Compulsive Sexual Behaviour disorder CSBD is “a persistent pattern of failure to control intense, repetitive sexual impulses or urges resulting in repetitive sexual behaviour” (1). Excessive sexuality has been discussed as a pathological phenomenon since the first centuries of the first millennium (2). The characteristics of the disorder and its elements have been the subject of a great deal of debate in the past few decades. Different terms were assigned to describe this phenomenon. CSBD, excessive sexual behaviour, hyper-sexuality or sexual addiction (SA), are all different labels for the same phenomenon, reflecting different theoretical frameworks for understanding non-paraphilic undue sexual behaviour. The behaviour those concepts represent is maladaptive sexual behaviour that takes a lot of time daily and persists despite adverse consequences and despite efforts to stop it (3). The literature is still inconclusive regarding the prevalence and classification of CSBD. The main goal of this paper is to empirically examine three major elements of addiction and whether they are present in CSB, namely mood modification, withdrawal, and sensitization. We will first review the evolution of this classified disorder, then portray the core elements of addiction, followed by a literature review that associates CSB with these elements.

Orford (4) challenged the traditional views on sexual behaviour, arguing against labelling it as normal or exaggerated. Rather he emphasized the conflict between sexual desires and life demands. Carnes (5) introduced the concept of sexual addiction, sparking ongoing theoretical debates. Others, like Coleman (6), rejected the idea of viewing sex as a drug of choice and saw the hyper-sexuality as a compulsive reaction to an inner obsessive state. Given the enjoyable element of sex, Barth and Kinder (7) rejected the compulsive perspective and viewed hyper-sexuality as an expression of impulse control disorder.

In 2010, Kafka (8) proposed the inclusion of hypersexual disorder as a distinct diagnosis in the upcoming Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; 9). When considering its categorical placement in the DSM-5, he suggested that the term “compulsive,” while suitable for describing certain features of the condition, deviates from the conventional DSM-based understanding of obsessive-compulsive spectrum disorders. Conversely, considering hypersexual disorder as a primary impulse disorder could potentially conflict with the current classification of other seemingly analogous biologically mediated appetitive behaviour disorders, such as bulimia nervosa (eating disorders) or hypersomnia (sleep disorders). In 2018 the ICD, in its 11th edition ([ICD-11]; 1) introduced the diagnosis of CSBD as “a persistent pattern of failure to control intense, repetitive sexual impulses or urges resulting in repetitive sexual behaviour”. Yet, this conceptualization didn’t put to rest the debates and different standpoints that are still being held, the addiction model is still being advocated, and the role of impulsivity over compulsivity is still being discussed (10–13). While compulsivity, impulsivity, and addiction, share some features, they differ in profound others. Addiction is characterized by withdrawal and tolerance phenomena that are not characterized by impulsivity and compulsivity (9).

Black et al. (14) discovered that intrusive and repetitive sexual fantasies were prevalent among individuals with CSBD, which supports classifying CSBD as a specific variant of OCD. Fuss et al. (15) found that CSBD in OCD tends to coexist with other impulsive, compulsive, and mood disorders, supporting its conceptualization as a compulsive–impulsive disorder. Bőthe et al. (16) found that impulsivity, rather than compulsivity, had a stronger correlation with hyper-sexuality in a large community sample. Levi et al. (17) revealed a connection between obsessive-compulsive symptoms and CSBD among individuals using websites to find sexual partners. They also found that impulsivity was linked to increased online pornography consumption, which in turn played a role in the development of CSBD among those who used the internet for pornography. The resulting picture remains ambiguous as to a specific concept that summarizes what is covered under the name CSBD. In the following set of studies, we will examine whether CSB fits the addiction model, namely whether CSB meets the core elements of ‘behavioural addiction’ as described below.

Traditionally, the concept of addiction was used to involve using exogenous substances, namely drugs (e.g. 18, 19). The release of the DSM-5 broadens the definition of addiction by reclassifying and relabelling Pathological Gambling. Previously categorized under impulse control disorders, Pathological Gambling is now recognized as Gambling Disorder (GD) and classified alongside drug addictions under the category of “Substance-Related and Addictive Disorders” (9, 20). A similar transition occurred in the ICD-11 (1), where Pathological Gambling was reclassified under “Disorders due to Addictive Behaviours,” thus acknowledging the presence of behavioural addictions. However, the ICD-11 (1) classified CSBD under impulse control disorders, and the phenomenology described seems to combine those of addiction, impulse control disorders, and compulsive characteristics. The unification of substance addiction and behavioural addiction under the same category has recently been reinforced by findings from Di Carlo et al. (21), who identified a significant association between frequent substance use and high-risk gaming. The authors conclude that there is a common neurobiological vulnerability for both gaming and substance use.

