Introduction

People struggle to resist various desires every day (Hofmann et al. 2012, 2013). A majority of these desires are temptations: short-term gains that conflict with one’s pursuit of long-term goals (Fujita 2011). Because temptations can elicit both appetitive and aversive responses, two fundamental motivational systems (Cacioppo et al. 1999; Gray 1987, 1994; Lang 1995)—the appetitive and aversive systems—regulate people’s responses to temptations. We argue that past research investigating individual differences related to the two systems have focused primarily on the behavioral regulation of the two systems (e.g., BIS/BAS; Carver and White 1994). Here, we consider how the two motivational systems interplay and regulate affective activation. Specifically, we aim to explore if the relative activation of one system over the other, captured by the Motivated Action Measure (MAM: Lang et al. 2005, miniMAM: 2011), may be a better predictor for people’s spontaneous affective reactions toward temptations than measures related to behavioral regulation.

Appetitive, aversive motivational systems, and temptations

Two motivational systems underlie our interactions with the social environment (Cacioppo et al. 1999; Gray 1987, 1994; Lang 1995). The appetitive system determines how readily one would approach and engage with the stimulus, deem it to be safe or appealing, and experience affective positivity towards it. The aversive system determines how defensively one would react and avoid the stimulus, deem it to be harmful or threatening, and experience affective negativity towards it (Lee and Lang 2009; Potter et al. 2006). Both systems are evolutionarily adaptive, with the appetitive system promoting exploration and foraging behavior and the aversive system enabling timely withdrawal from danger (Ito and Cacioppo 2005; Lang and Bradley 2013). Given the importance of the two systems in human survival, scholars have been interested in capturing individual differences in how they regulate behavior. The predominant measures in the literature are the behavioral inhibition and activation system (BIS/BAS) scales (Carver and White 1994). The BIS/BAS was designed to capture the regulation of approach and avoidance behavior by the appetitive and aversive systems, respectively (Carver and White 1994; Cooper et al. 2007; Leone et al. 2001). BIS measures the degree to which the person will move away from (avoid) an undesirable, unpleasant goal or punishment; BAS measures the degree to which a person will act towards (approach) a desirable goal or reward. BIS/BAS scales have been shown to predict impulsivity (Caseras et al. 2003; Dawe and Loxton 2004; Jorm et al. 1998), alcoholic tendencies (Franken 2002; Zisserson and Palfai 2007), as well as substance use and risky health behaviors (Franken and Muris 2006a). In general, BIS negatively and BAS positively predict impulsive, detrimental behaviors.

Individual differences in the activations of the two systems should contribute to people’s desires toward temptations because temptations can elicit both appetitive and aversive responses. Temptations have two key features: short-term gains that also conflict with one’s pursuit of long-term goals (Fujita 2011). Partaking in temptation offers immediate gratification (e.g., consuming alcohol), but may impede our long-term goals (e.g., being productive). Thus, temptations can activate two opposing affective responses. First, the hedonic and often visceral gratification associated with temptations (Loewenstein 1996) can activate appetitive responses. Second, because temptations are instrumentally incongruent with long-term goal pursuit and succumbing to them also reflects negatively on one’s self-efficacy (Rothman et al. 2011), they can activate aversive responses in the forms of negative evaluation (Ferguson and Bargh 2004; Fishbach et al. 2010) and negative emotions such as regret (Abraham and Sheeran 2003) and shame (Patrick et al. 2009). Consequently, temptations should evoke a strong appetitive response from people with high BAS and a strong aversive response from people with high BIS. How the appetitive and aversive systems regulate responses, thus, should predict where people’s desire toward temptations comes from.

