29.02.04 · psychology / neuroscience

Neuroplasticity and neurogenesis: LTP, critical periods, recovery from brain injury

stub3 tiersLean: nonepending prereqs

Anchor (Master): Bliss and Lomo — Long-lasting potentiation of synaptic transmission (1973)

Intuition Beginner

The brain was once thought to be fixed after childhood. We now know it changes throughout life — this is neuroplasticity. When you learn something new, the connections between neurons strengthen. Canadian psychologist Donald Hebb said it simply: "Neurons that fire together, wire together." Long-term potentiation (LTP) is the molecular basis of this — when two neurons communicate repeatedly, their connection gets stronger, sometimes for hours or days.

The brain has critical periods — windows of time when it is especially receptive to learning. A child can learn languages effortlessly during a language critical period, but adults struggle because that window has closed. After brain injury (like a stroke), healthy brain regions can sometimes take over the functions of damaged areas through reorganization.

Visual Beginner

The table below organizes the main forms of plasticity by the timescale over which they operate and a familiar example of each. Plasticity is not a single phenomenon but a hierarchy of mechanisms spanning milliseconds to years.

Form of plasticity	Timescale	Everyday example
Synaptic strengthening (LTP)	Minutes to hours	Memorizing a phone number
Synaptic weakening (LTD)	Minutes to hours	Forgetting a rarely used fact
Dendritic spine growth	Hours to days	Learning to juggle
Critical-period learning	Weeks to years	A toddler absorbing language
Cortical reorganization	Weeks to months	Recovering movement after stroke
Adult neurogenesis	Weeks to months	New neurons in the hippocampus

The same underlying machinery — activity-dependent changes in synaptic strength — operates across all these scales. What differs is the molecular depth of the change (receptor insertion versus new synapse growth) and the developmental stage of the brain.

Worked example Beginner

Consider a patient who suffers a stroke damaging the motor cortex controlling the right hand. In the first weeks the hand is paralyzed. With intensive rehabilitation, movement gradually returns — not because the dead neurons recover, but because neighbouring cortical areas reorganize to take control of the hand's muscle maps.

Neurologist Edward Taub discovered that forcing patients to use their impaired arm — by restraining the good arm — produced dramatic recovery. This constraint-induced movement therapy works because it counteracts learned non-use: patients stop trying with the bad arm, and the brain reallocates its cortical real estate. Forced use drives the remaining healthy cortex to rewire around the damage.

Check your understanding Beginner

Exercise (easy, multiple choice).

A child learns a second language with native fluency; an adult who studies the same language for years still speaks with an accent. Which concept best explains this difference?

A. Long-term depression B. Critical period C. Neurotransmitter depletion D. Synaptic pruning error

Hint

There is a developmental window during which the brain is maximally receptive to certain kinds of input, like the sounds of a language. After it closes, learning is still possible but less complete.

Answer

Option B. Critical periods are developmental windows during which the brain is maximally plastic for specific functions. Language sound discrimination has a critical period that peaks in infancy and closes by puberty. Within the window the brain absorbs phonetic categories effortlessly; after it closes, the brain can still learn grammar and vocabulary but cannot fully acquire native pronunciation.

Formal definition Intermediate+

Hebbian learning and the synaptic plasticity rule

Donald Hebb's 1949 postulate — "when an axon of cell A is near enough to excite cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place such that A's efficiency, as one of the cells firing B, is increased" — is the conceptual seed of all modern plasticity theory. Formally, a Hebbian plasticity rule strengthens the weight $w_{ij}$ of the synapse from presynaptic neuron $i$ to postsynaptic neuron $j$ in proportion to the correlation of their activities:

Δ w_{ij} = η x_{i} y_{j}

where $x_{i}$ is the presynaptic activity, $y_{j}$ is the postsynaptic activity, and $η$ is a learning rate. The pure Hebbian rule is unstable: without a counterbalancing mechanism, correlated firing drives weights to infinity. Two solutions dominate biology. Synaptic normalization (competitive) caps total synaptic weight per neuron. Homeostatic synaptic scaling (global, multiplicative) scales all of a neuron's weights up or down to keep its average firing rate in a target range (Turrigiano, 1998). Long-term depression provides the local, input-specific counterweight: synapses that are active when the postsynaptic cell is weakly depolarized are selectively weakened, not strengthened.

The NMDA receptor as coincidence detector

The molecular implementation of Hebb's rule centres on the NMDA-type glutamate receptor. Unlike the AMPA receptor, which opens whenever glutamate binds, the NMDA receptor is doubly gated. Its channel pore is blocked by a magnesium ion ( $Mg^{2 +}$ ) at the resting membrane potential. The block is voltage-dependent: it is expelled only when the postsynaptic membrane is already depolarized. The NMDA channel therefore opens only when two conditions are met simultaneously — glutamate is bound (presynaptic activity has occurred) and the postsynaptic membrane is depolarized (the postsynaptic cell is active). This makes the NMDA receptor a molecular coincidence detector, implementing an AND gate at the synaptic level.

When the NMDA channel opens, it admits calcium ions ( $Ca^{2 +}$ ) into the postsynaptic spine. Calcium is the master second messenger of plasticity: its concentration and time course determine whether the synapse is potentiated or depressed.

Early-phase LTP: CaMKII and AMPA receptor insertion

In early-phase LTP (E-LTP, lasting 1-3 hours), the calcium influx activates calcium/calmodulin-dependent protein kinase II (CaMKII). CaMKII autophosphorylates — once activated by calcium/calmodulin, it phosphorylates itself at threonine-286, rendering itself persistently active even after calcium falls. This molecular memory switch converts a transient calcium signal into a sustained kinase activity. Active CaMKII phosphorylates AMPA receptor subunits (GluA1 at serine-831), increasing their single-channel conductance, and drives the insertion of new AMPA receptors into the postsynaptic density from intracellular recycling endosomes. The net effect: the same amount of presynaptic glutamate release now produces a larger postsynaptic response. The synapse has been potentiated.

Late-phase LTP: protein synthesis and structural change

Late-phase LTP (L-LTP, lasting days to weeks and potentially the lifetime of the animal) requires de novo protein synthesis. The calcium signal, amplified through CaMKII, propagates via the cAMP/PKA and MAPK pathways to the nucleus, where it activates the transcription factor CREB (cAMP response element-binding protein). CREB drives the expression of immediate early genes — notably Arc and c-fos — whose protein products stabilize the potentiation. The synthesis of new proteins enables structural plasticity: the growth of new dendritic spines, the enlargement of existing spines, and the expansion of the postsynaptic density. Protein synthesis inhibitors applied during the induction of L-LTP abolish the late phase while sparing the early phase, cleanly dissociating the two.

An analogous distinction holds for the transition from short-term to long-term memory at the behavioural level, and this pharmacological dissociation is one of the strongest links between synaptic physiology and systems-level memory.

Long-term depression

Long-term depression (LTD) is the activity-dependent weakening of synapses — the necessary counterpart to LTP. In the hippocampus and cerebellum, LTD is also NMDA-dependent, but the distinguishing factor is the magnitude and time course of the calcium signal. A large, brief calcium rise (as in LTP) favours kinases; a smaller, prolonged calcium rise (as in LTD) favours phosphatases — specifically calcineurin (a calcium/calmodulin-dependent phosphatase) and PP1. Phosphatases dephosphorylate AMPA receptor subunits, triggering receptor internalization: AMPA receptors are removed from the postsynaptic membrane and recycled into endosomes. With fewer AMPA receptors, the same glutamate release produces a smaller postsynaptic response.

