Addictive Behaviors 44 (2015) 71–79

Contents lists available at ScienceDirect

Addictive Behaviors

Desire thinking: What is it and what drives it? Gabriele Caselli a,⁎, Marcantonio M. Spada b a b

Studi Cognitivi, Cognitive Psychotherapy School, Milano, Italy London South Bank University, London, UK

H I G H L I G H T S • • • • •

Desire thinking is a cognitive process that leads to the escalation of craving. Metacognitive beliefs may drive the activation and perseveration of desire thinking. We tested a metacognitive model of craving in four clinical and community samples. Results support interaction of desire thinking and metacognitive beliefs A key clinical implication may be: ‘desires don’t matter, the response to them does’.

a r t i c l e

i n f o

Available online 23 July 2014 Keywords: Addictive behaviors Craving Desire thinking Metacognitions Metacognitive model of desire thinking and craving Metacognitive theory

a b s t r a c t Introduction: The aim of this study was to provide an overview of the construct of desire thinking and test a metacognitive model of desire thinking and craving, based on the work of Spada, Caselli and Wells (2012; 2013), which aims to explain the perseveration of desire thinking. Method: We conducted two studies involving four clinical samples (total N = 493) and a community sample (N = 494) presenting with different addictive behaviors. The relationships among variables were examined by testing the fit of path models within each sample. Results: In the model presented it was proposed that positive metacognitions about desire thinking are associated with, in turn, imaginal prefiguration and verbal perseveration, marking the activation of desire thinking. Verbal perseveration is then associated to negative metacognitions about desire thinking and craving denoting the pathological escalation of desire thinking. Finally, a direct association between positive metacognitions about desire thinking and negative metacognitions about desire thinking would mark those occasions where target-achieving behaviour runs as an automatized schemata without the experience of craving. Results indicated a good model fit in the clinical sample and a variation in the model structure in the community sample. Conclusion: These findings provide further support for the application of metacognitive theory to desire thinking and craving in addictive behaviors. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Defining desire thinking In the Elaborated Intrusion (EI) theory of desire (Kavanagh, Andrade, & May, 2005; Kavanagh, May, & Andrade, 2009; May, Andrade, Panabokke, & Kavanagh, 2004), it has been suggested that the duration, frequency and intensity of craving, primarily as an affective and subjective response, may be the result of the combination of automatic (conditioned) and voluntary cognitive processes. According to the EI theory a variety of external and internal triggers lead to the ⁎ Corresponding author at: Studi Cognitivi, Cognitive Psychotherapy School, Viale Giardini 472/L, 41100 Modena, Italy. E-mail address: [email protected] (G. Caselli).

http://dx.doi.org/10.1016/j.addbeh.2014.07.021 0306-4603/© 2014 Elsevier Ltd. All rights reserved.

activation of automatic associations that contain information about a desired target or activity (e.g. its positive consequences or a felt sense of deprivation). When these associations intrude into awareness they are perceived as spontaneous and induce craving (Bywaters, Andrade, & Turpin, 2004; Witvliet & Vrana, 1995). The escalation and persistence of craving are dependent on the activation of a process of cognitive elaboration termed ‘desire thinking’ (Green, Rogers, & Elliman, 2000; Kavanagh et al., 2009; Tiffany & Drobes, 1990). Desire thinking can thus be conceptualized as a conscious and voluntary cognitive process orienting to prefigure images, information and memories about positive target-related experience (Caselli & Spada, 2010; Kavanagh, Andrade, & May, 2004; Kavanagh et al., 2005). Evidence indicates that desire thinking is multi-dimensional in nature, with imaginal prefiguration and verbal perseveration components (Caselli & Spada, 2011). The imaginal prefiguration component refers

72

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

to the allocation of attentional resources to target-related information and a multi-sensory elaboration in the form of anticipatory positive imagery or positive target-related memories recall. The verbal perseveration component refers to prolonged self-talk regarding worthwhile reasons for engaging in target-related activities and their achievement. Desire thinking appears to be similar to, but conceptually different from, other cognitive constructs involved in the genesis and maintenance of craving such as attentional biases, intrusive thoughts, thought suppression, rumination and worry. Firstly attentional biases are related to: (1) automatic tendencies of target-related cues to gain priority over attentional resources (Robinson & Berridge, 1993); and (2) the voluntary distribution of attentional resources in monitoring target-related cues (Field, Mogg, Zetteler, & Bradley, 2004). In both conceptualizations, attentional biases are likely to be in a mutual reciprocal relationship with further conscious elaboration, as in desire thinking, without implying it. Secondly, desire thinking, contrary to target-related intrusive thoughts or memories, implies a voluntary engagement in the elaboration of intrusive thought and it is not necessarily associated with an attempt to suppress them. Finally, desire thinking is a form of extended thinking which shares with rumination and worry features such as a self-focused attentional orientation and a perseverative nature, but it differs in terms of proportion of imagery-based elaboration and degree of focus centered on decision-making and planning instrumental behavior. 1.2. Desire thinking: a predictor of craving and a separate construct Research has shown that thinking about a desired target is closely linked to levels of craving (Green et al., 2000; Tiffany & Drobes, 1990) and induces physiological change similar to what is induced by direct experience (Bywaters et al., 2004; Witvliet & Vrana, 1995). Research has also demonstrated that desire thinking facets are active during a craving episode in individuals with alcohol abuse, nicotine dependence and problematic gambling (Caselli & Spada, 2010). In addition desire thinking has been found to have a significant effect on craving across a range of addictive behaviors in a community sample (Caselli, Soliani, & Spada, 2013), predict craving in alcohol abusers independently from level of alcohol use (Caselli & Spada, 2011), and play a role across the continuum of drinking and smoking behavior controlling for gender, age, negative affect and craving (Caselli, Ferla, Mezzaluna, Rovetto, & Spada, 2012; Caselli, Nikčević, Fiore, Mezzaluna, & Spada, 2012). Desire thinking has also been shown to be the strongest predictor of levels of problematic gambling in a clinical sample, independent from gender, negative affect and craving (Fernie et al., 2014). On similar lines desire thinking has been found to predict category membership as a problematic internet user and levels of problematic internet use in a community sample controlling for weekly internet use, negative affect and craving (Spada, Caselli, Slaifer, Nikčević, & Sassaroli, 2014). Taken together, these findings: (1) highlight the role of desire thinking in generating an escalation in frequency and intensity of craving; and (2) support the distinction between desire thinking and craving.