Evidence for this prediction, however, has been inconsistent. Individual differences in the affective and aversive systems as captured by BIS/BAS have shown mixed influence on people’s responses to desires. Although individuals with higher BAS experience desires with more intensity (Franken 2002; Zisserson and Palfai 2007), they are also more likely to resist the temptation (Dholakia et al. 2006). Hofmann and colleagues (2012) conducted the most comprehensive investigation of people’s daily experience and coping of desires (i.e., food, social, leisure) to date. Their findings suggest that BIS/BAS seemed to predict more of how people regulate desires than how desires are originated. First, BAS predicted people’s self-reported strength, but not frequency, of desires. Put another way, BAS had no bearing on people’s susceptibility in experiencing desires but only affected how intensely they experience it after the desire had occurred. Second, people’s self-reported desire strength may inherently encompass conscious control, for it was also predicted, negatively, by trait self-control. Again, BAS might have predicted self-reported desire strength by virtue of its association with behavioral regulation rather than affective origins of desires. Finally, BIS predicted neither frequency nor strength of desires. Instead, BIS, together with trait self-control, negatively predicted subsequent behavioral enactment of desires. This again indicates that BIS/BAS mostly taps into the way appetitive and aversive systems regulate people’s behavioral responses toward temptations or desires.

Affective activation of the motivational systems

The appetitive and aversive systems may, however, regulate not only behavioral responses but also affective activation toward temptations. The appetitive system may activate the initial appetitive feeling toward temptations as well as regulate the subsequent approach behavior; similarly, the aversive system may be in charge of both activating the initial aversive feeling and regulating the subsequent avoidant behavior toward temptations. Accordingly, because the conceptualization of BIS/BAS scales focuses on behavioral regulation (Carver and White 1994), it might not be capturing the way the two motivational systems regulate affective activation. Our distinction between affective activation and behavioral regulation is consistent with recent models of desire and self-control. Most notably, Kotabe and Hofmann (2015) distinguish between desire and control, whereby the final behavioral outcome toward the desirable target results from the relative potency of desire and control. High control does not necessitate abstinence if the desire is overwhelming. Conversely, low desire may be enacted if the control against it is weak. Here, we similarly argue that the appetitive and aversive systems are relevant to both affective activation (or desire) and behavioral control. Understanding individual differences in how the interplay of the appetitive and aversive systems regulate affective activation would thus complement past focus on behavioral regulation, painting a fuller picture of the relationship between the motivational systems and people’s desire toward temptations.

Past research suggests that the two motivational systems regulate affective activation through two signature activation patterns (Ito and Cacioppo 2005; Lang et al. 2005). The first signature pattern, termed the positivity offset, focuses on the relative activation of the appetitive system over the aversive system. This pattern of activation emerges when the person’s appetitive system is relatively more responsive (readily activated) than the aversive system in emotionally neutral situations (i.e., safe, peaceful and comfortable contexts). The person thus assumes a positive outlook when entering a social environment, ready to engage and explore the environment. The second signature pattern, termed as the negativity bias, depicts the relative activation of the aversive system over the appetitive system. This pattern emerges when the aversive system readily responds with more intensity than the appetitive system, when the situation increases in arousal. In other words, while the aversive system is less ready to fire up than the appetitive system in a neutral state, as soon as a threatening stimulus appears in the environment, the activation of the aversive system quickly overwhelms the appetitive system to motivate avoidant responses (Fig. 1). The signature activation patterns of the appetitive and aversive systems—positivity offset and negativity bias—allow a person to retain both curiosity and cautiousness about the surroundings. The person is motivated to explore in safety and prepared to avoid danger if the situation turns for the worse.

Fig. 1
figure 1

(adapted from Lang et al. 2005). Positivity offset is represented by a higher activation strength of the appetitive system over aversive system at intercept (low arousal). Negativity bias is represented by the steeper slope with which the aversive system rises in activation than the appetitive system as arousal increases

Relative activation of the appetitive and aversive motivational systems

Full size image

How can we measure these two signature activation patterns? The Motivation Action Measure (MAM; Lang et al. 2005) has been utilized to capture individual differences in positivity offset and negativity bias. In MAM, participants view a range of positive and negative pictures and report their feelings in terms of valence and arousal. Two indices, approach system activation (ASA) and defensive system activation (DSA), can be computed from participants’ feelings. The computation of ASA captures the signature activation pattern of the positivity offset, while the computation of DSA captures the signature activation pattern of the negativity bias. Hence, individual differences in ASA would show how readily the individual’s appetitive system activates over the aversive system at low levels of arousal, and higher ASA predisposes an individual with a greater likelihood of a resultant positive affective response. Conversely, individual differences in DSA would show how readily the individual’s aversive system activates over the appetitive system as arousal increases, and higher DSA predisposes the individual with a greater likelihood of a resultant negative affective response (Lang et al. 2007; Wang et al. 2011).

