Commentary on Heckmanet al. (2013): Negative affect increases craving-Questions about the relationship of affect, craving and smoking
Year: 2013
Type: editorial
Abstract: Heckman and colleagues 1 have done the field a service by carefully analyzing the relationship between affect induced in laboratory cue–reactivity (CR) settings and subsequent craving. Their finding that negative affect (NA) is associated with increased craving adds important information to our understanding of the inter-relationships among affect, craving and smoking. Each of these constructs and relationships is complex, and careful analysis is essential for understanding them. It is notable that the findings concern craving and not smoking itself. Although craving is a good proximal predictor of smoking 2, 3 the relationship is not perfect 4, so craving cannot necessarily be used as a proxy for smoking. Heckman et al. 1 describe CR studies as ‘laboratory analogues of drug-seeking’, yet CR studies seldom measure drug-seeking or drug use. Indeed, CR studies have been criticized for focusing on craving as a sole outcome, to the exclusion of smoking 5. Illustrating the issue, in a recent CR study we found that proximal smoking cues increased craving, but did not affect the probability of smoking, the latency to smoke or the amount of smoking 6. Ecological momentary assessment (EMA) data also illustrate some of the complexities of the relationship between craving and smoking. The observed relationship is curvilinear, such that initial increases in craving are associated with increased likelihood of smoking, but this relationship flattens at higher levels of craving 2, 3 (see also Shiffman et al. 6), implying that there are variations in craving intensity that do not translate into changes in smoking. Conversely, there are meaningful variations in craving intensity even when smoking is held constant: within smoking occasions, both NA and positive affect (PA) are associated with stronger craving (NA in a curvilinear fashion; 7), even though they are not associated with increased probability of smoking. These complexities suggest that one must attend to the actual level of NA, the level of craving and the smoking context to understand these interrelationships. These considerations also suggest that one cannot easily project findings about craving to conclusions about smoking. In this sense, the finding that NA increases craving does not directly support smokers' questionnaire reports that they smoke (not just crave) when experiencing emotional distress. Smokers' self-impressions may be incorrect 8, having been formed by the experience of noticing strong craving when smoking under distress. It is possible that NA enhances the affective salience of craving, and its subjective intensity, without affecting its motivational impact. As Heckman et al. 1 note, affective experience itself is complex, and not completely captured by an NA scale. According to the circumplex model of affect 9, the nature of NA is moderated by the accompanying level of arousal. High-arousal NA states produce feelings such as anxiety or anger, whereas low-arousal states yield feelings such as depression. As nicotine is a stimulant, with probable effects on arousal 10, these variations may be very important influences on the dynamics of affect and smoking. Heckman et al. note that most CR studies induce high-arousal NA states, so this is the affective state associated with craving in their meta-analysis. Consistent with this, EMA data on craving during smoking episodes 7 show that NA influences craving most when arousal is elevated. Studies that independently manipulated affective valence and arousal would be very informative. The CR literature's assessment of NA and PA as independent dimensions, as reflected in the meta-analysis, also implies a theory of the structure of affect—one that is neither intuitive nor universally agreed, although it is embodied in commonly used measures of affect 11. The circumplex model 9 posits that NA and PA are not independent dimensions, but rather opposite ends of a single bipolar dimension of affective valence, comporting with the intuition that one cannot feel very good and very bad at exactly the same time. From this perspective, it is interesting that the meta-analysis suggests that NA and PA may have opposite effects on craving, perhaps reflecting this underlying dimension of affective valence. Careful parsing of the structure of affect and its measurement is likely to be useful in understanding how affect influences craving and smoking. Heckman et al. 1 also raise another crucial issue about inferences from CR data. Although CR data are meant as laboratory models of the real-world dynamics of craving and smoking, there is no evidence that they accurately model real-world experience and behavior. It would be useful to have studies evaluating the validity of CR methods as assessment of real-world patterns of smoking or craving. In summary, Heckman et al.'s 1 meta-analysis gives us new insight into the relationship between experimentally induced affect and laboratory-assessed craving. At the same time, their paper raises many issues that could be fruitfully explored regarding the broader questions of how affect is best conceptualized and assessed, how craving relates to smoking and how these relationships are best assessed and understood. The author provides consulting to GlaxoSmithKline regarding smoking cessation, and is part of a group developing smoking cessation medications.
Source: Addiction
DOI: 10.1111/add.12363
Author: Saul Shiffman
Institution: University of Pittsburgh
Language: English
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2028429396