Journal of Research in Personality 45 (2011) 655–661 Contents lists available at SciVerse ScienceDirect Journal of Research in Personality journal homepage: www.elsevier.com/locate/jrp Appetitive and impulsive components in the Appetitive Motivation Scale Luigi Leone a,⇑, Fridanna Maricchiolo b, Fabio Presaghi a a b Department of Developmental and Social Psychology, ‘‘Sapienza’’ University of Rome, Italy Department of Cultural and Educational Studies, Experimental Psychology Laboratory, University of Roma Tre, Italy a r t i c l e i n f o Article history: Available online 25 September 2011 Keywords: BAS Impulsivity Reinforcement Sensitivity Theory Structural equation modeling Personality a b s t r a c t Measures of the Behavioral Activation System (BAS) often conflate reward-reactivity (the core of the BAS) and impulsivity-related variance. The Appetitive Motivation Scale (AMS), a recently-developed BAS measure, has been proposed as a one-dimensional scale which deals exclusively with core reward-reactivity. However, we show here how a two-dimensional interpretation of the scale reveals components which are linked with both the core of the BAS and impulsivity. A clear pattern of associations and dissociations with other BAS- and impulsivity-related measures provides support for such a two-dimensional interpretation. The results are discussed, highlighting the parallels between the AMS structure and other BASrelated measures, and emphasizing the theoretical challenges which are inherent in developing reliable BAS-related personality measures. Ó 2011 Elsevier Inc. All rights reserved. 1. Introduction Gray’s (e.g., 1982) Reinforcement Sensitivity Theory (RST) is a widely influential theory of personality, linking individual differences in behavioral activation and inhibition to underlying biological structures. The RST conceptualizes three main neurobiological systems which are allegedly linked to individual differences in personality but here we will focus solely on the Behavioral Activation System (BAS), which is activated by positive stimuli and mediates approach responses towards rewards. The BAS is a particularly relevant system for personality research because of its connections with action, goal-pursuit and positive emotionality (Carver & White, 1994). In addition, the system is theoretically appealing in clinical research, because it has been suggested that an unchecked hyperactive BAS may be conducive to various reckless behaviors (O’Connor, Stewart, & Watt, 2009; Voigt et al., 2009; Zisserson & Palfai, 2007). Finally, the BAS has been interpreted as the key motivational foundation of trait impulsivity and impulsive action (e.g., Pickering & Gray, 1999) and even as the basis of the impulsive component in different forms of psychopathology (Newman, MacCoon, Vaughn, & Sadeh, 2005; Urosevic, Abramson, Harmon-Jones, & Alloy, 2008). However, measuring BAS reactivity has proven to be a difficult and theoretically challenging task. A first attempt to design a purpose-built RST measure (Wilson, Gray, & Barrett, 1990) was hampered by serious psychometric shortcomings (e.g., Carver & White, 1994). At the end of the ⇑ Corresponding author. Address: Dipartimento di Psicologia dei Processi di Sviluppo e di Socializzazione, Via dei Marsi, 78, 00185 Roma, Italy. E-mail address: luigi.leone@uniroma1.it (L. Leone). 0092-6566/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.jrp.2011.09.003 1990s, the BAS scales developed by Carver and White (1994; CW-BAS) provided the first purpose-built BAS measure to gain widespread popularity, and to display satisfactory and replicable psychometric properties (e.g., Jorm et al., 1999; Leone, Perugini, Bagozzi, Pierro, & Mannetti, 2001). The CW-BAS comprises three scales which are related to different functions and definitions of the BAS (Carver & White, 1994). In spite of its success (it is indexed as being the most frequently used BAS-related measure), the CWBAS has been criticized for failing to adequately represent the BAS: it has been argued that the complex three-factor structure of the CW-BAS sets it apart from the theoretical tenets of RST, which conceptualizes the BAS as a tighter one-dimensional construct (e.g., Caseras, Avila, & Torrubia, 2003; Jackson, 2009). Furthermore, the ability of the CW-BAS scales to reflect the motivational properties of the BAS has been questioned (e.g., Jackson & Smillie, 2004). The intensity of the debate regarding how the BAS could and should be measured reflects a real conceptual problem, one that goes beyond psychometric quarrels. Indeed, there are theoretical difficulties in building a trait-level measure of the BAS (Smillie, Pickering, & Jackson, 2006). The BAS’ anatomical and neuro-chemical foundations are complex and not yet definitively identified, and its emotional, motivational and behavioral consequences are very broad and not necessarily homogeneous (Pickering & Smillie, 2008). In brief, it has proven to be difficult to translate the complex motivational processes regulated by the BAS into a trait measure, let alone a one-dimensional one (Jackson, 2003; Smillie & Jackson, 2006). A recent purpose-built BAS-related measure, the Appetitive Motivation Scale (AMS), was an attempt to directly tackle this difficulty (Jackson & Smillie, 2004). The authors intended to develop indicators of the construct by dealing exclusively with the motiva- 656 L. Leone et al. / Journal of Research in Personality 45 (2011) 655–661 tional role of appetitive stimuli in triggering approach behaviors aimed at obtaining rewarding experiences, thereby clearly defining the boundaries of the construct (Jackson & Smillie, 2004). Unlike the CW-BAS scales, the AMS was designed as a one-dimensional measure, thereby avoiding the possibility of the divergent concurrent and predictive validity results that are common when the three CW-BAS scales are considered separately (Carver, 2004; Smillie, Jackson, & Dalgleish, 2006). The AMS was also intended to avoid the conflation between BAS-related and impulsivityrelated indicators that is typical of other popular BAS measures (Cooper, Smillie, & Jackson, 2008; Smillie & Jackson, 2006), such as the Sensitivity to Reward (SR) scale (Caseras et al., 2003). This is a crucial point because a deep revision of the RST (r-RST; Gray & McNaughton, 2000) clarifies that BAS should be highly related with extraversion (e.g., Smillie, Pickering & Jackson, 2006) and more closely linked with functional forms of impulsivity (Dickman, 1990) rather than with dysfunctional or rash impulsivity (e.g., Jackson, 2009; Smillie & Jackson, 2006).1 Recent measures explicitly developed under the framework of the r-RST (Jackson, 2009) satisfy such requirements; however, the popularity of measures developed under the original RST (o-RST) among researchers makes it relevant to