Griffiths (22) has suggested using six components that constitute every addiction, whether related to substances or behaviours. Those components are salience, mood modification, tolerance, withdrawal, conflict, and relapse. Griffiths’ (2005) components overlap with the DSM-5 (9) criteria for GD. Since the only precedent for behavioural addiction is GD, we will use its criteria as a touchstone to evaluate CSBD. Amongst Griffiths’ six components of addiction, the ICD-11 requires for the diagnosis of CSBD, only ‘salience’ (“repetitive sexual activities becoming a central focus of the person’s life”), conflict (“Neglecting health and personal care or other interests, activities, responsibilities, and continued repetitive sexual behaviour despite adverse consequences”), and relapse (“Numerous unsuccessful efforts to significantly reduce repetitive sexual behaviour”), while ‘mood-modification’, ‘tolerance’ and ‘withdrawal’ are absent. Anecdotal reports based on clinical experience with CSBD patients support the idea of seeing CSBD as an addiction, but as described above the field suffers from a lack of systematic empirical evidence supporting the endorsement of those six components by CSBD patients. Below, we will review the three elements omitted by the ICD i.e., mood modification, tolerance, and withdrawal which are at the core of addiction disorder. We will discuss their meaning, and the current data attributing them to CSBD.

Mood often plays a pivotal role in substance abuse, with individuals using drugs or alcohol to self-medicate and alleviate emotional distress. The relationship between mood, mental state, and addiction is multifaceted. Individuals with higher levels of alexithymia (difficulty identifying and expressing emotions) and lower interoceptive sensibility (the ability to perceive internal body signals) are more prone to problematic internet use (PUI). This suggests that challenges in emotional awareness and bodily perception may contribute to internet addiction, as those with emotion identification difficulties may turn to the internet for coping (23). Additionally, higher levels of obsessional impulses and impulsivity are significantly associated with increased PUI (24). Mood modifications are not just precursors to substance use or dysfunctional behaviours but are fundamental components of the addiction framework, emerging and evolving as the addiction develops. According to Koob and Volkow, (25), the addiction cycle comprises three stages: binge or intoxication, withdrawal and negative affect, and preoccupation or anticipation. During the withdrawal stage, individuals experience negative emotions like anxiety and dysphoria, which can trigger relapse as they seek relief through substance use.

Studies have shown that adverse emotional states can heighten cravings for substances, such as nicotine, cocaine, alcohol, and heroin (26–29). Behavioural addictions like gambling or hypersexual behaviour may serve as coping strategies for managing unpleasant emotions, as highlighted by Carnes (30) and Griffiths (22). It is argued that individuals with CSBD frequently employ sex as a strategy for mood regulation or coping with negative emotions (31–33).

The DSM-5 defines tolerance, as either (a) a need for markedly increased amounts of a substance to achieve intoxication or a desired effect or (b) markedly diminished effect with continued use of the same amount of a substance. Karila et al. (34) describe a few features of CSBD that could suggest tolerance. Coleman-Kennedy and Pendley (35) claim that just as in substance addiction, individuals with CSBD experience the continual need to expand the time spent in sexual activity to ease the emotional pain. The related concept of ‘incentive salience’ suggests that people with addictive disorder experience intensified cravings or desire for addictive stimuli rather than increased pleasure from them, as outlined by Robinson and Berridge (36). This theory suggests that people with addictive disorder are triggered by minor cues associated with addictive stimuli, making the incentive more salient and placing them in a position of more ‘wanting’. Nevertheless, their level of ‘liking’ remains indifferent with not enough enjoyment to satisfy the increased ‘wanting’.

In their research, Brand et al. (37) found that ventral striatum activity, observed during engagement with preferred pornographic images, is correlated with symptoms of Internet pornography addiction. This brain region, recognized for its involvement in reward anticipation and craving, plays a pivotal role in linking the processes of wanting and liking, with incentives calibrated based on their interactive consequences (38–40). Extended drug use, as highlighted by Robinson and Berridge (39), may result in compulsive drug seeking (i.e., wanting) and drug taking without a corresponding enjoyment (i.e., liking) upon consumption.

“It is further proposed that sensitization of the neural systems responsible for incentive salience (for ‘wanting’) can occur independently of changes in neural systems that mediate the subjective pleasurable effects of drugs (drug ‘liking’) and of neural systems that mediate withdrawal. Thus, sensitization of incentive salience can produce addictive behaviour (compulsive drug seeking and drug taking) even if the expectation of drug pleasure or the aversive properties of withdrawal are diminished and even in the face of strong disincentives, including the loss of reputation, job, home and family (39)”. In accordance with this theory, regarding CSBD, Voon et al. (41) reported a dissociation in individuals with problematic pornography use, where a high level of wanting was not necessarily associated with an equivalent degree of liking. In their study, Gola et al. (42) observed a similar pattern, reinforcing the notion of a disconnection between wanting and liking in individuals with problematic pornography use. While some (43) view ‘tolerance’ and ‘incentive salience theory’ as two competing explanations for the need for increasing consumption by people with addictive disorder, Robinson and Berridge (36) see these two mechanisms as complementary and mutually compatible. The literature currently lacks a comprehensive analysis of the relationship between ‘liking’ and ‘wanting’ and their connection to sexual cue-reactivity. While previous studies have primarily concentrated on group differences observed in neuroimaging indices, our study aims to predominantly examine the behavioural response to sexual stimuli. Addressing these gaps through research may provide crucial insights for accurately identifying CSB.

