Trial and error. Discovering a way of learning

Scientists have discovered that the brain uses a dual mechanism to learn through trial and error, which explains how habits are formed and may help lay the groundwork for treatment strategies for conditions such as addictions or compulsions.

The research, led by neuroscientists at the Sainsbury Wellcome Centre (SWC) at University College London (UCL) and carried out in mice, could also help in the development of therapies for Parkinson's disease.

The trial-and-error learning method involves trying different options until you find one that works and learning from your mistakes along the way, Efe reported on Wednesday.

“We found a mechanism that we believe is responsible for habits. Once you develop a preference for a particular action, you can bypass the value-based system and simply rely on the default policy of what you’ve done in the past,” which also frees up the brain for other tasks, said Marcus Stephenson-Jones, group leader at SWC and lead author.

Researchers have discovered a dopamine (a neurotransmitter) signal in the brain that acts as a different type of learning signal than previously known.

It was known that dopamine signals in the brain form reward prediction errors (RPEs) that inform the individual whether an option is better or worse than expected (this is the circuit that leads individuals to seek pleasure).

In this new study, the authors found that, alongside reward prediction errors, there is an additional dopaminergic signal, called action prediction error (APE), that updates how often an action is performed.

These two teaching signals provide individuals (in this case, rats) with two distinct ways to learn to make a choice: learning to choose the most valuable option or the most common option, the study found.

“Imagine going to your local coffee shop. The first time you go, you might take a while to choose a sandwich, and depending on which one you choose, you may or may not like it. But if you go back to the coffee shop often, you won’t waste time thinking about which snack to choose and instead start choosing one that you like by default. We believe it’s the dopaminergic signal from the EPA that allows this default policy to be stored,” Stephenson-Jones explained.

Furthermore, this newly discovered learning system provides a much simpler way to store information than having to directly compare the value of different options, freeing up the brain to multitask.

For example, once you learn how to drive, you can also hold a conversation with someone during the ride. While your default system performs all the repetitive tasks of driving the car, your values-based system can decide what to talk about, the study explains.

The study's results show why it's so hard to break bad habits and why replacing one action with another may be the best strategy : If you consistently replace an action, such as chewing nicotine gum instead of smoking, the EPA system can take over and form a new habit on top of the other.

This research also has potential implications for Parkinson's disease , which is known to be caused by the death of midbrain dopamine neurons, those involved in movement, which may be responsible for encoding the EPR, which could explain why these patients have difficulty performing routine tasks such as walking, but can perform more flexible behaviors, such as ice skating.