investigate whether such measures could be refined to adequately capture the theoretical definitions put forward by the r-RST (e.g., Heym et al., 2008; Smillie, Jackson et al., 2006). A number of articles concerning the AMS scale validation and validity have appeared (Cooper et al., 2008; Jackson & Smillie, 2004; Smillie & Jackson, 2005) providing some support for the one-dimensionality of the scale – although the evidence is admittedly not decisive (Cooper et al., 2008) – and its predictive and construct validity. Nonetheless, the authors lamented a relative lack of replication of their results by independent researchers. We will attempt in this study to accomplish this goal, by adopting the same aims that were embraced by the AMS authors, shaping a BAS-related measure that: (1) addresses the motivational processes which are supervised by the BAS and (2) avoids any confusion and content-conflation with trait impulsivity, as required by the r-RST. We maintain that a further revision of the structure of the AMS will help to satisfy these two important theoretical and measurement criteria (Smillie, Pickering et al., 2006). 1.1. The AMS: theory, structure, validity Table 1 presents the 20 items comprised in the AMS (Jackson & Smillie, 2004). The authors stated that they intended to assess the BAS with the aim of ‘‘. . . impulsively approaching appetitive stimuli with the goal of obtaining a rewarding experience’’ (Jackson & Smillie, 2004, p. 1628, emphasis added), a definition consistent with the original RST. This emphasis on impulsivity faded away in the subsequent articles about the scale, as the authors recognized the need to differentiate impulsivity, at least in its dysfunctional form, from the BAS (e.g., Smillie & Jackson, 2005; Smillie & Jackson, 2006), more consistently with the r-RST (Smillie, Pickering et al., 2006). Nevertheless, the early inclusion of impulsivity as a defining feature of BAS-regulated behaviors substantially influenced the composition of the scale. Items 7, 10, 13, 16 and 18 are reminiscent of the typical sensation-seeking and lack of planning indicators which are found in several impulsivity measures. A different cluster of items (# 1, 5, 8) instead reflects a reluctance to adhere to norms and rules, a component that appears to be linked with psychoticism, or with the more general socially unde1 The revised RST proposes crucial changes in the definitions and functions of the Behavioral Inhibition System and of the Fight/Flight System (Gray & McNaughton, 2000; Heym, Ferguson, & Lawrence, 2008; Jackson, 2009; Smillie, Pickering et al., 2006). Here we focus only on predictions involving the BAS as conceptualized within the r-RST. Table 1 Items of AMS by Jackson and Smillie (2004) and items retained (⁄) in the Cooper et al. (2008) revision. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Believe rules are stifling Put plans into action Like to be busy Like to see how things work Choose which rules to follow Like things new and different⁄ Like to do things spontaneously⁄ Like to do things my way Do several things at same time⁄ Important to enjoy moment Actively look for new experiences⁄ Feel for how things work⁄ Look for new sensations⁄ Excited by what is new in my field⁄ Often lots of ideas⁄ Prefer not to plan too much Like to be rewarded⁄ Here and now is important New ideas all the time⁄ Enjoy new projects⁄ sirable consequences of impulsive behaviors aimed at obtaining personal advantage, rather than with BAS activity per se. In spite of the presence of distinct clusters of items, the authors opted for a one-dimensional structure for the scale. However, the presence of impulsivity- and psychoticism-related indicators shaped the pattern of associations with concurrent validity criteria. The AMS correlated with the Impulsivity scale of the Eysenck Personality Profiler (EPP; Eysenck, 1993), and with the Eysenck Personality Questionnaire (EPQ) Psychoticism scale. The AMS was also associated with EPQ-Extraversion, as BAS measures are expected to do from a theoretical standpoint (e.g., Smillie, Pickering et al., 2006). However, the magnitude of the AMS-Extraversion association was almost identical to the AMS-Psychoticism correlation, which appears problematic (e.g., Jackson, 2009). Concurrent associations with the CW-BAS scales also revealed the coexistence in the AMS of both impulsivity- and BAS-related indicators. Recent research has shown that the Drive and Reward-Responsiveness scales of the CW-BAS are more closely linked to the core BAS-concept, whilst the Fun scale is more closely linked to rash impulsivity (Heym & Lawrence, 2010; Leone, 2009; Smillie, Jackson et al., 2006). Thus, the AMS should be more closely linked to Drive and Reward-Responsiveness than to Fun scores. Instead, the AMS correlated slightly more strongly with Fun (Jackson & Smillie, 2004), and therefore to impulsivity-related manifestations, than with the other scales that more effectively represent the core functions of the BAS in sustaining efforts to attain appetitive outcomes (Smillie, Jackson et al., 2006). As such associations were computed using a small sample (N = 59), Smillie and Jackson (2005) sought replication with a bigger sample (N = 144). As it turned out, the AMS still correlated more strongly with the CW-BAS Fun scale than with the CW-BAS Reward-Responsiveness and Drive scales. Smillie and Jackson (2005) clarified an important theoretical point regarding the connections between the BAS and impulsivity. In contrast to Jackson and Smillie (2004), Smillie and Jackson (2005) clearly stated that the association between the BAS and trait impulsivity may be deemed arbitrary, at best. The descriptive correspondence between BAS and impulsivity may stem from the simple fact that being motivated by reward may prompt – when the right circumstances occur – quick impulsive action to seize the incentive (Corr, 2008). However, it may be argued that this kind of BAS-related impulsivity differs from the classical dysfunctional strain of impulsivity. The relatively under-researched forms of functional impulsivity (i.e., appropriate behavior intended to seize the reinforcer – Dickman, 1990) may reflect BAS sensitivity more L. Leone et al. / Journal of Research in Personality 45 (2011) 655–661 directly (Smillie & Jackson, 2006). Recent evidence confirms the similarities between functional impulsivity and the BAS (Leone & Russo, 2009; Smillie & Jackson, 2006), and the clear distinction between the core BAS measures and rash and dysfunctional impulsivity (e.g., Smillie, Jackson et al., 2006; Leone, 2009). The divergent associations of BAS measures with functional and dysfunctional impulsivity may be interpreted as a crucial test in developing and testing measures consistent with the revised RST (Jackson, 2009). In accordance with this