In view of the above the crucial question to emerge is ‘What makes desire thinking become perseverative and poorly regulated?’ Spada, Caselli and Wells (Caselli & Spada, 2011, 2013; Spada, Caselli, & Wells, 2012; Spada, Caselli, & Wells, 2013) argue that metacognition plays a central role in understanding dysregulation in desire thinking. Metacognition can be defined as any knowledge or cognitive processes that are involved in the appraisal, monitoring or control of cognition (Flavell, 1979; Wells, 2000). Theory and research in metacognition have been introduced, over the last twenty years, as a basis for understanding and treating psychological dysfunction (Wells, 2000; Wells & Matthews, 1994, 1996). Wells and Matthews propose that a style of managing thoughts and emotion that involves extended thinking (e.g. rumination and worry), threat monitoring, avoidance, and thought suppression leads to psychological dysfunction. This style is called the Cognitive Attentional Syndrome (CAS) and is problematic because it causes negative thoughts and emotions to persist, as it fails to modify dysfunctional self-beliefs, and increases the accessibility of negative information (Wells, 2000). The activation and persistence of the CAS in response to cognitive (e.g. intrusive thoughts) and affective (e.g. sense of deprivation) triggers are dependent on maladaptive metacognitions (or metacognitive beliefs). Metacognitions refer to the information individuals hold about their own cognition and about coping strategies that impact on it. Examples of metacognitions may include: “Worrying will help me cope” or “My thoughts are out of control”. In line with a metacognitive conceptualization, desire thinking may be considered as a coping strategy similar to rumination and worry and thus a central part of the CAS in addictive behaviors with maladaptive consequences including: (1) increased levels of craving and perception of being out of control; (2) increased accessibility of target-related information; and (3) interference with the regulation of craving (Caselli & Spada, 2011; Caselli et al., 2013). Recent research has also suggested that metacognitions may indeed play a role in desire thinking (Caselli & Spada, 2010, 2013). Metacognitions about desire thinking refer to the information individuals hold about their own desire thinking and desire-related thoughts. Positive metacognitions about desire thinking concern the usefulness of desire thinking in distracting from negative thoughts and emotions (e.g. “it helps not to be overwhelmed by my worries”), and in improved executive control over decisions and behaviors (e.g. “it helps to avoid bad decisions”, “it helps to have a greater control over my decisions”). Such metacognitions may be involved in the initiation of desire thinking when a target-related thought intrude into awareness. Negative metacognitions about desire thinking concern

PMDT

IP

1.3. Metacognition as a contributor to the perseveration of desire thinking Desire thinking is a common faculty of human beings and may be not maladaptive per se. Indeed desire thinking may be helpful to: (1) motivate individuals to make efforts and delay gratification in the presence of long term goals through a virtual anticipation of pleasant results; and (2) plan adequate action to reach goals despite obstacles. Desire thinking, however, can become maladaptive if and when it becomes perseverative and poorly regulated. This may happen when desire thinking is applied to non-realistic or non-reachable targets or when it is applied to targets whose achievement conflicts with other personal goals. An example of the latter is the perseveration of desire thinking about gambling when one's personal goal is the abandonment of this activity.

VP

NMDT

Craving

Fig. 1. Structure of a metacognitive model of desire thinking and craving. Note. PMDT = positive metacognitions about desire thinking; NMDT = negative metacognitions about desire thinking; IP = imaginal prefiguration; VP = verbal perseveration.

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

the uncontrollability of target-related thoughts (e.g. “I cannot stop thinking about my desires”) and loss of control over desire thinking (e.g. “thinking too much about my desires make me lose control”). These metacognitions may play a role in propagating perception of low control once a desire thinking episode has started which may possibly lead to an escalation of desire thinking and craving. 1.4. Test of a metacognitive model of desire thinking and craving Recently, Spada, Caselli and Wells (Caselli & Spada, 2011, 2013; Spada et al., 2012, 2013) have proposed a metacognitive model of desire thinking and craving (presented in Fig. 1) to explain the perseveration of desire thinking. In this model positive metacognitions about desire thinking are associated with, in turn, imaginal prefiguration (where attentional resources are allocated on target-related information) and verbal perseveration (which prolongs conscious self-talk about reaching the desired target) marking the activation of desire thinking. Verbal perseveration is then associated with negative metacognitions about desire thinking and craving denoting the pathological escalation of desire thinking. Finally, a direct association between positive metacognitions about desire thinking and negative metacognitions about desire thinking would mark those occasions where targetachieving behavior and perception of low control are not linked with the conscious experience of craving. The goal of the current study was to test the statistical fit of the metacognitive model of desire thinking and craving in both clinical and community samples across the following categories of addictive behavior: alcohol, gambling, internet, and tobacco. 2. Study 1: test of a metacognitive model of desire thinking and craving in clinical samples 2.1. Method 2.1.1. Participants A series of convenience samples presenting with addictive behaviors were recruited from either mental health services or community settings. All these samples shared three core inclusion criteria: (1) to be at least 18 years of age; (2) to speak fluent Italian; and (3) not to be diagnosed with progressive cerebral traumas and cognitive deficits. The majority of all samples were Caucasian (98.5%). Participants' characteristics for each sample are summarized in Table 1. One hundred and seventeen individuals (45 females) diagnosed with Alcohol Use Disorder (according to DSM-5 criteria; APA, 2013)