This calcium-threshold model — high calcium drives LTP via kinases, moderate calcium drives LTD via phosphatases — elegantly explains how a single receptor (NMDA) and a single ion (calcium) can produce bidirectional plasticity. The cerebellar form of LTD, at the parallel fibre–Purkinje cell synapse, is mechanistically distinct (it depends on postsynaptic climbing-fibre activation and mGluR signalling) but converges on the same endpoint of AMPA receptor internalization.

Spike-timing-dependent plasticity

The Hebbian rule $Δ w = η x_{i} y_{j}$ is rate-based: it depends on the average firing rates of the two neurons. Spike-timing-dependent plasticity (STDP), characterized by Bi and Poo (1998), revealed that the precise relative timing of individual spikes determines the sign and magnitude of plasticity. If the presynaptic neuron fires a few milliseconds before the postsynaptic neuron (pre-before-post, causally meaningful), the synapse is potentiated. If the presynaptic spike fires a few milliseconds after the postsynaptic spike (post-before-pre, causally meaningless or interfering), the synapse is depressed. The plasticity window is asymmetric and narrow: the magnitude of the change falls off exponentially with the inter-spike interval, vanishing beyond roughly 40 milliseconds.

STDP provides a far more temporally precise implementation of Hebb's principle than the rate-based rule. It is the dominant plasticity rule in computational models of self-organizing maps, receptive-field development, and reinforcement learning in spiking neural networks. Bi and Poo's experiments in cultured hippocampal neurons demonstrated it directly by pairing patch-clamp recordings of two connected neurons and controlling their relative spike timing with millisecond precision.

Critical periods in development

A critical period is a developmental window during which the brain is maximally susceptible to environmental influence for a specific function. Outside the window the same experience has little or no effect. The paradigmatic example is ocular dominance plasticity in the primary visual cortex (V1), established by David Hubel and Torsten Wiesel.

Hubel and Wiesel recorded from V1 neurons in cats and kittens. In a normal adult cat, V1 neurons respond preferentially to input from one eye or the other, and the cortex is organized into alternating ocular dominance columns — stripes of tissue devoted to the left eye alternating with stripes devoted to the right. When Hubel and Wiesel sutured one eye shut (monocular deprivation) in a kitten during a critical period (roughly 3 weeks to 3 months of age), the results were dramatic: cortical neurons that should have responded to the deprived eye shifted their allegiance almost entirely to the open eye. The deprived eye became functionally blind in the cortex, even though the eye itself and the retina were perfectly healthy. This effect was permanent if the deprivation spanned the critical period.

The same deprivation in an adult cat produced almost no shift in ocular dominance. The critical period had closed. Hubel and Wiesel also showed that the timing within the critical period mattered: a short deprivation early in the window produced larger effects than a longer deprivation later. They shared the 1981 Nobel Prize for this body of work.

Critical periods exist for many functions beyond vision: language phonology, absolute pitch, song learning in birds, filial imprinting, and binocular stereopsis each have their own windows. The windows differ in onset, duration, and molecular machinery, but the logic is the same: during the window the circuit is exquisitely sensitive to input, and after it closes the circuit crystallizes into its adult configuration.

Mechanisms that close critical periods

What terminates a critical period? The leading account implicates the maturation of inhibitory (GABAergic) circuitry, particularly parvalbumin-positive (PV+) fast-spiking interneurons. Early in development, inhibitory tone is low, and the cortex is highly excitable and plastic. As PV+ interneurons mature and strengthen their inhibitory synapses onto pyramidal cells, the balance of excitation and inhibition shifts. Strong inhibition raises the threshold for NMDA receptor activation and calcium influx, effectively shutting down the plasticity machinery.

A structural correlate of this closure is the deposition of perineuronal nets (PNNs) — dense extracellular matrix structures that envelop PV+ interneurons and mature synapses. PNNs form progressively during development and stabilize the synaptic architecture, rendering circuits resistant to further experience-dependent change. Enzymatic digestion of PNNs (with chondroitinase ABC) in adult animals can reopen critical-period-like plasticity, a finding with implications for recovering amblyopia in adults.

Additional molecular brakes on plasticity include myelin-associated growth inhibitors and the Otx2 homeoprotein, which is imported into PV+ interneurons and required for their maturation and PNN assembly. The emerging picture is that critical-period closure is an active process — a set of molecular "brakes" that stabilize the circuit — rather than a passive loss of plasticity.

Adult neurogenesis

For most of the twentieth century the dogma held that no new neurons are born in the adult mammalian brain. Santiago Ramón y Cajal, the founder of modern neuroscience, declared in 1913 that "once development was ended, the founts of growth of the axons and dendrites dried up irrevocably. In the adult centers, the nerve paths are something fixed, ended, and immutable."

This dogma was overturned in the 1990s. Peter Eriksson and colleagues (1998) examined postmortem brain tissue from cancer patients who had received bromodeoxyuridine (BrdU), a thymidine analogue that labels dividing cells, as a diagnostic marker for tumour proliferation. They found BrdU-labelled neurons in the dentate gyrus of the hippocampus, demonstrating that new neurons are generated in the adult human brain. Adult neurogenesis is now confirmed in two brain regions in mammals: the subgranular zone of the dentate gyrus (producing granule cell neurons that integrate into hippocampal circuits) and the subventricular zone of the lateral ventricles (producing interneurons that migrate via the rostral migratory stream to the olfactory bulb).

The functional significance and even the existence of adult hippocampal neurogenesis in humans remains contested. Sorrells and colleagues (2018), using improved histological methods, reported that neurogenesis in the human dentate gyrus drops to undetectable levels after childhood. Boldrini and colleagues (2018), using different methods, found neurogenesis persisting into old age. The disagreement persists, partly because postmortem tissue quality and fixation protocols profoundly affect the detection of immature neuronal markers (such as Doublecortin, DCX, and PSA-NCAM). What is clear is that in rodents, where the phenomenon is robust, adult-born granule cells contribute to pattern separation, the discrimination of similar but not identical experiences, and to the behavioural response to antidepressants and exercise.

Environmental factors robustly modulate neurogenesis in animal models. Exercise (voluntary wheel running) increases the proliferation of neural progenitors. Environmental enrichment (larger cages with toys, social interaction, and exercise opportunities) increases both the production and the survival of new neurons. Learning tasks that engage the hippocampus promote the survival of newly generated cells. Conversely, stress and elevated glucocorticoids, aging, sleep deprivation, and alcohol all reduce neurogenesis. The antidepressant effects of SSRIs, exercise, and electroconvulsive therapy are all correlated with increased hippocampal neurogenesis in rodents, raising the possibility that neurogenesis is a common pathway for antidepressant action — though the causal link in humans is unproven.

Recovery from brain injury

When a stroke or traumatic brain injury destroys neural tissue, the brain has limited but real capacity to reorganize. Recovery proceeds through several mechanisms operating on different timescales.

Spontaneous recovery dominates the first weeks to months. Three processes contribute. First, resolution of edema and reperfusion of the ischemic penumbra — tissue that was dysfunctional but not dead — restores function in regions adjacent to the infarct. Second, diaschisis resolves: the sudden loss of input from the damaged area depresses activity in remote but connected regions, and this depression lifts over weeks as the surviving circuits adapt. Third, unmasking of latent synapses — previously silent pathways that were functionally suppressed — rapidly expands the repertoire of surviving circuits.

