What is “Emotion”?

Since childhood I’ve been confused about my emotions. Clearly these inner feelings exist, and are strong, but what are they? Could they be controlled, or even defined? The mystery of emotion was the stimulus that drove me towards the study of the mind and, from there, neuroscience.  While I have never directly studied “emotion”, I continue to read about it and think about it.

In the past decade neuroscience and psychology have reached an apparent consensus. An important feature of the consensus is that emotions are conscious feelings of the inner state of the individual. While I appreciate, and largely agree with the consensus, I’m proposing an extension: Emotion as Motivation.

Rough thesis: Rather than “emotion” as simply a feeling of inner state (which it is), an emotion is best characterized as a feeling that permeates a motivational state with specific content. The motivational state is a framework for action that includes a representation of the self and a goal.  Think of the framework as a terrain with you and the goal represented. The motivational state (the framework) allows you to chart imaged paths to the goal and anticipate the consequences.

Prior to 1990, while there was general consensus that emotions were “inner feelings” there was no consensus on the processing that led to emotional states of the function of emotional states. While the topic was of interest to psychologists and philosophers, the burgeoning field of neuroscience was largely absent.  Dating back more than a century, theories came in two camps, illustrated by the example of seeing a bear, “feeling” fear and running to escape.

Theory #1 (an intuitive theory): emotion as feeling plus motivation

1. I see a bear
2. this percept leads to the inner state of fear, the emotion
3. the emotion of fear is a motivation that initiates escape behavior.

Theory #2: James-Lange (an unintuitive theory) emotion as feeling

1. I see a bear.
2. this percept elicits (consciously or unconsciously)
   1. escape behavior
   2. visceral activation (blood rushes to muscles, sweat, pupils dilate etc)
3. The sensory combination the bear plus the sensory feedback of internal state of the body leads to the feeling of fear, the emotion of fear.

Theory 1, perhaps the easiest to accept, is that emotion is both a “feeling” and a motivation. The feeling is part of the inner processing that may lead to successful escape behavior.

Theory 2 is that emotion is merely a feeling, a categorization of the bear percept plus running plus internal state. Emotion is not motivation, since it occurs after the bear threat, and after the initiation of escape behavior. It is the conscious residue of action.

In my reading, recent theories reject each of these models, but converge on something slightly different.  Descriptions by Joe LeDoux, Lisa Barrett  and Jerome Kagan are roughly consistent. I’ll summarize from Lisa Barrett.⁽¹⁾

1. You see the bear, creating a clear percept “bear”
2. This percept causes 3 things to happen
   1. escape behavior (running)
   2. autonomic arousal
   3. a large “affective” response. (“affect” is a simple one dimensional quality of good-for-me or bad-for-me).
3. Conscious Post-hoc analysis of the preceding 3 leads to the conscious categorization of the state as a “fearful event”: I am (or I was) afraid. This post-hoc categorization is the emotion. Emotion is a conscious, language-like categorization. The “function” of emotion is categorization. You can generalize across emotional experiences, and communicate with others.

According to Barrett, these emotion categorizations are not innate, are not tightly linked to specific facial expressions, and vary from culture to culture. According to Joe LeDoux, (who studies rats), “emotions” are “autonomic” conscious percepts and, because we have no clear evidence of “autonomic” consciousness  in animals, he posits that animals such as his rats have no emotions. LeDoux cites Morgan’s Canon: when explaining animal behavior, one should not posit a higher psychological process if a lower level process will suffice. According to LeDoux, when a rat freezes to a conditioned stimulus (perhaps a tone that predicts shock), this is not “fear” or the behavioral expression of fear, but the expression of a “survival circuit”. The rat could exhibit this behavior with or without the self-conscious feeling of fear. ⁽²⁾

My critique of the current consensus is not that it is wrong, but that it is incomplete. If for the moment we accept the notion that emotions are “inner feelings”, what is their value? What is their function? I feel that if we go beyond the quale of emotion, we may reach a coherent explanation.

The philosopher Martha Nussbaum has a different perspective⁽³⁾.  According to Nussbaum, emotions not only have characteristic inner kinds of feelings (quale) they also have characteristic content. Specifically, the content involves a representation of others — the objects of an emotion — who can facilitate or block a route to personal flourishing.  In other words, Nussbaum’s emotions are linked to motivation and behavior. Her description does not stop at inner feeling.