73

were recruited from alcohol services in the Modena and Milan (Italy) areas. The mean age was 46.6 years (SD = 11.8; range = 24–64). The mean alcohol consumption on the Quantity Frequency Scale (QFS; Cahalan, Cisin, & Crossley, 1969), just before the first contact with the service, was 64.8 units per week (SD = 37.8; range = 40–240). Eighty-two individuals (13 females) diagnosed with Gambling Disorder (according to DSM-5 criteria; APA, 2013) were recruited from gambling services in the Modena and Parma (Italy) areas. The mean age was 42.4 years (SD = 12.9; range = 21–65). The mean score on the South Oaks Gambling Screen (SOGS; Lesieur & Blume, 1987) was 7.3 (SD = 5.2; range = 1–17). One hundred and sixty-seven individuals (71 females) reporting problematic internet use were recruited from the community on the basis of whether their score on the Internet Addiction Test (IAT; Young, 1998) was equal or greater than the cut-off threshold of 40. The mean score on the IAT was 51.3 (SD = 7.8; range = 41–74). The mean age was 25.8 years (SD = 7.8; range = 18–44). The average weekly internet use was 16.7 h (SD = 11.4; range = −7–70). One hundred and forty individuals (72 female) that were active tobacco users were recruited from the community. The mean score on the Fagerström Test of Nicotine Dependence (FTND; Heatherton, Kozlowski, Frecker, & Fagerström, 1991) was 3.3 (SD = 2.5; range = 0–10). The mean age of the sample was 36.4 years (SD = 13.4, range: 18–65). The average duration of tobacco use was 17.2 years (SD = 12.5; range = 1–45).

2.1.2. Self-report instruments 2.1.2.1. Desire Thinking Questionnaire (DTQ; Caselli & Spada, 2011). The DTQ consists of 10 items, with two sub-scales of five items each, assessing desire thinking. The first sub-scale, verbal perseveration, concerns the perseveration of verbal thoughts about desire-related content and experience and includes items such as: “I mentally repeat to myself that I need to practice the desired activity”. The second sub-scale, imaginal prefiguration, concerns the tendency to prefigure images about desire-related content and includes items such as: “I imagine myself doing the desired activity”. Items are general in content and refer to the desired activity that may be specified in the instructions. Each item is rated on a 4-point Likert scale ranging from 1 (almost never) to 4 (almost always). Score ranges for each sub-scale are 4 to 20 with higher scores indicating higher levels of desire thinking. The DTQ total score and sub-scale scores have been shown to possess good factor structure, internal consistency, test–retest reliability, predictive and discriminative validity (Caselli & Spada, 2011).

Table 1 Summary of participants' characteristics. Study 1

Age M SD Range Gender Female Male Symptom measures M SD Range

Study 2

Alcohol Use Disorder (N = 117)

Gambling Disorder (N = 82)

Problematic internet use (N = 167)

Tobacco users (N = 140)

Community sample (N = 494)

46.6 11.8 24–64

42.4 12.9 21–65

25.8 7.8 18–44

36.4 13.4 18–65

33.3 3.8 18–65

45 72 QFS 64.8 37.8 40–240

13 69 SOGS 7.3 5.2 1–17

71 96 IAT 51.3 7.8 41–74

72 68 FTND 3.3 2.5 0–10

291 203 – – – –

Note. M = mean; SD = standard deviation; QFS = Quantity Frequency Scale, SOGS =South Oaks Gambling Screen; IAT = The Internet Addiction Test; FTND = Fagerström Test of Nicotine Dependence.

74

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

2.1.2.2. The metacognitions about Desire Thinking Questionnaire (MDTQ; Caselli & Spada, 2013). The MDTQ consists of 18 items, with three subscales, assessing metacognitions about desire thinking. The first 8-item sub-scale (range = 8–32), positive metacognitions about desire thinking, includes items such as “I need to think about what I desire in order to feel motivated”. The second 6-item sub-scale (range = 6–24), negative metacognitions about the uncontrollability of desire thinking, includes items such “I cannot stop thinking about a desired activity or object once I started”. The third four-item sub-scale (range =4–16), metacognitions about the importance of target-related intrusive thoughts and the need to control them, includes items such as “Thoughts about certain desires should always be avoided”. Each item is rated on a 4-point Likert scale ranging from 1 (do not agree) to 4 (agree very much). Higher scores indicate stronger agreement with metacognitive beliefs. The MDTQ total score and sub-scale scores have been shown to possess good factor structure, internal consistency, test–retest reliability, predictive, and discriminative validity (Caselli & Spada, 2013). For the present study only the first two sub-scales were administered. 2.1.2.3. The Penn Alcohol Craving Scale (PACS; Flannery, Volpicelli, & Pettinati, 1999). The PACS consists of 5 items assessing craving for alcohol. The first three questions are centered on the duration, frequency and intensity of craving. The fourth question asks to rate ability to resist drinking if alcohol were available. The final question asks to rate overall average craving for alcohol during the previous week. Each question is scaled from 0 to 6 (total range = 0–30). Higher scores indicate higher levels of craving for alcohol. The PACS has been shown to possess good psychometric properties (Flannery et al., 1999). 2.1.2.4. Gambling Craving Scale (GACS, Young & Wohl, 2009). The GACS consists of 9 items, with three sub-scales, assessing craving for gambling. Each sub-scale contains three items assessing (1) an intention to gamble that was anticipated to be fun and enjoyable (Anticipation); (2) strong, urgent desire to gamble (Desire); and (3) relief from negative experiences expected from gambling (Relief). Each item is rated on a 7-point Likert scale ranging from 1 (strong disagreement) to 7 (strong agreement) with a total range that goes from 9 to 63. Higher scores indicate higher levels of craving for gambling. The GACS has been shown to possess good psychometric properties (Young & Wohl, 2009). With respect to the fact that these three sub-scales are moderately associated, we decided to adopt the total score to represent the underlying construct of craving for gambling (Young & Wohl, 2009). 2.1.2.5. The Internet Use Craving Scale (IUCS). The IUCS is a modified version of the Penn Alcohol Craving Scale (PACS; Flannery et al., 1999). The item structure was maintained, but items were rephrased so as to refer to internet use. Higher scores indicate higher levels of craving for internet use. 2.1.2.6. Brief Questionnaire of Smoking Urges (QSU-Brief; Cox, Tiffany, & Christen, 2001). The QSU-Brief consists of 10 items, with two subscales of five items each, assessing urges for cigarettes. The first subscale represents a desire and intention to smoke with smoking perceived as rewarding, and includes items such as: “A cigarette would taste good now”. The second sub-factor represents an anticipation of relief from negative affect with an urgent desire to smoke and includes items such as: “I could control things better right now if I could smoke”. We adopted response categories rangingfrom 1 (strongly disagree) to 7 (strongly agree) according to Cappelleri et al. (2007) with a total range that goes from 10 to 70. Higher scores indicate a greater urge to smoke. The QSU-Brief has been shown to possess good psychometric properties (Cappelleri et al., 2007; Cox et al., 2001). 2.1.2.7. Quantity Frequency Scale (QFS; Cahalan et al., 1969). This QFS consists of 9 items assessing alcohol consumption levels, with three subscales of three items each, assessing the use of beer, spirits and wine.