Research has found that ASA and DSA exert independent and opposite effects on people’s affective responses. Whereby high ASA has been linked to high sensation seeking and greater alcohol use, high DSA tends to be found among low sensation seekers (Finn 2002; Joseph et al. 2009; Rubenking and Lang 2015). In the context of substance abuse, people with high ASA demonstrated more positive physiological responses (i.e., startle reflex, facial muscle activity) towards images of the addictive substances, while individuals with high DSA displayed stronger physiological responses of withdrawal from the images (Lang and Yegiyan 2011). Even in the context of media consumption, whereby strong physiological addiction is largely absent, ASA and DSA predict people’s preferences. People with higher ASA prefer more risqué, competitive, and violent content such as pornography and violent video games, while those with DSA were more inclined towards safer, traditional forms of entertainment such as news and puzzle games (Alhabash et al. 2014; Krcmar et al. 2015; Potter et al. 2011).

Despite the clear overlap in conceptualization and empirical evidence, the MAM and BIS/BAS have not been investigated together. In the present paper, we aim to bridge the gap in the literature by considering affective activation in addition to behavioral regulation. We included the MAM for additional predictive utility, as it directly captures individuals’ affective response to temptations. Specifically, ASA and DSA should contribute to the relative strength of the appetitive and aversive responses elicited by temptations. As discussed, temptations embody features that respectively activate the affective and aversive systems: hedonic positivity from immediate, often visceral gratification (Loewenstein 1996), versus instrumental negativity from being a detriment to one’s efforts to fulfil beneficial long-term goals (Abraham and Sheeran 2003; Ferguson and Bargh 2004; Fujita 2011; Patrick et al. 2009). The pronounced pattern of positivity offset associated with high ASA would signify that the hedonic positivity of a temptation activates stronger appetitive responses than instrumental negativity activates aversive responses. Hence, individuals with high, as compared to low, ASA should generate a larger offset of appetitive over aversive responses from temptations. This initial offset can be countered by the activation of the aversive system. The negativity bias allows the instrumental negativity of temptations to activate an intense aversive response. DSA indexes the effectiveness of this negativity bias. People with high DSA should thus generate a stronger aversive response to temptations than those with low DSA, and should be better able to refrain and withdraw from the temptation.

Hence, the MAM and the BIS/BAS differ in that the MAM measures the trait affective reaction toward the temptation targets, while the BIS/BAS measures the trait propensity to behaviorally consume those objects. Though they both deal with trait level motivational systems, MAM is uniquely suited in determining if a temptation target will elicit desire. It gauges the allure of the temptation object through the combined activations of appetitive and approach motivations to determine the resultant affective experience. In contrast, the BIS/BAS measures the propensity to act on the temptation. In sum, we predict that ASA and DSA guide people’s spontaneous desire towards temptations, albeit in opposite directions. On the other hand, we expect the BIS/BAS to be less instrumental in predicting such spontaneous responses, but more predictive of people’s behavioral tendency to indulge in temptations.

To examine whether the MAM has any predictive utility over the BIS/BAS, we measure both affective and behavioral responses to temptation. We aim to capture people’s spontaneous affective responses to temptations in the form of implicit attitude. Attitude is the evaluative summary of affective responses (Fazio and Olson 2014), and implicit attitude in particular may represent the computational outcome from affective activations of the appetitive and aversive systems. The dual-process models of self-regulation (e.g., Strack and Deutsch 2004) make the distinction between an association-based, resource-independent impulsive system and a rule-based, resource-dependent reflective system. The affective systems are thought to exert direct influence in the impulsive system; the reflective system, when the person has adequate cognitive resources and motivation, may further modulate the behavioral outcome. Implicit attitude has been deemed to represent the evaluative summary in the impulsive system. Indeed, implicit attitude predicts behavior especially when one’s ability for controlled processing is diminished (Hofmann et al. 2007, 2008). Hence, we posit that people’s implicit attitudes reflect the outcome of their impulsive affective activations of the two motivational systems. In addition, we measure people’s behavioral responses toward temptations through a preference task, where they choose between tempting or non-tempting options. We argue that like previous studies (Hofmann et al. 2012), people’s responses on such explicit measures may be the result of behavioral regulation rather than affective activation.

