Skip to Content

How emotions underlie even the coldest human calculations

Expert chess players run the gamut of feelings when solving chess problems, according to a study with significant implications for machine intelligence.

It’s easy to imagine that emotion gets in the way of the most difficult decisions. Get rid of this cumbersome human artifact and surely people would be able to make cold, calculating choices in the most exacting of situations.

Not so. Neuroscientists have long studied people with brain injuries that prevent them experiencing emotions. But instead of being precise, ruthless killers, these people are paralyzed by indecision.

The truth is that when it comes to everyday choices—deciding between cheese or ham in your sandwich, for example—it doesn’t matter how much cold hard logic you bring to bear; these decisions are ultimately emotional.

But what of more detailed calculations like those involved in mathematics or chess? Surely they can’t be governed by fickle human emotion?

Actually, they can, say Thomas Guntz at the University of Grenoble in France and a few colleagues. These guys have measured the changes in emotional state experienced by chess players as they tackle increasingly difficult problems. And they say that emotions play a key role in helping players solve complex problems.

The ability to automatically measure changes in human emotional states has advanced by leaps in bounds in recent years. Changes in pupil size are an indicator of concentration levels. Heart rate is a measure of arousal and can be monitored by looking for changes in the color of facial skin.

Body posture and gestures also indicate emotional changes, and these are straightforward to monitor with 3-D cameras such as the Kinect. All this can be correlated with the object of a person’s attention, as measured by head orientation and eye gaze.

Together, these indicators provide a comprehensive overview of an individual’s emotional state and how it changes from moment to moment.

Guntz and co turned this powerful gaze to the emotional state of 30 expert and intermediate chess players as they solved increasingly challenging chess puzzles. Each puzzle required the player to checkmate an opponent. Puzzles that can be solved in one to three moves are considered easy, while those that require four to six moves are considered challenging.

As the players tackled each problem, the team recorded changes in gaze, body posture, cardiac rhythm, facial expression, and so on. They then used this data to infer how each player’s emotional state changed during the task.

For example, the player’s basic emotional state—happiness, sadness, anger, fear, disgust, or surprise—can be judged from his or her microexpressions; changes in cardiac rhythm suggest changes in arousal; and the rate of self-touching is a measure of stress.

“[Our results] revealed an unexpected observation of rapid changes in emotion as players attempt to solve challenging problems,” the researchers say.

For this reason, they think emotions must play a role in the decision-making process. “Our current hypothesis is that the rapid changes in emotion are an involuntary display in reaction to recognition of previously encountered situations during exploration of the game state,” they say.

This must play a crucial role in pruning the decision tree of potential moves, think Guntz and co. The way advanced chess players do this pruning is very different from the thought process beginners use. Over time, expert players learn to recognize certain patterns of play or positions of strength and weakness.

This pattern recognition significantly simplifies the process of deciding on the next move. Instead of considering all the pieces separately, the top players consider them in groups called chunks. Top players are thought to store up to 100,000 of these chunks in long-term memory. When playing a game, they transfer these chunks into short-term memory, where the reasoning takes place.

And that’s where players ought to run into trouble. There is a well-known limit on the amount of information that humans can store in short-term memory. Back in the 1960s, the American psychologist George Miller showed that we can store between five and nine chunks that way. Beyond that, we are overwhelmed.

So how do chess players manage 100,000 chunks when they can only hold a handful in their working memory at any one time?

They use emotion, say Guntz and co. When a player spots a chunk he or she has seen before, the valence associated with it causes it to be brought to the fore for further analysis or rejected as a bad option.

In this way, top players use emotion to move relevant chunks from long-term to short-term memory and back again. And it is this change in emotional state that the team was able to record.

That has huge implications for our understanding of human decision-making and for machine intelligence in general. Guntz and co are careful to temper their result with the suggestion that their work is still in its early stages and more needs to be done.

But it provides a curious new way to think about the problem of decision-making and how machines could do it more effectively. Until now, machines have mainly used increasingly powerful computational resources to make decisions. That effectively drains the mystery from problems like checkers, chess, and more recently Go. But ask them to choose between ham and cheese in a sandwich and they’re stumped.

Emotions clearly provide some kind of indexing system that allows us to access certain memories more quickly. Understanding how that works and how it can be applied to machines is an important goal.

Ref: : The Role of Emotion in Problem Solving: First Results from Observing Chess


Keep Reading

Most Popular

How to opt out of Meta’s AI training

Your posts are a gold mine, especially as companies start to run out of AI training data.

The return of pneumatic tubes

Pneumatic tubes were supposed to revolutionize the world but have fallen by the wayside. Except in hospitals.

Why does AI hallucinate?

The tendency to make things up is holding chatbots back. But that’s just what they do.

How a simple circuit could offer an alternative to energy-intensive GPUs

The creative new approach could lead to more energy-efficient machine-learning hardware.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.