The total scores from the different sub-scales are added and an estimated daily or weekly level of alcohol consumption can be computed. This QFS has been extensively used and possesses good reliability and validity (Hester & Miller, 1995). 2.1.2.8. Fagerström Test of Nicotine Dependence (FTND; Heatherton et al., 1991). The FTND consists of 6 items assessing nicotine dependence. Higher scores indicate higher levels of nicotine dependence, with cutoff points of 3 and 5 respectively denoting moderate and high nicotine dependence and a total range score of 0 to 10. The FTND has been widely used to report smoking status in smoking populations and possesses good reliability and validity (Heatherton et al., 1991; Pomerleau, Carton, Lutzke, Flessland, & Pomerleau, 1994). 2.1.2.9. South Oaks Gambling Screen (SOGS; Lesieur & Blume, 1987). The SOGS consists of 20 items assessing gambling. Higher scores indicate higher levels of gambling, with total scores of 0 denoting ‘no problem with gambling’, 1–4 denoting ‘some problems with gambling’, and 5 above denoting ‘probable pathological gambling’. SOGS has been shown to possess good reliability and validity (Gambino & Lesieur, 2006). 2.1.2.10. The Internet Addiction Test (IAT; Young, 1998). The IAT consists of 20 items assessing the problematic internet use. Examples of items include: “How often do you choose to spend more time on-line over going out with others?”; “How often do you lose sleep due to latenight log-ins?” Each item is rated on a 5-point (range = 20–100) Likert scale covering the degree to which internet use affect daily routine, social life, productivity, sleeping pattern and feelings. Higher scores indicate higher levels of problematic internet use. The IAT has been widely used to report problematic internet use status and possesses good reliability and validity (Widyanto & McMurran, 2004). For the purposes of the current study a single factor model was used (Faraci, Craparo, Messina, & Severino, 2013). 2.1.3. Procedure Ethics approval for the study was obtained from London South Bank University Research Ethics Committee and from the local National Health Service Trust. Participants from the sub-sample diagnosed with Alcohol Use Disorder and Gambling Disorder were directly recruited from mental health services that agreed to collaborate in this research project. Clinicians were informed about study procedures by the investigator and invited to present the study just after diagnostic screening and no more than after one month from the first contact. Participants were informed that the purpose of the study was to examine their experience of craving for alcohol or for gambling. After giving informed consent participants were instructed to provide demographic details and complete the self-report instruments referring to the month before the first contact with the service. All participants and clinicians were debriefed following completion of the study. Participants from the sub-sample of problematic internet users and tobacco users were recruited from the general population through email contacts in a viral-like fashion, starting from an Italian web journal mailing-list (www.stateofmind.it). Participants received the e-mail and were requested to visit the website of the study. When participants first visited the website, the first webpage explained the purpose of the study: “To explore their craving for tobacco use” (or problematic internet use). Participants were then directed, if consenting to participate in the study, to a second webpage containing basic demographic questions and the self-report instruments. Data submitted was forwarded to a generic postmaster account. This ensured that participants' responses were anonymous. A second submission from the same IP address was not allowed so as to avoid multiple submissions from the same participant. All participants were then debriefed following the completion of the study.

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

75

Table 2 Between groups range, means, standard deviations (in parenthesis) and alphas of the study variables.

PACS

Range

Alcohol Use Disorder (N = 114)

Gambling Disorder (N = 78)

Problematic internet use (N = 158)

Tobacco users (N = 133)

Community sample (N = 482)

0–30

–

–

–

–

9–63

16.9 (5.7) .81 –

–

–

–

0–30

–

21.8 (8.6) .83 –

–

–

1–7

–

–

11.3 (4.4) .92 –

–

0–30

–

–

–

3.0 (1.2) .92 –

5–20

11.3 (3.9) .80 12.7 (4.2) .78 19.5 (5.1) .78 14.4 (4.7) .85

9.6 (4.4) .89 9.8 (4.9) .90 16.5 (4.4) .75 11.9 (4.9) .88

8.7 (2.4) .82 9.6 (2.7) .83 20.7 (4.3) .79 10.9 (3.6) .80

8.3 (2.8) .79 10.4 (3.8) .74 13.4 (4.4) .77 11.3 (4.2) .80

Alpha GACS Alpha IUCS Alpha QSU-brief Alpha GCS Alpha DTQ-IP Alpha DTQ-VP

5–20 Alpha

PMDT

8–32 Alpha

NMDT

6–24 Alpha

15.6 (5.7) .81 10.8 (3.4) .80 8.8 (3.4) .81 15.4 (4.7) .78 8.2 (3.4) .85

Note. PACS = Penn Alcohol Craving Scale; GACS = Gambling Craving Scale, IUCS = Internet Use Craving Scale; QSU-Brief = Questionnaire of Smoking Urge, brief version; GCS =General Craving Scale; DTQ-IP = Desire Thinking Questionnaire, imaginal prefiguration; DTQ-VP = Desire Thinking Questionnaire, verbal perseveration; PMDT = positive metacognitions about desire thinking; NMDT = negative metacognitions about desire thinking.

2.2. Results 2.2.1. Descriptive statistics, data configuration and correlation analyses For each sub-sample the same data configuration procedure was followed. An inspection of histograms, skewness and kurtosis coefficients was examined to identify the presence of univariate outliers, considering both symmetry and peakedness. We then tested for the presence of multivariate outliers by calculating the distance of Mahalanobis (D2). Subsequently, the coefficient of Mardia, the inspection of graphical distribution of D2 and Q–Q plots were examined to ensure a multivariate normal distribution. We then examined

Table 3 Inter-correlations of variables in all samples.