view, the AMS was more strongly connected with Functional Impulsivity than dysfunctional impulsivity scores (Jackson & Smillie, 2004); however, such an association is likely to have been biased by the classical impulsivity-related indicators included in the AMS. Unsurprisingly, the presence of classical impulsivity indicators leads to sizeable associations of the AMS with impulsivity scales, as well as with the Fun scale of the CW-BAS (Smillie & Jackson, 2005) that appears to better reflect dysfunctional impulsivity (Leone & Russo, 2009). The body of evidence regarding the original 20-item form of the AMS may be deemed mixed. The AMS was aligned with BAS criteria such as extraversion, the CW-BAS scales and functional impulsivity, but also with psychoticism and other dysfunctional impulsivity measures. Cooper et al. (2008) attempted to revise the scale to render it more internally coherent and more independent of the elusive concept of classical ‘‘dysfunctional’’ or ‘‘rash’’ impulsivity. Using a sizeable sample, Cooper et al. (2008) refined the AMS, eliminating 9 out of the 20 original items. The remaining 11 items (Table 1) were assumed to be a reflection of a single onedimensional factor, and yielded satisfactory internal consistency estimates. The items which were eliminated included most of the typical impulsivity-related indicators, as well as those referring to discomfort with adhering to the rules. In spite of the changes in the scale’s composition, validity correlations were noticeably similar for both the original and the revised scales (Cooper et al., 2008, Table 6). The correlations of the AMS (both in its original and revised form) with Impulsivity, Psychoticism and the Impulsivity-leaning CW-BAS Fun scale were only slightly weaker than the correlations between the AMS and the ‘‘core’’ BAS measures (Functional Impulsivity, CW-BAS Drive scale). It appears that a one-dimensional interpretation of both the original and the revised AMS scales still delivers a fuzzy set of relationships with core BAS and impulsivity measures. 657 dysfunctional outcomes (Corr, 2008). We will show that such components can be found in the AMS and that double dissociation patterns best represent the linkages between these two components and the criteria representing the core BAS and impulsivity, respectively. Our hypotheses could be more formally summarized as follows: Hypothesis 1. We anticipate that two dimensions reflecting the core BAS (appetitive motivation and impulsivity-related variance) would account more satisfactorily for variation among the AMS items, compared with the original one-factor structure. Hypothesis 2. Appetitive motivation will be related to extraversion (e.g., Jackson, 2009) and conscientiousness, as is expected of a BAS measure reflecting sensitivity to appetitive stimuli and persistence and conscientiousness in pursuing reward contingencies (Heym et al., 2008). The AMS–Impulsivity factor will be more weakly related to extraversion and significantly associated with classical impulsivity. Hypothesis 3. Appetitive motivation will more strongly predict CW-BAS Drive and Reward-Responsiveness scores than CW-BAS Fun scores. We expect the opposite pattern to materialize for AMS-Impulsivity. Hypothesis 4. Appetitive motivation will be linked with functional impulsivity, but not with dysfunctional impulsivity (e.g., Jackson, 2009; Smillie & Jackson, 2006). The reverse pattern is expected for the AMS-Impulsivity factor. Support for the hypotheses listed above will be interpreted as implying the conceptual and empirical value of recognizing the bidimensionality of the AMS. Results supportive of our hypotheses would be consistent with r-RST definition of the BAS (e.g., Jackson, 2009; Smillie, Pickering et al., 2006) and with empirical and conceptual evidence on the structure and function of the BAS (e.g., Corr, 2008; Leone, 2009; Smillie, Jackson et al., 2006) obtained using different measures (the AMS instead than the CW-BAS) and in different cultural contexts. 3. Method 2. Aims and hypotheses 3.1. Participants and procedure Although Cooper et al.’s (2008) revision enhanced the consistency and content validity of the AMS, the revised scale still did not attain the desired pattern of strong associations with other BAS measures and weak or null associations with rash impulsivity measures. We suspect that some conflation of impulsivity-related and BAS-related content still exists in the revised AMS. The authors’ preference for a one-dimensional structure for the AMS may have overshadowed the presence of a distinct impulsivityrelated component. Relaxing the one-factor-only assumption may allow us to disentangle the impulsivity and core-BAS components that could be present in the AMS, and that have been detected in other BAS measures (e.g., Smillie, Jackson et al., 2006). We anticipate that the AMS may include two components: a dimension which is closely linked with conscientiousness in pursuing reward contingencies, persistence in working towards future positive outcomes and a future-oriented focus (e.g., Heym et al., 2008) regulated by the BAS, and a dimension which clings to the impulsive behavioral components that in some instances (proximity to the reward) may be considered to serve BAS-motivated goal-pursuit, but that under different circumstances (when the reward is not yet in reach) may lead to The study was conducted in Italy. Students at a large University were asked to recruit participants from among their acquaintances. A group of 388 individuals agreed to participate (males = 179; females = 207; two missing values). Participants varied considerably in age (M = 35, SD = 13.35, range 18–76) and occupation. Importantly, psychology students were not included in the sample. The participants anonymously filled in booklets including the personality measures described below. In order to avoid systematic sequence and fatigue effects, the order of the measures in the booklets was pseudo-randomized across subjects. 