How Learning Works: What Neuroscience Says About Memory and the Brain

Interest in the neuroscience of learning has grown as students, teachers and professionals look for evidence-based ways to retain information. Decades of laboratory research on synapses, sleep and memory tests now offer a reasonably clear picture of what happens in the brain when something is learned, and which study habits actually make knowledge stick.

The Synapse: Where Memories Begin

At the cellular level, learning is thought to depend on long-term potentiation (LTP), a lasting increase in the strength of the connection, or synapse, between two neurons. A 2019 review in npj Science of Learning by Wickliffe Abraham, Owen Jones and David Glanzman describes the underlying rule as neurons that 'fire together, wire together': when a receiving neuron is activated at the same moment that the neurotransmitter glutamate reaches it, a receptor called NMDA opens and lets calcium flow into the cell, triggering a cascade that strengthens the synapse.

The same review distinguishes an early phase of potentiation, which lasts hours and needs no new proteins, from a late phase that depends on protein synthesis and gene activity and can persist for days or weeks — and, in some mammalian experiments the authors cite, for at least a year. A related but opposite process, long-term depression, weakens synapses through similar calcium signalling, so which outcome occurs depends on the precise pattern and amount of calcium entering the cell.

Harvard researchers led by Adam Cohen added detail to this picture in a study published in Nature Neuroscience on 6 May 2025, reported by the Harvard Gazette. Using a new imaging method called EPSILON, the team tracked AMPA receptor proteins — which govern how strongly a synapse responds — and found their movement correlated closely with markers of neuronal activation during memory formation. Cohen described the technique as giving researchers 'a lens into the synaptic architecture of memory, something previously unattainable in such detail.'

Sleep Consolidates What You Learned During the Day

Synaptic changes made while awake are only the first step. A July 2025 review in BMB Reports by Jaekyung Kim and Minjeong Park, both at KAIST in South Korea, describes how non-REM (NREM) sleep reorganizes new memories so the hippocampus can hand them off to the cortex for longer-term storage, a process called systems consolidation. Three brain rhythms coordinate this: slow oscillations (0.1–4 Hz) in the cortex, sleep spindles (10–15 Hz) generated in the thalamus, and hippocampal sharp-wave ripples (150–250 Hz), which compress and replay the day's neural activity patterns.

According to the same review, REM sleep then switches the brain into a different mode, dominated by theta rhythms (5–8 Hz), that helps integrate new memories into broader knowledge and tag them emotionally. Kim and Park describe this division of labor through the synaptic homeostasis hypothesis: NREM sleep strengthens the connections most tied to new learning while scaling down weaker ones globally, and REM sleep then prunes redundant synapses, which the authors say keeps the network from becoming saturated over time.

What Actually Improves Retention: Spacing and Retrieval Practice

Outside the lab bench, a separate line of cognitive-science research asks which study habits best use these biological mechanisms. A study by Butowska-Buczyńska and colleagues, published in the Proceedings of the National Academy of Sciences in October 2024, tested whether actively recalling information (retrieval practice) beats simply rereading it, and whether varying the context of each recall attempt adds further benefit. Across a battery of experiments, including one with a 24-hour delay before the final test, retrieval practice consistently produced better memory performance than restudying the same material.

The researchers describe a 'superadditive' effect when spacing study sessions apart is combined with varying the retrieval cue, meaning the two techniques together produced a larger benefit than either alone. They also report a mismatch in learners' own judgments: participants tended to rate constant, repeated cues as more effective, even though the data showed variable cues produced better actual recall — a gap the authors flag as an obstacle to getting students and teachers to adopt the method.

• Learning strengthens synapses through long-term potentiation, driven by calcium entering neurons via NMDA receptors (Abraham, Jones & Glanzman, 2019).
• Sleep is not passive: NREM rhythms consolidate new memories while REM sleep integrates and prunes them (Kim & Park, 2025).
• Actively recalling information beats rereading it, and the benefit grows when retrieval cues are varied and spaced out (Butowska-Buczyńska et al., 2024).