What are emotions?
There are several definitions of emotions and theoretical approaches to study them. In this course we will follow what is called the definition of emotions as core affect: Short-lived psychological states that consume the individual’s energy and strongly influence cognition and behavior, for example expression.
Emotional or affective behavior of an individual takes place at various timescales. Depending on the timescale we study, we refer to various types of behavior and their manifestation:
- Reflex reactions are the fastest components of emotions and manifest in very fast physiological responses. One example is jumping from your seat if you see a snake.
- Core affect operates at a slower timescale and can be assessed by asking people to answer questions about their emotional states. These states relax quickly and are triggered by a perceived stimulus (event) that is appraised (evaluation or interpretation).
- Mood is a slow and constant emotional state that can have various components (e.g. be positive and negative at the same time). Mood changes slowly throughout the day and does not need to change because of a stimulus.
- Personality traits are lifelong behavior patterns, some of them relate to the way we cope with our emotions or react to the emotions of others. It takes many years for a personality trait to change.
Affective Science is the (interdisciplinary) scientific study of emotions. Computational Affective Science applies methods from Computer Science and Data Science to Affective Science. Some examples are:
- Affective Computing: Development of systems that detect, process, and elicit emotion
- Cyberpsychology of Affect: Understanding the interplay between emotions and information and communication technologies
- Emotion Recognition: Identification of human emotion using any kind of modality: text, voice, facial expression, physiological signals (skin conductance, muscle activity, EEG, fMRI), etc
- Sentiment Analysis: Detection of subjective states from (textual) data, including emotion
To empirically study emotions, we need methods to measure them. Emotions can be measured through various signals and observable behaviors of individuals.
- Emotion self-reports: individuals answer a question about their emotional state or rate their emotions in a scale
- Facial expression: the image of the face of a person is interpreted by other people to annotate emotions or automatically processed with detection software
- Verbal expression: spoken expression with its prosody (e.g. volume of voice) or written text annotated by other people or with sentiment analysis
- Brain activity: based on magnetic resonance or electroencephalograms to measure activity in brain regions asociated with certain emotional states
- Other physiological signals: Muscle activation (e.g. frowning, smiling), heart rate, skin conductance (sweating), etc. Measured with physical sensors or other passive methods like wristbands.
In the following, we are going to cover four models of how to capture emotions in quantitative research. Some approaches are better for some modes or signals (e.g. text, facial expression) than others.
Ekman’s basic emotions model