Withdrawal symptoms encompass unpleasant physical or emotional sensations that arise upon the discontinuation or reduction of a specific activity or substance (44). In substance use disorder this is characterized by neuro-adaptive changes at the transmitter-receptor level. Withdrawal symptoms in behavioural addictions like pornography watching necessitate a distinct explanation, as they differ from the neuro-adaptive changes seen in substance dependence. Piper (45) suggested that addictive behaviours, such as tobacco (or pornography consumption), become ingrained daily rituals triggered by internal or external circumstances. Avoiding those rituals leading to emotional and somatic withdrawal symptoms particularly when used as a self-regulation method during stressful events. Wray and Dickerson (46) revealed significant mood and behavioural disturbances akin to alcohol addiction during the recovery phase from GD. Subsequent research, including studies by Blaszczynski et al. (47) Cunningham‐Williams et al. (48), and Rosenthal and Lesieur (49), has consistently identified withdrawal phenomena in individuals with GD.

In the context of CSBD, there is a widespread assertion that withdrawal symptoms manifest when individuals attempt to avoid or decrease their sexual activity (e.g., 34, 50–52). Those assertions are often based on general impressions rather than methodological research, suggesting a need for further empirical investigation in this area.

The literature so far describes a debate over whether to classify CSB as an impulse control disorder or as a behavioural addiction. Sassover and Weinstein (13) have presented a debate paper and argued that, due to a lack of validated experiments and data, there is currently insufficient evidence to support CSB as a behavioural addiction. The following studies will examine three components that are essential for a behavioural addiction model, namely mood modification, withdrawal and sensitization or tolerance, in the hope of clarifying the issue of the classification of CSB.

The Institutional Review Board of Ariel University approved the study, approval number 2020930. The research sample was recruited through open “sexaholic-anonymous” meetings and social media groups. For their participation and time loss, a small monetary compensation was promised. All participants signed an informed consent prior to their participation. Overall, 44 people participated in the study. However, since only one participant identified herself as a woman, she was excluded from the study. Thus, the final sample included 43 participants between the ages of 18-41 with an average age of 25.81 (SD = 5.48), all of whom males, as mentioned. Regarding the participants’ sexual orientation, 30 (69.8%) participants identified as heterosexuals, 7 (16.3%) identified as homosexuals, and 6 (14%) chose not to disclose their orientation. Participants were divided into two groups based on their scores on the Hebrew Bergen-Yale Sex Addiction Scale (HBYSAS). The HBYSAS, which reflects Griffiths (22) six addiction components, assessed sexual behaviours and cognitions over the past year using a five-point Likert scale (0 = very rarely to 4 = very often). Those scoring 18 and above classified as CSB and those scoring below 18 classified as non-CSB. The CSB group consisted of 22 participants with a mean score of (M = 22.64) and a standard deviation of (SD = 2.15). The non-CSB group consisted of 21 participants with a mean score of (M = 9.76) and a standard deviation of (SD = 5.59). Among the participants in the non-CSB group, 3 received a score of 0 and were classified by the questionnaire as “no sex addiction.” 2 participants received a score between 1 and 6 and were classified as “low sex addiction risk,” while 16 participants received a score between 7 and 17 and were classified by the questionnaire authors as “moderate sex addiction risk”. No significant differences between groups were found regarding years of education, F (1, 41) = 0.96, p = .33, η2p = .02 (M = 11.81, SD = 3.31 for non-CSB participants; M = 12.59, SD = 1.71 for CSB participants). The participants allocated to the CSB groups were significantly older (M = 27.59, SD = 4.35) than the non-CSB participants (M = 23.85, SD = 6.01), F (1, 40) = 5.41, p = .03, η2p = .12. In the non-CSB group, 15 participants (71.4% of the group’s participants) identified as heterosexuals, 3 (14.3%) identified as homosexuals, and the rest 3 (14.3%) chose not to identify their orientation. In the CSB group, 15 participants (68.2%) identified as heterosexual, 4 (18.18%) identified as homosexual, and the remaining 3 (13.64%) chose not to identify. The results of the Fisher’s exact test, p = 1, indicated a non-significant association between group affiliation and sexual orientation.

Hebrew Bergen-Yale Sex Addiction Scale (HBYSAS; 53). The HBYSAS questionnaire is the Hebrew version of the Bergen Yale Sex Addiction Scale (BYSAS) developed by Andreassen et al. (54). The BYSAS was based on the six components of addiction proposed by Griffiths (22) as mentioned above. Both the HBYSAS and ICD-11 criteria for CSB focus on uncontrolled sexual behaviour, repeated failed attempts to reduce it, and the negative consequences of such behaviour. However, the HBYSAS uniquely addresses tolerance (increased urges over time) and withdrawal (distress when unable to engage in sexual behaviour), which are not part of the ICD-11 definition. The ICD-11 emphasizes the behaviour’s centrality in the individual’s life and its persistence despite lack of satisfaction, while the HBYSAS includes the use of sex as a coping mechanism for negative emotions. The scale assessed sexual behaviours and cognitions during the past year using a five-point Likert scale ranging from 0 = very rarely to 4 = very often. A sample item asked how often during the past year the participant spent a lot of time thinking about sex, masturbation, or planned sex.