3.2. Measures An Italian version of the AMS (Jackson & Smillie, 2004) was used. The scale includes 20 items and a five-point (disagree/agree) response scale (Cooper et al., 2008). AMS items have been translated by the first author and then back-translated by a native English speaker. The original and back-translated versions closely matched. Impulsivity was measured using the Italian version (Russo, Leone, & De Pascalis, 2011) of the I7 scale (Eysenck, Pearson, 658 L. Leone et al. / Journal of Research in Personality 45 (2011) 655–661 Easting, & Allsopp, 1985), which is a widely-used 19-item (YES/NO response) measure of trait impulsivity (a = .80). The Big Five Questionnaire (short version, Caprara, Schwartz, Capanna, Vecchione, & Barbaranelli, 2006) is a 60-item five-point response scale of the Big Five traits. Due to the objective of this study (see Hypothesis 2), only Energy (E, 12 items, a = .73) and Conscientiousness (C, 12 items, a = .82) scores were used for the analyses. The CW-BAS (Carver & White, 1994; Italian version Leone, Pierro, & Mannetti, 2002) includes 13 BAS-related items (5-point disagree/agree scale). Three facets are measured: positive responses to the occurrence or anticipation of rewards or RewardResponsiveness (five items, a = .78); persistence in the pursuit of reward or Drive (five items, a = .75); and a desire for novel rewards and a tendency to act quickly in the pursuit of appetitive goals or Fun (four items, a = .79). An Italian version (Leone et al., 2002) of the Dickman Impulsivity Inventory (Dickman, 1990) was used. Twelve items measured dysfunctional impulsivity (a = .83), and 11 items measured functional impulsivity (a = 71). Five-point disagree/agree scales were used. 4. Results Prior to performing a factor analysis of the AMS, a formal test (Shapiro–Wilk test) of the normality of the distribution of the 20 items was performed. We found significant departures from normal distributions, as has already been reported (Cooper et al., 2008). As a first step towards testing the tenability of a two-factor structure for the AMS (Hypothesis 1), exploratory factor analysis (EFA) was conducted. The full information maximum likelihood method, assuming multivariate ordinal data, was used as method of estimation (Jöreskog & Moustaki, 2006). We used Parallel analysis (PA; Horn, 1965) and Velicer’s MAP test (Velicer, 1976) to determine the number of factors to retain (Velicer, Eaton, & Fava, 2000; Zwick & Velicer, 1986). The results of PA and Velicer’s MAP test indicated that the number of factors ranged from a minimum of two (Velicer’s MAP) to a maximum of seven (as suggested by PA). Alternative solutions to the one-factor model put forward by Cooper et al. (2008) do exist and can be further explored. The first EFA model investigated includes all 20 AMS items. A one-factor solution was inspected but several items loaded relatively poorly (<.35). A two-factor solution was then investigated. Two interpretable factors emerged after Promax rotation. Items 3, 2 and 4 marked the first factor. These items reflect the subjects’ approach in terms of action orientation; the second extracted factor was marked by items 8, 10, 5 and 18, which reflect a mixture of present orientation, fun-seeking and discomfort with external rules. Finally, a three-factor solution was considered. In this solution, the first factor replicated its counterpart in the two-factor solution, while the second and third factors were a split of the second factor in the previous solution. The first ‘‘slice’’ of the split was fairly close to the ‘‘present orientation and sensation-seeking’’ (items 10, 11 and 18) interpretation suggested for the two-factor solution, whilst the other ‘‘slice’’ (items 5, 8 and 1) reflected items pertaining to adherence to the norm and a ‘‘personal way of doing things’’. It is doubtful whether the latter component fits conceptually within the BAS construct. Therefore, the items marking this component were eliminated from further analyses. Of the remaining items, those with unsatisfactory factor loadings (i.e., <.35) were recursively eliminated from the analysis and different solutions (i.e., the number of factors) were compared at each step. At the end of this process, three items were eliminated, leading to a 14item set (Table 2) and a two-factor correlated solution (r = .39, p < .01) that outperformed competing structures. Cronbach’s alpha values for the ordered category items (Zumbo, Gadermann, & Zeisser, 2007) were satisfactory (Factor 1 = .83; Factor 2 = .79). Confirmatory factor analyses (CFA) were then performed in order to provide a further test of the 14-item two-factor solution. CFA models were estimated using a polychoric correlation matrix and the robust maximum likelihood estimation function (Jöreskog, 2005). In order to obtain correct standard errors and to test the statistics under non-normality, the Satorra-Bentler chi-square statistic (1988) was used. Following Hu and Bentler’s (1999) suggestions, values of root mean square error of approximation (RMSEA) and of the standardized mean square residual (SRMR) 6 .08 (better if 6.05) and comparative fit index (CFI) and non-normed fit index (NNFI) values above P.95 are interpreted as indicative of good fit. Indices for the two-factor solution were then compared with alternative one-dimensional structures: a 20-item one-dimensional model (Jackson & Smillie, 2004), and the revised 11-item one-dimensional model (Cooper et al., 2008). The 20-item one-factor model fitted unsatisfactorily [v2(170) = 1272.61, p < .01; SB-v2(170) = 1155.74, p < .01; RMSEA = .093; SRMR = .092; CFI = .90; NNFI = .89], replicating results from Cooper et al. (2008), with all indices exceeding the acceptable cut-offs. Even though the 11-item revised AMS showed a consistent improvement in terms of CFI and NNFI (.94 and .92), the RMSEA still indicated an unacceptable fit (.10) [v2(44) = 417.74, p < .01; SB-v2(44) = 220.41, p < .01; SRMR = .074]. Instead, the two-factor model comprising 14 items presented a very encouraging fit, with all indices within acceptable thresholds [v2(76) = 486.10, p < .01; SB-v2(76) = 258.44, p < .01; RMSEA = .079; SRMR = .077; CFI = .95; NNFI = .95]. It appears that the results of both EFA and CFA converge in suggesting the presence of two related yet distinct facets of appetitive motivation on the AMS. The first facet (items 3, 2, 4, 20, 14, 19, 15 and 12, see Table 2) refers to a goal-committed appetitive motivation, possibly regulating complex approach behaviors and entailing some degree of future oriented focus, planning and perseverance. Tentatively, we prefer to refer to this factor using the original label ‘‘Appetitive Motivation’’. Items on the second facet (11, 10, 6, 13, 18 and 16) refer instead to behavioral tendencies targeting the immediate enjoyment of sensations and emotions without much planning; tentatively, we have simply labeled this factor ‘‘Impulsivity’’ or ‘‘Impulsive hedonism’’.2 Hypotheses 2-4 concern various issues of convergent and discriminant validity. We adopted a structural equation approach to test for the convergence, divergence and discriminant hypotheses on the associations between approach motivation and general and more specific personality measures. In the three subsequent models the two AMS factors were defined by their respective items (8 and 6 items, respectively), whilst the criteria were defined by a single indicator each (average scale-score). In the first model we regressed extraversion, conscientiousness and impulsivity (I7) on the two AMS factors. Impulsivity is perhaps misleadingly (e.g., Depue & Collins, 1999; Smillie & Jackson, 2006) connected