The problem Psychologically plausible ways of differentiating emotions
The problem Psychologically plausible ways of differentiating emotions from one another and from non-emotions Why care? Emotion theorists: Better understanding of causes and consequences Clinicians: Better prospects for diagnosis and treatment Modelers: Better KR models for reasoning, better artificial characters for simulations, games, (social) robotics
BUT What is an emotion? What isn’t an emotion? Other affective states/conditions Moods? Traits? Preferences?
Where to start? How do we “get at” emotions? A typical (but neglected) Cognitive Science problem
What should we look at? Language? Linguistics Culture? Anthropology/psychology The real world? Anthropology/psychology Self-reports? Social Psychology Faces? Anthropology/psychology Bodily changes? Physiology Behavior? Behavioral Psychology Individuals? Personality Psychology Animal models? Animal Psychology Interacting Processes? Artificial Intelligence
How do we “get at” emotions? How can we identify the things we want to study? Language seems like obvious starting point?
But in any given language, which words refer to emotions? English > 2000 candidates Difficult to properly sample emotion space e. g. Russell’s (1980) circumplex model
? The model proposes two orthogonal dimensions: (un)pleasantness or valence, and activation, arousal, or intensity Examples of problems: 1. In the model, angry is as similar to afraid as to annoyed, even though angry and afraid are quite different emotions. Makes no sense.
intense afraid angry annoyed angry and annoyed are qualitatively similar, angry and afraid are not lost in a circumplex representation unpleasant
Examples of problems: 2. anger is less intense than afraid, but more unpleasant Makes no sense
intense afraid angry annoyed unpleasant anger is less intense than afraid, but more unpleasant. model loses important information
Problem partly due to taking too simplistic approach Probably, intensity dimension should be potency Probably, need a third dimension for intensity Problem of calibration of intensity.
Further evidence that intensity dimension isn’t intensity Direct intensity ratings show “anger” is more intense than “afraid” (Frijda, Ortony, Sonnemans, & Clore, 1992)
n typical value for “anger” 7. 3
Mean typical value for “afraid” 6. 3
Language? emotions words mapping problem In any language, which words refer to emotions? Different emotions-to-lexical-item mappings: 1. one many (anger emotion 19 lexical items) 2. one (relief emotion) 3. one none (“fears confirmed” emotion)
Starting with words is a problem Criteria for sampling items? Many are not emotions words Confounds words referring to different emotion types (anger vs fear) with words referring to same type but with different intensity (anger vs annoyed) Garbage in, garbage out Conclusion that emotions vary in valence and intensity neither interesting or informative
A linguistic approach to determining the referents of words in the affective lexicon If words like “sleepy, ” “droopy, ” “violent, ” and “abandoned” don’t refer to emotions, we shouldn’t use them or expect a theory to account for them How can we tell to what an affective word refers? we need a theory, and we need to test it
conditio ns internal non-mental external mental subjective evaluations objective evaluations e. g. , sexy e. g. , abandoned physical and bodily states e. g. , aroused, tired affective states such e. g. , happy, sad affect focal behavior focal cognition focal affectivebehavioral conditions affectivecognitive conditions behavioralcognitive conditions e. g. , cheerful, e. g. , encouraged e. g. , careful cognitive conditions e. g. , certain
A linguistic approach to determining the referents of words in the affective lexicon Test theory: Linguistic context: “being x” vs. “feeling x” adds affect that simply “being x” might not have, so x might seem emotional even when it isn’t
Linguistic analysis “feeling x” = to experience the feelings typically associated with being x = the emotions one has when (a) one believes that one is x, and (b) one cares that one is x “feeling abandoned” adds affect that simply “being abandoned” does not have, so abandoned seems emotional even though it isn’t
Linguistic judgment data (Ortony, Clore, & Foss, 1987) To what degree do you think “feeling x” refers to an emotion ? To what degree do you think “being x” refers to an emotion ? For good examples of emotions, there should be no difference Discriminant analyses confirmed eight distinct patterns of scale values
conditio ns internal non-mental external mental subjective evaluations objective evaluations e. g. , sexy e. g. , abandoned physical and bodily states e. g. , aroused, tired affective states such e. g. , happy, sad affect focal behavior focal cognition focal affectivebehavioral conditions affectivecognitive conditions behavioralcognitive conditions e. g. , cheerful, e. g. , encouraged e. g. , careful cognitive conditions e. g. , certain
An empirical criterion for what conditi words refer to emotions Some examplesinterna better than others extern al l nonmental subjective evaluations objective evaluations e. g. , sexy e. g. , abandoned physical and bodily states e. g. , aroused, tired affective states such e. g. , happy, sad affect focal behavior focal cognition focal affectivebehavioral conditions affectivecognitive conditions behavioralcognitive conditions e. g. , encouraged e. g. , careful e. g. , cheerful, glum better examples worse examples cognitive conditions e. g. , certain
- Slides: 24