Sadness, but not all negative emotions, heightens addictive substance use

Abstract

Do negative feelings in general trigger addictive behavior, or do specific emotions play a stronger role? Testing these alternative accounts of emotion and decision making, we drew on the Appraisal Tendency Framework to predict that sadness, specifically, rather than negative mood, generally, would 1) increase craving, impatience, and actual addictive substance use and 2) do so through mechanisms selectively heightened by sadness. Using a nationally representative, longitudinal survey, study 1 (n = 10,685) revealed that sadness, but not other negative emotions (i.e., fear, anger, shame), reliably predicted current smoking as well as relapsing 20 years later. Study 2 (n = 425) used an experimental design, and found further support for emotion specificity: Sadness, but not disgust, increased self-reported craving relative to a neutral state. Studies 3 and 4 (n = 918) introduced choice behavior as outcome variables, revealing that sadness causally increased impatience for cigarette puffs. Moreover, study 4 revealed that the effect of sadness on impatience was more fully explained by concomitant appraisals of self-focus, which are specific to sadness, than by concomitant appraisals of negative valence, which are general to all negative emotions. Importantly, study 4 also examined the topography of actual smoking behavior, finding that experimentally induced sadness (as compared to neutral emotion) causally increased the volume and duration of cigarette puffs inhaled. Together, the present studies provide support for a more nuanced model regarding the effects of emotion on tobacco use, in particular, as well as on addictive behavior, in general.

Keywords: addictive behavior; appraisal tendency; emotion; impatience; smoking.

Fig. 1.

Across a longitudinal, population-based dataset, only sadness reliably predicted smoking status. The x axis displays the self-reported trait emotions measured in each dataset. The y axis displays the unstandardized beta in simultaneous regressions with smoking status as the dependent variable after controlling for age, gender, and SES. Smoking status was defined by self-reported daily smoking. Error bars represent 1 SE. *P < 0.05.

Fig. 2.

Individuals randomly assigned to the sadness (vs. neutral) condition experienced greater state sadness, which in turn predicted impatient choices for cigarette puffs via enhanced self-focus. Numbers indicate standardized betas. Solid lines indicate significant paths. Dashed lines indicate nonsignificant paths. **P < 0.01, ***P < 0.001.

Fig. 3.

Participants randomly assigned to the sadness condition inhaled 30% greater volume (mL) per puff than did participants in the neutral condition. Error bars represent 1 SE at the participant level.