A Targeted Review of the Neurobiology and Genetics of Behavioral Addictions: An Emerging Area of Research

Abstract

This review summarizes neurobiological and genetic findings in behavioral addictions, draws parallels with findings pertaining to substance use disorders and offers suggestions for future research. Articles concerning brain function, neurotransmitter activity and family history/genetics findings for behavioral addictions involving gambling, internet use, video game playing, shopping, kleptomania and sexual activity were reviewed. Behavioral addictions involve dysfunction in several brain regions, particularly the frontal cortex and striatum. Findings from imaging studies incorporating cognitive tasks have arguably been more consistent than cue-induction studies. Early results suggest white and gray matter differences. Neurochemical findings suggest roles for dopaminergic and serotonergic systems, but results from clinical trials seem more equivocal. While limited, family history/genetic data support heritability for pathological gambling and that those with behavioral addictions are more likely to have a close family member with some form of psychopathology. Parallels exist between neurobiological and genetic/family history findings in substance and non-substance addictions, suggesting that compulsive engagement in these behaviors may constitute addictions. Findings to date are limited, particularly for shopping, kleptomania and sexual behavior. Genetic understandings are at an early stage. Future research directions are offered.

Introduction

Classes of behaviors having hedonic qualities (at least initially) including gambling, shopping, sexual behaviors, internet use and video game play may lead to compulsive engagement among a minority of individuals. At excessive levels, these behaviors are considered “impulse control disorders not elsewhere classified” in the DSM-IV-TR¹, However, they may also be considered non-substance or “behavioral” addictions^2-7. As gambling, shopping, sex, gaming and internet use are normative behaviors, it may be challenging to differentiate between normal and excessive participation⁵. Further challenges may stem from greater heterogeneity in the syndromes of behavioral addictions, complicating their categorization⁸. Mechanisms underlying behavioral (versus substance) addictions are relatively poorly understood, in part because animal models that have facilitated insight into substance use disorders^9,10 are less straightforward or advanced for behavioral addictions^8,11,12.

Behavioral addictions share important elements with substance addictions. These include impaired control over engagement, continued engagement despite negative consequences and urges or cravings^6,13. Behavioral and substance addictions frequently co-occur^14,15 and there are similarities in the progression of the disorders (e.g., high rates of the conditions in adolescents and young adults, negative reinforcement motivations and a “telescoping” phenomenon observed in females^6,16).

Similar neurobiological features underlie both substance and behavioral addictions^8,17,18, with common features involving cross-sensitization, brain function and neurochemistry⁸. Cross-sensitization involves neuro-adaptations in which repeated exposure to one drug leads to a more robust response to another⁸. With respect to non-substance addictions, exposure to a substance of abuse can lead to sensitization to a natural reward and vice-versa^8,19-21. The extent to which these findings extend to behaviors like gambling warrants additional investigation. All drugs of abuse affect the brain's “reward circuit,” with the mesolimbic dopamine pathway being of particular importance. This pathway includes dopaminergic neurons extending from the ventral tegmental area to the nucleus accumbens (NAc)^22-25. Dopamine levels that are either too high or too low are suboptimal and may lead to impulsive and risk-taking acts including excessive substance use²⁶. Natural rewards and abused substances appear to induce similar activity in reward circuitry and connected regions, including the amygdala, hippocampus and frontal cortex⁸.

Genetic and family history findings, albeit limited for behavioral addictions, provide further evidence of commonality between behavioral and substance addictions²⁷. Co-morbidity among behavioral and substance addictions and other psychiatric conditions appears to involve shared genetic factors^15,27-30.

The present review considers neurobiological and genetic/family history evidence pertaining to behavioral addictions. After describing our methods, we discuss brain function (Table 1), neurotransmitter systems (Table 2) and family history/genetic findings (Table 3) relating to six behavioral addictions: pathological gambling primarily; problematic Internet use and video-game playing secondarily; and thirdly, compulsive shopping, kleptomania and hypersexuality. We highlight similarities and differences with substance addiction findings, describe conclusions and offer suggestions for future research. Epidemiology and clinical findings are addressed briefly; however, several recent reviews^2,31 and an edited volume¹⁴ have addressed these topics. We excluded studies involving only healthy or Parkinson's Disease (PD) participants. While PD studies provide a useful model for behavioral addictions, the extent to which these findings apply to the larger population of non-PD patients is uncertain (see ^32,33).