Cravinga

DTQ-IP

DTQ-VP

PMDT

DTQ-IP

DTQ-VP

PMDT

NMDT

AUD: .52⁎⁎ GD: .50⁎⁎ PIU: .51⁎⁎ TU: .37⁎⁎ CS: .36⁎⁎

AUD: .67⁎⁎ GD: .53⁎⁎ PIU: .53⁎⁎ TU: .42⁎⁎ CS: .42⁎⁎ AUD: .68⁎⁎ GD: .80⁎⁎ PIU: .70⁎⁎ TU: .68⁎⁎ CS: .46⁎⁎

AUD: .35⁎⁎ GD: .21⁎

AUD: .53⁎⁎ GD: .44⁎⁎ PIU: .32⁎⁎ TU: .22⁎⁎ CS: .27⁎⁎ AUD: .50⁎⁎ GD: .46⁎⁎ PIU: .39⁎⁎ TU: .57⁎⁎ CS: .24⁎⁎ AUD: .59⁎⁎ GD: .50⁎⁎ PIU: .42⁎⁎ TU: .77⁎⁎ CS: .40⁎⁎ AUD: .57⁎⁎ GD: .34⁎⁎ PIU: .32⁎⁎ TU: .56⁎⁎ CS: .45⁎⁎

PIU: .07 TU: .37⁎⁎ CS: .20⁎⁎ AUD: .43⁎⁎ GD: .24⁎ PIU: .22⁎ TU: .44⁎⁎ CS: .30⁎⁎ AUD: .45⁎⁎ GD: .24⁎ PIU: .21⁎⁎ TU: .55⁎⁎ CS: .29⁎⁎

Note. AUD = Alcohol Use Disorder; GD = Gambling Disorder, PIU = problematic internet users, TU = tobacco users; CS = community sample; DTQ-IP = Desire Thinking Questionnaire, imaginal prefiguration; DTQ-VP = Desire Thinking Questionnaire, verbal perseveration; PMDT = positive metacognitions about desire thinking; NMDT =negative metacognitions about desire thinking. a =The craving measure was different for each sample (see Self-report Instruments sub-section). ⁎ p b .05. ⁎⁎ p b .01.

multicolinearity using the Tolerance Index (Ti) and the Variance Inflation Factor (VIF). A Ti over .02 and a value under 5.0 for VIF are considered reliable cut-off points for the absence of multicolinearity. An analysis of residual statistics (residual Q–Q plots, skewness, kurtosis, correlations with variables) was performed to identify any indication of non-linearity, to support their homoscedasticity and the absence of significant correlation. The Durbin–Watson statistic was then performed to ensure the absence of autocorrelation. Finally, Cook's distances were examined to identify participants' data that would significantly change the regression analysis coefficients. On the basis of these analyses, outliers and influential data points were removed for three participants from the Alcohol Use Disorder sub-sample, four participants from Gambling Disorder sub-sample, nine participants from problematic internet use sub-sample, and seven participants from the tobacco use sub-sample. Descriptive statistics for all variables from the final sub-samples are presented in Table 2. All of the correlations showed that all dimensions significantly correlated with each other with the exception of the association between craving and positive metacognitions about desire thinking in the problematic internet use sample (see Table 3). 2.2.2. Path analysis In this study, our principal aim was to test the fit of a metacognitive model of desire thinking and craving suggested from previous studies (see Caselli & Spada, 2011, 2013; Spada et al., 2012, 2013). To reach this aim, a path analysis using the program LISREL 8.8 (Jöreskog & Sörbom, 1996) was employed. This technique represents a group of multiple regression equations, and estimates simultaneously the coefficients of the whole system. The advantage of path analysis over a series of multiple regression analysis is that the estimates are simultaneously adjusted for each other, thus allowing for greater separation of the effects of different independent and mediating variables. A model of such kind is considered to be a good fit to the data set if chi-square is non-significant, the Root Mean Square Error of Approximation (RMSEA) is between .08 and .05 (adequate fit) or below .05 (good fit), and the Standardized Root Mean Square Residual (SRMR) is below .09. The Comparative Fit Index (CFI), the Goodness of Fit Index (GFI) and the Non-normed Fit Index (NNFI) were also employed as incremental fit indices; their values need to be close to or above .95 in order to support a good fit. Moreover, we were interested in exploring any additional paths which could significantly improve the fit of the model. For this

76

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

Fig. 2. Path Analysis of Metacognitions, Desire Thinking and Craving in all Clinical Samples. Note. PMDT = positive metacognitions about desire thinking; NMDT = negative metacognitions about desire thinking; IP = imaginal prefiguration; VP = verbal perseveration.

purpose, we explored the modification indices that have been reported in LISREL. The metacognitive model of desire thinking and craving for the Alcohol Use Disorder sub-sample resulted in non-significant chisquare (Χ2 = 2.37, df = 3, p = .50), a RMSEA of less than .01 (p of test of close fit = .61), a SRMR of .03, a CFI of 1.00, a GFI of .99 and a NNFI of .99. The metacognitive model of craving for the Gambling Disorder sub-sample resulted in non-significant chi-square (Χ2 = .99, df = 3, p = .80), a RMSEA less than .01 (p of test of close fit = .84), a SRMR of .02, a CFI of 1.00, a GFI of .99 and a NNFI of .99. The metacognitive model of craving for the problematic internet use sub-sample resulted in a non-significant chi-square (Χ2 = 3.03, df = 3, p = .39), a RMSEA of .02 (p of test of close fit = .54), a SRMR of .02, a CFI of .1.00, a GFI of .99 and a NNFI of .98. The metacognitive model of craving for the tobacco use sub-sample resulted in a non-significant chi-square (Χ2 = 4.92, df = 3, p = .17), a RMSEA of .07 (p of test of close fit = .29), a SRMR of .03, a CFI of 1.00, a GFI of .99 and a NNFI of .98. A closer inspection of t-values showed that all paths were significant with the exception of the path between positive metacognitions about desire thinking and verbal perseveration that were not significant for the Gambling Disorder and problematic internet use sub-samples. This path was therefore removed. The final model for the GD sub-sample

resulted in a non-significant chi-square (Χ2 = 1.23, df = 4, p = .87), a RMSEA of less than .01 (p of test of close fit = .89), a SRMR of .01, a CFI of 1.00, a GFI of 1.00 and a NNFI of 1.00. The final model of craving for the problematic internet use sub-sample resulted in a nonsignificant chi-square (Χ2 = 5.44, df = 4, p = .24), a RMSEA of .04 (p of test of close fit = .31), a SRMR of .03, a CFI of .99, a GFI of .99 and a NNFI of .97. Additional paths were also tested, one by one, but none turned out to be significant at the .05 level. Thus, the model was retained. It was not possible to test feedback loops (between craving and the reactivation of desire thinking components and related metacognitions) simultaneously as this would have resulted in the model being not identified. Results from path analysis are presented in Fig. 2. 3. Study 2: test of a metacognitive model of desire thinking and craving in a community sample 3.1. Method 3.1.1. Participants A convenience community sample of 494 individuals (291 females) agreed to take part in the study. For purposes of inclusion participants

