The Brain Science of Chess

How does the brain play chess?

Computers can play chess. Can this tell us about how the human brain plays chess?

Computers are Turing Machines. Turning Machines simply means performing one algorithm after another. Alan Turning came up with the concept.

However the brain does not do this. It works in parallel. A computer must execute an instruction serially one at a time. But a brain can do many things simultaneously.

Consciously its serial, like a computer. But unconsciously its parallel.

Also neurons contain noise. Not like a computer which deals in pure voltages. This make it harder to envision the mind like a computer.

The FFA (Fusiform Face Area) (What is it?)

It's a part of the brain (inferior temporal lobe) that activates when viewing faces. It's near the back of the brain.

That's why its called the face area (its not the only area that activates though)

A part of your brain that recognizes faces also activates when looking at chess positions (if you're good at chess)

There is a debate as to whether this area is evolutionarily adapted to represent faces specifically or is a general purpose module to recognize individual objects within a single category, (e.g. different types of cars).

Another view is that it is for decoding complex objects with multiple items and relations. In this case the activation to faces is just one example of a complex objects with multiple items and relations (eyes, nose, lips). This would mean that it not specifically for only faces.

Bilalić (2016) looked at brain activation in the Fusiform Face Area (FFA) when viewing chess positions. Interestingly, there greater modulation of the FFA when looking at chess position in experts (compared to looking at images of tools or rooms), but not in amateurs.

So chess skill means greater recognition of chess positions, which is indicated by the FFA Activation. FFA Activation was not shown when they simply looked at images of individual chess pieces, even when there was a basic relation between two pieces in an image (image of a rook checking the king).

So the FFA only allows complex relations between multiple objects to be recognized, and to form a holistic understanding of the positions.

Advanced chess players focus more on relations between pieces according to eye movement studies. They immediately switch to these relations even though it is in their peripheral vision. Amateurs on the other hand, don't have a structured way of looking at the board. Simon and Chase (1973).

The build up of the relation search in experts allows a overall understanding of the position.

Suppression of the Default Network

The Default Network is a group of brain regions that activate when you start mind wandering of daydreaming.

If you are focused on a activity with high levels of concentration, the Default Network is suppressed.

If you have trouble sleeping due to not being able to ease your mind, the Default Network is over-activated, which is why you can't stop thinking. (Mind-breaking mental activity will help you if you have insomnia).

Stronger Chess players, show greater level of Default Network suppression (they have greater concentration because they know what to do, if you are a beginner and don't know what to do then you can't really concentrate).

Beginners can't concentrate, imagine being asked to concentrate on a game that you don't know much about. You can't concentrate because you don't what to concentrate on cause you don't know how to play at a high level.

Christoff et al. (2016). Mind-wandering as spontaneous thought: a dynamic framework. This is a side profile view. The PFC parts are at the front where your forehead is on the first diagram (first diagram shows inside the brain, second the surface of the brain). The second diagram is reversed, IFG is at the front.

In your brain there's a lot of info. But how can something new be created by taking two pieces of info together? How can your brain create a plan involving a position its never seen. This involves the Frontal Parietal Network:

Activation of the Frontal-Parietal Network when looking at difficult chess positions

The Frontal-Parietal Network is a group of brain regions that activate when you are thinking about something that requires integration from many different areas (when you have a difficult chess problem you need more info to solve the problem)

The Global Workspace Theory posits that this network holds your conscious experience. Dehaene et al. (1998).

When you see a chess board position you can:

1. Play a move physically
2. Imagine different possible moves
3. Know how good the move is
4. Describe the position in words
5. Remember the move

That's seems obvious. Think about creating a robot that can do that.

No one ever did it, genius minds failed. Why? Because it's basically magic.

It sounds obvious, but how can it be obvious if no-one managed to create a robot that could do that (create a human robot).

How can a machine do so many things?

The Global Workspace Theory posits that there are brain regions (modules) dedicated to different tasks and that a copy of your conscious experience (holistic understanding) is sent to every module which can then perform its task related to what is in the Frontal-Parietal Network (aka your understanding on whats going on).

The Frontal-Parietal Network and the Global Workspace are the same thing.

The modules then compete for a share of the Global Workspace (some modules may suppress other modules through the Frontal-Parietal Network, corresponding to a change in what you're doing e.g. when you're finding a move intently, you may not remember stuff which happened around you, this would be the semantic chess memory areas of the brain sending signals to the Frontal-Parietal Network that it needs more attention, and the episodic memory module (memory of things happening around you) would be suppressed resulting in less memory of what happened).

That explains why you can do all those things listed above. This also explain why its hard to do multiple things at the same time.

This requires a neural code, so that the messages can be understood.

The neural code is a mystery, maybe forever.

Example of modules that can communicate with the Global Workspace. As an example, If you are busy trying to find the best move, the Semantic memory module connects with the Global Workspace. The Episodic memory (recreation of life events) module could get suppressed here, resulting in not remembering the things that happened around you as you were busy concentrating. The Global Workspace is your conscious experience.