Table 1.

Overview of brain function/neuroimaging results for six types of behavioral addiction and similarities to and differences from key results in behavioral addictions and substance use disorders (SUDs), with a focus on fronto-striatal findings.

Behavioral addiction	Key results	Similarities to/differences with key results in substance use disorders
gambling	Frontal areas and striatum:
	D2-like Dopamine Receptor PET Studies: Do not support between-group differences from control comparison subjects in striatal binding	Between-group differences observed in substance-dependent and control comparison subjects in striatal binding
	Cue-induction: Difference from control subjects in frontal cortical areas and striatum but precise nature of differences seemingly inconsistent	Difference from control subjects in frontal cortical areas and striatum but precise nature of differences seemingly inconsistent
	Cognitive tasks: Reduced frontal activity in most studies; typically reduced ventral striatal activity in PG groups; stronger ventral striatal activity in less severe groups compared to controls	Most findings suggest reduced activity in frontal areas, similar findings of reduced ventral striatal activity compared to controls, but also findings suggesting increased activity and negative findings
	White matter: Poor integrity in multiple regions including corpus callosum	Poor white matter integrity in multiple regions including corpus callosum
Internet use	Frontal areas and striatum:
	D2-like Dopamine Receptor PET Studies: reduced D2-like receptor availability in dorsal striatum, no differences in ventral striatum	Between-group differences observed in substance-dependent and control comparison subjects in striatal binding, particularly in dorsal striatum
	Resting state: Increased regional homogeneity in multiple regions including frontal areas and ACC	No findings utilizing the same regional homogeneity method were located.
	White and gray matter: Poor white matter integrity and decreased gray matter volumes in multiple regions.	Poor white matter integrity and decreased gray matter volumes in SUDs.
Video-game playing	Frontal areas and striatum:
	Resting state: Increased metabolism in middle orbitofrontal gyrus, reduced metabolism in left precentral gyrus, increased metabolism in left caudate	Reduced activity at resting state typically found in frontal areas, but some exceptions; reduced ventral striatal activity and increased dorsal activity typically found.
	Cue induction: Increased activity compared to controls in multiple frontal areas including OFC, dlPFC; increased activity in NAc and right caudate compared to controls.	Differences from control comparison subjects in frontal areas and ventral striatum but precise nature of differences inconsistent; evidence of increased dorsal activity compared to controls.
	Other regions: Increased activity in ACC in response to cues, decreased activation in loss trials of risk/reward task; increased activity in insula.	Increased activity in ACC in response to cues, also implicated in risky decision making in SUD; increased activity in insula.
	Gray matter: Increased volume in left thalamus; decreased volume in multiple regions (e.g., inferior temporal gyri)	Decreased gray matter volume in SUDs in multiple regions (e.g., orbitofrontal cortex, cerebellum).
shopping	Striatum: Increased activity in ventral striatum upon product presentation.	Dysfunction in ventral striatum but precise nature of dysfunction differs per task
	Other regions: Reduced activity in insula and ACC during price presentation, ACC activated during decision phase	Insula activated in response to substance cues; proposed role for ACC activity in risky decision-making in SUDs
kleptomania	White matter: Poor integrity in ventral-medial-frontal regions	Poor white matter integrity in SUDs
sex	White matter: Higher integrity in lower superior frontal region	Poor white matter integrity in SUDs

Table 2.

