| dc.description.abstract | Simultaneous experience of multiple emotions has been experimentally documented but there is a lack of systematic research coverage as to why and how emotions are blended. An appraisal theory view to understand emotional blends treats each blend as a coherent psychological state with a distinctive cognitive foundation described by appraisals. In two studies, I attempted to answer the following questions: 1) what are the major emotional blends, 2) what are the key appraisals associated with these blends, 3) how are appraisals from each constituting emotion in a blend synthesized, 4) can emotional blends and their associated appraisals be reliability induced, and 5) can a voice based emotion detection algorithm automatically measure multiple emotions at the same time? The first study, which answered the first three questions, used clustering algorithms on an existing dataset to extract emotional blends. The second study, which answered question 3 to 5, experimentally induced a few emotional blends in an autobiographical retelling task. The results from the two studies supported the hypothesis that the emotional blends represented coherent units of emotional experience that were observable and inducible, each with a characteristic profile of appraisals that could explain its existence. The appraisals from different emotions were either retained when blended together, or modified by the presence of other emotions in the blend. The voice-based emotion detection was able to capture changes in intensity when happy and sad voices were blended in a bittersweet blend, although the general resolution was lacking, and failed to reliably classify individual emotions. As a first systematic attempt on an appraisal theory view to emotional blends, this dissertation covers multiple aspects of emotional blends, with new questions inspired by the findings discussed as future directions. | |