with BAS sensitivity; BAS measures should be aligned with extraversion instead (e.g., Jackson, 2009); therefore, we expected stronger linkages between appetitive motivation and extraversion, compared with the association between extraversion and impulsive hedonism. The r-RST conceptualization of the BAS point instead to a 2 We are aware that different labels could have been proposed. For instance, it could be argued that the Appetitive Motivation factor also reflects sensation-seeking plus action orientation and/or quick refocusing of attention (Dickman, 1990), plus possibly future oriented motivations (Corr, 2008; Heym et al., 2008). Therefore, a different label may have been chosen. Similarly, the ‘‘Impulsivity’’ factor could have been more narrowly labeled in order to refer to present orientation and a lack of planning. Nonetheless, these alternative interpretations are still consistent with reward sensitivity and therefore with the core BAS on the one hand, and to impulsivity-related manifestations on the other. L. Leone et al. / Journal of Research in Personality 45 (2011) 655–661 Table 2 Factor loadings (after Promax rotation) of the two factors solution. 2. Put plans into action 3. Like to be busy 4. Like to see how things work 12. Feel for how things work 14. Excited by what is new in my field 15. Often lots of ideas 19. New ideas all the time 20. Enjoy new projects 6. Like things new and different 10. Important to enjoy moment 11. Actively look for new experiences 13. Look for new sensations 16. Prefer not to plan too much 18. Here and now is important Factor 1 Factor 2 Uniqueness 0.689 0.803 0.632 0.423 0.536 0.473 0.496 0.584 0.325 0.009 0.212 0.220 0.174 0.137 0.188 0.026 0.053 0.185 0.310 0.281 0.314 0.327 0.598 0.600 0.674 0.569 0.534 0.563 0.591 0.371 0.624 0.726 0.487 0.593 0.534 0.403 0.386 0.636 0.389 0.530 0.757 0.724 Note: factor loadings exceeding 0.35 are presented in boldface. looser association with impulsivity. Therefore, appetitive motivation should be more weakly linked with impulsivity, whilst the Impulsive component of the AMS was expected to be closely related to classic impulsivity. We partialled out impulsivity effects on extraversion and conscientiousness in order to avoid spurious associations between the AMS factors and extraversion and conscientiousness due to their association with impulsivity. Considering the overall information stemming from the different fit indices, the model fitted satisfactorily: v2(112) = 365.52, p < .001, RMSEA = .076 (90% CI for RMSEA: .068–.085), NNFI = .92, CFI = .93. As expected, the two AMS factors show a complex dissociation pattern with the criteria. Appetitive motivation was posip < .01) and tively related to extraversion (c = .443, conscientiousness (c = .44, p < .01) and negatively to impulsivity (c = .34, p < .01). Instead, the impulsive hedonism facet was positively related to impulsivity (c = .52, p < .01) and to a lesser extent to extraversion (c = .20, p < .05), and was non-significantly associated with conscientiousness (c = .02, ns). Clearly, the two facets relate very differently to extraversion, conscientiousness and impulsivity. In addition, it appears that only the Appetitive Motivation facet behaved as a core BAS measure as defined within r-RST, whilst the impulsive hedonism facet was aligned with impulsivity, and therefore it is doubtful whether it should be regarded as an indicator of core BAS sensitivity (e.g., Smillie & Jackson, 2006). In order to gain further information regarding which AMS component best reflects core BAS sensitivity, we regressed in a Structural Equation Model the three CW-BAS components on the two AMS factors. Hypothesis 3 anticipated a closer association between the Approach Motivation facet of AMS and Drive and RewardResponsiveness, the two CW-BAS scales which most closely approximate BAS sensitivity (Leone, 2009; Smillie, Jackson et al., 2006). The fit can be considered acceptable: v2 (112) = 351.18, p < .001, RMSEA = .074 (90% CI for RMSEA: .066–.083), NNFI = .93, CFI = .94. As anticipated, a clear dissociation pattern was found: the Appetitive Motivation factor was connected to Drive (c = .37, p < .01) and Reward-Responsiveness (c = .33, p < .01), but not to Fun (c = .08, ns), whilst the Impulsive facet of the AMS was closely aligned with Fun (c = .63, p < . 01), but not connected to Drive (c = .11, ns) or Reward-Responsiveness (c = .07, ns) (the CW-BAS facets considered to be markers of the core BAS; Smillie, Jackson et al., 2006). Previous evidence (Leone & Russo, 2009) has shown that the CW-BAS facets of Drive and Fun are related to functional and dysfunctional impulsivity in different ways, suggesting that only drive is aligned with the functional impulsive functions supervised by 3 Completely standardized parameters are reported. 659 the BAS (Smillie & Jackson, 2006). In an analogous fashion, Hypothesis 4 hypothesized that only the Appetitive Motivation factor of the AMS would lead to functional impulsivity (Jackson, 2009), whilst the Impulsive facet of the AMS should instead be uniquely associated with dysfunctional impulsivity. In order to control for the association between the two forms of impulsivity, dysfunctional effects on functional impulsivity were partialled out. When functional and dysfunctional impulsivity were regressed on the two AMS factors, a barely acceptable fit was found: v2 (100) = 343.85, p < .001, RMSEA = .079 (90% CI for RMSEA: .070– .089), NNFI = .91, CFI = .93. As anticipated, appetitive motivation was suitably aligned with functional impulsivity (c = .31, p < .01) and negatively connected with dysfunctional impulsivity (c = .42, p < .01); such divergent pattern appears consistent with an ideal BAS measure (Jackson, 2009; Leone & Russo, 2009). In turn, the Impulsivity items of the AMS were positively related to dysfunctional impulsivity (c = .44, p < .01), and unrelated with functional impulsivity (c = .16, ns). In accordance with results from the previous models, this pattern suggests that only the subset of items of the AMS loading on the Appetitive Motivation facet represents the core BAS. 5. Discussion We reasoned that the AMS could be better modeled by specifying two factors, one related to core BAS functioning and the other to impulsivity. Such a model appears to outperform the onedimensional alternatives. We anticipated that the Appetitive Motivation facet would show a pattern of associations which were highly consistent with the empirical definitions of core BAS sensitivity (i.e., linkages with the Extraversion, Drive and RewardResponsiveness facets of the CW-BAS, and with functional impulsivity). Instead, the Impulsive facet of the AMS was expected to relate to classical and dysfunctional impulsivity and with BAS indicators known to share significant variance with impulsivity. These hypotheses were empirically supported. Furthermore, the impulsive cluster within the AMS was weakly related with extraversion, compared with its Appetitive Motivation counterpart. In brief, as is the case for the CW-BAS, the AMS represents a complex operationalization of BAS sensitivity. Nevertheless, from our perspective, such complexity is not necessarily a pitfall of the measure; instead, it can be interpreted as a useful feature of the AMS, as it allows researchers to test different hypotheses that take into account different BAS-related processes which occur in different stages of reward pursuit (Corr, 2008; Heym et al., 2008). We believe that BAS-related scales often fail to show simple onedimensional structures because the complex construct they intend to target is difficult to fit into a one-dimensional Procrustean bed, and not because of banal psychometric misspecifications. We will discuss these issues in greater depth below. 