← Concept Vault
Concept · Neuroscience · Maps to the Law of Gestation

Long-Term Potentiation

The synapse is where the past edits the future. LTP is the edit.
Read · 22 minPillar · Neuroscience

Long-Term Potentiation (LTP) is the closest thing neuroscience has to a physical mechanism for memory. When you learn something and it sticks, the leading explanation is that specific synapses — the junctions where one neuron speaks to the next — got physically stronger, and stayed stronger. That strengthening is LTP. It is among the most-studied phenomena in the cellular neuroscience of learning, the workhorse model behind half a century of memory research, and — this is the part most popular accounts skip — its existence is not the same thing as proof that LTP is memory. Hold that tension the whole way through. LTP is the best mechanistic candidate we have, it is real, it is measurable, and the link from "this synapse got stronger in a dish" to "you remember your tenth birthday" is still partly inferred rather than fully demonstrated. The work is respecting both the power and the gap.

Let's pull it all the way down — from a single calcium ion to a behaving animal — and then say honestly where the story is solid and where it's still load-bearing assumption.

01

What it is

LTP is a long-lasting increase in the strength of a synapse following a brief, high-frequency burst of activity. Strip it to its bones: if neuron A fires onto neuron B repeatedly and intensely in a short window, the connection from A to B becomes more efficient — A's future signals will produce a bigger response in B — and that change persists for hours, days, in some preparations weeks.

"Strength" here has a precise electrophysiological meaning. When neuron A releases neurotransmitter, it produces a small voltage change in neuron B called an excitatory postsynaptic potential (EPSP). LTP means that the same input now produces a larger EPSP. The synapse has been turned up. Crucially, it stays turned up after the triggering stimulus is gone. That persistence is the whole point — a memory that fades in seconds is not a memory. The "long-term" in the name is the load-bearing word.

Two properties make LTP a credible memory mechanism rather than just a curiosity:

  • Input-specificity. Only the synapses that were active during the triggering event get strengthened. The neighboring synapses on the same neuron, if they were quiet, stay unchanged. This is essential — if learning one thing strengthened every connection a neuron had, memory would be useless mush. Specificity means a single neuron can store information in a pattern of which of its thousands of synapses are potentiated.
  • Associativity. A weak input that would not normally trigger LTP on its own can be potentiated if it's active at the same time as a strong input onto the same neuron. Fire weak-and-strong together and the weak one gets strengthened too. This is the cellular face of association — the mechanism by which "things that happen together get linked together." It is, almost literally, the synapse learning a correlation.

These two properties are not incidental. They are exactly the properties a learning device would need. That's a large part of why LTP captured the field.

02

What it does

At the level of the whole brain, LTP is the proposed substrate for encoding — the laying-down of new information into the nervous system's physical structure. The frame that dominates modern neuroscience is that memory is stored in the pattern and strength of synaptic connections. You don't have a "file" of your grandmother somewhere; you have a distributed web of neurons whose connection strengths, sculpted by experience, reconstruct "grandmother" when partially activated. Learning, on this view, is the rewiring of those strengths. LTP is the leading candidate for how a connection gets stronger; its counterpart LTD (Long-Term Depression, below) is how a connection gets weaker. Together they give the brain a bidirectional knob on every synapse — turn up, turn down — and that bidirectional control is what makes a network capable of storing and updating information rather than just accumulating it.

The functional consequences cascade outward:

  • It makes co-activation durable. Whatever neurons happened to fire together during an experience leave behind strengthened links, so that re-activating part of the pattern tends to pull up the rest. This is the cellular logic of recall by cue: a smell brings back the whole afternoon because the smell-neurons and afternoon-neurons were potentiated into one another.
  • It enables prediction. Once correlations are wired in, the network anticipates. A wired-in association is a bet about what tends to follow what — the physical residue of the world's regularities.
  • It is metabolically expensive and therefore selective. Strengthening a synapse permanently requires building new proteins (more below). The brain doesn't do this for everything — it gates LTP behind thresholds of activity and attention, which is one reason emotionally salient or repeated events stick and idle background noise doesn't.

The honest framing: LTP is a mechanism for changing connection strength. The claim that this is how the brain learns and remembers is extremely well-supported as a correlation and a plausibility argument, and partially demonstrated causally — but the field still distinguishes "LTP is a learning mechanism" (strong) from "this specific memory lives in these specific potentiated synapses" (much harder, and only recently being shown).

