Research
- Impulsive and reflective influences on self-regulatory behavior
- Relationship between implicit and explicit measures (implicit-explicit correspondence)
- Predictive validity of implicit measures
- Working memory capacity and impulse control
- Consciousness and introspection
- Person × situation interactions
- Health behavior
- Evaluative conditioning
Impulsive and Reflective Influences on Self-Regulatory Behavior
In typical self-control situations, people are torn between their long-term goals to restrain behavior (i.e., weight control) and their immediate impulses that promise hedonic fulfillment (i.e., being attracted by a high-caloric snack). A good metaphor for such internal tugs-of-war is that of a wild horse on the one hand and a rider whose job it is to tame the horse. Such internal conflicts are very common in everyday life, and the unfettered expression of impulses has far-reaching implications such as caused by overeating, drug abuse, violent aggression, or sexual harassment. 
Previous social psychological research has primarily focused on the situational determinants that are detrimental for self-control. Personality and individual differences research on the other hand has identified those traits that appear to be particularly relevant for self-control outcomes. However, neither the situational nor the dispositional approach has so far been much concerned with conceptualizing the notion of impulse.
In order to overcome this limitation, my colleagues and I proposed an integrative approach to the study of self-control (Hofmann, Friese, & Strack, 2009; Hofmann, Friese, & Wiers, 2008). The approach is consistent with a dual-systems perspective (e.g., Strack & Deutsch, 2004) and enables to investigate the interplay of three important elements: (a) reflective precursors such as deliberate evaluations or restraint standards, (b) impulsive precursors such as automatic affective reactions or approach-avoidance tendencies, and (c) situational and dispositional boundary variables that shift the relative influence of reflective versus impulsive precursors on behavior.
So far, the suggested framework has been applied to the prediction of a large range of self-control outcomes in the domains of prejudice (Hofmann, Gschwendner, Castelli, & Schmitt, 2008), eating behavior (Hofmann, Friese, & Roefs, 2009; Hofmann, Rauch, & Gawronski, 2007), drinking behavior (Friese, Hofmann, & Wänke, 2008; Hofmann & Friese, 2008), sexual behavior (Hofmann, Gschwendner, Friese, Wiers, & Schmitt, 2008), consumer choice behavior (Friese et al., 2008), addiction (Wiers, Beckers, Houben, & Hofmann, 2008), and aggression (Hofmann et al., 2008). In this research program we identified a number of important situational and dispositional boundary variables such as ego depletion (Hofmann et al., 2007), cognitive load (Friese et al., 2008), alcohol consumption (Hofmann & Friese, 2008), mortality salience (Friese & Hofmann, 2008), low working memory capacity (Hofmann et al., 2008), low behavioral inhibition (Hofmann et al., 2009), and low trait self-control (Friese & Hofmann, 2008) which all increase the impact of impulses on behavior (for a reviews, see Friese, Hofmann & Schmitt, 2008; Hofmann et al., 2009). This conclusion was driven by a stronger predictive validity of implicit measures under these conditions as compared to control conditions. This body of evidence (a) lends direct support for dual-system conceptions of human behavior determination, (b) indicates the need to specify both reflective and impulsive precursors in predictive models of self-control outcomes, and (c) draws attention to the situational and dispositional risk factors involved in self-control but also to potentials for its improvement.
Relationship between Implicit and Explicit Measures (Implicit-Explicit Correspondence)
In analogy to earlier consistency debates in social and personality research, the development of implicit measures has brought up the question of how strongly implicit and explicit measures of attitudes, self-esteem, and personality traits correspond or diverge. We addressed this issue in a meta-analysis on the correlation between IAT measures and explicit self-report questionnaires (Hofmann, Gawronski, Gschwendner, Le, & Schmitt, 2005). Across studies, there was evidence of a slight positive overall relationship. More importantly, the variability in correlations was substantial, indicating that the implicit-explicit relationship depends on a variety of moderators or boundary conditions.
Drawing on a dual-systems framework of associative and propositional representations (e.g., Strack & Deutsch, 2004), my colleagues and I (Hofmann, Gschwendner, Nosek, & Schmitt, 2005) proposed a model of implicit-explicit correspondence (see figure) that organizes the empirical evidence into five broad factors: (a) translation between implicit and explicit representations (e.g., representational strength, introspective awareness), (b) additional information integration for explicit representations (e.g., low time pressure; high need for cognition), (c) properties of explicit assessment (e.g., social desirability concerns), (d) properties of implicit assessment (e.g., situational malleability), and (e) research design factors (e.g., sampling bias, measurement correspondence).
In past and ongoing work, I am particularly interested in the conditions that moderate the degree to which inner mental experiences (Hofmann, Gschwendner, & Schmitt, 2005) or self-observed behavior stemming from implicit processes may enter conscious awareness so that they can be integrated into one’s explicit representation of an attitude or a personality disposition (for a review, see Hofmann & Wilson, in press). In a series of recent studies (Hofmann, Gschwendner, & Schmitt, 2009), we found that people appear to have a blind spot with regard to the nonverbal behavioral manifestations of their implicit dispositions, even though neutral observers may readily detect these cues and use them as a basis for dispositional inferences.
Predictive Validity of Implicit Measures
Just as implicit-explicit correspondence, the predictive validity of implicit measures appears to be highly context-dependent (see also the above evidence in the self-regulation domain). Recently, my colleagues and I conducted a review of the available evidence on the predictive validity of implicit measures (Friese, Hofmann, & Schmitt, 2008) and suggested a classification system of moderators with two dimensions that is based on contemporary two-process models: One dimension distinguishes moderators according to their influence on the opportunity to control, the motivation to control, or the reliance on either automatic or controlled processes without changes in opportunity or motivation. The second dimension classifies moderators according to whether they pertain to a disposition of the acting person, the situation the behavior occurs in, or the behavior itself. The general view that emerges from the empirical work we reviewed is that implicit measures seem to predict behavior well under certain (but not all) situations, for some (but not all) people, and with regard to some (but not all) types of behavior. Furthermore, implicit measures seem to complement explicit measures in that they often appear to work best in cases where explicit measures fail to yield satisfactory predictive validity.
Evaluative Conditioning
Recently, I got interested into the way by which likes and dislikes are acquired. A classic paradigm to study these processes is Evaluative Conditioning (EC). EC can be defined as a change in the liking of a stimulus (Conditioned Stimulus; CS) that result from pairing that stimulus with other positive or negative stimuli (Unconditioned Stimulus; US). Despite decades of research on the phenomenon, there is still a lot of debate regarding the generality of the phenomenon, its similarity or difference with regard to Pavlovian conditioning, and the processes underlying EC. In an attempt to address these questions from a summative perspective, my colleagues and I (Hofmann, De Houwer, Perugini, Baeyens, & Crombetz, in press, Psychological Bulletin) recently conducted a large-scale meta-analysis spanning more than 50 years of EC research. Across a total of 214 studies included in the main sample, we obtained a medium size mean EC effect of d = .52. More importantly, about 70% of the variance in effect sizes across studies were attributable to true systematic variation. We conducted moderator analyses to partially explain this variation as a function of concrete aspects of the procedural implementation and abstract aspects of the relation between CS and US. Among a range of other findings, EC effects were stronger for high versus low contingency-awareness, for supraliminal versus subliminal US presentation, for post-acquisition as compared with post-extinction effects, and for self-report as compared with implicit measures. The meta-analysis is available here.