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

77

were required to be at least 18 years of age and speak fluent Italian. The mean age for the sample, which consisted primarily of Caucasians, was 33.3 years (SD = 3.8; range 18–65; see Table 1). 3.1.2. Self-report instruments 3.1.2.1. Desire Thinking Questionnaire (DTQ; Caselli & Spada, 2011). See description in Study 1. For the present study participants were invited to think about a specific activity for which they usually felt a moderate or high level of desire toward and to describe it on a blank line (Caselli & Spada, 2011). 3.1.2.2. The metacognitions about Desire Thinking Questionnaire (MDTQ; Caselli & Spada, 2013). See description in Study 1. 3.1.2.3. The General Craving Scale (GCS, Caselli & Spada, 2011). The GCS is a modified version of the PACS (Flannery et al., 1999) — see description in Study 1. In the general version adopted in this study the 5 self-report item structure was maintained but items were rephrased so as to not refer to a specific desire target but to that which participants had previously indicated. For example item 1 of PACS “How often have you thought about drinking or how good a drink would make you feel?” was modified to: “How often have you thought about your desired target or how good achieving it would make you feel?” As in PACS the scale consists of 5 items of assessed craving each scaled from 0 to 6 (range = 0–30). 3.1.3. Procedure Ethics approval for the study was obtained from London South Bank University Research Ethics Committee. Participants were recruited through e-mail contacts in a viral-like fashion, leaflets and advertisements in work environments and public places in the regions of Emilia-Romagna, Lombardia and Marche (Italy). Dissemination of information was achieved through the support of the Cognitive Psychotherapy School, Studi Cognitivi which has branches in Milan, Modena and San Benedetto del Tronto (Italy). Interested individuals were invited to make contact with Studi Cognitivi and received the study pack consisting of information and debriefing sheets, consent form, demographic questions and the self-report instruments with the instructions to: (1) complete in one session; and (2) return to the Studi Cognitivi within two weeks. On the first page of the study pack, participants were also asked if they had previously participated to a research on the same topic and if this were the case their data would have been removed. No participants were eliminated on this basis. In total 712 packs were distributed and 494 of these were returned completed to the investigator. The study packs were coded with a pre-generated id number on the front page which would be the only link between participants' individual details and their data. This system would protect confidentiality and allow participants to withdraw from the study at any time by using their id number. 3.2. Results 3.2.1. Descriptive statistics, data configuration and correlation analyses A data configuration procedure identical to the one adopted in Study 1 was followed. A total of twelve multivariate outliers were identified and eliminated from further analysis to ensure a linear relationship between variables on the basis of distance of Mahalanobis. The inspection of skewness, kurtosis, coefficient of Mardia and the graphical distribution of D2 and Q–Q plots supported a linear symmetrical distribution and a multivariate normal distribution. The inspection of Tolerance Index and Variance Inflation Factor supported the absence of multicolinearity between variables. Finally, the inspection of residual statistics and Cook's distance did not identify any indication of nonlinearity or influential data point. The final sample consisted of 482

Fig. 3. Path analysis of metacognitions, desire thinking and craving in a community sample. Note. PMDT = positive metacognitions about desire thinking; NMDT = negative metacognitions about desire thinking; IP = imaginal prefiguration; VP = verbal perseveration.

individuals (287 females). Descriptive statistics for all variables are shown in the last column of Table 2. In order to examine the basic pattern of relationships between measures of metacognitions, desire thinking and craving, two-tailed Pearson Product–moment correlations were run. These are presented in Table 2. Consistent with predictions, craving was found to be positively correlated with all predictors. Positive and negative metacognitions about desire thinking were also positively and significantly correlated with desire thinking. 3.2.2. Path analysis To test the metacognitive model of craving suggested from Study 1 a path analysis using the program LISREL 8.8 (Jöreskog & Sörbom, 1996) was employed. The analysis revealed that the model depicted in Fig. 1 was not a good fit to the data with a significant chi-square (Χ2 =24.31, df = 3, p b .01), a RMSEA of .12 (p of test of close fit = .001), a SRMR of .05, a CFI of .96, a GFI of .98 and a NNFI of .88. The inspection of tvalues and modification indices suggested improvements of this model by removing a path between craving and negative metacognitions about desire thinking and adding a path between imaginal prefiguration and craving. The results of the revised model showed that this was a good fit to the data with a non-significant chi-square (Χ2 = 4.66, df = 3, p = .19), a RMSEA of .03 (p of test of close fit = .60), a SRMR of .02, a CFI of 1.00, a GFI of 1.00 and a NNFI of .99. The final model is shown in Fig. 3. 4. Discussion The aim of this study was to provide an overview of the rapidly emerging construct of desire thinking and test a metacognitive model of desire thinking and craving (see Fig. 1), based on the work of Spada, Caselli and Wells (Caselli & Spada, 2011, 2013; Spada et al., 2012, 2013), which aims to explain the perseveration of desire thinking. In this model positive metacognitions about desire thinking are associated with, in turn, imaginal prefiguration (where attentional resources are allocated on target-related information) and verbal perseveration (which prolongs conscious self-talk about reaching the desired target) marking the activation of desire thinking. Verbal perseveration is then