The total score was acquired by summation and ranged between 0 and 24. A total score of 0 is classified as “no risk of sex addiction”, 1 to 6 as “low risk of sex addiction”, 7 to 17 “moderate risk for sex addiction” and score above 18 as “high risk of sexual addiction”. The score of 18 was designated as the cutoff by the questionnaire authors and was used in the current study. According to the authors’ recommendation, a cut-off score of 18 or above was set to ensure relatively high scores (“often” or “very often”) on most items, combined with sufficiently significant scores or at least some presence on the remaining items. Thus, a composite score of 18 or above indicates a more pervasive presence of the symptoms, aligning with the operational definition of addiction in the literature. The original BYSAS presented with good internal consistency (Cronbach’s α =.83), the internal consistency of the HBYSAS was acceptable (Cronbach’s α = 0.79; 53), and in the current study, the internal consistency was excellent (Cronbach’s α = .95).

Positive and Negative Affect Schedule (PANAS; 55). The PANAS is a widely used self-report measure developed to assess two major domains of affect: Positive Affect (PA) and Negative Affect (NA). The PANAS comprises 10 items measuring PA (e.g. alert, inspired, and enthusiastic) and 10 items measuring NA (e.g. distressed, upset, and guilty). Participants rate the degree to which they endorse each item on a five-point Likert scale (1 = very slightly or not at all; 5 = extremely). Items relevant for each affect are summed separately and yield PA and NA total scores. Higher scores represent greater endorsement of the relevant construct.

The validity and reliability of PANAS has been demonstrated in several studies Cronbach’s alpha in the original version was 0.89 for the PA scale and 0.85 for the negative scale (55). In the current study we used the Hebrew version that was developed and validated by Anaby et al. (56). Internal consistency in our study ranged between 0.81 and 0.85 for PA and between 0.76 and 0.80 for the NA.

Pornography Craving Questionnaire (PCQ; 57). The 12-item PCQ measures present craving for pornography thorough five elements (i.e., perceived control over use, current desire to use, psychophysiological reactivity, intentions to use, and mood changes). Sample items include “If the situation allowed, I would watch porn right now” and “If I watched porn right now, I would have difficulty stopping.” Respondents are asked to indicate how strongly they agree with each item using seven points Likert scale, from 1- “disagree completely”, to 7- “agree completely”. The general scores of the PCQ were obtain by summation of the scores of each item, ranging from 12 to 84. Higher scores indicate greater current craving for pornography. The internal consistency for the PCQ was excellent (Cronbach’s α = .91) in the original as well as in subsequent samples (57). The PCQ was also shown to have convergent, criterion, and discriminant validity (58). We translated the questionnaire to Hebrew for the purpose of the current study using a translation back translation method. In the current study, the internal consistency was excellent for all the four times the PCQ was used (Cronbach’s α ranging between.91 and.94).

All participants filled in their demographics regarding their, age, gender, sexual orientation, and years of education. Then they were asked to fill out the HYBSAS, PCQ and the PANAS for baseline measurement. Participants with a score of 18 or above in the HYBSAS were assigned to the CSB group, and the rest were assigned to the non-CSB group. The PCQ was used to assess the level of craving for pornography, while the PANAS was employed to determine whether the mood induction was successful. The participants underwent mood induction manipulation, by watching three short movies of two minutes length each. One movie included a comic routine to induce pleasant mood; another movie was an interview with relatives to casualties after a terror attack to induce an unpleasant mood; and another movie was a media report regarding a nature resort, in order to induce a more natural mood. The stimulus movies’ order was counterbalanced. After each movie, participants filled out the PCQ and PANAS. After completion, the participants were thanked, debriefed, and were given the promised compensation.

Prior probing the research question, the manipulation effectiveness was evaluated, by conducting two mixed design 4 by 2 [condition by CSB affiliation] ANOVAs, which in both the condition was the within-subject factor. Each ANOVA was conducted separately for the negative and the positive mood scores obtained by the PANAS. For the research question, a mixed design 4 by 2 [condition by research group] ANOVA was conducted, in which the condition was the within-subject factor, and the average score in the PCQ was the dependent variable. The data were analysed using IBM SPSS Statistics (Version 29). Post hoc analyses utilizing paired t-tests with Bonferroni correction were performed for every main or interaction effect. Additionally, for each effect, post-hoc power analyses were computed using G*Power software, version 3.1 (59).

Prior to testing the research hypotheses, the manipulation effectiveness was evaluated. This was achieved by conducting two mixed design 4 X 2 [condition X CSB affiliation] ANOVAs, which in both the condition was the within-subject factor. Each ANOVA was conducted separately for the negative and the positive mood scores obtained by the PANAS.