Overview of current research

We test our predictions in a dieting context. Dieting revolves around achieving an ideal body appearance or healthy body in the long run, but dieters often meet with temptations in the form of tasty, unhealthy food. The consumption of such food gives immediate pleasure, but holds negative consequences such as weight gain or health issues contrary to the individual’s long-term goal. As such, we are interested in how dieters regulate their affective activation and behavioral responses toward tasty, unhealthy food.

We employed the miniMAM, a brief and validated version of the MAM (Lang et al. 2011), and the BIS/BAS (Carver and White 1994) to index affective activation and behavioral regulation of the appetitive and aversive systems, respectively. As another index of behavioral regulation to contrast with the MAM, we measured participants’ trait self-control using the Brief Self-Control scale (BSC: Tangney et al. 2004). Trait self-control has been construed as the ability to overcome instinctual response tendencies for self-regulatory purposes (Metcalfe and Mischel 1999; Baumeister and Vohs 2004), focusing on the effortful promotion of desirable responses and inhibition of undesirable responses (Baumeister et al. 1994; Muraven and Baumeister 2000; Tangney et al. 2004). Past research has shown that the BSC, like the BIS/BAS, is strongly related to behavioral regulation (e.g., Hofmann et al. 2012). As such, showing that MAM predicts implicit attitude toward temptations over and above both the BIS/BAS and BSC, two trait individual differences of behavioral regulation, would lend strong support to our prediction that the MAM uniquely indexes affective activation. Conversely, if both the BIS/BAS and BSC are more predictive of participants’ behavioural responses than the MAM, we would be more confident that the MAM is much less involved in behavioural regulation. Hence, the BSC would allow us to better elucidate how MAM accounts for the affective origin of temptations.

Participants’ implicit attitude towards tempting foods (e.g., cakes, burgers) was measured using an Affect Misattribution Procedure (Payne et al. 2005). Participants’ behavioral intention was measured through a food choice task. They had to choose between receiving healthy or unhealthy foods as possible parting gifts. The choice of an unhealthy food indicates indulgence in temptation, against the dieting goal. We predict that participants’ implicit attitude toward tempting foods is positively predicted by ASA and negatively predicted by DSA, over and above BIS/BAS and BSC. ASA and DSA, however, should be less effective with regard to participants’ behavioral intention towards temptation.

Method

Participants

From a mass survey during a university lecture, 358 students who indicated they had a dieting goal were recruited for the current study. Forty-two participants did not complete the full study and four additional participants had computer errors. They were hence excluded from the analyses. Of the remaining 312 participants (M = 18.9 years, SD = 1.3 years), 213 were female. Because correlations between individual differences stabilize as the sample size approaches 250 (Schönbrodt and Perugini 2013), our sample size is appropriate for our investigation. Similarly, power analysisFootnote 1 indicates that we have high power to detect small to medium effect sizes (Cohen 1988).

Procedure

Participants turned up for two sessions, separated by about a week (5–9 days). During the first session, participants completed the mini Motivation Activation Measure (Lang et al. 2011). Participants were presented with 35 pictures (none of the pictures are about food) of varying arousal (categorized into 6 arousal levels), selected from the International Affective Picture System (IAPS). For each picture, they rated its arousal first, followed by positive and negative content in random order using a 9-point scale, 1 (not at all) to 9 (extremely). The pictures were randomly presented to prevent order-effects.

We computed ASA and DSA according to Lang et al. (2005, 2011). The computation of ASA follows the characterization of the positivity offset: the readiness of activation of the appetitive over the aversive system in a neutral setting. Two mean positivity ratings were calculated. First, participants’ mean positivity rating for the seven positive images highest in arousal (level 6) was calculated. These mean positivity ratings indicated the maximum activation of the appetitive system towards highly arousing pleasant stimuli. Second, participants’ mean positivity ratings for all 14 positive and negative images lowest in arousal (level 1) were calculated. These mean positivity ratings indicated the resting activation of the appetitive system. The difference between the first and second mean positive ratings is then taken as the readiness of activation of the appetitive system.