78

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79

associated to negative metacognitions about desire thinking and craving denoting the pathological escalation of desire thinking. Finally, a direct association between positive metacognitions about desire thinking and negative metacognitions about desire thinking would mark those occasions where target-achieving behavior and perception of low control are not linked with the conscious experience of craving. The results of Study 1 provided preliminary cross-sectional support for the validity of the central structure of the model (see Fig. 2). These results were generally confirmed across four different addictive behaviors: alcohol, gambling, internet and tobacco, suggesting a transdiagnostic impact of the interplay of metacognitions and desire thinking in the escalation of craving. The only exception was in the connection between positive metacognitions about desire thinking and the verbal perseveration component of desire thinking in the Gambling Disorder and problematic internet use samples. A possible explanation for this may lie in the fact that the desired target is not a psychoactive substance. In the Alcohol Use Disorder and tobacco use samples the chain may be simply shortened in order to avoid the physiological withdrawal effect generated by the neurochemical features of the drugs being used. The results from Study 2 supported the existence of a somewhat structurally different model in the community sample compared to that obtained in the four clinical samples examined in Study 1. Fig. 3 suggests that positive metacognitions about desire thinking may be associated with the tendency to dwell upon positive thoughts about nature and effect of a desired activity. In addition, differentto clinical samples, both the imaginal and verbal components of desire thinking contribute to craving directly. The impact of imaginal prefiguration on craving is supported by research that has highlighted how the manipulation of imagery heightens craving and negative affect (e.g. Bywaters et al., 2004; Witvliet & Vrana, 1995). It is plausible to assume that imaginal prefiguration induced-craving has a transient nature that is common in community samples and everyday desires but it becomes perseverative and clinically significant once verbal perseveration is activated as a habitual metacognition-driven response. Fig. 3 also indicates a lack of activation of negative metacognitions about desire thinking as a response to the escalation of craving. This entails a qualitative difference between community and clinical samples that probably lies in the way escalating distress is appraised as a signal that may confirm (in clinical samples) uncontrollability of thoughts and behavior (Spada, Caselli, & Wells, 2013). Taken together, these findings suggest that desire thinking, and metacognitions associated to its activation and perseveration, may be a core process that can discriminate a slight experience of desire/deprivation from excessive craving. In a nutshell they imply that: (1) targetrelated intrusions are not the problem, but rather the way individuals cognitively react to them is (‘desires don't matter, but the response to them does’); (2) desire thinking, as a cognitive response, may be detrimental if applied to dangerous targets; that is targets we want to remain abstinent from (‘it's not good to think too much about a target we do not want to achieve’), or if aimed at reducing negative internal experience and sense of deprivation (‘We cannot crave less by thinking more about it’); and (3) improving on-line awareness that a decision has already been taken and consciously postponing any further elaboration may be a core therapeutic direction for treatment and relapse prevention of addictive behaviors. Following this line of reasoning, the results of these studies have a number of possible implications for the assessment, conceptualisation and treatment of craving. Firstly, in terms of assessment, information should be gathered not only in relation to the content of target-related thoughts or craving, but also to desire thinking and associated metacognitions. Secondly, the model of desire thinking can be used to guide the development of idiosyncratic case conceptualizations as well as to socialize patients to the idea that both metacognitions and desire thinking contribute to the escalation of craving and perception of uncontrollability. In terms of interventions, the primary therapeutic target would be interrupting desire thinking and modifying associated

metacognitions. This could be achieved by shifting to a metacognitive mode of processing, gaining a flexible control over attention and thinking style, modifying metacognitive beliefs and developing new plans of processing. Metacognitive therapy techniques such as attention training technique (ATT), detached mindfulness, situational attentional refocusing (Wells, 2008) as well as visuospatial tasks (May, Andrade, Panabokke, & Kavanagh, 2010) may help improve cognitive flexibility, provide disconfirmatory evidence of metacognitive beliefs and support verbal reattribution strategies to drive their modification. The results of these studies must be considered with regard to design limitations. First social desirability, self-report biases, context effects and poor recall may have contributed to errors in self-report measurements. Second the clinical samples are limited in size and future replications are needed to ensure the reliability of the model. Third, given the cross-sectional nature of this study caution is recommended in presuming causal inferences. A number of important questions need to be addressed in future research studies. The phenomenology of desire thinking requires further exploration, particularly in relation to its possible neurobiological basis and behavioral outcomes (e.g. the decision to achieve a given target). For instance, it would be useful to verify whether both desire thinking and related metacognitions can longitudinally predict levels of craving and frequency of start signals for engaging in a given behavior, even after a period of abstinence from the desired target. It could be speculated that desire thinking may play a role in generating internal states that people use as signals for their decision making process on the basis of specific metacognitive knowledge (Spada et al., 2012, 2013). In addition the link between negative metacognitions about desire thinking and intention to achieve a target has not been tested yet. Finally, it would be also useful to test the mediating role of desire thinking and related changes in metacognitions during treatment. In conclusion, the studies presented provide support for the potential value of a metacognitive model of desire thinking and craving, and innovative transdiagnostic lines of treatment for addictive behaviors. Role of funding source The study was completely self-funded.

Contributors Gabriele Caselli and Marcantonio M. Spada, co-wrote the manuscript.

Conflict of interest There are no conflicts of interest to declare.

References American Psychiatric Association (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Bywaters, M., Andrade, J., & Turpin, G. (2004). Intrusive and non-intrusive memories in a non-clinical sample: The effects of mood and affect on imagery vividness. Memory, 12, 467–478. Cahalan, D., Cisin, I., & Crossley, H. (1969). American drinking practices: A National Survey of Drinking Behaviors and Attitudes. Rutgers Center for Alcohol Studies, monograph no. 6, New Brunswick, USA. Cappelleri, J. C., Bushmakin, A. G., Baker, C. L., Merikle, E., Olufade, A. O., & Gilbert, D.G. (2007). Multivariate framework of the Brief Questionnaire of Smoking Urges. Drug and Alcohol Dependence, 90, 234–242. Caselli, G., Ferla, M., Mezzaluna, C., Rovetto, F., & Spada, M. M. (2012). Desire thinking across the continuum of drinking behaviour. European Addiction Research, 18, 64–69. Caselli, G., Nikčević, A., Fiore, F., Mezzaluna, C., & Spada, M. M. (2012). Desire thinking across the continuum of nicotine dependence. Addiction Research and Theory, 20, 382–388. Caselli, G., Soliani, M., & Spada, M. M. (2013). The effect of desire thinking on craving: An experimental investigation. Psychology of Addictive Behaviors, 27, 301–306. Caselli, G., & Spada, M. M. (2010). Metacognitions in desire thinking: A preliminary investigation. Behavioural and Cognitive Psychotherapy, 38, 629–637. Caselli, G., & Spada, M. M. (2011). The Desire Thinking Questionnaire: Development and psychometric properties. Addictive Behaviors, 36, 1061–1067. Caselli, G., & Spada, M. M. (2013). The metacognitions about Desire Thinking Questionnaire: Development and psychometric properties. Journal of Clinical Psychology, 69, 1284–1298.

