Greetings everyone! The second reading group meets us with an overview of emotion and predictive processing in music. The stage is set with a general introduction to the ITPRA theory of expectation, and the evolutionary adaptation of expectation and emotion in chapter one of David Huron’s Sweet Anticipation (2006). Koelsch et al. (2015) provide an in depth, comprehensive account of the neurobiological foundations of human emotions with a generous sprinkling of evidence from the neuroscience of music, dubbed “The quartet theory of human emotions”. This is followed by a concise review from Gebauer, Kringelbach and Vuust (2015) which links the quartet theory and predictive coding as a plausible account of emotion in music. Below is a summary of the articles. I’d recommend reading the first and third texts first, then tackle the neurological article with concepts of expectation, prediction, and emotion from those texts in mind. A review of our discussion points will be in this subsequent blog post.
Huron notes that we have a pretty well agreed-upon understanding of musical emotion and expectation– for example, in Western music minor chords are sad, a diminished seventh can signify suspense– but the principles of these folk-psychological generalizations are by no means cross-culturally universal. If we want a clear picture of how we process emotion and expectation in music, then we need to appeal to “psychology proper”, including neuroscience, biology, evolution, and culture. Expectation is a biological adaptation making use of specific physiological structures (such as the endocrine (hormone) system), and our culture is the environment in which we learn, apply, and refine our expectations. There are obvious biological advantages to forming accurate expectations (predictions), and emotions can be thought of as the ‘motivational amplifiers’ either enhancing or diminishing those expectations in a given situation. Referencing Meyer’s Emotion & Meaning in Music, Huron goes on to introduce the ITPRA theory of expectation.
“The principle emotional content of music arises through the composer’s choreographing of expectation. “
At at least a superficial level, the ITPRA model resembles the hierarchical predictive processing framework we reviewed last time, however it encompasses a more broad range than seemingly brain-centered prediction units. ITPRA consists of five response systems: Imagination and Tension (pre-outcome responses) occur before an (un)expected event’s onset, and Reaction response, Prediction response, and Appraisal (post out come responses) occur after the event onset.
- Imagination– Thinking and feeling about future possibilities. Simulating the future event as if it had already happened.
- Tension– Preparation for an anticipated event, including motor preparation (arousal) and perceptual preparation (attention) with the goal of matching the appropriate levels of arousal and attention just in time for the expected event.
- Prediction Response– Generates expectation-related emotion: if stimulus is expected, the emotional response is positively valenced; if the stimulus is unexpected, the emotional response is negatively valenced.
- Reaction Response– Immediate, non-conscious somatic response. Similar to reflex, but importantly, can be learned/trained over time (i.e. through schemas). Can be associated with system 1 in dual process theories of emotion.
- Appraisal– Slower, context-dependent, cognitive response. Takes into account complex social/biological factors. Can be associated with system 2.
Koelsch et al. provide a very detailed look at the ‘brainy bits’ of expectation and emotion. If you’re new to neuroscience or haven’t memorized the various parts of the brain and their functions, don’t worry. We’ll just start with a basic overview.
The Quartet Theory of Human Emotions
First, Koelsch et al. differentiate between effector systems and affect systems. The effector systems are the biological systems which act in combination with the affect systems to generate four different categories of neurological affect. In their diagram above, you can see how the effector systems and affect systems have reciprocal interactions amongs themselves, with each other, as well as with the conscious appraisal system (here defined as the language system). Music, interestingly, can perhaps play a part in the conscious appraisal system (think prosody), or can rather hold a place in a nonconsious system prior to any bias which arises from translating an emotion into words.
A Quartet of Affect systems
- Brainstem-centered affect system- this controls ascending activation, or the subjective feeling of being energized. The brainstem controls the autonomic nervous system (ANS) as well as expression of emotions, modulation of pain, maiting behavior, and coordinating behavioral ANS activity (as in freeze, flight, or fight). The brainstem also generates both somatomotor and neuroendocrine activity in response to stimuli with emotional valence.
- Music can already evoke autonomic and motor responses at the level of the brainstem. This may be one avenue through which music can (subconsciously or consciously) modulate our general level of excitement, say, when we want to make doing chores or working out more enjoyable (I’m thinking of jymmin).
- Diencephalon-centered affect system- this processes pain/pleasure regarding urges and homeostatic emotions. Koelsch et al. note that ‘the thalamus imbues sensory information with affective valence’, even before that information is consciously perceived.
- The dopaminergic reward system can be activated by listening to music, of course explaining why music can make us feel happy through release of dopamine in the brain. This happens especially during anticipation and experience of peak emotion in music, aka when music gives you the chills.
- Hippocampus-centered affect system- this system regulates survival behaviors, and is associated with attachment-related affects such as love, and is involved with the social human motivation for group-inclusion.
- Orbitofrontal-centered affect system- this system handles a lot of activity. This is where ‘primary appraisal’— fast, automatic, nonconscious appraisal of sensory information—occurs, as well as automatic shifts of attention and ‘stimulus evaluation checks’, where sensory information is imbued with emotional valence and takes part in decision making and motivation or inhibition of behavior. The orbitofrontal cortex (OFC) also generates expectancies, and controls emotional behavior. This control is shaped by social cultural norms.
- Music is of course mediated by social and cultural norms, so perhaps the OFC control of emotional behavior can be modulated by musical activity, not just the emotions themselves.
Koelsch et al. describe how the integration of affect and effector systems forms an emotion percept, or unverbal (pre-verbal) subjective feeling, via four components: an affective component, a sensory-interoceptive component which integrates the physiological condition of the body, a motor component consisting of action tendencies, and a cognitive component detailing cognitive (though not necessarily conscious) appraisal.
In section four, the authors describe how their quartet theory interacts with the language system. Interestingly, they note that while emotion percepts need to be reconfigured into propositional linguistic expressions,
“Music…has the advantage of evoking a feeling sensation (i.e., an emotion percept) without this sensation being biased by the use of words…although music seems semantically less specific than language, music can be more specific when it conveys information about sensations that are problematic to express with words because music can operate prior to the reconfiguration of emotion percepts into words. “
Gebauer, Kringelbach, and Vuust now give us a short and concise commentary on how we can bring all of this emotion and expectation talk together under the framework of predictive coding. In Bayesian models of predictive coding, perception, action, emotion, and learning have the following definitions:
- Perception – the process of minimizing prediction errors between higher-level ‘prediction units’ and lower level ‘error units’
- Action– active engagement of the motor system to resample the environment in order to reduce prediction error
- Emotion– weight/modulator of prediction error itself, guiding behavior, action and learning
- Learning– long term influence on the prediction units
Through conscious and/or subconsious processing of statistical regularities in music, we form expectations (predictions) that are fulfilled or violated to varying degrees. Unexpected (unanticipated, unpredicted) events are met with heightened physiological arousal and attention, and can be further modulated by (conscious or unconscious) cognitive appraisal. Pleasure in music arises from the interplay between our expectations and the actual musical event– “the predictive motion between tension and release”. It seems that Meyer had the right idea all along, the science just needed to catch up.
The Quartet Theory of Human emotion seems to provide the neurophysiological grounding mechanisms by which both Huron’s ITPRA theory and predictive coding accounts could actually occur in the brain. What do you think? See you Thursday at 1pm! 7 Bristo Square, 1.210