Regarding the positive mood scores, Mauchly’s test indicated that the assumption of sphericity had been retained, Mauchly’s W = .78, χ2 (5) = 10.11, p = .07. The main effect for condition was significant, F (3, 123) = 19.48, p <.001, η2p = .32, with power value of 1. Post hoc analyses using t tests for paired samples with Bonferroni correction (α = .05/6 = .008) showed that the average positive affect score reported after watching the unpleasant movie (M = 25.98, SD = 8.05) was significantly lower than the other conditions (M = 31.44, SD = 5.85 for baseline measurement; M = 32.30, SD = 6.85 after pleasant movie; M = 31.16, SD = 6.93 after neutral movie), t(42) = -5.01, p <.001, Cohen’s d = -.76 for the comparison between the average PA score measured after the unpleasant stimulus and the one measured at the baseline level; t(42) = -6.33, p <.001, Cohen’s d = -.96 for the comparison between the average PA score measured after the unpleasant stimulus and the one measured after the pleasant stimulus; and t(42) = -6.04, p <.001, Cohen’s d = -.92 for the comparison between the average PA score measured after the unpleasant stimulus and the one measured after the neutral stimulus. The comparison between the average PA score measured at the baseline and the one measured after the positive stimulus was non-significant, t(42) = -1.12, p = .27, Cohen’s d = -.17, as well the comparison between the average PA score measured at the baseline and the one measured after the neutral stimulus, t(42) = 0.30, p = .76, Cohen’s d = .05. The comparison between average PA score measured after the neutral stimulus and the score measured after the pleasant stimulus proved also to be non-significant, t(42) = -1.47, p = .15, Cohen’s d = -.22.

The analysis also showed main effect for the between subjects factor, F (1, 41) = 5.51, p < 0.05, η2p = .12, with power value of.90, meaning that non-CSB participants reported for a significantly more positive affect in average (M = 32.29, SD = 6.08), compared to CSB participants (M = 28.25, SD = 5.18). No significant interaction of condition X group was found, F (3, 123) = 0.08, p = .97, η2p = .002, with power value of.04. The results are illustrated in Figure 1.

Figure 1. Average PANAS Positive Affect Scores by Mood Induction Conditions. CSBD, compulsive sexual behavior disorder (error bars show standard deviations).

Regarding the negative mood scores, Mauchly’s test of sphericity was non-significant, Mauchly’s W = .80, χ2 (5) = 8.72, p = .12. The mixed design ANOVA yielded a significant main effect for condition, F (3, 123) = 3.00, p <.05, η2p = .07, with power value of 1. Post hoc analyses using t tests for paired samples with Bonferroni correction (α = .05/6 = .008) showed a trend toward significance when the average NA score measured during baseline measurement (M = 22.91, SD = 7.79) and the one measured after the neutral stimulus (M = 19.98, SD = 4.82), with the baseline score was higher, t (42) = 2.60, p = .01, Cohen’s d =.40.

The rest of the comparisons were non-significant. Specifically, The comparison between the average NA score measured after the unpleasant stimulus (M = 22.37, SD = 7.64) was not significantly different than the one measured at baseline, t(42) = -0.40, p = .69, Cohen’s d = -.06; than the one measured after the positive stimulus (M = 20.53, SD = 7.85), t(42) = 1.53, p = .13, Cohen’s d = .23; or from the one measured after the neutral stimulus (M = 19.98, SD = 4.82), t(42) = 2.36, p <.05, Cohen’s d = .36. Non-significant resulted also emerged when the average NA score, measured after the positive score was compared with either the score measured at baseline, t (42) = 2.39, p <.05, Cohen’s d = .37; or the score measured after the neutral stimulus, t (42) = -0.49, p = .63, Cohen’s d = -.08.

The ANOVA revealed a significant main effect for the between-subjects factor, F (1, 41) = 15.46, p <.001, η2p = .27, with power value of 1, meaning that the participants who affiliated with CSB reported an average score of NA (M = 24.22, SD = 4.82), which was higher from the scores of the participants who had not CSB (M = 18.55, SD = 4.62). No significant interaction of condition X CSB affiliation was found, F (3, 123) = 0.56, p = .65, η2p = 0.01 with power value of.12.

A mixed design 4 X 2 [condition X research group] ANOVA was performed, in which the condition was the within-subject factor, and the average score in the PCQ was the dependent variable. Mauchly’s test of sphericity was non-significant, Mauchly’s W = .86, χ2 (5) = 6.16, p = .29. A significant main effect for the research groups was found, F (1, 42) = 65.68, p <.001, η2p = .62, with a power value of 1, meaning that the average PCQ score for the participants in the CSB group was higher than the score of the participants in the non-CSB group. The means and standard deviations regarding this main effect are listed in Table 1, presented in the next page.

Table 1. Means and standard deviations of PCQ scores for research groups by experimental condition.