The computation of DSA follows the characterization of the negativity bias: the readiness of activation of the aversive over the appetitive system at rising levels of arousal. Two mean negativity ratings were calculated. First, participants’ mean negativity rating for the seven negative images of medium level of arousal (level 3/4) was calculated. These mean negativity ratings indicated the activation of the aversive system towards stimuli which are slightly more negative than neutral. Second, participants’ mean negativity ratings for all 14 positive and negative images lowest in arousal (level 1) were calculated. These mean negativity ratings indicated the resting activation of the aversive system. The difference between the first and second mean negative rating for negative images is taken as the readiness of activation of the aversive system.

Participants’ activations of the behavioral inhibition and approach systems were measured using the BIS/BAS scale (Carver and White 1994). For the current study, 13 items (α = 0.80) measuring BAS-related factors; drive, fun seeking, and reward responsiveness were averaged for each factor and summed together as participants’ overall BAS. The correlations for these subscales can be found in Table 1. Seven items were averaged as participants’ BIS (α = 0.72), following the normal procedure. Participants also completed the brief self-control scale (α = 0.81, BSC: Tangney et al. 2004), and other unrelated measures that included demographic questions.

Table 1 Correlations between ASA, DSA, other predictor variables and food primes
Full size table

During the second session, participants completed the Affect Misattribution Procedure (Payne et al. 2005) to examine their implicit attitudes towards tempting foods. Participants were presented with pairs of images flashed sequentially, first with an image of the attitude object (prime stimuli) followed by a Chinese pictograph. They were asked to do nothing with the first image and only judge the visual pleasantness of the Chinese pictograph. For each pictograph, they were instructed to press keys to indicate if the pictograph was more or less visually pleasant than average. Their judgments are assumed to reflect the (misattributed) affective response to the prime (i.e., attitude object) presented before (Blaison et al. 2012). The prime stimuli in the present study were images of tempting foods (e.g., pizza, chocolate, cake), which have been shown to be adequate in eliciting consumption responses (Bailey 2014), and images of gray squares and neutral objects (e.g., buildings, trees) were used as controls. The inter-stimulus interval (ISI) between the prime and the pictograph was varied to be 100 and 1000 ms for each half of the stimuli respectively, as this ensured the spontaneity of each response (Davis 1970; Kornmeier et al. 2007). Participants’ proportion of “pleasant” responses to the pictographs following the food primes was taken to be their implicit attitude towards foods. Participants’ proportion of “pleasant” responses to the pictographs following the gray and neutral images was also measured.

Next, participants underwent a gift-choosing task as an explicit measure of participants’ behavioral intention to indulge or resist the temptation of unhealthy food. As a cover story, participants were told that the experimenters were thinking of giving presents to subsequent participants of future experiments and were asked to aid in choosing one out of two possible gifts for ten different pairs of gifts. Participants were asked to indicate which gift they would take, should they be presented with each pair at the end of the experimental session. Out of the ten pairs, three choices pitted unhealthy foods with healthy ones: apple versus chips, tea versus soda, chocolate versus granola bar (reverse scored).

Participants completed other unrelated tasks before going through funneled debriefing (Bargh and Chartrand 2000), and no one guessed the purpose of the experiment.

Results

The correlations between all other variables can be found in Table 1.Footnote 2 ASA and DSA were not significantly correlated, r(310) = 0.08, p = .17, suggesting that they were distinct and orthogonal constructs (Gable et al. 2003). ASA was positively correlated with BAS, r(310) = 0.15, p = .008, and DSA was positively correlated with BIS, r(310) = 0.16, p = .005, consistent with the notion that ASA/DSA and BAS/BIS are conceptually related. The modest correlations also indicate that ASA/DSA and BAS/BIS are distinct constructs.