G. Caselli, M.M. Spada / Addictive Behaviors 44 (2015) 71–79 Cox, L. S., Tiffany, S. T., & Christen, A. G. (2001). Evaluation of the brief Questionnaire of Smoking Urges (QSU-Brief) in laboratory and clinical settings. Nicotine & Tobacco Research, 3, 7–16. Faraci, P., Craparo, G., Messina, R., & Severino, S. (2013). Internet Addiction Test (IAT): Which is the best factorial solution? Journal of Medical Internet Research, 15, e225. Fernie, B.A., Caselli, G., Giustina, L., Donato, G., Marcotriggiani, A., & Spada, M. M. (2014). Desire thinking as a predictor of gambling. Addictive Behaviors, 39, 793–796. Field, M., Mogg, K., Zetteler, J., & Bradley, B. P. (2004). Attentional biases for alcohol cues in heavy and light social drinkers: The roles of initial orienting and maintained attention. Psychopharmacology, 176, 88–93. Flannery, B.A., Volpicelli, J. R., & Pettinati, H. M. (1999). Psychometric properties of the Penn Alcohol Craving Scale. Alcoholism: Clinical and Experimental Research, 23, 1289–1295. Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitivedevelopmental inquiry. American Psychologist, 34, 906–911. Gambino, B., & Lesieur, H. (2006). The South Oaks Gambling Screen (SOGS): A rebuttal to critics. Journal of Gambling Issues, 17. Green, M., Rogers, P., & Elliman, N. (2000). Dietary restraint and addictive behaviour: The generalisability of Tyffany's cue reactivity model. International Journal of Eating Disorder, 27, 419–427. Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., & Fagerström, K. O. (1991). The Fagerström Test for Nicotine Dependence: A revision of the Fagerström Tolerance Questionnaire. British Journal of Addiction, 86, 1119–1127. Hester, R. K., & Miller, W. R. (1995). Handbook of alcoholism treatment approaches: Effective alternatives. Needham Heights, USA: Allyn & Bacon. Jöreskog, K. G., & Sörbom, D. (1996). LISREL 8: User's reference guide. Chicago: Scientific Software International. Kavanagh, D. J., Andrade, J., & May, J. (2004). Beating the urge: Implications of research into substance-related desires. Addictive Behaviours, 29, 1359–1372. Kavanagh, D. J., Andrade, J., & May, J. (2005). Imaginary relish and exquisite torture: The elaborated intrusion theory of desire. Psychological Review, 112, 446–467. Kavanagh, D. J., May, J., & Andrade, J. (2009). Tests of the elaborated intrusion theory of craving and desire: Feature of alcohol craving during treatment for an alcohol disorder. British Journal of Clinical Psychology, 48, 241–254. Lesieur, H., & Blume, S. (1987). The South Oaks Gambling Screen (SOGS): A new instrument for the identification of pathological gamblers. American Journal of Psychiatry, 144, 1184–1188.

79

May, J., Andrade, J., Panabokke, N., & Kavanagh, D. (2004). Images of desire: Cognitive models of craving. Memory, 12, 447–461. May, J., Andrade, J., Panabokke, N., & Kavanagh, D. (2010). Visuospatial tasks suppress craving for cigarettes. Behaviour Research and Therapy, 48, 476–485. Pomerleau, O. F., Carton, S. M., Lutzke, M. L., Flessland, K. A., & Pomerleau, C. S. (1994). Reliability of the Fagerström Tolerance Questionnaire and the Fagerström Test for Nicotine Dependence. Addictive Behaviors, 19, 33–39. Robinson, T., & Berridge, K. (1993). The neural basis of craving: An incentive-sensitisation theory of addiction. Brain Research Reviews, 18, 247–291. Spada, M. M., Caselli, G., Slaifer, M., Nikčević, A., & Sassaroli, S. (2014). Desire thinking as a predictor of problematic internet use. Social Science Computer Review, 32, 474–483. Spada, M. M., Caselli, G., & Wells, A. (2012). The metacognitive therapy approach to problem drinking. In S. Hayes, & M. Levin (Eds.), Mindfulness and acceptance for addictive behaviors: Applying contextual CBT to substance abuse and behavioral addictions. New York, USA: New Harbringer. Spada, M. M., Caselli, G., & Wells, A. (2013). A triphasic metacognitive formulation of problem drinking. Clinical Psychology & Psychotherapy, 20, 494–500. Tiffany, S. T., & Drobes, D. J. (1990). Imagery and smoking urges: The manipulation of affective content. Addictive Behaviours, 15, 531–539. Wells, A. (2000). Emotional disorders and metacognition: Innovative cognitive therapy. Chichester, UK: Wiley. Wells, A. (2008). Metacognitive therapy for anxiety and depression. London, UK: Guilford Press. Wells, A., & Matthews, G. (1994). Attention and emotion: A clinical perspective. Hove, UK: Erlbaum. Wells, A., & Matthews, G. (1996). Modelling cognition in emotional disorder: The S-REF model. Behaviour Research and Therapy, 34, 881–888. Widyanto, L., & McMurran, M. (2004). The psychometric properties of the Internet Addiction Test. CyberPsychology & Behavior, 7, 443–450. Witvliet, C. V., & Vrana, S. R. (1995). Psychophysiological responses to indices of affective dimensions. Psychophysiology, 32, 436–443. Young, K. S. (1998). Internet addiction: The emergence of a new clinical disorder. CyberPsychology & Behavior, 1, 237–244. Young, M. M., & Wohl, M. J. A. (2009). The Gambling Craving Scale: Psychometric validation and behavioral outcomes. Psychology of Addictive Behaviors, 23, 512–522.