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

My first and most general definition is the following: to learn is to form an internal model of the external world.

Random exploration, stochastic curiosity, and noisy neuronal firing all play an essential role in learning for Homo sapiens.

knowledge from its environment.

In this view, learning becomes similar to programming: it consists of selecting the simplest internal formula that fits the data, among all those available in the language of thought.

They store each episode through synaptic changes, so we can remember it later.

it. It is simply a matter of rewarding curiosity instead of punishing it: encouraging questions (however imperfect they may be), asking children to give presentations on subjects they love, rewarding them for taking initiative. . . . The neuroscience of motivation is extremely clear: the desire to do action X must be associated with an expected rew

to learn is to form an internal model of the external world.

the active mode, where children constantly experiment and question themselves like good budding scientists, and the receptive mode, where they simply record what others teach them. School often encourages only the second mode—and it may even discourage the first, if children assume that teachers always know everything better than students do.

The immense number of parameters that neural networks possess often leads to a second obstacle, which is called “overfitting” or “overlearning”: the system has so many degrees of freedom that it finds it easier to memorize all the details of each example than it is to identify a more general rule that can explain them.

Alerting, which indicates when to attend, and adapts our level of vigilance. Orienting, which signals what to attend to, and amplifies any object of interest. Executive attention, which decides how to process the attended information, selects the processes that are relevant to a given task, and controls their execution.