Impulsive desire

In order to examine the unique effect of ASA and DSA on implicit attitudes toward the tempting food primes, a hierarchical regression was performed. Participants’ ratings for neutral and grey primes were first entered in Block 1, R 2 = 0.22, F(2, 309) = 43.77, p < .001. Participants’ ratings of neutral primes positively predicted ratings of the food primes, B = 0.18, t(309) = 8.55, p < .001, 95% CI [0.137, 0.219], but the grey primes did not, B = −0.02, t(309) = −0.39, p = .70, 95% CI [−0.114, 0.076]. In Block 2, measures related to behavioral regulation: BIS, BAS, and trait self-control (BSC), were added, R 2 change = 0.01, F(3, 306) = 1.39, p = .25. They did not yield any significant effects, ps > 0.103. In Block 3, ASA and DSA were added, R 2 change = 0.055, F(2, 304) = 11.62, p < .001. Consistent with predictions, ASA positively predicted implicit attitudes towards food primes, B = 0.03, t(304) = 4.09, p < .001, 95% CI [.014, 0.039]. Participants with higher affective activation of the appetitive system found tempting food to be more pleasant. DSA negatively predicted implicit attitudes towards food primes, B = −0.02, t(304) = −2.91, p = .004, 95% CI [−0.040, −0.008]. Participants with higher affective activation of the aversive system found tempting food to be less pleasant. BIS, BAS, and BSC measures remained non-significant, ps > 0.114.

Behavioral intent

Hierarchical regression was also conducted on participants’ unhealthy food choices (M = 1.36, SD = 0.92). In Block 1, the regression model containing BIS, BAS and BSC achieved significance, R 2 = 0.03, F(3, 307) = 2.89, p = .035. Specifically, BSC significantly predicted participants’ behavioral intention, B = −0.22, t(307) = −2.71, p = .007, 95% CI [−0.384, −0.061]. BIS and BAS did not yield any significant effects, ps > 0.519. In Block 2, ASA and DSA were added, R 2 change = 0.001, F(2, 305) = 0.22, p = .80, indicating that ASA and DSA offered little predictive utility with regard to participants’ behavioral intentions. Other than BSC, all predictors were non-significant, ps > 0.500.

Discussion

In this paper, we distinguish affective activation from behavioral regulation as a distinct way appetitive and aversive systems regulate people’s responses. In accordance to past research, affective activation involves the relative activation patterns of the two motivational systems, the positivity offset and negativity bias (Ito and Cacioppo 2005). The key question is whether these relative patterns of activation, as measured by ASA and DSA in the MAM (Lang et al. 2005, 2011), provide additional predictive utility over and above measures related to behavioral regulation (i.e., BIS/BAS, BSC) in people’s response towards temptation.

We measured participants’ implicit attitudes towards the temptation as an indication of their spontaneous affective responses towards the tempting stimuli. As hypothesized, the individuals’ positivity offset, as captured by ASA, positively predicted participants’ implicit attitudes towards tempting foods. At the default level of processing, the hedonic gratification of indulgent foods (Loewenstein 1996) elicits a stronger appetitive response than the instrumental negativity (Ferguson and Bargh 2004; Fishbach et al. 2010) does an aversive response. The more easily activated one’s appetitive over the aversive system is, the larger the positivity offset it generates, and the more positive the person evaluates temptations. On the other hand, the individuals’ negativity bias, captured by DSA, negatively predicted participants’ implicit attitudes towards temptations. As predicted, an easily activated aversive (than appetitive) system diminishes the appetitive responses towards tempting foods by activating an intense response to the instrumental negativity of unhealthy foods. The more activated the aversive system is relative to the appetitive system, the greater the aversive response that is generated, and the more negative the person views temptations. Thus, affective activation, involving the relative patterns of appetitive and aversive activation, independently determine the strength of affective responses toward temptations.

More importantly, these relative patterns of activation predicted implicit attitudes over and beyond the BIS/BAS and the BSC scale, testifying to the importance of measuring affective activation in addition to behavioral regulation. As predicted, the non-significant finding of the BIS/BAS suggests that unlike the positivity offset and negativity bias, the BIS/BAS is less involved in the affective origin of desires. This corroborates past research indicating its conceptualization of and involvement in behavioral regulation instead (Carver and White 1994; Dholakia et al. 2006; Hofmann et al. 2012). It concurs with the similarly non-significant findings for BSC, suggesting that unlike past research assessing desires in an explicit manner, implicit attitude more purely captures the spontaneous affective responses or instinctual desires toward temptations.