The ANOVA yielded also a significant main effect for condition, F (3, 123) = 20.09, p <.001, η2p = .33 with a power value of 1. Post-hoc analyses of the simple effects, using t tests for paired samples with Bonferroni correction (α = .05/6 = .008) revealed that the average PCQ score obtained during the baseline measurement was significantly higher than the average score obtained after both the pleasant stimulus, t (42) = 3.27, p <.01, Cohen’s d = .50; and the unpleasant stimulus, t (42) = 6.34, p <.001, Cohen’s d = .97. However, the comparison between the average PCQ score obtained during the baseline measurement and the average score measured after the neutral stimulus was not significant, t(42) = 2.17, p <.05, Cohen’s d = .33. Post-hoc analyses revealed also that the PCQ average score, measured after the presentation of a pleasant stimulus was significantly higher than the score obtained after the unpleasant stimulus, t(42) = 4.28, p <.001, Cohen’s d = .65, but not significantly different from average score obtained after the neutral stimulus, t(42) = -1.13, p = .26, Cohen’s d = -.17. Finally, the post-hoc analyses revealed that the PCQ average score, measured after the neutral stimulus were significantly higher than the score obtained after the unpleasant stimulus, t(42) = -5.73, p <.001, Cohen’s d = -.87. The means and standard deviations regarding this main effect are also listed in Table 1.

Lastly, the ANOVA revealed a significant condition by CSB interaction effect, F (3, 123) = 3.80, p = 0.01, η2p = .09 with a power value of.76. Post-hoc analyses of the simple effects, using t tests for paired samples with Bonferroni correction (α = .05/12 = 0.004), revealed that regarding the participants who did not qualify for CSB, only one comparison was significant: The average PCQ score measured during baseline was higher than the score measured after the unpleasant stimulus, t(20) = 4.09, p <.001, Cohen’s d = .89. The average PCQ score measured during baseline was not significantly different than PCQ score measured either after the pleasant or the neutral stimulus, t (20) = 3.05, p <.01, Cohen’s d = .67; and t (20) = 1.90, p = .07, Cohen’s d = .42 accordingly. A non-significant difference was found between the average PCQ scores measured after the pleasant stimulus and the score measured after both the unpleasant stimulus, t(20) = 2.85, p = .01, Cohen’s d = .62; and the neutral stimulus, t(20) = -1.39, p = .18, Cohen’s d = -.30. Finally, for the non-CSB participants, A non-significant difference between the average PCQ scores measured after the unpleasant stimulus and the scores measured after the neutral stimulus, t(20) = -3.08, p = .01, Cohen’s d = -.67.

For the participants who qualified to CSB, post-hoc analyses revealed that the average PCQ score measured the unpleasant stimulus was significantly lower than the average score obtained after each of the three other conditions – pleasant stimulus, t(21) = 4.06, p = .001, Cohen’s d = .87; neutral stimulus, t(21) = -5.63, p <.001, Cohen’s d = -1.20; and baseline measurement, t(21) = 5.32, p <.001, Cohen’s d = 1.13. The remaining post-hoc comparisons of the average PCQ score did not demonstrate a significant difference. Specifically, the difference between the average PCQ score obtained during baseline was not statistically different than the scores obtained both after the pleasant and the neutral stimulus, with values of t(21) = 1.89, p = .07, Cohen’s d = .40; and t(21) = 1.26, p = .22, Cohen’s d = .27 accordingly to each comparison. The comparison between the average PCQ scores obtained after the pleasant and neutral conditions proved to be non-significant as well, t (21) = -0.66, p = .52, Cohen’s d = -.14. The means and standard deviations regarding the interaction’s simple effects are also listed in Table 1, mentioned above. The results are illustrated in Figure 2.

Figure 2. Average PCQ Scores by CSBD diagnosis for Each Mood Induction Condition. PCQ, pornography craving questionnaire; CSBD, compulsive sexual behavior disorder (error bars show standard deviations). ***p <.001.

The findings of this study indicate a discernible causal link between positive affect and pornography craving (PC) in both groups. Furthermore, both CSB and non-CSB individuals exhibited reduced PC following exposure to the negative movie. This reduction was statistically significant for the non-CSB group when compared to baseline. However, for the CSB group, this effect was significant not only compared to baseline but also in comparison to other conditions. Moreover, the impact of the negative movie was notably more pronounced in the CSB group, leading to PC levels that were no longer higher than those of the non-CSB group.

The positive and negative affect scales (PA, NA) as represented in the PANAS are mutually independent and can represent two different aspects of bad mood (55). Low PA is traditionally associated with a lack of liveliness, loss of pleasure, and depressed experience, while NA represents irritability and nervousness as reflected in an anxious state of mind (55, 60). The current research does not allow us to draw conclusions regarding the relationship between an anxious mood and PC, However, it can be demonstrated that a low PA leads to a decrease in PC in both groups.