Overview of neurotransmitter system involvement in six types of behavioral addiction and similarities to and differences from key results in substance use disorders

Behavioral addiction	Key results	Similarities to/differences with key results in substance use disorders
gambling	Dopamine: Dopamine implicated but precise nature unclear; limited findings negative comparing PG with controls in D2-like receptor availability; individual differences in dopamine release and function; some results suggest differential response to agonist and antagonist administration in PG; clinical trial results with antagonists have been negative	Data suggest reduced numbers of D2-like receptors in SUDs compared to controls; substance use has been related to release in some studies but also individual differences; some clinical trial findings with antagonists positive, while others negative, with limited clinical utility demonstrated
	Serotonin: Varied neurochemical findings suggest differential function in PG compared to controls; negative and mixed clinical trial findings with reuptake inhibitors and a receptor antagonist suggest possible individual differences in function	Neurochemical studies suggest differential function; some clinical trial results with reuptake inhibitors have been positive while others negative, suggesting possible individual differences in activity
	Opioids: Multiple positive clinical findings suggest role for opioidergic systems in PG	Multiple positive clinical findings suggest role for opioidergic systems, particularly for opiates and alcohol
	Glutamate: Preliminary positive clinical findings suggest it may have a role, particularly in impulsive and compulsive behaviors	Preliminary positive clinical findings suggest it may have a role, particularly in impulsive and compulsive behaviors
	Norepinephrine: Elevated at resting state and found to increase during gambling; blunted growth hormone response to clonidine	Elevated during use of some substances, particularly cocaine
Internet use	Dopamine: Low levels of dopamine transporter expression in striatum	Low levels of dopamine transporter expression in striatum in some studies though higher levels in other studies
Video-game playing	Dopamine: The role of dopaminergic activity has not been investigated directly; limited and preliminary findings related to genotypes associated with possible alterations in dopamine signaling suggest possible differences from controls in dopaminergic activity.	Evidence suggests differences in dopaminergic activity between substance dependent individuals and controls
shopping	Serotonin: Positive results in an open-label clinical trial but negative results in controlled trials with serotonin reuptake inhibitors	Clinical results with reuptake inhibitors have been positive in some studies and negative in others.
kleptomania	Serotonin: Some neurochemical results suggesting involvement but negative clinical results	Neurochemical studies differences from controls in serotonergic function; clinical results inconclusive and suggest possible individual differences
	Opioids: Positive preliminary clinical results suggest a possible role for opioidergic system involvement	Positive clinical results suggest a role for opioidergic systems in SUDs
sex	Serotonin: Limited positive clinical results suggest possible role for serotonergic activity	Clinical results with reuptake inhibitors have been positive in some studies and negative in others.

Table 3.

Overview of genetic results for six types of behavioral addiction and similarities to and differences from key results in substance use disorders

Behavioral addiction	Key results	Similarities to/differences with key results in substance use disorders
gambling	Behavior genetics: PG highly heritable, equivalent heritability between males and females	SUDs highly heritable, equivalent heritability between males and females
	Molecular: Small, additive effects across genes; associations with polymorphisms of dopamine receptor genes, but also negative findings; also preliminary findings associating serotonin transporter and MAO-A polymorphisms; no genome-wide association studies to date for PG or other behavioral addictions	Small, additive effects across genes; associations with polymorphisms of dopamine receptor genes, but also negative findings; also findings associating serontoin transporter and MAO-A polymorphisms; genome-wide associations indicate some different contributions to individual SUDs
Internet use	Molecular: Harm avoidant subgroup showed over-expression of SS-5HTTLPR in preliminary studies	Findings associating serotonin transporter to SUDs
Video-game playing	Molecular: Taq 1A polymorphism of DRD2 receptor gene and low activity COMT alleles more prevalent in compulsive video game players in preliminary studies	Some findings linking SUDs to Taq 1A polymorphism but negative findings as well; COMT has been linked to nicotine dependence and the low activity variant has been associated with early-onset alcoholism
shopping	Behavior genetics: Compulsive shoppers likely to have close family members with various psychopathology	Those with SUDs likely to have close family members with various psychopathology
	Molecular: No evidence of abnormality in two 5-HTT polymorphisms that were investigated	Findings associating serotonin transporter to SUDs
kleptomania	Behavior genetics: Kleptomaniacs likely to have close family members with various psychopathology	Those with SUDs likely to have close family members with various psychopathology
sex	Behavior genetics: Relationship to parental history of sexual compulsion, more likely to have first degree relatives with SUDs	Those with SUDs more likely to have first-degree relatives with SUDs

Methods

Literature searches were conducted in May 2012 using Medline and Google Scholar. Each search was conducted using a general search term (neuro*, MRI, PET, imaging and genet*) and a search term for one of the following behavioral addictions (search terms in parentheses): gambling (gambl*), shopping (compulsive shopping, shopping addict*, compulsive buying), kleptomania (kleptomania, steal), sexual behavior (compulsive sex*, sex* addict*), internet (internet addict*, compulsive internet) and video game play (video gam*). Given space limitations and the multiple topics reviewed, data deemed most relevant are covered.

