How do synapses change during learning?

Cellular and molecular mechanisms

Learning reshapes synapses, the contact points where neurons transmit signals, by altering both the strength and composition of synaptic connections. Long-term potentiation and long-term depression are enduring increases and decreases in synaptic efficacy that were first characterized in the hippocampus by Timothy Bliss and Terje Lømo and later elaborated by many researchers. At the receptor level, NMDA-type glutamate receptors act as coincidence detectors: when presynaptic glutamate release coincides with postsynaptic depolarization, calcium enters through NMDA receptors and triggers signaling cascades. Robert Malenka at Stanford University and Mark Bear at Massachusetts Institute of Technology have documented how calcium-dependent enzymes such as CaMKII promote the rapid insertion of AMPA-type glutamate receptors into the postsynaptic membrane, making the synapse more responsive to future inputs. Conversely, different calcium dynamics and phosphatase activation lead to AMPA receptor removal during long-term depression, weakening transmission.

Synaptic tagging, gene expression, and neuromodulation

Some forms of lasting memory require not only fast receptor trafficking but also new gene expression and protein synthesis. Eric Kandel at Columbia University demonstrated in invertebrate models that short-term synaptic changes can be consolidated into long-term changes through transcriptional programs and growth of new synaptic contacts. The synaptic tagging and capture hypothesis explains how transient activity can mark synapses to receive later-synthesized proteins that stabilize structural changes. Neuromodulators such as dopamine gate plasticity by signaling the behavioral relevance of events. Wolfram Schultz at the University of Cambridge showed that phasic dopamine release carries reward prediction information that influences which synapses are strengthened, linking synaptic change to learning driven by outcomes.

Structural and systems-level changes

Beyond receptors and signaling, learning alters synapse structure. Dendritic spines grow, shrink, appear, and disappear on timescales from minutes to weeks; Rafael Yuste at Columbia University and others have visualized spine dynamics with two-photon microscopy. Training can expand synaptic contacts and reorganize cortical maps, as demonstrated by Michael Merzenich at the University of California San Francisco in studies of sensory training and skill learning. At the network level, repeated synaptic changes lead to circuit reconfiguration and systems consolidation, transferring memory reliance across brain regions over time.

Relevance, causes, and consequences

Synaptic plasticity underlies habit formation, language acquisition, motor skill learning, and the formation of episodic memory. Causes of plastic change include patterned neuronal activity, neuromodulatory signals that indicate salience, and environmental influences such as enriched experience or stress. Gina Turrigiano at Brandeis University has emphasized that homeostatic mechanisms counterbalance Hebbian changes to maintain network stability, preventing runaway excitation or loss of function. When plasticity is dysregulated, consequences include cognitive impairment in aging and neurodegenerative disorders, and maladaptive learning in addiction where reward-related synapses are pathologically reinforced. Understanding synaptic change bridges molecular neuroscience and practical domains such as education, rehabilitation after injury, and public health interventions that shape the environments in which learning occurs.