Our brain anticipates all of the new situations that it may encounter in a lifetime by creating a special kind of neural network that is “pre-adapted” to face any eventuality, scientists say.
Researchers at the French Institute of Health and Medical Research (INSERM) studied the diversity and adaptability of primate behaviour.
Human and non-human primates can learn an astonishing variety of novel behaviours that could not have been directly anticipated by evolution.
We now understand that this ability to cope with new situations is due to the “pre-adapted” nature of the primate brain, the researchers said.
The study shows that this seemingly miraculous pre-adaptation comes from connections between neurons that form recurrent loops where inputs can rebound and mix in the network, like waves in a pond, thus called “reservoir” computing.
This mix of the inputs allows a potentially universal representation of combinations of the inputs that can then be used to learn the right behaviour for a new situation. The researchers showed this by training a reservoir network to perform a novel problem solving task.
They then compared the activity of neurons in the model with activity of neurons in the prefrontal cortex of a research primate that was trained to perform the same task.
Remarkably, there were striking similarities in the activation of neurons in both the reservoir model and the primate.
The breakthrough shows that we have taken big step towards understanding the local recurrent connectivity in the brain that prepares primates to face unlimited situations, researchers said.
The research shows that by allowing essentially unlimited combinations of internal representations in the network of the brain, one of them is always on hand for the given situation, they said. The study was published in the journal PLOS Computational Biology.