Pathological gambling (PG)

Neurobiological responses to cue induction and behavioral tasks assessing cognitive control, simulated gambling, impulse control, risk/reward decision-making and reward processing have been reported in PG. Findings demonstrating similarities and differences between PG and substance addictions have been reviewed recently¹⁸.

Brain function in PG

Most neuroimaging studies have implicated frontal cortical areas and the striatum, as well as other regions. Generally, findings regarding brain function underlying cognitive tasks have been more consistent than cue-induction findings.

Cue-induction studies suggest dysfunction in frontal areas, although the precise nature of the dysfunction is unclear. In cue-exposure tasks, PG (versus control) participants have shown reduced activation in ventrolateral and ventromedial prefrontal cortices (vlPFC and vmPFC^7,34), although other cue-presentation studies in problem gamblers³⁵ and PG³⁶ have shown increased frontal activations. Apparent differences in findings across studies may relate to task design and analytic approaches. Studies with imaging conducted during cognitive tasks have more consistently shown decreased activity in frontal areas such as the vmPFC in PG^37-40 although increased frontal activation in problem/PG has also been reported^41,42.

Multiple studies implicate the striatum in PG. Decreased ventral striatal glucose metabolism and increased metabolism in the dorsal striatum at resting state have been found among PG patients with co-morbid bipolar disorder⁴³. However, in PET (positron emission tomography) studies at resting state, no significant differences have been found between PG and healthy controls in D2-like receptor^44,45 or serotonin 1B receptor availability in the ventral and dorsal striata, although in the latter case receptor availability correlated with problem-gambling severity in ventral striatum/pallidum⁴⁶. In functional-magnetic-resonance-imaging (fMRI) studies during gambling-cue exposure, decreased activation has been observed in the ventral⁷ and dorsal striatum⁴⁷ in PG (versus controls); however, there have also been negative results in the ventral striatum in PG/problem gambling samples^35,36. Regarding activity associated with task performance, most findings indicate decreased ventral activity in PG (versus non-PG)^38,40,48 with some evidence of elevated dorsal activity^42,48. Some differences in findings among studies are likely attributable to the specific tasks used. Also, differences pertaining to ventral striatal activity may relate to subject groups as some studies involve problem gambling⁴⁹ or mixed problem gambling/PG groups⁴¹ who may have different biological responses. Findings from Linnet et al.^44,45 suggest individual differences in that the PG sample was divided about evenly between those who showed and did not show elevated dopamine release in the ventral striatum during the Iowa Gambling Task. Limited findings with tasks related to impulsivity have not shown significant differences in striatal activation between PG and controls^50,51.

Regarding other brain regions, PG subjects (versus controls) differ in ACC activity following gambling-cue exposure^7,34. Relatively diminished insular activation in PG during cue presentation⁷ and reward processing has been reported⁴⁰. Relatively poor white matter integrity has been related to impulsivity⁵² and has been found among those with PG compared to controls in areas including the corpus collosum^53,54. Negative results have been found for white and gray matter volume differences between PG and controls⁵³.

In summary, most imaging findings in PG have implicated frontal cortical areas and the striatum. Tasks relating to risk/reward, gambling and cognitive control typically show reduced activity in PG in frontal areas and ventral striatum more consistently. Early results suggest reduced insula activity and poor white matter integrity in PG.