Rehabilitation drives activity-dependent plasticity that consolidates and extends spontaneous recovery. The most evidence-based approach is constraint-induced movement therapy (CIMT), developed by Edward Taub. Taub's insight came from studies of deafferentation in monkeys: when a single limb was deafferented (its sensory input surgically removed), the monkey stopped using it — not because motor circuits were damaged, but because the animal learned to compensate with the intact limbs. Taub called this learned non-use. He found that restraining the good limb forced intensive use of the impaired limb, and that this forced use drove dramatic cortical reorganization and functional recovery. CIMT, now widely applied in stroke rehabilitation, combines restraint of the unaffected limb with intensive, shaped practice of the affected limb over several hours per day for two weeks. Neuroimaging shows that CIMT produces measurable enlargement of the cortical representation of the affected hand.

Beyond CIMT, recovery is promoted by task-specific repetitive practice, robotic-assisted therapy, virtual reality training, and non-invasive brain stimulation (transcranial magnetic stimulation and transcranial direct current stimulation, covered in the master tier). The common principle is that recovery requires intensive, repetitive, task-relevant activity that drives the surviving cortex to reorganize — passive movement without active effort produces far less plasticity.

Phantom limb and cortical remapping

Plasticity is not always beneficial. After amputation, most patients experience a phantom limb — vivid sensations, often painful, that appear to originate from the missing limb. The neuroscientific basis, worked out by V. S. Ramachandran, is cortical remapping. In the primary somatosensory cortex, the body is represented as a map (the sensory homunculus). The hand area is adjacent to the face area. When the hand is amputated, the cortical territory that represented the hand no longer receives input and is rapidly invaded by neighbouring representations. Touching the face of an arm amputee can evoke sensations in the phantom hand, because the face area has expanded into the vacated hand territory.

This remapping illustrates that plasticity is a double-edged sword: the same mechanisms that enable learning and recovery also produce maladaptive sensations. Phantom limb pain may arise when the remapped cortex generates spontaneous, aberrant activity that the brain interprets as pain in the missing limb. Ramachandran's mirror box therapy — using a mirror to create the visual illusion that the missing limb is present and can be moved — exploits visual feedback to "retrain" the remapped cortex and can relieve phantom pain in some patients.

Key mechanism: NMDA-dependent long-term potentiation Intermediate+

Of all the mechanisms described above, the most consequential for understanding neuroplasticity is the NMDA receptor–dependent LTP cascade. It is the best-characterized molecular implementation of Hebb's principle, the leading candidate cellular mechanism for learning and memory, and the point of departure for virtually all modern plasticity research.

Bliss and Lomo's discovery

In 1973, Timothy Bliss and Terje Lomo, working in Per Andersen's laboratory in Oslo, delivered brief, high-frequency electrical stimulation (a tetanus: 100 Hz for 3-4 seconds) to the perforant path — the fibre tract entering the dentate gyrus of the rabbit hippocampus. They recorded the population excitatory postsynaptic potential (EPSP) from the granule cell layer. After the tetanus, the synaptic response to a single test pulse was dramatically and persistently enhanced — it remained elevated for hours. This was long-term potentiation: activity-dependent, synapse-specific, and enduring. The phenomenon displayed three properties that made it an attractive memory mechanism: input specificity (only the stimulated pathway was potentiated; neighbouring unstimulated inputs to the same cells were unchanged), associativity (a weak input could be potentiated if it was active at the same time as a strong input to the same postsynaptic cell), and persistence (lasting hours in vitro, and in vivo potentially weeks).

The molecular cascade, step by step

The induction of NMDA-dependent LTP at a hippocampal CA3-to-CA1 synapse (Schaffer collateral input) proceeds through a stereotyped cascade.

Step 1: Coinidence detection. A high-frequency tetanus depolarizes the postsynaptic membrane sufficiently to expel the magnesium block from NMDA receptor channels. With glutamate bound (from presynaptic release) and the membrane depolarized, NMDA channels open and calcium floods into the postsynaptic spine. The NMDA receptor's voltage dependence is what makes LTP associative and input-specific: only synapses whose activity coincides with postsynaptic depolarization admit calcium.

Step 2: CaMKII activation. The calcium binds calmodulin, and the calcium-calmodulin complex activates CaMKII. At high local calcium concentrations, CaMKII subunits are phosphorylated at threonine-286. This autophosphorylation is critical: it renders CaMKII constitutively active, so the kinase continues to operate after the calcium transient has subsided. A few CaMKII holoenzymes (each a dodecamer) anchored in the postsynaptic density can thereby maintain a persistent enzymatic signal from a brief trigger — a molecular switch.

Step 3: AMPA receptor potentiation. Active CaMKII phosphorylates GluA1 subunits of AMPA receptors at serine-831, increasing their single-channel conductance. Simultaneously, CaMKII drives the exocytosis of intracellular AMPA receptor-containing vesicles and their insertion into the postsynaptic density. The net effect is more AMPA receptors, each more conductive, at the potentiated synapse.

Step 4: Structural consolidation (late phase). For LTP to persist beyond a few hours, the calcium signal must reach the nucleus via the cAMP/PKA and MAPK pathways. CREB is activated and drives transcription of plasticity-related proteins, including Arc (which locally regulates AMPA receptor trafficking at potentiated synapses), BDNF (brain-derived neurotrophic factor, which promotes spine growth and synaptic stabilization), and structural proteins that enlarge the postsynaptic density and stabilize new dendritic spines. Protein synthesis inhibitors block L-LTP without affecting E-LTP, establishing that the late phase is transcription- and translation-dependent.

Why this mechanism matters

The NMDA-LTP cascade is the most completely characterized molecular implementation of a Hebbian learning rule in the brain, and it is the bridge between synaptic physiology and behavioural memory. Genetic manipulations that disrupt specific components of the cascade produce parallel deficits in LTP and in learning: mice lacking CaMKII autophosphorylation (T286A mutants) are impaired in both hippocampal LTP and spatial learning in the Morris water maze. Mice with knocked-out NMDA receptor subunits specifically in CA1 lose both LTP and the ability to form new spatial memories. These convergent genetic, pharmacological, and electrophysiological results make LTP the strongest known correlate of memory at the synaptic level.

The mechanism also defines the boundary of what is known. While the induction cascade is mapped in molecular detail, how the persistent change is maintained for the lifetime of a memory — days to decades — remains an open question. The synthesis of new proteins during L-LTP stabilizes potentiation for days, but proteins turnover on the timescale of hours to days, so the physical trace of a long-term memory cannot reside in any individual molecule. The leading hypothesis (synaptic tagging and capture, and the clustering of synaptic molecules) posits that memory is stored in the structure and molecular composition of the synapse as a population, maintained by self-sustaining local interactions rather than by any single long-lived molecule.

Exercises Intermediate+

Exercise 1 (medium, multiple choice).

A synapse receives low-frequency stimulation (1 Hz for 15 minutes) at a hippocampal CA1 neuron, producing a modest, prolonged rise in postsynaptic calcium. What is the most likely outcome at this synapse?

A. Long-term potentiation, because calcium always drives LTP B. Long-term depression, because the moderate calcium signal favours phosphatases C. No change, because only high-frequency stimulation affects synapses D. Excitotoxicity, because calcium is always neurotoxic

Hint

The calcium-threshold model distinguishes LTP from LTD by the magnitude and kinetics of the calcium signal. Consider which enzymes dominate at moderate versus high calcium.

Answer

Option B. The calcium-threshold model holds that a large, brief calcium rise activates kinases (CaMKII, PKA) and drives LTP, while a modest, prolonged calcium rise preferentially activates phosphatases (calcineurin, PP1) and drives LTD. Low-frequency stimulation produces the moderate calcium profile, so LTD is the expected result: AMPA receptors are dephosphorylated and internalized, weakening the synapse. Option A is wrong because calcium does not always drive LTP. Option C is wrong because low-frequency stimulation is the standard LTD induction protocol.

