It was predicted that emotional states affect the way, how individuals build their performance expectancies. Emotional states were assumed to affect the mean level of the performance expectancies and to influence whether the expectancies relate to general or specific self-concepts. Positive mood should lead to more intensive information processing than negative mood. More specifically, individuals with positive mood should simply infer their performance expectancies concerning a specific task from their general self-concept (as a result of intensive information processing), whereas for participants in negative mood, performance expectancies should be inferred from the relevant specific self-concept (as a result of less intensive information processing). In an experiment, positive or negative mood was induced in N = 158 university students. General and specific self-concept as well as task specific performance expectancies and task performance were assessed. As predicted, mean expectancies were higher in the positive mood-condition than in the negative mood-condition. Furthermore, specific self-concept was predictive of expectancies when participants were in negative mood. When participants were in positive mood, expectancies could only be predicted on the basis of the general self-concept. The findings support the idea that the formation of performance expectancies can be understood as information processing. We discuss how emotional states affect learning processes.

Performance expectancies are a key factor in explaining achievement-related variables like choice, persistence, and achievement. Performance expectancies are assumed to depend – to a large extent – on the academic self-concept of an individual. The impact of self-concept on expectancies of success has been assumed to be mainly due to subject- or task-specific rather than a general academic self-concept. In contrast, this paper argues, that inferring expectancies from the self-concept is a cognitive activity that requires cognitive resources and that variables influencing the type of information-processing can also influence the link between self-concepts and expectancies. Based on dual-process-theories of information processing, the present paper argues that inferring expectancies from the general self-concept requires less cognitive capacity compared to inferring expectancies from the specific self-concept. The first theoretical reasoning underlying our hypotheses is therefore that, given peripheral information-processing, expectancies will depend on general academic self-concept whereas given central information-processing expectancies will depend on task-specific self-concepts. The type of information-processing depends on different factors (like relevance, distraction, or cognitive motivation). As another determinant, mood has been found to strongly influence the type of information-processing: Positive mood leads to peripheral information processing whereas negative mood leads to rather central information processing. This is the second theoretical reasoning underlying our hypothesis. Bless, Bohner, Schwarz and Strack (1990) for example found that attitudes of participants in positive mood were not affected by the quality of arguments, whereas participants in negative mood changed their attitudes as a function of argument quality. In combining the first an the second theoretical reasoning, we predicted that performance expectancies will depend on the general self-concept given positive mood. In contrast, given negative mood, performance expectancies will depend on the specific self-concept.

Emotional states may not only affect the type of information processing. They can also affect judgments in a direct way in that individuals can use their actual emotional states as heuristic (“How do I feel about it?”) for judging an object: The use of this heuristic results in more positive judgments given positive compared to negative mood (Schwarz & Clore, 1983; Schwarz & Clore, 2003). Therefore, mood may also have a main effect on performance expectancies: Positive mood should lead to higher expectancies than negative mood.

The predictions were tested in an experiment with N = 158 university students. In a single-test situation general academic self-concept was assessed. Then, the self-concept concerning a specific task (analogy-task) was assessed (specific self-concept). We then included a distraction task in order to achieve a clearer separation between the items assessing self-concepts and the items assessing performance expectancies. After this, the mood manipulation was applied. Two different pieces of music were used to induce positive or negative mood. The two pieces were found to be appropriate for mood induction in university students in a pilot study. After listening to five minutes of the music, participants answered a manipulation check concerning their actual emotional state and then reported their performance expectancies concerning a specific set of analogy tasks. They were told that they will receive 25 analogy tasks. As performance expectancies, the participants reported, how many tasks they expected to comlpete within five minutes. Finally, the participants were given 25 tasks and five minutes time to complete them. Task performance was recorded as the number of correct tasks.

Results first showed that the mood induction was successful: Self-reported emotional states were more positive in the positive mood-condition than in the negative mood condition. In line with our hypothesis, in a regression predicting performance expectancies, mood had a main effect on performance expectancies: Higher performance expectancies were found in the positive than in the negative mood condition. Additionally, as predicted, the interaction term of mood and general self-concept as well as the interaction term of mood and specific self-concept had statistically significant effects on the performance expectancies. These interactions were due to the fact that in the positive mood-condition, only the general self-concept (but not the specific self-concept) had an effect on performance expectancies. In contrast, in the negative mood-condition, only the specific self-concept (but not the general self-concept) was related to performance expectancies. In a regession predicting task performance, mood had no statistically significant effect. Only performance expectancies predicted task performance.

The results are important for the understanding of the factors underlying students' performance expectancies. They support the idea that inferring expectancies from the self-concept is a cognitive activity requiring cognitive capacity. Whether general or specific self-concepts are used to infer the expectancies seems to be a function of the type of information-processing used. Contrary to the assumption of a greater predictive power of specific compared to general components of the self-concept, given positive mood, expectancies can better be predicted from the general academic self-concept. Our explanation is in line with research on the effects of mood on the processing of social information. Positive mood increases the likelihood of relying on general knowledge structures, whereas negative mood is associated with the use of more case-specific information (see Bless, 2001). Implications of the findings for self-concept models will be discussed. Furthermore, it will be outlined, that positive mood does not necessarily have positive consequences of learning outcomes.