We also compared the predictive utility of considering affective activation with BIS/BAS and BSC towards explicit behavioral intention to indulge in temptation. We posited that MAM, which primarily indices spontaneous affective activation, would have less predictive power over participants’ behavioral intention, compared to the BIS/BAS and BSC. Our findings were consistent with expectations in that individuals’ positivity offset and negativity bias did not predict participants’ tempting food choices. On the other hand, BSC predicted their food choices, validating the task as an indicator of participants’ behavioral intention towards temptation. However, despite its theoretical relevance, the BIS/BAS was not significant in predicting regulatory behavior towards temptations.

We offer possible explanations for why the BIS/BAS, compared to BSC, was not predictive of the food choice task. First, the BIS/BAS may not be as sensitive as BSC when the temptation is not salient. A recent meta-analysis suggests that BSC is uniquely related to long-term formation of good habits (de Ridder et al. 2012), which may be effectively activated even with minimal salience of the temptation. The food choice task was fairly indirect in tapping into participants’ indulgence in tempting foods, as it was designed to reduce demand characteristics regarding the temptation. The cover story and filler options should have diluted the salience of tempting foods. Hence, tempting food options might not have elicited strong responses from the BIS/BAS. If the task had been framed such that participants were to make a choice using information provided about the gifts, in this case, all the nutritional value (etc., calories) of each food item, both the BSC and the behavioral regulatory measure of BIS/BAS may have emerged as significant predictors. Second, the present way of indexing participants’ behavioral intention may be methodologically suited to the BSC. The meta-analysis showed that the self-control scale is particularly predictive when behavior is measured in hypothetical terms and through self-report (de Ridder et al. 2012). This is similar to the current food choice task, where participants have to project their gift preferences and list them down.

Although the BAS, as a unitary construct, was not predictive of participants’ implicit attitudes and food choice, we do not rule out the possibility that its subscales (drive, fun-seeking and reward responsiveness) may capture certain aspects of behavioral regulatory activation toward temptations. Past studies have employed BAS as a unitary construct to investigate temptation and desires (Hoffman et al. 2012; Zisserson and Palfai 2007) and we retained the same approach for the ease of comparison. However, there can be meaningful distinctions between its subscales (Voigt et al. 2009). Fun-seeking, for example, has been linked to impulsivity, while reward responsiveness and drive have been tied to sensitivity for reward (Dawe et al. 2004; Dawe and Loxton 2004; Franken and Muris 2006b). In particular, the drive subscale has been shown to correlate strongly with neural activity in response to pictures of appetitive food (Beaver et al. 2006), which suggests that it may be particularly sensitive to the presence of temptations. However, additional analyses using the BAS subscales instead of the BAS composite yielded no novel insights.Footnote 3 Our present study design may not be sensitive enough to detect such differences, and certain BAS subscales may emerge important in a different context. Even so, it is clear from our findings that the MAM remains the stronger predictor of implicit attitudes towards temptation beyond the BIS/BAS. Though the diversity in BAS taps into different aspects of behavioural activation, we reason that they still operate on a different level than the MAM (behavioural and not affective activation). The associated impulsivity, motivations, and neural correlates with these subscales, though operating in an implicit manner, may still be directed more towards regulating action than affect. Future research is encouraged to further differentiate the roles of the BAS subscales in affective and behavioral regulation.

Our paper demonstrates the predictive utility of individual differences in affective activation, as charted by the relative activation of the appetitive and aversive systems. This has important implications. Past research has focused on how implicit attitudes towards temptations, once formed, are especially potent in guiding consumption behaviors (Hofmann et al. 2007, 2008). Less research is devoted to what underlies implicit attitudes toward temptations in the first place. By showing that dieters’ implicit attitudes towards tempting foods are predicted by affective activation as fundamental as the appetitive and aversive systems (Elliot and Covington 2001; Gable et al. 2003), we can begin to make novel predictions about individuals’ reactions to temptations across domains. For example, we can predict that individuals with an easily activated appetitive system relative to the aversive system (i.e. high trait positivity offset), are particularly prone to experiencing strong desires for temptations. These strong desires would apply across multiple life domains such as dieting, monetary spending, and indulging in leisure and have behavioral consequences. Past research has shown that the more frequently people feel and inhibit unwanted desires, the more likely they were to experience self-control failures in the later parts of the day (Hofmann et al. 2012), perhaps due to a depletion of limited self-control resources (Baumeister et al. 2007). An easily activated appetitive system may thus dispose the individual to exercise more self-control to handle the allure of temptations and leave him/her too depleted to cope with other important tasks. On the other hand, individuals who are equipped with an easily activated aversive system relative to the appetitive system (i.e. high trait negativity bias) may experience less unwanted desires. Without the need to expand much effort to resist temptations, they could focus their cognitive resources on important pursuits in life.

