The role of emotions in humans is one of the oldest debates in philosophical history. From groups claiming that emotions are useless and need to be conquered, to those that claim that emotions are better than reason, and everything in between, a diverse range of opinions has been debated.

One part of this debate aims to answer what emotions even are. Are emotions the feelings you get when you receive a good grade, or when you look at someone you love? Or are emotions what motivate you to respond in anger when you’re annoyed?

To answer these questions, philosophers have developed theories of emotion with the goal of accounting for the intuitions we’ve developed regarding emotions and the role they play in our lives.

The feelings theory of emotion

One popular theory, the feelings theory of emotions, claims that emotions are the feelings themselves. This is the theory most commonly accepted by psychologists, and is also known as the James-Lange theory, since it is a synthesis of the works of psychologists William James and Carl Lange.

The theory includes both mental and physical feelings, including physiological changes that occur and can cause changes to our mental states. These emotions are distinctive conscious experiences and are things we feel. 

Intuitively, this theory seems to make sense. It would be fair to assume that emotions are constitutive of physiological changes and feelings that go along with certain mental states. The theory also seems to pick up on the component of emotions that we think to be most central, in that they involve feelings. 

But is there any reason beyond intuition to believe this theory? 

According to James, an important benefit of the theory is that it is able to accurately predict outcomes. We can predict reasonably well how certain biological and physiological changes will impact our emotions. This theory has even given us the ability to develop tools for dealing with emotions, like breathing in for four seconds, holding your breath for four seconds, and then breathing out for four seconds.

But an important objection to the feelings theory is that it arguably fails to separate emotions from other kinds of mental states, such as general thoughts and intuitions. A way to respond to this objection would be to claim that we consider bodily feelings emotions insofar as they only lead to specific mental states, but this has the danger of excluding unconscious perceptions and emotions that still impact the way we feel. 

So if emotions are not feelings, what are they?

AI and the evaluative theory of emotions

A theory that some are more partial to — and one that may be important when considering the development of artificial intelligence (AI) — is an evaluative theory of emotions. According to this theory, emotions are processed information and sensations and can be used to justify our beliefs about the world.

Consider the disgust you might feel eating your least favourite food or the fear you feel coming across a snake in the woods. Your emotions here would be incredibly helpful with providing you evaluative information that can affect your decisions, which could serve as a good evolutionary shortcut for the brain as well.

When it comes to the development of AI, the next step is going to be to provide AI with tools to figure out what to do with the large amount of data they already have.

Evolutionarily, this is a similar concern humans would have faced upon the development of stronger sensory intake mechanisms. Evaluation via emotion was a shortcut that prevented our brains from having to process all this sensory information and come to immediate conclusions, and the same might be true for AI. 

And this is something that’s already being done. 

AlphaZero, Google’s chess AI, uses evaluative emotion in its play strategy. Instead of calculating every possible move and outcome, which every past chess bot did — an incredibly computationally intensive process — AlphaZero creates a value system to figure out which paths of processing are worth exploring, and which should be left behind. An evaluative theory, as opposed to the feeling theory, leaves space for machines to develop emotions.

Evaluative theories of emotion can also account for the slight pause there tends to be in between a reaction and deliberate action to emotional stimuli in both humans and animals. Because these evaluations are largely unconscious, they can occur incredibly quickly and allow us to make decisions. In cases of fear, for example, animals tend to freeze before running away or fighting; that pause is where an emotional evaluation could take place.

Through looking at AI and human evolution, evaluatory theories of emotion become more appealing and useful in application.