Pereira et al. (2020) showed that the left Prefrontal cortex activation in people solving puzzles increased as the difficulty of a chess puzzle increased. A simple puzzle like smothered mate doesn't need much Fronto-Parietal activation as it can be handled well by a dedicated module (recognition of chess tactics).

But harder puzzles require Fronto-Parietal activation, which takes info from many modules to try to find a solution. This corresponds to the feeling of effort when trying to solve a hard puzzle, the brain digs through memory and tries combining memory patterns to find a solution, this requires more activation of the Fronto-Parietal network.

Christoff et al. (2016). Mind-wandering as spontaneous thought: a dynamic framework. Red is the Fronto-Parietal Network.

Emotions are expanded Thoughts

An emotion is a big thought. What does a thought do? It make you perform actions (think mate in 2, so play mate in 2)

What does an emotion do? It makes you perform actions (feel move is bad, don't play move)

The difference is an illusion. Emotions feel strong, because it makes you do something more readily.

Large scale actions, require large scale transmission. Large scale transmission requires compression of information, that's why there's only 6 or so emotions.

Guntz et al. (2018) published a paper about the role of emotions in chess. They had two groups: experts and intermediates and had them solve chess puzzles with increasing difficulty.

They recorded their facial expressions through Facial Activation Units with a software which identified emotions through facial expressions and found that the amount of emotion changes increased as the puzzles got harder for the experts.

For the intermediates though, the emotion changes increased until a certain point and then it dropped. It dropped because the puzzle got too hard for them. Emotion change represents different states of thinking. Since the puzzles got too hard for the intermediates, they thought less, and then had less emotional change.
Guntz et al. (2018). The Role of Emotion in Problem Solving: First Results from Observing Chess

They came up with a cognitive theory of chess involving emotions, playing chess involves these factors:

Valence: Assessment of position (positive or negative)
Arousal: Urgency in the position (if there are immediate threats or opportunities, tactical positions are high in arousal)
Dominance: Familiarity with a position (confidence)

There are four phases of chess reasoning according to De Groot: Orientation (chunk recognition), Exploration (searching different lines), Investigation (choosing the move to play), Validation (checking the move works).

Guntz et al. (2018) proposed that emotions help to choose between the large amount of different chunks (chess vocabulary). Emotions help the player recognize what is bad and what is good.

Chunks are patterns of chess pieces which are held in long term memory. A castled Kingside is one example of a chunk. The chunking theory seeks to explain why chess masters are able to have a greater memory for chess positions. The remember more because of their conceptual organization of the pieces.

But there is also a theory of templates, which proposes that a large template can store different chunks within it which explains why masters remember more when memorizing multiple boards then the chunking theory would suggest. (e.g. Ruy Lopez pawn structure with pieces in typical positions as a template, with slots (empty squares) for chunks). The Science of Chess Patterns.

Here's how they measured Valence and Arousal:

Valence: Intensity of positive emotions - negative emotions (in Facial Actions Units)
Arousal: Average of the combination of Facial Actions Units
Guntz et al. (2018). The Role of Emotion in Problem Solving: First Results from Observing Chess

The Flip Side of Emotions

So emotions help simplify the task through vibes.

But it can also backfire. Lack of confidence causes anxiety. Too much anxiety means less tendency to play sharp moves (manifested in less info transfer from/to modules and the Fronto-Parietal network). This means that less chunks are searched or integrated as the Fronto-Parietal network is less juiced up.

This results in worse chess as you have less info to work with.

This is a irrational response, a manifestation of evolutionary pressures. Obviously feeling anxiety helps avoid death. But it doesn't help chess, humans weren't designed for chess.

The brain being less willing to risk results in the feeling of anxiety. It feels that anxiety is a certain 'feeling' that causes you to stop doing things, but this an illusion.

The Fronto-Parietal network becoming more activated and having more connection with the chess semantic memory modules results in the feeling of confidence.

Summary

1. Fusiform Face Area (complex relation recognition) activates when looking at chess positions in strong players.

2. Default Network (mind-wandering) is suppressed in strong players.

3. Fronto-Parietal Network (Global Workspace) is necessary for integrating info from different brain areas to play chess.

4. Emotions help to recognize what moves should be played.

5. Emotions can be a double-edged sword.

Whenever you do something it feels like it's because you 'just can'.

That's an illusion.

Sources

Simon and Chase (1973). Skill in Chess
Dehaene et al. (1998). A neuronal model of a global workspace in effortful cognitive tasks
Hänggi et al. (2014). The architecture of the chess player's brain
Bilalić (2016). Revisiting the Role of the Fusiform Face Area in Expertise
Christoff et al. (2016). Mind-wandering as spontaneous thought: a dynamic framework
Guntz et al. (2018). The Role of Emotion in Problem Solving: First Results from Observing Chess
Pereira et al. (2020). Dynamics of the Prefrontal Cortex during Chess-Based Problem-Solving Tasks in Competition-Experienced Chess Players: An fNIR Study
Williams et al. (2025). Neural correlates of chess expertise: A systematic review of brain imaging studies comparing expert versus novice players