Learning induces synaptic plasticity that strengthens or weakens connections between neurons based on experience. Long-term potentiation (LTP) increases synaptic strength through NMDA receptor calcium influx and postsynaptic changes, while long-term depression weakens synapses. Learning also recruits new neurons into circuits, expands cortical maps, and promotes dendritic spine formation. These cellular mechanisms explain how neurons change to represent learned associations, supporting new perceptual, motor, and cognitive abilities.
Neuroplasticity — your prerequisite concept — is the brain's general capacity to change in response to experience. But "the brain can change" is a very broad claim. Experience-dependent plasticity makes it specific: it describes *what changes*, *why it changes*, and *how those changes support learning*. The key bridge between the two concepts is long-term potentiation (LTP), the synaptic mechanism that converts experience into durable structural change.
You know the cellular mechanism of LTP: coincident pre- and postsynaptic activity opens NMDA receptors, allowing Ca²⁺ influx into the postsynaptic spine. This calcium signal triggers kinase activity (particularly CaMKII), which phosphorylates existing AMPA receptors (making them more responsive) and drives trafficking of new AMPA receptors into the synapse. The result is a strengthened connection — the same presynaptic input now produces a larger postsynaptic response. What connects this to *learning* is the Hebbian insight: cells that consistently fire together (because they're activated by the same stimulus or sequence of events) repeatedly co-activate their shared synapse, meeting the coincidence condition that opens NMDA receptors and inducing LTP. The synapse that supports the learned association literally grows stronger.
LTP produces both functional and structural changes. The functional change — more AMPA receptors — is fast. The structural change — new dendritic spines, growth of existing spines, sometimes even new axonal boutons — takes hours but is more durable. This structural consolidation is what makes memories persist beyond the window of kinase activity. Learning also drives cortical map expansion: when a body region is repeatedly stimulated (as in Braille reading), the cortical area devoted to that finger expands at the expense of adjacent representations. This map plasticity is experience-dependent in the most literal sense — use the finger more, the map grows; amputate the finger, the map is invaded by neighbors. The same principle applies to motor learning: pianists show enlarged cortical representation of the finger movements they have practiced most.
Long-term depression (LTD) is the complement of LTP and equally important for learning. LTD occurs when a synapse is repeatedly activated without coincident postsynaptic firing — the presynaptic cell fires, but the postsynaptic cell doesn't reach threshold. Moderate Ca²⁺ influx (insufficient to trigger CaMKII) instead activates phosphatases that remove AMPA receptors from the synapse, weakening the connection. LTD ensures that not all synapses strengthen simultaneously — only the ones that are predictively correlated with outcomes are preserved. LTD is the pruning mechanism at the circuit level, analogous to synaptic pruning during development but operating throughout adulthood.
The full picture of experience-dependent plasticity is therefore a coordinated multi-level process: individual synapses strengthen (LTP) or weaken (LTD) based on activity patterns, spines grow or retract to support these changes, cortical maps reorganize to reflect patterns of use, and in the hippocampus, adult neurogenesis adds new neurons that can be recruited into newly formed memories. Learning doesn't just use the brain — it physically reshapes it. The learner who has practiced a skill for thousands of hours has a brain that is measurably different from a novice's, not because of intelligence, but because sustained experience-dependent plasticity has built a more efficient, more responsive circuit for that domain.