How novelty speeds up learning
Brain scientists led by Sebastian Haesler (NERF, empowered by IMEC, KU Leuven and VIB) have identified a causal mechanism of how novel stimuli promote learning. Novelty directly activates the dopamine system, which is responsible for associative learning. The findings have implications for improving learning strategies and for the design of machine learning algorithms.
A fundamental type of learning, known as associative learning, is commonly observed in animals and humans. It involves the association of a stimulus or an action with a positive or negative outcome. Associative learning underlies many of our every-day behaviors: we reward children for doing their homework, for example, or limit their TV time if they misbehave. Scientists have known since the 1960's that novelty facilitates associative learning. However, the mechanisms behind this phenomenon remained unknown.
"Previous work suggested that novelty might activate the dopamine system in the brain. Therefore we thought that dopamine activation might also promote associative learning." says Prof. Sebastian Haesler, who led the study.
To demonstrate that novelty indeed activates dopamine neurons, the researchers exposed mice to both new and familiar smells. In a second step, the mice were trained to associate novel and familiar smells with reward.
The findings demonstrate that dopamine activation by novel stimuli promotes learning. They further provide direct experimental support for a group of theoretical frameworks in computer science, which incorporate a 'novelty bonus' to account for the beneficial effect of novelty. Incorporating such a bonus can speed up machine learning algorithms and improve their efficiency.
From a very practical perspective, the results remind us to break our routine more often and seek out novel experiences to be better learners.
(Content Courtesy: http://www.vib.be/en/news/Pages/Novelty-speeds-up-learning-thanks-to-dopamine-activation.aspx)