Change Synaptic Weights to Adapt and Learn

The effort to uncover principles of learning and memory used to be called the search for the engram, the memory trace in the brain that constituted a memory. Overwhelming evidence now shows that learning in the nervous system occurs because neurons modify the strength of the synapses between them. This synaptic modification allows small neural circuits to become highly selective for stimuli that have been learned.

Learning is a less extreme version of changes in neural circuits that occur during embryonic and early postnatal development. During development, an exuberance of synapses and neurons form and compete for synaptic connections. Unfit synapses are pruned, and neurons that don't emerge from synaptic pruning with enough connections die off entirely.

This situation is the basis of many critical periods in development in which certain types of experience, such as having useful visual output from both eyes, is necessary for some cortical connections to form. After the critical period, the winning synapses and circuits are stabilized, and later experience can't produce large-scale neural rewiring.

After development, synapses in a few selected areas of the brain are still malleable, but not everywhere, and neuronal death caused by the lack of appropriate synapses either stops or proceeds very slowly. In addition, synaptic alteration occurs in a much more restricted manner. Learning in the hippocampus is unique because it consists first of changes in synaptic strength, followed by growth of new neural connections, and then new neurons.

This doesn't mean that the adult nervous system lacks adaptation mechanisms. The eye adapts to dim versus bright light by using rods rather than cones, and people manage to sleep on rocking boats and in noisy environments. And it is unclear how much most of the adult nervous system is capable of changing, given the right trigger. It could be far more than initially thought.