond and Abercrombie1995; Isgor, Kabbaj, Akil, & Watson, Reference Isgor, Kabbaj, Akil and Watson2004; Katz, Roth, & Carroll, Reference Katz, Roth and Carroll1981). Similarly, among humans, exposure to childhood trauma predicts aberrant tonic HPA axis functioning and physiological sensitization to stress, which in turn predict symptoms of depression (see Heim & Binder, Reference Heim and Binder2012; Heim, Newport, Mletzko, Miller, & Nemeroff, Reference Heim, Newport, Mletzko, Miller and Nemeroff2008). Such findings suggest that early-life stressors can elicit neural and physiological changes that affect responses to acute stressors, which may then increase risk for the emergence of psychiatric symptoms (Beauchaine, Neuhaus, Zalewski, Crowell, & Potapova, 2011).

Neural vulnerabilities for psychopathology that emerge early in development can also be exacerbated by the psychopathology that they portend once psychiatric illness develops. As reviewed above, for example, low mesolimbic function, an important neural substrate of trait impulsivity, confers vulnerability to all externalizing disorders, including SUDs (Zisner & Beauchaine, Reference Zisner, Beauchaine, Beauchaine and Hinshaw2015). Strong and recurrent exposure to stimulants provokes downregulation of tonic DA activity in the NAcc, but sensitized phasic DA release specifically to these stimulants (see Thomas, Beurrier, Bonci, & Malenka Reference Thomas, Beurrier, Bonci and Malenka2001; Vezina, Reference Vezina2004). These changes, combined with the diminished PFC regulatory functioning that occurs in addiction, lead to greater impulsive decision making and higher risk of relapse (see Goldstein & Volkow, Reference Goldstein and Volkow2011; Kalivas, Reference Kalivas2008; Schoenbaum & Shaham, Reference Schoenbaum and Shaham2008). Among individuals with tonic DA levels that are inherently low, neural dysfunction elicited by substance dependence may be particularly devastating.

Interim conclusions

Heterotypic comorbidity emerges at least in part from a common neural vulnerability (low tonic DA and blunted DA-mediated responses to anticipation of incentives) with concurrent impairments in other neural circuits (e.g., hyperresponsitivity of the septohippocampal system to negative stimuli), which manifest as both overlapping and independent psychopathological symptoms. However, subcortically mediated deficits in approach and avoidance behavior are not necessarily impairing in the absence of co-occurring PFC dysfunction given critical roles of specific PFC regions in self-regulation and emotion regulation. Without PFC dysfunction, individual differences in trait impulsivity produce ordinary variation in personality, not psychopathology (Beauchaine, Reference Beauchaine2001, Reference Beauchaine2015a; Beauchaine & Gatzke-Kopp, Reference Beauchaine and Gatzke-Kopp2012; Beauchaine et al., Reference Beauchaine, Gatzke-Kopp and Mead2007). Similarly, trait anxiety and negative affectivity, which are vulnerabilities to unipolar depression, are less likely to result in functional impairment in the presence of strong emotion regulation capabilities. Furthermore, cumulative effects of neurobiological vulnerabilities, such as the degree to which an individual produces blunted mesolimbic responses or excessive septohippocampal responses, within the context of environmental risk factors, such as childhood trauma and substance abuse, influence the ways in which such neural deficits develop, manifest, and are maintained (see Beauchaine, Reference Beauchaine2015a; Beauchaine et al., Reference Neuhaus, Beauchaine, Beauchaine and Hinshawin press; Beauchaine & McNulty, Reference Beauchaine and McNulty2013; Sauder et al., Reference Sauder, Beauchaine, Gatzke-Kopp, Shannon and Aylward2012; Shannon, Beauchaine, Brenner, Neuhaus, & Gatzke-Kopp, Reference Shannon, Beauchaine, Brenner, Neuhaus and Gatzke-Kopp2007).