Exercise 2 (medium, conceptual).

Explain why the NMDA receptor's magnesium block is essential for Hebbian learning. What would happen to synaptic plasticity if NMDA receptors lacked the magnesium block and opened whenever glutamate bound?

Hint

The magnesium block imposes a requirement for postsynaptic depolarization. Consider what this dual requirement (glutamate binding AND depolarization) achieves, and what would be lost if only glutamate binding were required.

Answer

The magnesium block makes the NMDA receptor a coincidence detector: the channel opens only when presynaptic glutamate release (glutamate bound) and postsynaptic activity (depolarization expelling the Mg $^{2 +}$ block) co-occur. This dual requirement implements Hebb's rule at the molecular level — only synapses whose activity correlates with postsynaptic firing are strengthened. If the magnesium block were removed, any glutamate release would permit calcium entry regardless of postsynaptic state, so all active glutamatergic synapses would be potentiated indiscriminately. Input specificity and associativity would be abolished: the synapse could no longer distinguish correlated from uncorrelated input, and learning would lose its selectivity.

Exercise 3 (medium, multiple choice).

In Hubel and Wiesel's monocular deprivation experiments, suturing one eye of a kitten shut during the critical period caused cortical neurons to shift their responses toward the open eye. If the same deprivation is performed on an adult cat, what happens?

A. The same dramatic shift in ocular dominance occurs B. Almost no shift occurs; the critical period has closed C. The cortex becomes entirely unresponsive to both eyes D. The open eye becomes permanently blind

Hint

The critical period is defined by the developmental window during which experience can reshape cortical organization. Consider what "critical period closure" means for the same manipulation applied later.

Answer

Option B. The defining feature of a critical period is that the same experience has drastically different effects inside versus outside the window. Monocular deprivation in an adult cat produces little to no shift in ocular dominance, because the inhibitory circuitry, perineuronal nets, and other molecular brakes have stabilized the cortical architecture. This is what it means for the critical period to have closed: the cortex is no longer susceptible to experience-driven reorganization at the same scale. The closure is why adult amblyopia (lazy eye) is far harder to treat than childhood amblyopia.

Exercise 4 (medium, conceptual).

Bi and Poo (1998) found that pairing presynaptic and postsynaptic spikes with precise relative timing determines whether a synapse is strengthened or weakened. Describe the STDP rule and explain why the sign of plasticity reverses when the temporal order of the spikes is flipped.

Hint

Consider what pre-before-post means causally (the presynaptic spike contributed to firing the postsynaptic cell) versus post-before-pre (the presynaptic spike arrived too late to have contributed). Which temporal order should be rewarded, and which punished?

Answer

In STDP, if the presynaptic neuron fires a few milliseconds before the postsynaptic neuron (pre-before-post), the synapse is potentiated: the presynaptic input was causally involved in triggering the postsynaptic spike, so the connection is strengthened to reinforce that causal link. If the presynaptic neuron fires a few milliseconds after the postsynaptic spike (post-before-pre), the synapse is depressed: the presynaptic input arrived too late to have contributed to the postsynaptic firing, so weakening the connection prunes a non-causal input. The sign reversal reflects a causal-inference principle — the brain strengthens inputs that predict postsynaptic firing and weakens inputs that do not. The window is narrow (tens of milliseconds) and asymmetric, with potentiation typically having a narrower window than depression.

Exercise 5 (hard, experimental design).

A researcher applies a protein synthesis inhibitor (anisomycin) to a hippocampal slice during the induction of LTP and finds that the potentiation is normal at 1 hour but absent at 24 hours. A second researcher finds that a CaMKII inhibitor (KN-62) abolishes the potentiation at both 1 and 24 hours. Interpret these results in terms of early-phase and late-phase LTP and identify the molecular step each drug disrupts.

Hint

Consider the cascade: CaMKII activation is an early, kinase-dependent step that does not require new protein synthesis. Late-phase LTP requires CREB-driven transcription and translation. Which drug targets which phase?

Answer

Anisomycin (the protein synthesis inhibitor) blocks late-phase LTP but spares early-phase LTP. At 1 hour, the potentiation is driven by CaMKII-mediated AMPA receptor phosphorylation and insertion — processes that use existing proteins and do not require new synthesis. At 24 hours, the potentiation depends on plasticity-related proteins synthesized via the CREB pathway; without protein synthesis, the late phase fails. This double dissociation confirms that E-LTP and L-LTP are mechanistically distinct.

KN-62 (the CaMKII inhibitor) abolishes both phases because CaMKII activation is upstream of both the early AMPA receptor potentiation and the signalling to the nucleus. Without CaMKII, the calcium signal cannot be converted into either the early kinase cascade or the downstream transcriptional activation, so neither phase is induced. This establishes CaMKII as a necessary and early node in the LTP cascade, upstream of the protein-synthesis-dependent step.

Exercise 6 (hard, multiple choice).

Eriksson et al. (1998) identified BrdU-labelled neurons in the postmortem human dentate gyrus. Which finding provides the strongest evidence that these BrdU-labelled cells are genuinely new neurons rather than pre-existing neurons that incorporated BrdU through DNA repair?

A. The BrdU-labelled cells were located only in the cerebral cortex B. The BrdU-labelled cells co-expressed neuronal markers (NeuN) and had morphological features of granule neurons C. The BrdU-labelled cells expressed glial markers (GFAP) instead of neuronal markers D. The BrdU-labelled cells were found only in patients younger than 20

Hint

BrdU labels dividing cells by incorporating into newly synthesized DNA. To prove the labelled cells are neurons (not glia, not DNA-repair artefacts), what markers and morphology would you need to show in the same cells?

Answer

Option B. BrdU is incorporated into DNA during S-phase, so it labels dividing cells. To prove the labelled cells are neurons, one must demonstrate that the BrdU-positive cells co-express neuron-specific markers (such as NeuN, a neuronal nuclear antigen) and display the morphology and location of granule cell neurons in the dentate gyrus. Eriksson et al. showed exactly this co-localization. Option A is wrong because neurogenesis is found in the dentate gyrus, not cortex. Option C would indicate glia, not neurons. Option D is incorrect because Eriksson found labelled neurons in adults of various ages, which was precisely the point — neurogenesis persists beyond childhood. The co-expression of BrdU with NeuN and the granule-cell morphology are what distinguish genuine adult neurogenesis from DNA repair or glial proliferation.

Advanced topics Master

The LTP molecular cascade in full: CaMKII, PKA, MAPK, CREB, and immediate early genes

The master-level account of LTP requires tracing the full signalling cascade from the NMDA receptor to gene expression. The early phase centres on CaMKII, whose autophosphorylation at T286 creates a persistent molecular switch. But the transition to late-phase LTP engages a second tier of kinases. Calcium, acting through adenylyl cyclase, raises cAMP and activates protein kinase A (PKA). PKA in turn activates MAPK (mitogen-activated protein kinase, also called ERK) via the Rap1/B-Raf pathway. MAPK translocates to the nucleus, where it phosphorylates and activates CREB at serine-133. Phosphorylated CREB binds to cAMP response elements (CREs) in the promoters of target genes and drives their transcription.