In addition, although previous research using fMRI has examined the process underlying the desire for temptations, they focused on the neural correlates of self-control failures as the basis of how people fall to temptation (Lopez et al. 2014). The study found that greater reactivity of brain regions associated with rewards predicted the strength and enactment of desire, while activation of the brain region associated with self-regulation predicted less susceptibility to the temptation. However, unlike the current study, their findings shed light on people’s differential response in handling the desire, and not the origin of desire in the first place. Hence, our findings offer unique contribution in that the MAM captures the trait-level reactivity of the appetitive and aversive systems that predispose individuals to the extent of desires they will feel, independent of their ability to exert conscious self-control.

Our research also contributes to existent literature, by focusing on the conceptualization of the positivity offset and negativity bias as the relative activation of one motivation system over another, rather than a linear relationship between each system and response. Past studies concerning the neural correlates of BIS/BAS have discovered a consistent link of BAS with brain areas associated with appetitive responses and BIS with brain areas involved with aversive responses (Balconi et al. 2009; Herwig et al. 2006; Reuter et al. 2004), signalling a biological predisposition in approach and withdrawal behavioral activations. However, these activations of specific brain structures in isolation do not shed light on how the systems interplay, and how the interaction guides one’s experience and response to temptation. Our findings suggest that the relative activation of the motivational systems may be more relevant in understanding one’s affective response to temptations, as the joint influence of the hedonic gratification and instrumental negativity will activate both appetitive and aversive motivations. Hence, the relative degree each motivational system is activated will ultimately guide the individual’s eventual affective response towards the temptation. This concept of relative activation has not been applied to temptation in past research, nor captured in the traditional BIS/BAS measure.

We also encourage future research to explore the ways in which situational factors modulate the relative activation of the appetitive and aversive systems. Although the present paper focuses on individual differences in positivity offset and negativity bias, we believe that situational factors can differentially activate the two systems to influence implicit attitudes. This is in line with the view that while certain attitudes may be relatively stable and unchanging across contexts (Fazio 1995), many attitudes are constructed on the spot based on relevant input from the surrounding circumstances (Schwarz 2007). According to the constructionist view, how the relative activation of the appetitive and aversive systems contribute to one’s construction of attitude toward temptations depends on the specific context.

This perspective sheds new light on how people’s implicit attitude towards temptations is altered by situational factors. For example, past research finds that when an abstract mindset is induced (Fujita et al. 2006) or the long-term goal is made salient (Fishbach et al. 2010), people’s implicit attitude towards temptations become more negative. Our findings suggest that this common outcome of devalued temptations from the two situational factors may work through different mechanisms. When an individual focuses on abstract, high-level concepts, the hedonic positivity of temptations may become less salient (Mischel et al. 1996). Consequently, the temptations are less able to activate the appetitive system to generate a large positivity offset, hence reducing the allure of temptations. On the other hand, a long-term goal (e.g., academic achievement) may increase the salience of the negative instrumentality of temptations and upregulate the activation of the aversive system. The intense aversive response from the negativity bias then undermines or overwhelms any initial appetitive response to reduce the overall attitude towards temptations.

In sum, our findings highlight that the addition of affective activation to behavioral regulation of the two motivational systems provides a fuller understanding of people’s responses towards temptations. The relative activation patterns of the appetitive and aversive systems, in the form of positivity offset and negativity bias, may allow better predictions on how desires originate, rather than based on the activation strength of each system alone. We suggest that a more inclusive and nuanced conceptualization of the appetitive and aversive systems is in order and future studies should consider including MAM as an index for affective activation in addition to BIS/BAS measures in their investigations.