Neurotransmitter activity in PG

Most findings relate to dopamine and serotonin, although other neurotransmitters have been implicated. While dopamine dysfunction has been hypothesized for PG⁵⁵, findings have been less conclusive. Data^44,45 suggest individual differences in PG and control groups in dopamine release during the Iowa Gambling Task but no baseline between-group differences regarding D2-like receptor availability. Although PG and control groups showed similar dopamine release during slot-machine-task performance, dopamine release correlated with problem-gambling severity in PG.⁵⁶ Amphetamine administration increased motivations to gamble among problem gamblers⁵⁷. The D2-like antagonist haloperidol has also been associated with increased gambling motivations in PG⁵⁸, although individual differences appear important⁵⁹. Individual differences may explain negative clinical-trial findings with D2-like antagonist drugs^60,61.

Findings from neurochemical studies with varied methods suggest differences in serotonergic function between PG subjects and controls^18,62-67. Clinical-trial findings involving serotonin-reuptake inhibitors (SRIs) and a 5HT2-receptor antagonist have been negative or mixed though^60,61,68-72. While neurochemical studies indicate serotonergic dysfunction in PG, mixed clinical findings suggest important individual differences.

Regarding other neurotransmitters, multiple positive clinical-trial findings with opiate antagonists^73-76 (see ⁷⁷ for negative results) suggest opioidergic involvement in PG. Preliminary evidence of efficacy for medications that alter glutamate neurotransmission^78,79 suggest that glutamate may contribute to impulsive and compulsive behaviors and treatment outcome in PG⁷⁹. Elevated levels of adrenergic agents and their metabolites have been observed in PG^80,81. Norepinephrine levels increase in problem gamblers during gambling⁸². Blunted growth hormone responses to clonidine has been observed in PG⁸³, which may reflect elevated noradrenergic secretion.

Family history/genetics in PG

Twin studies suggest that genetic factors may contribute more than environment factors to gambling problems^15,84,85. PG heritability estimates range from 50-60%¹⁵, with increasing genetic contributions seen with greater problem-gambling severity⁸⁶. Molecular studies find small, additive effects across multiple genes⁸⁷. Associations between PG and genetic variants related to dopamine transmission (e.g., DRD2) have been found^88-92 (but see⁹³ for negative results). A variant in the serotonin-transporter-gene promoter region (5-HTTLPR) has been associated with PG in males⁹⁴ and monoamine oxidase A (MAO-A) among males with severe PG^95,96. These studies have multiple limitations relating to sample size, sample characterization and analytic approaches, and these factors may relate to inconsistencies in replication.

Compulsive internet use

Brain function in compulsive internet use

In a resting-state fMRI study, increased regional homogeneity was found among compulsive internet users in frontal areas (e.g., superior frontal gyrus) and other regions (e.g., parahippocampus). Increased regional homogeneity may reflect greater synchronization among these regions. Given that many of the implicated regions are components of the “reward circuit,” these findings intimate enhanced sensitivity to reward among compulsive internet users⁹⁷.

In a small, resting-state fMRI and PET study, reduced D2-like receptor availability was found in the dorsal striatum, with negative correlations between binding potential in this region and self-reported internet addiction measures. No evidence of dysfunction in the ventral striatum was found⁹⁸.

Regarding other brain regions, the ACC was implicated in the aforementioned study of increased resting-state regional homogeneity among compulsive internet users⁹⁷. Poor white-matter integrity and gray-matter density/volume differences have been seen in compulsive internet users (versus controls). Using diffusion-tensor imaging (DTI), lower FA in orbitofrontal cortex, corpus collosum and cingulum was seen in compulsive internet users (versus controls)⁹⁹. Using MRI, lower gray-matter density was found in regions tied to emotion regulation including the ACC, posterior cingulate, insula and lingual gyrus¹⁰⁰. In a separate study, reduced FA values were found in the parahippocampal gyrus¹⁰¹ and decreased volume observed in the cerebellum, orbitofrontal cortex, dorsolateral prefrontal cortex (dlPFC) and ACC. Regional gray matter volumes correlated inversely with duration of internet addiction¹⁰¹. These findings intimate that compulsive internet use may induce gray-matter reductions or that individuals with low gray-matter volumes may be predisposed to internet addiction.