The most important CREB targets for plasticity are the immediate early genes (IEGs) — genes whose transcription is rapidly induced (within minutes) without requiring new protein synthesis, because their transcription factors (including CREB itself) are constitutively present. Arc (Activity-regulated cytoskeleton-associated protein) is the best-characterized plasticity-related IEG. Arc mRNA is rapidly transported to active synapses, where it is locally translated and regulates AMPA receptor endocytosis — bidirectionally tuning synaptic strength. c-fos and Zif268 (Egr1) are additional IEGs used as markers of neural activity and plasticity. Genetic deletion of Arc impairs long-term memory and causes a striking inability to maintain LTP beyond a few hours, while leaving initial learning intact.

The full cascade therefore operates as a feedforward amplification system: a brief NMDA-mediated calcium signal activates CaMKII (a local, fast switch), which propagates through PKA and MAPK to the nucleus (a slower, amplifying tier), which drives CREB-dependent transcription of Arc and other IEGs (a genomic, enduring tier) whose protein products act locally at the active synapse to consolidate the change. The temporal architecture — fast local switch, slower amplifying cascade, genomic consolidation — maps onto the psychology of memory consolidation: short-term memory persists for minutes (E-LTP), and consolidation into long-term memory requires protein synthesis over hours (L-LTP).

Structural plasticity: spine dynamics and two-photon imaging

LTP is not only a change in receptor number and conductance but also a structural change in the synapse. Dendritic spines — the tiny protrusions on dendrites that receive excitatory input — change shape and number with plasticity. Spines are not static: they form, grow, shrink, and disappear on a timescale of minutes to days. LTP induces the growth of new spines ("filopodia" maturing into stubby or mushroom spines) and the enlargement of existing spine heads. LTD induces spine shrinkage and elimination. The morphology of a spine is correlated with its strength: larger, mushroom-shaped spines with well-developed postsynaptic densities are stronger and more stable; thin, filopodium-like spines are weaker, more transient, and more plastic.

This dynamic picture was revealed by two-photon laser scanning microscopy, which allows imaging of individual dendritic spines in the living brain (through a cranial window) over days and weeks. Two-photon microscopy uses infrared light (which penetrates tissue more deeply than visible light and causes less phototoxicity) to excite fluorescent dyes in a tiny focal volume, producing optical sections without physical sectioning. Using this technology, Svoboda, Bonhoeffer, and others showed that motor learning induces the formation of new dendritic spines in the motor cortex, that a fraction of these new spines persist for days and stabilize the learned skill, and that the same spines are stabilized when the same task is relearned — suggesting that the physical trace of a motor memory resides in the subset of spines that were created and preserved during the initial learning.

The rules governing spine dynamics have been refined by time-lapse imaging combined with optogenetic or electrophysiological induction of plasticity. New spines formed during LTP tend to appear in clusters along the dendrite, suggesting that plasticity is not purely synapse-by-synapse but involves dendritic compartmentalization. The clustered plasticity hypothesis proposes that spatially clustered synapses on the same dendrite cooperate, so that the potentiation of one synapse lowers the threshold for potentiating its neighbours. This local cooperativity may be the substrate for associating related inputs.

Metaplasticity: the Bienenstock-Cooper-Munro sliding threshold

Plasticity itself is plastic. Metaplasticity — the plasticity of synaptic plasticity — refers to the fact that the history of a neuron's activity changes the rules governing future plasticity at its synapses. The most influential formalization is the Bienenstock-Cooper-Munro (BCM) theory (1982).

In the BCM model, the sign of plasticity depends on the postsynaptic firing rate $y$ relative to a modification threshold $θ_{M}$ . For low postsynthetic firing rates ( $y < θ_{M}$ ), LTD is induced. For high firing rates ( $y > θ_{M}$ ), LTP is induced. The nontrivial feature is that the threshold $θ_{M}$ is itself a function of the neuron's recent average activity: $θ_{M} = \overset{y}{ˉ}^{p}$ for some $p > 1$ , where $\overset{y}{ˉ}$ is a time-averaged postsynaptic activity. When the neuron has been firing at a high rate (high $\overset{y}{ˉ}$ ), the threshold $θ_{M}$ shifts upward, making LTP harder to induce and LTD easier — a homeostatic brake. When the neuron has been quiet, $θ_{M}$ shifts downward, lowering the bar for LTP.

The BCM sliding threshold solves two problems simultaneously. First, it provides homeostatic stability: a neuron that is chronically overactive raises its LTP threshold and lowers its LTD threshold, driving synaptic weights down and restoring normal firing; the reverse for a chronically underactive neuron. This prevents the runaway potentiation that pure Hebbian rules produce. Second, it generates selectivity: in a model of visual cortical development, BCM neurons converge on orientation-selective receptive fields that match the statistics of natural images, providing a principled account of how experience shapes cortical tuning during development.

The molecular correlate of the BCM threshold is thought to involve the balance of kinase and phosphatase activation as a function of calcium level — the same calcium-threshold mechanism described for LTP/LTD above. Metaplasticity may also involve changes in NMDA receptor subunit composition (a shift from NR2B-containing to NR2A-containing receptors during development, which alters the kinetics and calcium permeability of the channel and changes the plasticity rules).

Homeostatic synaptic scaling

While Hebbian and BCM plasticity are synapse-specific (they modify individual connections based on local activity), homeostatic synaptic scaling (Turrigiano, Leslie, Desai, Rutherford, Nelson, and Turkis, 1998) operates globally across all of a neuron's synapses. When a neuron's firing rate is chronically driven above its target (for example, by blocking inhibition), the neuron scales down all of its excitatory synaptic strengths by a multiplicative factor. When firing is chronically suppressed, the neuron scales up all of its excitatory synapses. The scaling is multiplicative rather than additive, so it preserves the relative strengths of synapses (and thus the information stored in their pattern) while adjusting the overall gain.

The mechanism involves AMPA receptor insertion or removal across the entire dendritic tree, regulated by the neuron's average firing rate over hours. The signalling pathway involves retinoic acid synthesis and the transcription factor NPAS4, among other components. Homeostatic scaling is distinct from LTD/LTP in being slow (hours, not minutes), global (all synapses, not just active ones), and multiplicative (proportional, not absolute). Together with metaplasticity, it provides the stabilizing counterweight that keeps Hebbian learning from destroying the stored information it creates.

Reactivating critical periods in adults

If critical-period closure is an active process (inhibition maturation, PNN deposition), then reversing those molecular brakes might reopen critical-period-like plasticity in adults. Several strategies have been explored.

Valproate is a histone deacetylase (HDAC) inhibitor and mood-stabilizing drug. Gervain and colleagues (2013) showed that administering valproate to adult rats during auditory training restored their ability to learn to discriminate tones that are only discriminated during a developmental critical period — a form of "critical period reopening" for auditory plasticity. HDAC inhibitors promote chromatin relaxation (acetylation opens chromatin, making genes accessible for transcription) and thereby lower the barrier to experience-driven gene expression. The finding has motivated trials of valproate for restoring absolute-pitch learning and for enhancing auditory rehabilitation in adults.

Enzymatic digestion of perineuronal nets with chondroitinase ABC, injected into the adult visual cortex, restores ocular dominance plasticity: monocular deprivation in chondroitinase-treated adult rats produces shifts in ocular dominance that are normally seen only in juveniles. This approach has been used to promote recovery from amblyopia (lazy eye) in adult animals, a condition previously considered untreatable after the critical period.

Levetiracetam (Keppra, an antiepileptic drug) has been reported to restore binocular vision in adult animals with amblyopia by reducing the excessive inhibition in visual cortex that normally prevents plasticity. Clinical trials are exploring whether the same approach can restore vision in human adults with amblyopia.

