Advancedadvanced biology
Neurobiology: Action Potentials, Synapses, and Plasticity
Learning objectives
- Explain the ionic basis of the resting membrane potential and the action potential.
- Describe the Hodgkin-Huxley model and the roles of voltage-gated sodium and potassium channels.
- Trace synaptic transmission from calcium influx to postsynaptic response.
- Explain long-term potentiation and the NMDA receptor's role as a coincidence detector.
- Survey key methods in neuroscience and identify major open questions.
The neuron as an electrical device
Neurons exploit the physics of ions across membranes. At rest, a neuron sits near -70 millivolts, inside negative, because potassium leak channels let K+ drift out down its concentration gradient while the sodium-potassium ATPase continuously pumps 3 Na+ out for every 2 K+ in, maintaining the gradients. The Goldman-Hodgkin-Katz equation predicts the resting voltage from each ion's permeability and gradient. This resting state is stored energy: the membrane is a charged capacitor awaiting discharge. Dendrites collect graded potentials -- small, decaying voltage changes from synaptic inputs -- which summate at the axon hillock. If depolarization reaches threshold, around -55 mV, the neuron fires. Myelin sheaths, wrapped by glia, insulate axon segments so the signal leaps between nodes of Ranvier in saltatory conduction, reaching 100 meters per second; unmyelinated fibers manage barely one. Demyelinating diseases like multiple sclerosis reveal how critical this insulation is.
The action potential: Hodgkin and Huxley
In 1952, Alan Hodgkin and Andrew Huxley, working with the squid giant axon -- thick enough to thread with electrodes -- combined voltage-clamp experiments with mathematical modeling to explain the action potential (Nobel 1963). Their model showed that depolarization opens voltage-gated sodium channels, and Na+ influx depolarizes the membrane further: explosive positive feedback driving the rising phase. Sodium channels then inactivate within a millisecond while slower voltage-gated potassium channels open, repolarizing the membrane, often with a brief hyperpolarizing undershoot. The absolute refractory period, when inactivated sodium channels cannot reopen, enforces one-way propagation and caps firing rate. Decades later, molecular cloning confirmed their predicted gating particles as real channel domains. The action potential is all-or-none -- its amplitude does not encode stimulus strength -- so neurons encode intensity in firing frequency and timing, a digital-like code built from analog ion flows.
Synaptic transmission
When the action potential invades the presynaptic terminal, voltage-gated calcium channels open, and the resulting Ca2+ microdomain triggers synaptic vesicles to fuse with the membrane via SNARE proteins, releasing neurotransmitter in discrete packets -- the quantal release Bernard Katz demonstrated at the neuromuscular junction. Glutamate, the main excitatory transmitter, opens AMPA receptors for fast signaling; GABA opens chloride channels that inhibit. Ionotropic receptors act in milliseconds; metabotropic GPCR-type receptors modulate over seconds to minutes. Postsynaptic potentials summate in space and time: thousands of inputs are weighed each moment, excitation against inhibition. Neuromodulators -- dopamine, serotonin, norepinephrine, acetylcholine -- tune this computation globally, adjusting gain, signal-to-noise, and plasticity rules. Nearly every psychoactive drug, from SSRIs to anesthetics, works by tweaking this synaptic machinery, which is why understanding transmission in molecular detail remains central to psychiatry and neurology.
Plasticity: how synapses learn
In 1973, Bliss and Lomo showed that brief high-frequency stimulation of hippocampal inputs produces long-term potentiation (LTP): synapses that stay stronger for hours or longer. The NMDA receptor is the molecular coincidence detector behind it -- blocked by magnesium at rest, it opens only when glutamate binding coincides with postsynaptic depolarization, admitting the calcium pulse that activates CaMKII and drives more AMPA receptors into the synapse. Long-term depression (LTD) weakens synapses through related but distinct calcium signaling, and spike-timing-dependent plasticity refines the rule: inputs that fire just before the postsynaptic spike strengthen, those after weaken. Structural changes follow -- dendritic spines grow, new synapses form -- and systems-level consolidation transfers memories from hippocampus to cortex over days. LTP remains the leading cellular model of memory, though the field is careful: blocking LTP impairs learning, but memory is a network property, not a single synapse's.
Tools and open questions
Neuroscience advances through its instruments. The patch clamp (Neher and Sakmann, Nobel 1991) records currents through single ion channels; calcium indicators like GCaMP let researchers watch thousands of neurons fire in behaving animals; optogenetics -- microbial opsins expressed in defined neurons -- provides causal control with millisecond light pulses; and connectomics maps every synapse, beginning with the 302-neuron C. elegans wiring diagram (1986) and now scaling to flies and cubic millimeters of mouse brain. Yet fundamental questions persist. How does the brain encode information -- the neural code -- across millions of noisy neurons? How do circuits implement cognition, and where does conscious experience arise? How do genes build brains that then rewire themselves? The field's trajectory is from correlation to causation: from watching neurons fire during behavior to controlling them and predicting the behavioral consequence, closing the loop between molecule, circuit, and mind.
The human brain holds about 86 billion neurons and perhaps 100 trillion synapses. Mapping and understanding this network -- the connectome -- is a project measured in decades, with the full fly brain connectome completed only in 2024.
Key points
- The resting potential (~-70 mV) reflects K+ gradients maintained by the Na+/K+ ATPase.
- Hodgkin and Huxley's 1952 model explains the action potential through voltage-gated Na+ and K+ channels.
- Myelin enables saltatory conduction, speeding signals up to ~100 m/s.
- Calcium-triggered, SNARE-mediated vesicle fusion releases neurotransmitter in quanta.
- NMDA receptors act as coincidence detectors underlying LTP, the leading cellular model of memory.
- Patch clamp, calcium imaging, optogenetics, and connectomics drive the field from correlation to causation.
Common misconceptions
✓ Truth: The action potential is all-or-none, but neurons encode information in firing rate, precise spike timing, bursting patterns, and graded dendritic integration. The code is far richer than binary.
✓ Truth: Imaging shows virtually all brain regions active across a day; even simple tasks engage widespread networks. The myth misreads early neuroscience -- neurons are metabolically expensive, and evolution does not maintain idle tissue.
✓ Truth: Adult neurogenesis occurs in the hippocampus of many mammals, including debated evidence in humans, and synaptic rewiring continues throughout life. What declines with age is the rate, not the capacity for plasticity itself.
Frequently asked questions
What exactly is the refractory period for?
It has two jobs. The absolute refractory period, caused by inactivated sodium channels, prevents the action potential from traveling backward and limits maximum firing rate. The relative refractory period, from lingering potassium efflux, means stronger stimuli are needed to fire again, shaping firing patterns.
Why does myelin speed up conduction so dramatically?
Myelin insulates axon segments, reducing capacitance and ion leakage, so depolarization spreads passively and rapidly to the next node of Ranvier, where voltage-gated channels regenerate the spike. The signal effectively jumps node to node instead of being rebuilt continuously.
How can the NMDA receptor detect coincidence?
At resting voltage, a magnesium ion blocks its pore. Glutamate binding alone is not enough; the postsynaptic membrane must also depolarize to expel the magnesium. Only when presynaptic activity coincides with postsynaptic firing does calcium flood in -- the physical basis of Hebbian learning.
Is LTP really the same thing as memory?
Not exactly. LTP is the best cellular model of memory storage: blocking it impairs learning, and learning induces LTP-like changes. But memories are network-level phenomena involving consolidation, retrieval circuits, and systems interactions -- LTP is a mechanism, not the whole story.