In summary, early findings suggest regional homogeneity in frontal areas, reduced D2-like receptor availability in the dorsal striatum, poor white-matter integrity and gray-matter density/volume differences affecting regions implicated in reward and emotion processing.

Neurotransmitter activity in compulsive internet use

In a small SPECT study, the dopamine transporter appeared to be expressed at lower levels in the striatum among young adult males with compulsive internet use, compared to controls¹⁰². In terms of clinical-trials results, there have been no controlled pharmacotherapy studies⁵.

Family history/genetics in internet use

Harm-avoidant problem internet users more frequently carried the short allele of a variant in the promoter region of the gene coding for the serotonin transporter (SS-5-HTTLPR), an allele also common among depressed patients¹⁰³.

Compulsive video-gaming

We have separated findings concerning video-gaming from those pertaining to internet use. However, neurobiological research on compulsive video-gaming typically involves web-based games; thus, video-game findings cannot be separated clearly from internet findings.

Brain function in compulsive video-gaming

Using resting-state PET, increased metabolism was found in the middle orbitofrontal gyrus, which might reflect compensatory cognitive processing¹⁰⁴. Reduced metabolism was found in the precentral gyrus, which might reflect insensitivity to negative consequences¹⁰⁴. In cue-exposure studies, greater pre- and post-cue changes indicative of increased activity were observed in compulsive internet users (versus controls) in the orbitofrontal cortex (OFC), medial frontal cortex and dlPFC¹⁰⁵. In a subsequent study, greater pre-/post-cue changes were observed in the dlPFC among current compulsive players compared with controls¹⁰⁶. Pre- and post-treatment fMRI during cue-induction was incorporated into an open-label bupropion trial¹⁰⁷. Similar to other studies, stronger activity was found in the dlPFC (versus controls), with dlPFC activity declining after the 6-week treatment period. In an fMRI study tied to a computer-based guessing task involving monetary wins and losses, greater activation in the OFC was found on win trials among compulsive internet users, attributed to higher reward sensitivity¹⁰⁸.

Regarding striatal activity, increased metabolism was found in the left caudate¹⁰⁴. Greater activity post-cue induction was found in the right NAc and right caudate in compulsive game-players compared to controls during fMRI¹⁰⁵.

The ACC and insula have also been implicated in compulsive video-gaming. In a cue-induction fMRI study¹⁰⁶, greater activity post-cue was found in the ACC among compulsive gamers. During a reward processing guessing task, decreased ACC activation was found during loss trials in compulsive video-gamers (versus controls), suggesting hypo-sensitivity to loss¹⁰⁸. Increased insular activity was found at rest¹⁰⁴. Compulsive game-players demonstrated increased volume in the thalamus but decreased volume in the inferior temporal, right middle and left inferior occipital gyri¹⁰⁹.

In summary, findings in samples of predominantly young male compulsive game-players suggest increased activity at rest, to cues and during reward processing in frontal areas, the striatum and other regions, and reduced sensitivity to loss outcomes. Findings of increased activity seem to run counter to multiple PG study findings. Areas implicated in compulsive video-gaming appear to contribute to reward processing, impulse control and memory.

Neurotransmitter activity in compulsive video-gaming

A role for dopaminergic dysfunction has been proposed¹¹⁰. Genetic findings reported below are consistent with dopaminergic contributions to compulsive video-game-playing¹¹⁰.

Family history/genetics in compulsive video-gaming

Limited molecular genetic research has been performed. Allelic variants of the DRD2 Taq1A1 allele that have been associated with altered dopamine signaling have been suggested to contribute to compulsive video-gaming. Amongst male gamers, the Taq1A1 allele was related to higher self-reported reward dependence¹¹⁰. Variants of the gene encoding catechol-o-methyl transferase (COMT) that have been implicated in dopamine transmission and addictions¹¹¹ have also been reported to be more prevalent among compulsive gamers¹¹⁰.