The general principle is that critical periods are not strictly time-limited windows that close forever; they are states of the circuit that can, in principle, be re-entered by manipulating the molecular brakes (inhibition, PNNs, epigenetic state). Whether this can be done safely and selectively in humans — without destabilizing other circuits — is an open clinical question.

The critical period for absolute pitch

Absolute pitch (perfect pitch) — the ability to name or produce a musical note without a reference — is a striking example of a critical-period phenomenon. The prevalence of absolute pitch is far higher among individuals who began musical training before age 6 than among those who began later, controlling for years of training. The condition also has a substantial genetic component (higher concordance in identical twins, overrepresentation in certain populations), but the genetic predisposition appears to require early musical exposure during a developmental window to manifest. This gene-environment interaction is a textbook case of how a critical period gates the expression of a latent capacity.

Stem cell therapy and neural regeneration

The ambition to replace neurons lost to injury or disease has driven decades of stem cell research. Embryonic stem cells and induced pluripotent stem cells (iPSCs) can be differentiated into neural progenitors in vitro and transplanted into the brain. In animal models of Parkinson's disease, transplantation of fetal dopaminergic neurons into the striatum has produced functional integration and dopamine release, and some clinical trials have shown modest benefit. The results are inconsistent, however, and the variability has been attributed to differences in graft preparation, patient selection, and immunological factors.

The challenges of stem cell therapy are substantial. Transplanted neurons must survive, differentiate into the correct cell type, migrate to the correct location, form functional synapses with the host circuitry, and avoid forming tumours or aberrant connections. In the central nervous system, the environment after injury is actively hostile to regeneration: glial scars form around the lesion, and myelin-associated growth inhibitors (including Nogo-A, MAG, and OMgp) signal through the NgR1/p75/LINGO-1 receptor complex to collapse growth cones and prevent axon extension.

Spinal cord injury and the Nogo-A antibody

The limited regeneration of central nervous system axons stands in sharp contrast to the robust regeneration of peripheral nerves. The discovery of Nogo-A — a myelin protein that inhibits axon growth — by Martin Schwab and colleagues in 1989-2000 provided a molecular explanation and a therapeutic target. Anti-Nogo-A antibodies neutralize the inhibitory signal and promote sprouting of surviving axons, serotonergic fibre growth, and functional recovery in rodent and primate models of spinal cord injury.

An anti-Nogo-A antibody (nerinelimab) entered human clinical trials for spinal cord injury. The results have been mixed: while animal models showed dramatic recovery, translating this to humans has proven difficult, partly because human spinal cord injuries are heterogeneous (contusion, compression, complete versus incomplete transection) and partly because the recovery depends on intensive rehabilitation to drive the newly sprouted axons into functional circuits. The general lesson — that promoting regeneration requires both removing inhibitors and providing activity-dependent guidance — has shaped the field's current strategy of combining growth-promoting interventions with intensive rehabilitation.

Brain-computer interfaces for stroke recovery

Brain-computer interfaces (BCIs) record neural activity (from electrocorticography, EEG, or intracortical arrays) and use it to control an external device — a cursor, a robotic exoskeleton, or a functional electrical stimulation system that activates paralyzed muscles. For stroke recovery, the rationale is that decoding the patient's intention to move from the surviving cortex, and providing sensory feedback (via muscle movement or direct stimulation), closes the sensorimotor loop and drives activity-dependent plasticity in the peri-infarct cortex. The patient attempts to move, the BCI detects the attempt and moves the exoskeleton, the patient sees and feels the movement, and this congruent feedback reinforces the cortical circuits that generated the attempt.

Clinical trials of BCI-based stroke rehabilitation have shown measurable gains in motor function in patients with chronic stroke who had plateaued with conventional therapy. The approach is still experimental, but it exemplifies a general principle: recovery from brain injury is maximized when the rehabilitation engages the same intentional and feedback circuits that normal learning uses.

Transcranial direct current stimulation and motor learning

Transcranial direct current stimulation (tDCS) delivers a weak (1-2 mA) constant current through scalp electrodes, shifting the resting membrane potential of cortical neurons by a fraction of a millivolt. Anodal tDCS over the motor cortex modestly depolarizes cortical neurons, increasing their excitability; cathodal tDCS hyperpolarizes them, decreasing excitability. Applied during motor training, anodal tDCS has been reported to accelerate skill acquisition.

The effects are small and variable across individuals, and the field has been plagued by replication failures and methodological concerns. The mechanism was long assumed to be direct modulation of neuronal excitability, but recent work suggests that much of the scalp-applied current is shunted through the cerebrospinal fluid and skull, with only a small fraction reaching the cortex. The current understanding is that tDCS effects are real but modest, depend critically on the behavioural context (tDCS enhances learning only when paired with active training), and are modulated by the individual's baseline cortical excitability and genetic factors (such as BDNF Val66Met polymorphism). The clinical potential — for stroke rehabilitation, depression, and cognitive enhancement — is actively investigated but unproven at scale.

BDNF, exercise, and epigenetic regulation

Brain-derived neurotrophic factor (BDNF) is the central molecular link between experience and plasticity. BDNF promotes neuronal survival, dendritic growth, synapse formation, and LTP. It is upregulated by exercise, environmental enrichment, learning, antidepressant treatment, and — in the developing brain — by the maternal and environmental factors that shape critical periods. The BDNF Val66Met polymorphism (a single nucleotide change producing a valine-to-methionine substitution at position 66) impairs the intracellular trafficking and activity-dependent release of BDNF. Carriers of the Met allele show reduced hippocampal volume, impaired episodic memory, and reduced responsiveness to the cognitive benefits of exercise and the therapeutic effects of antidepressants.

Exercise robustly increases hippocampal BDNF and, in rodents, increases adult neurogenesis and improves spatial learning. In humans, aerobic exercise increases hippocampal volume and improves memory performance, effects that are mediated at least partly by BDNF. The exercise-induced BDNF increase is one of the strongest and most reproducible links between a modifiable lifestyle factor and structural brain plasticity, and it underlies the recommendation of exercise for cognitive health and as an adjunct treatment for depression.

The regulation of BDNF by experience is partly epigenetic. Environmental enrichment and exercise alter the methylation state of the BDNF gene promoter, changing its accessibility to transcription factors. Chronic stress and early adversity produce the opposite epigenetic changes, suppressing BDNF expression and impairing plasticity. Meaney and Szyf's work on maternal care and DNA methylation (covered in 29.02.01) demonstrated that early experience produces persistent epigenetic marks on stress-related genes; analogous mechanisms operate on BDNF, linking experience to the molecular capacity for plasticity across the lifespan.

Mindfulness meditation and cortical thickness

Long-term mindfulness meditation practice is associated with measurable changes in brain structure. MRI studies by Lazar, Holzel, and colleagues found increased cortical thickness in the prefrontal cortex and the insula in experienced meditators, regions involved in attention, interoception, and emotional regulation. Eight-week mindfulness-based stress reduction (MBSR) programs produce detectable changes in grey matter density in the hippocampus, posterior cingulate cortex, and temporoparietal junction.

The interpretation of these findings requires caution. Meditation-associated structural differences could reflect self-selection (people with certain brain structure may be drawn to meditation), and the effect sizes are small. But the direction of the changes is consistent with the hypothesis that sustained attentional practice drives experience-dependent plasticity in the circuits it engages. The findings are a contemporary example of the same principle Hubel and Wiesel established: the brain reorganizes in response to what it repeatedly does.