If you accept the notion that an emotion must have a target, we can reset both the definition and function of emotion. Emotions are no longer simply a category of “feeling” or felt states; an emotion becomes a category of “feeling” linked to a specific behavioral framework. Emotion are no longer a passive sensory qualia, the are linked to motivation and action.

Proposed definition:

“Emotion is the conscious feeling of  motivation”  

Emotion sets the stage for action.

The value (function) of emotion

If you’re with me so far, great. The next section is more speculative, and depends heavily on novel ideas about consciousness. Since I haven’t laid out these ideas, and won’t do that  here, this will only be a sketch.

In brief, emotions permit deliberative, strategic action. Emotions enhance the ability to evaluate options to reach a goal or avoid negative consequences.

What do I mean by “framework”, and what benefits does this provide? I’m thinking of two modes of action, that can be called “model-based” and “model free” (terms borrowed from “learning theory”⁽³⁾). As I used the term, “model” is roughly synonymous to “schema”, “framework”, “context”, “map” or “action frame”. Behavior can be “model dependent” or “model-free”. Model free actions are immediate stimulus response actions, where the context has no bearing. Model free: If I see a bear, I run. Model Based: If I see a bear in the woods, I run; if I see a bear in the circus or zoo, I watch.  Context matters.

I won’t argue the point here, but posit (for future argument) that all consciousness is model-based. The principal value of conscious processing is that it permits deliberation. Deliberation means, roughly, setting up an internal representation of the action space, imagining various paths (options)  through the action space and choosing one. Clearly this is a major proposal, and won’t be defended here. However, if you accept this proposal, and accept that emotions are aspects of consciousness, then, ipso facto, an emotion brings with it is a specific action space.

Rather than asking you to accept the broad brush of this proposal for consciousness, I will ask you to accept a narrower notion: that each emotion is tied to a specific action frame, the motivational representation. Think of the space as a 2d representation of self and goal set upon a terrain. This brain representation permits imagined action along several paths. Each path is evaluated, and the optimal path selected.

The clear advantage of this notion is that it permits emotional states to underly careful deliberative decision making. Deliberation involves imagining the cost-benefit of a variety of actions before selecting one. Deliberation permits strategic planning of indirect and distant routes towards difficult goals.  Finally, deliberation permits the avoidance of routes that would lead to disaster.

 

Summary

• While an emotion is characterized as a conscious “feeling”, it is something more. Each emotion has an specific content that includes objects and a framework.
• “Motivational state” is conscious awareness of the contents of emotion, the objects plus framework. A particular feeling (quale) permeates this space.
• The “framework” permits deliberation. Imagined paths between self and goal. The selected path is the one that maximizes benefits while minimizes costs.
• Emotion, therefore, is consciousness of a motivational state which lays out possible paths for action.

 

---

⁽¹⁾ Current theories of emotion that largely overlap:

• Barrett, L. F., Mesquita, B., Ochsner, K. N., & Gross, J. J. (2007). The experience of emotion. Annual Review of Psychology, 58, 373–403. http://doi.org/10.1146/annurev.psych.58.110405.085709
• L Barrett, How Emotions Are Made: The secret life of the brain. Houghton Mifflen. 2017
• LeDoux, J. Anxious: Using the Brain to Understand Fear and Anxiety. Penguin Books, 2015
• Kagan, J. What is Emotion? History, Measures and Meaning. Yale University Press, 2007

⁽²⁾ June 20. I received an email from Joe LeDoux that led to corrections of an earlier version of this paragraph.  LeDoux feels that animals have “noetic” consciousness, but not “autonoetic” consciousness. Noetic consciousness can be described as an awareness of what is out there, such as objects in the world. “Autonoetic” consciousness is a similar awareness that includes reference to self. According to LeDoux, emotions are autonomic. Therefore he and Brown conclude that animals do not have emotions.

I also disagree with Joe. My strong guess is that animal’s such as rats have emotions. Rather than Morgan’s Cannon, I apply the principle of  parsimony. If the behaviors bear strong resemblance to human behaviors, and the brain organization, although much smaller is almost identical, why not infer that similar states are occurring? Why is consciousness unique to the human brain?

⁽³⁾ Nussbaum, Martha (2001). Upheavals of thought: the intelligence of emotions. Cambridge New York: Cambridge University Press. ISBN 9780521531825.

⁽⁴⁾  States versus Rewards: Dissociable Neural Prediction Error Signals Underlying Model-Based and Model-Free Reinforcement Learning. Neuron, vol 66, 2010.