5.1. The complex structure of BAS measures: Is it really so bad? Cooper et al.’s (2008) revision of the AMS acknowledged that empirical results may cast some doubts on the one-dimensionality of the scale. Thus, an alternative strategy which allows for multidimensionality is feasible, as our results suggest, and – arguably – provides a better account of the data. The two factors corresponding to ‘‘core’’ appetitive motivation and to impulsivity depart significantly from a one-dimensional perspective. However, we are not suggesting that there is something wrong with the AMS; instead, we are convinced that researchers can find in the AMS a sensitive and efficient tool with which to disentangle impulsivityrelated variance in BAS criteria from core appetitive motivation variance. As has been shown for the CW-BAS scales (Smillie, 660 L. Leone et al. / Journal of Research in Personality 45 (2011) 655–661 Jackson et al., 2006), it is possible to detect within the AMS a reflection of the reward-reactivity versus trait impulsivity distinction (e.g., Quilty & Oakman, 2004). It would be interesting to ascertain whether an appetitive motivation score devoid of almost all conflation with impulsivity is able to predict the behavioral criteria which have been investigated in previous research relying on the AMS full score (Smillie & Jackson, 2005; Smillie & Jackson, 2006). Our results suggest that the 20-item score is conflated with variance relating to impulsivity, and lacks specificity. It could thus be argued that previous results underestimate trait effects on behavioral criteria. The leaner and more specific factor we propose as indexing appetitive motivation may reveal stronger associations. The impulsivity-related cluster of AMS items may turn out to be useful as well. First, AMS impulsivity may act as an efficient statistical control for any expected effect governed by appetitive motivation. Second, as some behavioral criteria may stem from a mixture of SR and impulsive action orientation, the two factors of the AMS may together provide a richer account of complex behaviors. Finally, the Impulsive facet may also reveal how, under some circumstances in need of theoretical definition, behaviors, which are usually governed by the BAS, may turn out to be dysfunctionally influenced by impulsivity. Such interactions between behaviors, situational or personal contingencies and traits (appetitive motivation vs. impulsivity) may help to improve models of maladaptive behaviors such as addictions, risk-taking and delinquency which have been rather simplistically traced back to the BAS or to impulsivity. 5.2. Implications for the conceptualization of the BAS and impulsivity Over the past few years, the r-RST and most empirical research have converged on the notion that extraversion (and not impulsivity or psychoticism) can be considered as the appropriate BAS at the surface personality level (e.g., Depue & Collins, 1999; Smillie, Pickering et al., 2006; Jackson, 2009). Nevertheless, empirical evidence has consistently shown that most purpose-built BAS measures include some impulsivity variance. For instance, the Fun scale of the CW-BAS is more closely aligned to impulsivity than to the core BAS (Leone, 2009; Smillie, Jackson et al., 2006). In a remarkably similar vein, the AMS can be modeled with two components which are aligned to the core BAS and to impulsivity. Different degrees of theoretical and empirical misspecifications of these purpose-built BAS measures could account for such pervasive conflation. A first source of confusion traces back to earlier versions of the RST that explicitly connected BAS and impulsivity. Nevertheless, it appears that theoretical reasons may account for the BAS–impulsivity conflation even within the revised RST framework. Even though impulsivity and the BAS may correspond to independent physiological substrata, their functional outcomes may be related because of some common mediator (Smillie, Jackson et al., 2006). Another possibility is that the BAS and impulsivity manifestations have become associated because their manifestations stem from the joint effects of several systems regulating activation and inhibition (Corr, 2002). As the behavioral outcomes of the BAS are affected by the activity in other systems which are related to conflict detection, inhibition and flight or fight, it may be possible that the outcomes share variance with impulsivity measures and indices of impulsive tendencies via indirect paths of association (Smillie, Jackson et al., 2006). A different account of BAS-impulsivity associations has conceptualized the BAS as encompassing different functions during goal pursuit: a behavior maintenance function and a consumatory function (Carver & Miller, 2006; Corr, 2008). Impulsivity would be part of the consumatory function, and would kick in once the likelihood of attaining the reward became favorable and the incentive was within reach. In such instances, unleashing impulsive behavior in order to attain the goal in question may be considered to be adaptive and consistent with a definition of BAS as appetitive motivation and functional impulsivity (Smillie & Jackson, 2006). A similar account has been proposed in order to tentatively describe the different functions of the three CW-BAS scales during different phases of complex incentive-pursuit: Drive and Reward-Responsiveness provide the behavioral maintenance motivation needed during complex approach behavior involving multiple sub-goals, whilst the Fun-Seeking component kicks in when the incentive needs to be seized swiftly (Corr, 2008; Leone, 2009). Our pattern of results is consistent with the interpretation of the Appetitive Motivation facet that views the AMS as tapping into the behavioral maintenance function of BAS. Support for this notion relies on the negative association between appetitive motivation