London taxi drivers and hippocampal volume

One of the most striking demonstrations of experience-dependent structural plasticity in humans is Eleanor Maguire's study of London taxi drivers (2000). To obtain a licence, London taxi drivers must pass "The Knowledge" — a gruelling examination requiring memorization of the layout of over 25,000 streets and thousands of landmarks within a six-mile radius of Charing Cross, typically taking 3-4 years of study. Maguire used MRI to compare the hippocampi of licensed taxi drivers with age-matched controls and found that the taxi drivers had significantly larger posterior hippocampi, a region critical for spatial memory. The volume of the posterior hippocampus correlated with years of taxi-driving experience. A follow-up study (Woollett and Maguire, 2011) confirmed that successful acquisition of The Knowledge was accompanied by measurable growth in grey matter in the posterior hippocampus — a prospective demonstration of structural plasticity in the adult human brain driven by a specific, intensive learning experience.

The taxi-driver studies are among the strongest evidence that adult human brain structure is shaped by experience, and they provide a human parallel to the enrichment-induced hippocampal plasticity demonstrated in rodents. They also illustrate the reciprocal nature of plasticity: the taxi drivers showed enlargement of the posterior hippocampus and a corresponding reduction in the anterior hippocampus, suggesting that the total hippocampal volume is roughly conserved while its internal organization is reshaped by use.

Connections Master

Neuroscience: brain and behaviour 29.02.01. The prerequisite unit introduced neuroplasticity at a survey level. This unit formalizes the molecular machinery (NMDA-dependent LTP, critical-period biology, adult neurogenesis) that 29.02.01 described qualitatively. The Hebbian principle, the action-potential biophysics, and the synaptic physiology on which LTP depends are all treated in the prerequisite.
Brain regions and function 29.02.02 pending. The plasticity mechanisms treated here operate on the anatomical structures catalogued there. Cortical reorganization after stroke reshapes the motor and somatosensory maps of 29.02.02. Hippocampal neurogenesis and the London taxi-driver findings are plasticity of the hippocampus, whose anatomy and role in memory were established in the regions unit.
Neurotransmitter systems 29.02.03 pending. LTP and LTD are glutamatergic phenomena — they depend on the NMDA and AMPA receptors that are the molecular substrate of excitatory transmission. GABAergic inhibition gates the NMDA receptor and closes critical periods. Dopamine and acetylcholine modulate which synapses are potentiated. The two units are most naturally read as a pair, with 29.02.03 supplying the receptor pharmacology and this unit supplying the plasticity it produces.
Learning and memory 29.04.01. LTP is the leading candidate cellular mechanism for the psychological phenomena of learning and memory treated there. The double dissociation between early-phase and late-phase LTP maps onto the distinction between short-term and long-term memory. The memory unit builds the systems-level theory (working memory, episodic memory, procedural memory) on this synaptic foundation.
Developmental psychology 29.03.01. Critical periods are the developmental-biology counterpart to the attachment and cognitive-development milestones treated there. The language critical period links directly to language acquisition; the visual critical period links to the development of perception.
Psychological disorders 29.09.01. Depression is correlated with reduced hippocampal volume and neurogenesis; the BDNF hypothesis links stress, depression, and plasticity. Addiction is maladaptive plasticity in reward circuits. Schizophrenia is associated with abnormal synaptic pruning during adolescence. The disorders unit reads these phenomena through the clinical lens.
Therapy and treatment 29.10.01. Stroke rehabilitation, constraint-induced therapy, tDCS, and the pharmacological manipulation of critical periods are the clinical applications of the mechanisms treated here. Antidepressant action may depend on BDNF-mediated neurogenesis. These are applied neuroplasticity.
Biology: cellular electrophysiology 17.09.01. The action potentials, voltage-gated channels, calcium signalling, and receptor biophysics underlying the NMDA-LTP cascade are treated in the biology strand. Two-photon imaging and optogenetics, the causal and measurement tools of this unit, are built on that biophysics.

Historical and philosophical context Master

Ramón y Cajal and the fixed-brain dogma

Santiago Ramón y Cajal, using the Golgi stain, established the neuron doctrine — the principle that the nervous system is composed of discrete cells (neurons) communicating at specialized junctions (synapses), rather than being a continuous reticulum. For this he shared the 1906 Nobel Prize. But Cajal also proclaimed that the brain's circuitry, once laid down in development, was immutable. His 1913 declaration — that "in the adult centers, the nerve paths are something fixed, ended, and immutable. Everything may die, nothing may be regenerated" — set the dogma that dominated neuroscience for most of the twentieth century.

The dogma was not irrational. Unlike peripheral nerves, which regenerate robustly after injury, central nervous system axons in mammals do not regrow, and no one had observed new neurons forming in the adult brain. The absence of evidence was interpreted as evidence of absence. The dogma began to crack in the 1960s and 1970s, with Joseph Altman's reports of new neurons in the adult rat hippocampus (largely dismissed at the time), was further challenged by Fernando Nottebohm's demonstration of seasonal neurogenesis in the adult songbird brain (1980s), and was definitively overturned for the human brain by Eriksson's 1998 BrdU study. The history is a case study in how a reasonable inference hardens into a dogma that then resists contrary evidence for decades.

Hebb's postulate and the cell assembly

Donald Hebb's 1949 book The Organization of Behavior proposed what would become the foundational principle of plasticity theory. Hebb, a Canadian psychologist working at McGill, was trying to explain how associations are learned and represented in the brain. His postulate — that correlated firing strengthens the connection between neurons — was a theoretical conjecture made two decades before the molecular machinery was discovered. Hebb went further: he proposed that learning forms cell assemblies — recurrent networks of neurons that, once strengthened by correlated firing, could sustain their own activity and thus represent a concept or memory. The cell assembly was the first computational theory of memory, and it anticipated modern ideas of attractor networks and Hebbian learning in artificial neural networks.

Hebb's influence extends beyond neuroscience. The backpropagation algorithm and the learning rules of modern deep learning are distant descendants of the Hebbian principle, though they use different update rules (error-driven rather than purely correlation-driven). The phrase "neurons that fire together, wire together" — a compact restatement of Hebb's postulate, coined by Carla Shatz and popularized by Löwel and Singer (1992) — has become one of the most widely known ideas in neuroscience.

Bliss and Lomo: the discovery of LTP

Timothy Bliss and Terje Lomo's 1973 experiment in the rabbit hippocampus was a watershed. By delivering high-frequency stimulation and observing persistent synaptic enhancement, they provided the first robust demonstration that synaptic strength could be persistently increased by neural activity — a phenomenon that had the properties (input specificity, associativity, persistence) demanded of a memory mechanism. The discovery launched a field: over the following five decades, the mechanisms of LTP were dissected in molecular detail, the conditions for its induction and expression were mapped, and its relationship to behaviour was probed through genetic and pharmacological manipulations.

The link between LTP and memory, while strongly supported by convergent evidence, has never been definitively "proven" in the sense that one could point to a single experiment showing that LTP is memory. The evidence is inferential and cumulative: LTP occurs in brain regions (hippocampus, amygdala, cortex) implicated in memory; manipulations that block LTP impair learning; manipulations that enhance LTP enhance learning; and the same molecular components (NMDA receptors, CaMKII, CREB) are required for both. A small but persistent community of sceptics has questioned whether the LTP induced in slices is the same phenomenon that stores memory in the living brain, and the field has taken these challenges seriously, refining its methods and claims. The honest position is that LTP is the best-characterized candidate mechanism for memory, and no alternative has comparable empirical support — but the inferential leap from synaptic physiology to experienced memory remains a gap, not a bridge.

