Affective and Semantic Representations of Valence: A Conceptual Framework

The current article discusses the distinction between affective valence—the degree to which an affective response represents pleasure or displeasure—and semantic valence, the degree to which an object or event is considered positive or negative. To date, measures that reflect positivity and negativity are usually placed under the same conceptual umbrella (e.g., valence, affective, emotional), with minimal distinction between the modes of valence they reflect. Recent work suggests that what might seem to reflect a monolithic structure of valence has at least two different, confounding underlying sources, affective and semantic, that are fundamentally distinct, dissociable, and that obey different, recognizable rules. The current work discusses this distinction and provides implications for affective science from both the theoretical and the empirical perspective.

Semantic and affective manifestations of ambi (valence)

People sometimes report both pleasant and unpleasant feelings when presented with affective stimuli. However, what is reported as "mixed emotions" might reflect semantic knowledge about the stimulus (Russell, J. A. (2017). Mixed emotions viewed from the psychological constructionist perspective. Emotion Review, 9(2), 111-117). The following research examines to what degree self-reported mixed emotions represent actual feelings compared to knowledge about the stimulus. In a series of three experiments, participants reported either their feelings or their knowledge in response to affective stimuli. In Experiment 1, we sampled the entire IAPS pictorial space and examined the proportion of mixed emotion ratings using feelings-focused and knowledge-focused self-reports. We found a higher degree of mixed emotions under knowledge-focused than feelings-focused self-reports. In Experiment 2, we used a priori selected pictures to elicit mixed emotions. The proportion of mixed emotions was again higher under knowledge-focused instructions. In Experiment 3, we used movie clips that were previously used to elicit mixed emotions. In contrast to Experiments 1 and 2, there was no difference between feelings-focused and knowledge-focused self-reports. The results suggest a strong semantic component and a weak experiential component of self-reports in the case of pictorial stimuli. However, ambivalent movie clips elicited a stronger experiential component, thus supporting the existence of mixed emotions at the level of feelings.

Dissociating affective and semantic valence

We examined the possible dissociation between two modes of valence: affective valence (valence of e emotional response) and semantic valence (stored knowledge about valence of an object or event). In Experiment 1, 50 participants viewed affective pictures that were repeatedly presented while their facial electromyography (EMG) activation and heart rate response were continuously recorded. Half of the participants provided self-report ratings about the valence of their feelings and half about the valence of the stimulus. Next, all participants performed an affective Simon task. In Experiment 2, 30 new participants performed the affective Simon task with the repeated exposure embedded within the task. The results showed that measures related to affective valence (feelings-focused self-reports, heart rate, and facial EMG activations) attenuated with repeated exposure to pleasant and unpleasant pictures, whereas measures related to semantic valence (knowledge-focused self-reports and congruency effect of affective Simon task) did not. These findings strongly suggest that affective and semantic valence represent two distinct psychological constructs. (PsycINFO Database Record

Processing negative valence of word pairs that include a positive word

Previous research has suggested that cognitive performance is interrupted by negative relative to neutral or positive stimuli. We examined whether negative valence affects performance at the word or phrase level. Participants performed a semantic decision task on word pairs that included either a negative or a positive target word. In Experiment 1, the valence of the target word was congruent with the overall valence conveyed by the word pair (e.g., fat kid). As expected, response times were slower in the negative condition relative to the positive condition. Experiment 2 included target words that were incongruent with the overall valence of the word pair (e.g., fat salary). Response times were longer for word pairs whose overall valence was negative relative to positive, even though these word pairs included a positive word. Our findings support the Cognitive Primacy Hypothesis, according to which emotional valence is extracted after conceptual processing is complete.