Hidden discreetly in our brain is a small area called the ventral tegmental area, or VTA. This region is essential in our daily lives, as it governs our relationship with rewards and motivation. Its secret? Dopamine, a chemical substance that the brain releases when anticipating something pleasant.
An international team of researchers from the University of Geneva (UNIGE), Harvard, and McGill has just revealed that this small brain region has even more astonishing capabilities than previously thought. Thanks to an innovative algorithm, they discovered that the VTA not only predicts the arrival of a reward but also precisely indicates when it is expected.
The team led by Alexandre Pouget, in collaboration with Naoshige Uchida (Harvard) and Paul Masset (McGill), reveals that the VTA has extremely precise predictive capabilities.
© Adobe Stock This exciting research, published in the journal
Nature, highlights the growing importance of collaboration between neuroscience and artificial intelligence.
Until now, scientists believed that the VTA merely indicated whether a reward would arrive or not. Indeed, studies conducted in the 1990s had shown that this area releases dopamine as soon as a signal (such as a light or sound) announces a future reward, rather than at the moment of the reward itself.
An even more elaborate prediction
This ability of the brain to anticipate future rewards is also the foundation of many techniques used in artificial intelligence, such as the one that enabled AlphaGo to defeat the world champion of the game of Go. The researchers therefore wanted to explore this similarity in greater depth.
Their discovery was surprising: the VTA does not just predict an overall reward, it also knows exactly when each reward will be obtained. Some neurons in this area focus on near-term rewards (within seconds), others on those arriving in a few minutes, and still others anticipate events much further in the future.
This diversity allows our brain to be very flexible, capable of prioritizing either immediate satisfaction or delayed rewards, depending on the circumstances and current needs.
When neuroscience and artificial intelligence collaborate
To achieve these astonishing results, the team combined two very different approaches. Alexandre Pouget and his team had developed a mathematical algorithm capable of precisely predicting the timing of rewards. The Harvard researchers, meanwhile, had recorded the actual activity of the VTA in animals awaiting rewards.
By applying the algorithm to this real data, they observed a perfect match between theory and reality. This experiment demonstrates how artificial intelligence techniques can not only draw inspiration from the workings of the human brain but also help neuroscience better understand our own internal mechanisms.
A discovery that opens the door to new avenues for understanding how we make decisions and how we manage our expectations of future events.