Inducing depression has already been found to reduce sexual desire (61). Previous studies using PANAS in general population have all shown that low PA is associated with lowered sexual desire. The findings regarding NA are less conclusive but all indicate a weak to no relationship or even a positive relationship so that anxiety leads to sexual interest (62–65). Grov et al. (66) examined the relationship between mood and sexual interest using self-report diaries, and additionally compared between CSB and non-CSB individuals. Grov et al. (66) found that anxiety and nervousness were associated with increase in sexual activity, particularly among those without CSB. Like in our study, Grov et al. (66) found that depression and discouragement were associated with reduced sexual activity for both groups but unlike our study the effect was less pronounced for the CSB group. Grov et al. (66) assumed that the depressive effect is less relevant to CSB individuals due to their pervasive sexual desire. It is important to note that Grov’s (2010) study dealt with sexual activity with other partners while our study dealt with watching pornography. Sexual activity can entail a wide range of emotions. Out of anxiety and nervousness, sex can have a calming effect of sexual release as well as possible “excitation transfer” of the anxious arousal. In depressed mood, seeking sexual activity can be utilized for intimacy and self-assurance (67). It might be intuitively argued that pornography provides no intimacy and therefore in our study low positive mood diminished PC to the minimum level for both groups, CSB and non-CSB participants. The same idea can be found by Grov et al. (66) that searching multiple partners was associated with elevated mood among non-CSB individuals and associated with depressive mood among CSB individuals.

To sum up, our study, like previous studies (68) shows that the relationship between mood and sexual activity takes different shapes depending on the specific mood, the specific sexual act. At the same time, the extent to which this relationship between mood and sexual activity distinguishes between CSB and non-CSB remains unproven.

Lastly, the results of the experiment indicate that the highest level of craving for pornography occurred in the initial state, where participants had not watched any films. It is possible that the absence of external stimuli in the initial state might have led participants to focus more on their internal desires and thoughts, thus intensifying their cravings. When occupied with watching films, their attention would have been partially diverted, potentially diminishing the immediate intensity of their cravings for pornography.

A significant limitation lies in the fact that, while the experiment specifically examined the impact of mood on the desire for pornography, the grouping of participants and the data collected do not provide information specific to pornography-related sexual behaviour but rather assess sexual behaviour in general.

HBYSAS (53). Details about the HBYSAS were mentioned in Study 1. In the current study, the internal consistency was good (Cronbach’s α = .83).

Participants filled in demographic details on a designated questionnaire, which was the same as the first study. Afterwards, they completed the HBYSAS, and as in the first study, a total score of 18 or above was set as the criteria for assignment under the CSB group. Then, participants were explained that they would be presented with an erotic movie divided into three segments with an increasing level of explicitness. Before viewing the first segment, the participant reported their desired level for consuming videotaped pornography in the present moment on a scale from 1 to 100. After watching each video segment, the participants rated both their level of enjoyment (liking measurement) and their desire to watch erotic movies (wanting measurement) on a scale ranging from 1 to 100, hence rating the liking measurement three times and the wanting measurement four times.

For the research question, a mixed design 3 by 2 [Liking measurement by research group] ANOVA and a mixed design 4 by 2 [Wanting measurement by research group] ANOVA were conducted, both in which the condition was the within-subject factor and the wanting or liking scores served as the dependent variables. The data were analysed using IBM SPSS Statistics (Version 29). Additionally, for each effect, post-hoc power analyses were computed using G*Power software, version 3.1 (59).

The second research question postulated whether the attraction towards pornography, measured by either liking or wanting it, would differ by CSB status. Regarding the liking measurement, a 3 by 2 [level of movie explicitness by CSB affiliation] mixed-design ANOVA was conducted, with the level of explicitness as the within-subject factor. Mauchly’s test of sphericity was non-significant, Mauchly’s W = 1, χ2 (2) = 0.25, p = .88.

The ANOVA yielded a significant main effect of the level of movie explicitness, F (2, 180) = 30.62, p <.001, η2p = .25, with a power value of 1. Non-significant effect for CSB affiliation was found, F (1, 90) = 1.14, p = .29, η2p = .01, meaning that the average liking score of the non-CSB participants (M = 41.07, SD = 19.07) was not significantly different than the average liking score of the CSB participants (M = 45.84, SD = 23.77). Finally, no significant interaction effect of levels of explicitness by CSB affiliation was found, F (2, 180) = 1.52, p = .22, η2p <0.05, with a power value of.38. Post hoc analyses for the explicitness level were conducted using t tests for paired samples with Bonferroni correction (α = .05/3). The analyses revealed that the average liking score obtained after watching the movie with the lowest explicitness level (M = 28.46, SD = 24.78) was lower than the average liking scores obtained after watching both the movie with intermediate explicitness level (M = 47.36, SD = 29.91), t(1, 91) = -5.73, p <.001, Cohen’s d = -.60; and high explicitness levels (M = 54.24, SD = 30.67), t(1, 91) = -7.49, p <.001, Cohen’s d = -.78. A non-significant difference was found between the liking scores of the movies with intermediate and higher explicitness levels, t (1, 91) = -1.98, p = .05, Cohen’s d = -.21. Figure 3 shows average liking scores by CSB affiliation according to movie explicitness level.

