We learn physical skills by feeling rewarded, even in the absence of a reward, finds new study

Using a high-tech virtual reality pool setup, researchers from Surrey and Imperial College London had 32 participants play pool on a physical table while wearing a virtual reality (VR) headset that slightly changed the cue ball’s trajectory, forcing players to adapt. 

In the study, published in npj science of learning, two different sessions took place, with different feedback given to each player taking part. In one session, players only got “error feedback”, with the balls disappearing before they could pocket them. In the other session, participants only got “reward feedback”, with successful shots being artificially shown pocketing, while failed shots gave no visual results. The goal was to see whether the brain would only use the type of feedback it was given. 

However, the brain behaved differently. Even when the reward – pocketing the ball – was taken away in the “error feedback” condition, players still showed strong signs of using reward-based learning. They were seen to be finding their own “mini rewards”, such as observing the ball taking a correct-looking path. The brain wouldn’t ignore that key information, much like seasoned basketball players can anticipate a good shot before it goes through the hoop.  

Measurements of brain activity confirmed this complex interaction, showing a mixed learning pattern, rather than separate processes as seen in past lab experiments. While players were seen to learn faster when focusing on errors, the underlying process in both conditions was always a blend of the two processes.