Few moments in a soccer game are more electrifying than the penalty kick. The goalkeeper stands, waiting for the kick – and even before the ball is struck they must predict where it is going and spring into action.
Their prediction will be built on years of reading run-ups, body positions and foot angles. But if you asked them after the game how they knew where the ball was going, they would struggle to describe the process. If the kick fooled them, however – if they were surprised by where it went – they will be able to replay the events with near perfect clarity.
We wanted to know why this happens. Why does the brain skim over expected events, but store the finest details of what catches it off guard? This is part of a bigger debate among neuroscientists, about whether the brain saves energy by prioritising what it expects or what surprises it.
In new research published in the Journal of Neuroscience, we showed that the brain doesn’t have to choose. It does both, just at different moments, and for different jobs.
Bringing the penalty kicks into the lab
To study this, we recruited 40 people and asked them to watch flashes of dots popping up at different locations around a circle in a semi-predictable pattern. They were told to press a button, indicating the side on which the dot appeared, as fast as they could.
Think of it as a stripped-down penalty kick: sometimes the dot shows up in the top-right where you (the goalkeeper) would expect, and sometimes it doesn’t. Pressing the right button is when you catch the ball, and the wrong one is when you miss it, when your prediction fails you.
Hu, Tran & Rideaux, CC BY-SA
While participants completed the task, we recorded how fast they reacted to each dot, how accurate their response was, and how well they remembered its location afterwards. Would they remember the location more clearly when the prediction was wrong?
At the same time, we also recorded their brainwaves (via electroencephalogram, or EEG) and pupil changes (eye tracker), tracking both neural and physiological activities going in the brain.
The tricky part, however, was to make sure we were really measuring prediction, not just the brain getting bored of watching the same thing over and over again. So, we specifically designed the patterns in where the dots appeared to keep expectation separate from that kind of simple fatigue.
The brain in its two acts
We found people reacted faster to dots that appeared where they expected, just like our goalkeeper, already moving preemptively.
But when they were asked to recall exactly where that expected flash was, their memory was worse. Instead, they located surprising ones more sharply.
The brain activities told us why. It seems prediction works in two acts.
First, before the flash even appears, the brain gets a head start, priming a response. That’s the speed. And those milliseconds matter. Imagine a swerving car, a falling glass, or a ball already flying past you.
Then, when the expected thing arrives, the brain shrugs. If I already knew that, why bother to look closely? Instead, it spends more energy on surprises, capturing them in crisp detail to refine future predictions of the world.
So, the decades-old debate over whether the brain favours the expected or the unexpected is a false dichotomy. It does both.
One process is motor: it acts fast on what it predicts, getting the body moving before the moment arrives. The other is sensory: it looks closely at what it failed to see coming and encodes those instances in sharp details. Each process, one after another.
The brain doesn’t just predict the world. It also learns from the moments when its predictions fail.
Beyond the penalty kick
Why does any of this matter? Because getting the balance right between prediction and surprise may be one of the brain’s most fundamental operating principles.
When it goes wrong, the consequences can show up clinically. A brain that leans too hard on what it expects, or fails to properly learn from what surprises it, is thought to underlie symptoms in a diverse range of conditions (such as schizophrenia and autism spectrum disorders).
Understanding how a healthy brain manages this balance is an important step toward understanding what happens when it breaks down.
It’s also a blueprint for building more biologically grounded artificial neural networks. The same trick – acting fast on the expected and encoding the surprising in detail – is exactly what could teach a neural network when to coast and when to pay attention, which could mean they use less energy to do the same job.