Hubel and Wiesel: critical periods and the Nobel Prize

David Hubel and Torsten Wiesel's collaboration, beginning at Johns Hopkins in 1958 and continuing at Harvard, produced the foundational work on the organization of the visual cortex and the critical period for its development. Their single-unit recordings in the cat and monkey visual cortex revealed that neurons are organized into orientation columns and ocular dominance columns — that the cortex has a systematic, map-like architecture. Their monocular deprivation experiments showed that this architecture is profoundly shaped by visual experience during a critical period and that the same experience has little effect in adults.

Hubel and Wiesel shared the 1981 Nobel Prize in Physiology or Medicine with Roger Sperry (for his split-brain work). The critical-period concept that emerged from their work has become central to developmental neuroscience, pediatrics (the treatment of congenital cataracts and amblyopia), and education (the timing of language and music instruction). The finding that the brain's architecture is shaped by early experience, within windows that later close, reframed the nature-nurture debate: the brain is neither fully specified by genes nor indefinitely plastic, but is specified by genes through experience-dependent mechanisms that operate during defined developmental windows.

The neurogenesis controversy

The claim that new neurons are born in the adult human brain was met with deep scepticism when it was first advanced. The 2018 exchange between Sorrells et al. (who reported that human hippocampal neurogenesis drops to undetectable levels after childhood) and Boldrini et al. (who reported that neurogenesis persists into old age) brought the controversy into public view. The disagreement is not settled, and it reflects a deeper methodological challenge: the reliable detection of newborn neurons in postmortem human tissue depends on the preservation of fragile markers (DCX, PSA-NCAM, NeuroD) that degrade rapidly after death. Different laboratories use different fixation protocols, antibodies, and counting methods, and the results are exquisitely sensitive to these choices.

The controversy has a philosophical dimension. The appeal of adult neurogenesis — the idea that the brain can literally renew itself, that we are not stuck with the neurons we are born with — is powerful, and it has been enthusiastically embraced by the popular press and the wellness industry. The actual evidence, in humans, is uncertain. The scientific task is to hold the appealing story and the uncertain evidence in the right relationship: to follow the evidence wherever it leads, without letting the desirability of the conclusion determine its acceptance.

Bibliography Master

Hebb, D. O., The Organization of Behavior: A Neuropsychological Theory (Wiley, 1949). The founding statement of the Hebbian postulate and the cell-assembly theory of memory.
Bliss, T. V. P. and Lomo, T., "Long-Lasting Potentiation of Synaptic Transmission in the Dentate Area of the Anaesthetized Rabbit Following Stimulation of the Perforant Path," Journal of Physiology 232 (1973), 331-356. The discovery of long-term potentiation.
Hubel, D. H. and Wiesel, T. N., "The Period of Susceptibility to the Physiological Effects of Unilateral Eye Closure in Kittens," Journal of Physiology 206 (1970), 419-436. The classic demonstration of critical-period ocular dominance plasticity.
Bienenstock, E. L., Cooper, L. N., and Munro, P. W., "Theory for the Development of Neuron Selectivity: Orientation Specificity and Binocular Interaction in Visual Cortex," Journal of Neuroscience 2 (1982), 32-48. The BCM sliding-threshold theory of metaplasticity.
Bi, G.-Q. and Poo, M.-M., "Synaptic Modifications in Cultured Hippocampal Neurons: Dependence on Spike Timing, Synaptic Strength, and Postsynaptic Cell Type," Journal of Neuroscience 18 (1998), 10464-10472. The characterization of spike-timing-dependent plasticity.
Eriksson, P. S., Perfilieva, E., Bjork-Eriksson, T., Alborn, A.-M., Nordborg, C., Peterson, D. A., and Gage, F. H., "Neurogenesis in the Adult Human Hippocampus," Nature Medicine 4 (1998), 1313-1317. The BrdU study demonstrating adult neurogenesis in the human dentate gyrus.
Sorrells, S. F., Paredes, M. F., Cebrian-Silla, A., Sandoval, K., Qi, D., Kelley, K. W., James, D., Mayer, S., Chang, J., Auguste, K. I., Chang, E. F., Gutierrez, A. J., Kriegstein, A. R., Mathern, G. W., Oldham, M. C., Huang, E. J., Garcia-Verdugo, J. M., Yang, Z., and Alvarez-Buylla, A., "Human Hippocampal Neurogenesis Drops Sharply in Children to Undetectable Levels in Adults," Nature 555 (2018), 377-381. The study questioning adult human neurogenesis.
Turrigiano, G. G., Leslie, K. R., Desai, N. S., Rutherford, L. C., Nelson, S. B., and Turkis, L. F., "Activity-Dependent Scaling of Quantal Amplitude in Neocortical Neurons," Nature 391 (1998), 892-896. The discovery of homeostatic synaptic scaling.
Taub, E., Uswatte, G., and Pidikiti, R., "Constraint-Induced Movement Therapy: A New Family of Techniques with Broad Application to Physical Rehabilitation — A Clinical Review," Journal of Rehabilitation Research and Development 36 (1999), 237-251. Constraint-induced movement therapy for stroke recovery.
Maguire, E. A., Gadian, D. G., Johnsrude, I. S., Good, C. D., Ashburner, J., Frackowiak, R. S. J., and Frith, C. D., "Navigation-Related Structural Change in the Hippocampi of Taxi Drivers," Proceedings of the National Academy of Sciences 97 (2000), 4398-4403. The London taxi driver study of experience-dependent hippocampal plasticity.
Ramachandran, V. S. and Hirstein, W., "The Perception of Phantom Limbs. The D. O. Hebb Memorial Lecture," Brain 121 (1998), 1603-1630. Phantom limb phenomena and cortical remapping.
Pizzorusso, T., Medini, P., Berardi, N., Chierzi, S., Fawcett, J. W., and Maffei, L., "Reactivation of Ocular Dominance Plasticity in the Adult Visual Cortex," Science 298 (2002), 1248-1251. Chondroitinase ABC digestion of perineuronal nets reopens critical-period plasticity.
Kandel, E. R., Schwartz, J. H., Jessell, T. M., Siegelbaum, S. A., and Hudspeth, A. J., Principles of Neural Science, 6th ed. (McGraw-Hill, 2021). Ch. 65-67 cover synaptic plasticity, learning, and recovery from injury in depth.
Myers, D. G. and DeWall, C. N., Psychology, 13th ed. (Worth, 2021). Ch. 2 provides the introductory treatment of brain plasticity and neurogenesis.
Bear, M. F., Connors, B. W., and Paradiso, M. A., Neuroscience: Exploring the Brain, 4th ed. (Wolters Kluwer, 2016). Ch. 23-25 cover neural plasticity, learning, memory, and development.

Prerequisites

29.02.01 pending

Tier anchors

beginner: Myers and DeWall, Psychology (13e), Ch. 2
intermediate: Kandel et al., Principles of Neural Science (6e), Ch. 65-67
master: Bliss and Lomo — Long-lasting potentiation of synaptic transmission (1973)

References

Myers, D. G. & DeWall, C. N. — Psychology, 13th ed. (Worth, 2021) · Ch. 2 Brain plasticity
Kandel, E. R. et al. — Principles of Neural Science, 6th ed. (McGraw-Hill, 2021) · Ch. 65-67 Plasticity and recovery
Bliss, T. V. P. & Lomo, T. — Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit · J. Physiol. 232 (1973) 331-356; LTP discovery
Hubel, D. H. & Wiesel, T. N. — The period of susceptibility to the physiological effects of unilateral eye closure in kittens · J. Physiol. 206 (1970) 419-436; critical periods

Estimated time

beginner: 15m
intermediate: 30m
master: 45m