How We Learn: Why Brains Learn Better Than Any Machine . . . for Now

Learning, in both brains and machines, thus requires searching for an optimal combination of parameters that, together, define the mental model in every detail.

“curse of dimensionality”—learning can become very hard when you have millions of potential levers to pull.

The more an axon is used, the more layers this sheath develops, thus insulating it better and better, allowing it to transmit information at a higher speed.

To cement the event, the neurons that were recently activated undergo major physical changes. They modify the strength of their interconnections, thus increasing the group support and making it more likely that this set of neurons will fire in the future. Some synapses become physically larger and may even get duplicated.

What is the most effective time interval between two repetitions of the same lesson? A strong improvement is observed when the interval reaches twenty-four hours—probably because sleep,

parameters: the quality and accuracy of the feedback we receive determines how quickly we learn.2

The idea is simple: while synaptic plasticity makes the brain malleable—especially in humans, where childhood lasts for fifteen or twenty years—our brain circuits remain subject to strong anatomical constraints, inherited from our evolution.

systems. But attention in humans exhibits a unique feature that further accelerates learning: social attention sharing.

not. Some research suggests that video games, especially action games that play with life and death, provide a particularly effective means of engaging our attentional mechanisms. By mobilizing our alerting and reward systems, video games massively modulate learning.

Consolidation, which renders what we have learned fully automated and involves sleep as a key component. Far from being unique