Compulsive shopping

Brain function in compulsive shopping

In a recent study¹¹², compulsive shoppers and healthy controls were compared on a multi-phase purchasing task¹¹³ during fMRI. During an initial product presentation phase, compulsive shoppers showed stronger activity in the NAc than did controls. During a subsequent price presentation phase, compulsive shoppers showed less activation of the insula and ACC than did controls, the latter of which was activated more strongly by compulsive buyers during the concluding decision phase.

Neurotransmitter activity in compulsive shopping

Favorable results were seen with citalopram in a small open-label trial¹¹⁴. A subsequent small trial beginning with an open-label period followed by double-blind, placebo-controlled administration among responders yielded additional positive results for citalopram¹¹⁵. These findings provided tentative support for possible serotonergic dysfunction in compulsive shopping. However, negative results with other SRIs (e.g., fluvoxamine,^116,117 escitalopram¹¹⁸) raise questions about the clinical utility of SRIs for compulsive shopping.

Family history/genetics in compulsive shopping

Limited data suggest that compulsive shoppers are more likely to have close family members with psychopathology^119,120. No differences were seen in the frequencies of two serotonin transporter gene (5-HTT) polymorphisms in individuals with and without compulsive shopping¹²¹.

Kleptomania

Brain function in kleptomania

Relatively poor white-matter integrity in ventromedial prefrontal cortical regions was seen in kleptomania¹²².

Neurotransmitter activity in kleptomania

Findings regarding serotonergic dysfunction have been inconsistent. Lower numbers of platelet-derived serotonin transporters have been reported in kleptomania^123,124, suggesting serotonergic dysfunction; however, negative findings from a small double-blind, placebo-controlled clinical trial involving open-label responders were reported for escitalopram¹²⁵. Positive results in a small double-blind trial of naltrexone¹²⁶ suggest possible opioidergic involvement.

Family history/genetics in kleptomania

Similar to compulsive shopping, limited findings indicate familial links to various psychopathologies^127,128.

Compulsive sexual behavior

Brain function in compulsive sexual behavior

Studies of sexual compulsivity have been limited. In a DTI study¹²⁹, individuals with sexual compulsivity had relatively low superior frontal region mean diffusivity compared to controls. These findings did not follow patterns of results from studies of other behavioral addictions^{53,54,99,101,122}.

Neurotransmitter activity in compulsive sexual behavior

Positive results for citalopram in a double-blind placebo controlled study of compulsive sexual behavior in homosexual and bisexual men suggest possible serotonergic dysfunction¹³⁰.

Family history/genetics in compulsive sexual behavior

Limited findings suggest a high proportion of those with compulsive sexual behavior had a parent with a similar condition¹³¹. Findings indicate tendencies for sexually compulsive individuals to have first-degree relatives with substance-use disorders (SUDs)¹³¹.

Similarities and Differences with Substance-Use-Disorder Findings

Neurobiological findings in the behavioral addictions remain scant and data are particularly sparse for compulsive shopping, kleptomania and compulsive sexual behaviors. However, available data provide evidence of underlying neurobiological impairment overall, which parallels SUD findings. Tables 1, 2 and 3 contain information comparing behavioral addiction with SUDs.

Findings of poorer white-matter integrity have perhaps been the most complementary between substance^132,133 and behavioral addictions^{53,54,99,101,122} (but see¹²⁹ for seemingly conflicting results). Cognitive-task results in SUDs^{50,51,134,135} and PG^40,50,51,136 have suggested reduced activity in frontal areas. Findings involving aspects of risk/reward decision-making (including reward processing), but arguably less so from response-impulsivity tasks, have tended to show reduced ventral-striatal activity in PG^38,40,48 and SUDs^137-140, although there have been seemingly opposing results^41,141,142. Findings have tended to show increased activity in the dorsal striatum in behavioral addictions^43,48 and SUDs^143,144.