Figure 3. Average Wanting Scores by CSBD diagnosis According to the Movie Explicitness Level. CSBD, compulsive sexual behavior disorder (error bars show standard deviations).

Regarding the wanting measurement, a 4 by 2 [level of explicitness by CSB affiliation] mixed-design ANOVA was conducted, with the level of explicitness as the within-subject factor. Mauchly’s test of sphericity was non-significant, Mauchly’s W = .91, χ2 (5) = 7.96, p = .15.

The ANOVA revealed a significant effect for CSB: F (1, 90) = 6.42, p = .01, η2p = .07, with a power value of.98, meaning that the average wanting score of the non-CSB participants (M = 48.58, SD = 23.01) was lower than the average wanting score of the CSB participants (M = 59.47, SD = 24.30). The ANOVA yielded a significant main effect regarding the level of explicitness: F (3, 270) = 22.93, p <.001, η2p = .20, with a power value of 1. No significant interaction effect was found for explicitness by CSB affiliation was found, F (3, 270) = 0.96, p = .41, η2p = .01, with a power value of.25.

Post-hoc analyses for the explicitness level were conducted, using t tests for paired samples with Bonferroni correction (α = .05/6). These analyses revealed that the average wanting score obtained prior to watching any movie (M = 37.67, SD = 28) was lower than the average scores obtained after watching the movies with either intermediate (M = 56.91, SD = 28.99), t(1, 91) = -5.71, p <.001, Cohen’s d = -.60; or higher explicitness levels (M = 61.14, SD = 30.49), t(1, 91) = -7.17, p <.001, Cohen’s d = -.75. No significant differences were detected between the average wanting score obtained prior to watching any movie and the average wanting score measured after the movie with the lowest explicitness level (M = 43.32, SD = 29.52), t(1, 91) = -1.58, p = .12, Cohen’s d = -.16. The average wanting score measured after the movie with the lowest explicitness level was lower than the average scores obtained after watching with either intermediate explicitness level, t(1, 91) = -4.52, p <.001, Cohen’s d = -.47; or higher explicitness levels, t(1, 91) = -5.29, p <.001, Cohen’s d = -.55. No significant differences were found between the scores obtained after watching the movies with intermediate or higher explicitness levels, t(1, 91) = -1.43, p = .16, Cohen’s d = -.15. Figure 4 shows average wanting scores by CSB affiliation according to movie explicitness level.

Figure 4. Average Wanting Scores by CSBD diagnosis According to the Movie Explicitness Level. CSBD, compulsive sexual behavior disorder (error bars show standard deviations).

Our hypotheses in this study were based on the theory that there are two parallel processes in addiction, i.e., incentive-salience and tolerance. We postulated that if CSB is an addiction, increased explicitness of sexual content would trigger more “wanting” for both groups and that this effect would be significantly higher for the CSB group due to their sensitized system. We also postulated that in response to the same level of sexual stimuli, CSB individuals will feel less satisfaction, namely less ‘liking’, compared to control participants due to tolerance. Regarding the ‘wanting’ hypothesis, our results show that CSB individuals do experience more wanting than non-CSB individuals, but that the difference between groups was not affected by explicitness level, both groups showed elevated wanting in response to more explicit content. As for the ‘liking’ hypothesis, our results show no more ‘liking’ in CSB individuals compared with non-CSB individuals and no increase in liking as a function of increased explicitness in CSB. We found that both groups showed the same pattern of change across levels of explicitness, namely an increase in liking and wanting as the sexual material became more explicit. Our results are supportive of those reported by Voon et al. (41) who identified a similar dissociation between wanting and liking in CSB individuals. They based their analysis on the incentive sensitization theory proposed by Robinson and Berridge (39) who argued for a dissociation between wanting and liking in addiction. According to their theory, as addiction progresses, individuals desire rewards but do not find them pleasurable. However, the self-reported data of their study demonstrated a linear regression between wanting and liking that was not significantly different for CSB and non-CSB. In our study, participants wanted pornography, although their wanting did not increase as a function of the level of explicitness. They did not differ from control participants in their rating of their liking nor was there an interaction between liking and increased level of explicitness in CSB.

There are other studies that have investigated the relationships between wanting and liking in CSB. Gola et al. (42) suggested a potential discordance between liking and wanting in CSB. Their results based heightened expectations for erotic cues among individuals with CSB compared to those without, but they found no between group differences in the rated hedonic value of the actual stimuli presented later. Although sexual cues might elicit greater desire in CSB individuals it does not mean that they would experience higher satisfaction with any form of sexual stimuli. In Gola et al. (42) study, the delay and the limited sexual stimuli might have been disappointing in a way for the CSB group.

A more recent study, File et al. (69) has also studied the mechanism of incentive sensitization focusing on substance use and potentially problematic behaviours. Their study is built upon the notion put forth by Berridge et al. (70) that vivid imagery of reward could evoke measurable properties without the presence of the actual stimuli. For instance, participants were prompted to envision scenarios involving their preferred alcoholic beverage, gambling or sexua