and impulsivity and its positive linkage with conscientiousness; this pattern is compounded by the significant association between the appetitive motivation facet and the CW-BAS Drive and Reward-Responsiveness scales, and its lack of an association with the CW-BAS Fun scale. Further support is provided by the unique effect of appetitive motivation on functional impulsivity. In turn, the impulsive facet within the AMS parallels fun-seeking in the CW-BAS, and it is related to both classical and dysfunctional impulsivity. Applying Corr’s (2008) reasoning to the AMS, the impulsive component may represent a theoretically significant part of BAS activity when the incentive needs to be seized. Instead, when the maintenance function of BAS is weakened, premature unleashing of the impulsive function may eventually disrupt goal attainment. Of course, such time-lagged and interactive interplay of the different BAS components during different phases of goal pursuit needs to be tested via experimental research, by manipulating the attainability and proximity of the incentive, as well as the complexity of the sub-goal structure leading to the final reward. If the BAS-impulsivity conflation does indeed stem from complex interactions at the physiological and behavioral levels, it should not be surprising that it is difficult to construct BAS measures which are devoid of impulsivity-related content and which are orthogonal to impulsivity measures. Responses on personality questionnaires may be considered as one instance of outcomes of the BAS (Smillie, Jackson et al., 2006), which often implies a degree of overlap with impulsivity-related outcomes (Smillie & Jackson, 2005). One-to-one correspondence between pure BAS measures and BAS sensitivity on the one side, and pure impulsivity measures and the basic process governing impulsivity on the other, appears to be a difficult standard to attain, and no single scale developed under the original RST framework satisfactorily respond to these requirements. Measures explicitly developed under the assumptions of the r-RST may do a better job, as suggested by recent evidence (Jackson, 2009). Nevertheless, it appears to be possible to devise within ‘‘classical’’ RST measures as the AMS and the Carver and White scales relatively clean factors satisfactorily matching the measurement requirements of the r-RST (Heym et al., 2008). In the case of classical BAS measures, it appears feasible to identify components more closely related to the behavioral maintenance function of the BAS and components aligned with the consumatory component of the BAS (Corr, 2008). Such distinct factors may be used to test discriminant, dissociation, mediation or interaction hypotheses that cannot be tested when the BAS and impulsivity components are collapsed into a single score. Some limitations should also be outlined. Our results rely on a single sample, and replications are warranted. However, the sample is not small by the average standards of personality research, and – unlike the typical samples in personality research – does not include psychology students. In addition, the AMS items are all directed ‘‘pro-trait.’’ This method factor may play a pervasive role in all the analyses; of course, this method bias is common amongst any previous research studies which have also used the L. Leone et al. / Journal of Research in Personality 45 (2011) 655–661 AMS. Common method bias artificially inflates the strength of onefactor structures, and makes it more difficult to devise alternative solutions. In spite of any effects of method bias, our two factors were clearly preferable over the one-factor solutions and provided a clearer pattern of associations with a wide array of criteria. Finally, our results may be, to some extent, dependent on the specific characteristics of our Italian sample of participants; however, this could be interpreted as a positive feature, because our results replicate and generalize previous findings using different measures and different cultural contexts (e.g., Leone & Russo, 2009; Smillie, Jackson et al., 2006; Smillie & Jackson, 2006). It appears unlikely that such conceptual consistency would result from any specific bias occurring in our sample because of methodological artifacts or culture-specific variance. More likely, consistency of results across samples, cultures and methods points to robust and generalizable effects. References Caprara, G. V., Schwartz, S., Capanna, C., Vecchione, M., & Barbaranelli, C. (2006). Personality and politics: Values, traits and political choice. Political Psychology, 27, 1–28. Carver, C. S. (2004). Negative affects deriving from the behavioral approach system. Emotion, 4, 3–22. Carver, C. S., & Miller, C. J. (2006). Relations of serotonin function to personality: Current views and a key methodological issue. Psychiatry Research, 144, 1–15. Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS scales. Journal of Personality and Social Psychology, 67, 319–333. Caseras, X., Avila, C., & Torrubia, R. (2003). The measurement of individual differences in behavioral inhibition and behavioral activation systems: A comparison of personality scales. Personality and Individual Differences, 34, 999–1013. Cooper, A. J., Smillie, L. D., & Jackson, C. J. (2008). A trait conceptualization of Reward-Reactivity. Psychometric properties of the Appetitive Motivation Scale (AMS). Journal of Individual Differences, 29, 168–180. Corr, P. J. (2002). J.A. Gray’s Reinforcement Sensitivity Theory: Tests of the joint subsystems hypothesis of anxiety and impulsivity. Personality and Individual Differences, 33, 511–532. Corr, P. J. (2008). Reinforcement sensitivity theory (RST): Introduction. In P. J. Corr (Ed.), The reinforcement sensitivity theory of personality (pp. 1–43). Cambridge: Cambridge University Press. Depue, R. A., & Collins, P. F. (1999). Neurobiology of the structure of personality: Dopamine, facilitation of incentive motivation, and extraversion. Behavioural and Brain Sciences, 22, 491–569. Dickman, S. J. (1990). Functional and dysfunctional impulsivity: Personality and cognitive correlates. Journal of Personality and Social Psychology, 58, 95–102. Eysenck, S. B. G. (1993). The I7: Development of a measure of impulsivity and its relationships to the superfactors of personality. In W. G. McCown, J. L. Johnson, & M. B. Shure (Eds.), The impulsive client: Theory research and treatment (pp. 141–149). Washington, DC: American Psychological Association. Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, F. J. (1985). Age norms for impulsiveness, venturesomeness and empathy in adults. Personality and Individual Differences, 6, 613–619. Gray, J. A. (1982). The neuropsychology of anxiety: An enquiry into the functions of the septo-hippocampal system. Oxford: Oxford University Press. Gray, J. A., & McNaughton, N. (2000). The neuropsychology of anxiety: An inquiry into the functions of the septo-hippocampal system (2nd ed.). Oxford: Oxford University Press. Heym, N., Ferguson, E., & Lawrence, C. (2008). An evaluation of the relationship between Gray’s revised RST and Eysenck’s PEN: Distinguishing BIS and FFFS in Carver and White’s BIS/BAS scales. Personality and Individual Differences, 45, 709–715. Heym, N., & Lawrence, C. (2010). The role of Gray’s revised RST in the Ppsychopathy continuum: The relationships of psychoticism with a lack of fear and anxiety, and increased impulsivity. Journal of Personality and Individual Differences, 49, 874–879. Horn, J. L. (1965). A rationale and test for the number of factors in factor analysis. Psychometrika, 32, 179–185. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modelling, 6, 1–55. Jackson, C. J. (2003). Gray’s Reinforcement Sensitivity Theory: a psychometric critique. Personality and Individual Differences, 34, 533–544. Jackson, C. J. (2009). Jackson-5 scales of the revised Reinforcement Sensitivity Theory (r-RST) and their application to dysfunctional real world outcomes. Journal of Research in Personality, 43, 556–569. 661 Jackson, C. J., & Smillie, L. D. (2004). Appetitive motivation predicts the majority of personality and an ability measure: A comparison of BAS measures and a reevaluation of the importance of RST. Personality and Individual Differences, 36, 1627–1636. Jöreskog, K. (2005). Structural equation modelling with ordinal variables using LISREL. Chicago: Scientific Software. Jöreskog, K., & Moustaki, I., (2006) Factor analysis of ordinal variables with full information maximum likelihood. <http://www.ssicentral.com/lisrel/techdocs/ orfiml.pdf>. Jorm, A. F., Christensen, H., Henderson, A. S., Jacomb, P. A., Korten, A. E., & Rodgers, B. (1999). Using the BIS/BAS scales to measure behavioural inhibition and behavioural activation: Factor structure, validity and norms in a large community sample. Personality and Individual Differences, 26, 49–58. Leone, L. (2009). Testing conceptual distinctions among Carver and White’s (1994) BAS scales: A comment and extension on Smillie, Jackson, and Dalgleish (2006). Personality and Individual Differences, 46, 54–59. Leone, L., Perugini, M., Bagozzi, R. P., Pierro, A., & Mannetti, L. (2001). Construct validity and generalizability of the carver white behavioural inhibition system/ behavioural activation system scales. European Journal of Personality, 15, 373–390. Leone, L., Pierro, A., & Mannetti, L. (2002). Validity of the Italian version of the BIS/ BAS Scale of Carver and White (1994): Generalizability of structure and relationships with related constructs. Giornale Italiano di Psicologia, 29, 413–434. Leone, L., & Russo, P. M. (2009). Components of the behavioral activation system and functional impulsivity: A test of discriminant hypotheses. Personality and Individual Differences, 43, 1101–1104. Newman, J. P., MacCoon, D. G., Vaughn, L. J., & Sadeh, N. (2005). Validating a distinction between primary and secondary psychopathy with measures of Gray’s BIS and BAS constructs. Journal of Abnormal Psychology, 114, 319–323. O’Connor, R. M., Stewart, S. H., & Watt, M. C. (2009). Distinguishing BAS risk for university students’ drinking, smoking, and gambling behaviors. Personality and Individual Differences, 46, 514–519. Pickering, A. D., & Gray, J. A. (1999). The neuroscience of personality. In L. Pervin & O. John (Eds.), Handbook of personality (2nd ed., pp. 277–299). New York: Guilford. Pickering, A. D., & Smillie, L. D. (2008). The behavioural activation system: Challenges and opportunities. In P. J. Corr (Ed.), The reinforcement sensitivity theory of personality (pp. 120–153). Cambridge: Cambridge University Press. Quilty, L. C., & Oakman, J. M. (2004). The assessment of behavioural activation the relationship between impulsivity and behavioural activation. Personality and Individual Differences, 37, 429–442. Russo, P. M., Leone, L., & De Pascalis, V. (2011). Cross-cultural validity of the I7 impulsiveness – Venturosomeness – Empathy scales: Evidence from the Italian, I7. Comprehensive Psychiatry, 52, 446–452. Satorra, A., & Bentler, P. M. (1988). Scaling corrections for chi-square statistics in covariance structure analysis. Proceedings of the Business and Economic Statistics Section of the American Statistical Association, 308, 313. Smillie, L. D., & Jackson, C. J. (2005). The Appetitive Motivation Scale and other BAS measures in the prediction of approach and active avoidance. Personality and Individual Differences, 38, 981–994. Smillie, L., & Jackson, C. J. (2006). Functional impulsivity and reinforcement sensitivity theory. Journal of Personality, 74, 47–83. Smillie, L. D., Jackson, C. J., & Dalgleish, L. I. (2006). Conceptual distinctions among Carver and White’s (1994) BAS scales: A reward reactivity versus trait impulsivity perspective. Personality and Individual Differences, 40, 1039–1050. Smillie, L. D., Pickering, A. D., & Jackson, C. J. (2006). The new reinforcement sensitivity theory: Implications for personality measurement. Personality and Social Psychology Review, 10, 320–335. Urosevic, S., Abramson, L. Y., Harmon-Jones, E., & Alloy, L. B. (2008). Dysregulation of the behavioral approach system (BAS) in bipolar spectrum disorders: Review of theory and evidence. Clinical Psychology Review, 28, 1188–1205. Velicer, W. F. (1976). Determining the number of factors from the matrix of partial correlations. Psychometrika, 41, 321–327. Velicer, W. F., Eaton, C. A., & Fava, J. L. (2000). Construct explication through factor or component analysis: A review and evaluation of alternative procedures for determining the number of factors or components. In R. D. Goffin & E. Helmes (Eds.), Problems and solutions in human assessment: Honoring Douglas N. Jackson at seventy (pp. 41–72). Norwell, MA: Kluwer Academic. Voigt, D. C., Dillard, J. P., Braddock, K. H., Anderson, J. W., Sopory, P., & Stephenson, M. T. (2009). BIS/BAS scales and their relationship to risky health behaviours. Personality and Individual Differences, 47, 89–93. Wilson, G. D., Gray, J. A., & Barrett, P. T. (1990). A factor analysis of the Gray–Wilson personality questionnaire. Personality and Individual Differences, 11, 1037–1045. Zisserson, R. N., & Palfai, T. P. (2007). Behavioral activation system (BAS) sensitivity and reactivity to alcohol cues among hazardous drinkers. Addictive Behaviors, 32, 2178–2186. Zumbo, B. D., Gadermann, A. M., & Zeisser, C. (2007). Ordinal versions of coefficients alpha and theta for likert rating scale. Journal of Modern Applied Statistical Methods, 6, 21–26. Zwick, W. R., & Velicer, W. F. (1986). Factor influencing five rules for determining the number of components to retain. Psychological Bulletin, 99, 432–442.