Evidence regarding neurotransmitter activity in behavioral addictions and SUDs has tended to be complementary. Neurochemical evidence has suggested reduced dopamine transporter and D2-like receptor availability at rest^{98,102,145,146} and dopamine release during activity related to addictive behavior^147,148, although there have been seemingly conflicting results at resting state in PG^44,45 and SUDs¹⁴⁹, and individual differences seem relevant to dopamine release^44,45,150. Neurochemical findings suggest differential serotonergic function compared to controls among those with behavioral addictions^62-66,124 and SUDs^151-153. Clinical results with dopamine antagonists^{60,61,154-156} and medications targeting serotonin systems (primarily SRIs^{68-72,157-159}) have demonstrated negative or mixed findings in behavioral addictions and SUDs. Clinical results involving opioid antagonists have tended to be positive for both types of conditions^{40,45,73-76,126,160-162}. Limited results with pharmacologic probes suggest a role for glutamatergic activity in PG^78,79 and SUDs^163,164. Neurochemical and clinical findings suggest a possible role for norandrenergic activity in PG^80-83 and SUDs^165-167.

Genetic (especially molecular) and family-history evidence is limited for behavioral addictions. However, available evidence suggests substantial heritability for PG^15,84. For other behavioral addictions, there is evidence suggesting familial risk across psychiatric conditions^{71,110,119,120,127,128,131}. SUDs appear highly heritable as well^27,168.

Evidence from cue-induction and resting-state imaging studies have been less clear and seemingly more conflicting. Resting-state and cue-induction findings in compulsive video-gaming have suggested increased activity across multiple brain regions^104-106,169. There have been seemingly conflicting results in problem/PG and SUD cue-induction studies for both ventral striatal (gambling^7,35; SUD^{7,143,144,170}) and frontal activity^171,172. Differences across studies in participant characteristics and other methodological details may contribute to these differing results^171,172. In addition, declines in dopamine release in response to drug consumption as dependence worsens¹⁷³ may also lead to heterogeneity in ventral-striatal activity across participants in SUD studies.

In summary, data suggest neurobiological dysfunction in behavioral addictions and SUDs. Some of the more complementary results have involved white-matter integrity, brain function during cognitive-task performance, neurotransmitter activity and overall heritability.

Conclusions and future research

Research on the neurobiology and genetics of behavioral addictions has accelerated in recent years, particularly in PG, compulsive internet use and compulsive video-gaming. Gaps in knowledge remain and research on other behavioral addictions has been limited. Existing research suggests parallels between behavioral addictions and SUDs. Additional genetic research, particularly molecular, would be valuable in delineating similarities and differences among individual behavioral addictions and between behavioral addictions and SUDs. Neuroimaging has begun to provide insight regarding similarities and differences. Additional research is needed, incorporating a broader variety of cognitive tasks¹⁷⁴. While conventional approaches have been valuable, alternative analytic methods such as computational modeling¹⁷⁴ may further illustrate parallels with SUDs.

Research testing medications and therapies indicated for SUDs has only begun. Studies involving individuals with co-occurring behavioral and substance addictions could enhance our understanding of addiction and advance treatment development. Females are often excluded from or under-represented in behavioral addiction studies, particularly in existing genetic studies and research on compulsive video-gaming. Future studies should include females and examine the extent to which various phenomena pertaining to behavioral addictions apply to both genders.

Given that behavioral addictions, particularly those relating to gambling, internet use and video-gaming, appear relevant to adolescents and young adults^2,101,110, longitudinal studies would be valuable. Epidemiologic data are limited for behavioral addictions with the possible exception of PG. National and international studies assessing prevalence of multiple behavioral addictions would enhance our knowledge regarding the extent to which these conditions affect people across the lifespan. Uniformly agreed-upon diagnostic criteria and assessment instruments would facilitate comparisons across studies.

Clinical Implications.

• ■ Behavioral addictions are characterized by dysfunction in multiple brain areas and neurotransmitter systems.

• ■ Family history/genetic findings suggest heritability for pathological gambling and psychopathology risk among families of individuals with behavioral addictions

• ■ Findings suggest parallels between neurobiological and genetic findings in substance and behavioral addictions.

• ■ Data support the conceptualization of excessive engagement in non-substance behaviors as addictions.

Limitations:

• ■ Existing data in non-substance or behavioral addictions are limited.

• ■ Data are particularly limited for compulsive shopping, kleptomania and compulsive sexual behavior.

• ■ Genetic findings are particularly preliminary and sparse.