03

The mechanism

This is where LTP earns its place as the most beautiful piece of machinery in cellular neuroscience, because the molecular story is unusually complete. We'll walk the canonical version — LTP at the Schaffer collateral synapses of the hippocampus (CA3 neurons onto CA1 neurons), which is the most-studied case by an enormous margin. Other forms exist with different mechanics, and that caveat matters later.

The coincidence detector: the NMDA receptor

The pivotal molecule is the NMDA receptor (N-methyl-D-aspartate receptor), a glutamate receptor sitting on the postsynaptic membrane. Glutamate is the brain's main excitatory neurotransmitter, and most fast signaling runs through a different receptor — the AMPA receptor — which opens the moment glutamate binds and lets sodium in. The NMDA receptor is the clever one. It has a property that makes it function as a literal AND gate, a coincidence detector requiring two things to be true at once:

  1. Glutamate must be present (the presynaptic neuron must have fired and released it), AND
  2. The postsynaptic membrane must already be depolarized (the receiving neuron must itself be active).

The reason is physical and elegant. At rest, the NMDA receptor's channel is physically plugged by a magnesium ion (Mg²⁺) sitting in the pore. Glutamate can bind all it wants — the channel stays blocked by the magnesium, so nothing flows. Only when the postsynaptic cell is already depolarized does the voltage change shove the magnesium ion out of the pore. Now, and only now, with glutamate bound and the magnesium expelled, the channel opens.

So the NMDA receptor fires only when the sender is talking AND the receiver is already listening loudly. That is coincidence detection built out of a magnesium plug and voltage. It is, mechanistically, how the synapse knows that two events were simultaneous — and simultaneity is the raw material of association.

Calcium: the trigger signal

When the NMDA channel finally opens, the critical thing that flows through it is calcium (Ca²⁺). This is the master signal. Calcium influx into the postsynaptic spine is the trigger that says "potentiate this synapse." And the amount of calcium matters enormously — this single fact is the hinge of the whole bidirectional system:

  • A large, fast rise in calcium → triggers LTP (strengthen).
  • A modest, sustained, lower rise in calcium → triggers LTD (weaken).

The same ion, same receptor family, different magnitude and timing, opposite outcome. The synapse reads the shape of the calcium signal as an instruction. This is one of the deepest and most established findings in the field — calcium is the second messenger that converts an electrical coincidence into a biochemical decision.

AMPA receptor trafficking: the expression mechanism

Here's the part that turns a signal into a stronger synapse. The large calcium rise activates a cascade of enzymes — most famously CaMKII (calcium/calmodulin-dependent protein kinase II), which many researchers consider the central "memory molecule" of early LTP. CaMKII has a remarkable feature: once activated by calcium, it can phosphorylate itself (autophosphorylation) and stay active after the calcium is gone. It is, in effect, a molecular switch that flips on and holds — a tiny piece of persistence built into the chemistry, which is exactly what a memory trace needs.

Activated CaMKII (and partner kinases like PKA, PKC) does two main things that make the synapse stronger:

  1. It phosphorylates existing AMPA receptors, making them conduct more current per channel.
  2. It drives more AMPA receptors into the synaptic membrane — pulled from intracellular pools and inserted into the postsynaptic density. This is AMPA receptor trafficking, and it is now considered the primary mechanism for expressing LTP. More AMPA receptors in the membrane means the same amount of glutamate produces a bigger response. The synapse is turned up because it now has more receivers.

This matters conceptually: the NMDA receptor triggers LTP, but it is the AMPA receptors that carry the strengthened signal afterward. Trigger and expression are different molecules. The NMDA receptor is the gatekeeper; the AMPA receptors are the volume knob it adjusts.

Early vs. late LTP: the protein-synthesis cliff

There are two phases, and the boundary between them is one of the most important facts in memory neuroscience.

  • Early LTP (E-LTP) lasts roughly one to three hours. It runs entirely on modifying proteins that already exist — phosphorylating receptors, trafficking AMPA receptors already made. No new genes, no new proteins. It's fast and it's temporary.
  • Late LTP (L-LTP) can last many hours to days, and it requires new protein synthesis and gene transcription. The signal travels to the nucleus, transcription factors like CREB (cAMP response element-binding protein) switch on memory-related genes, new proteins are built, and the synapse is structurally rebuilt — dendritic spines physically enlarge, and in some cases entirely new synapses grow.

This maps onto a behavioral fact you already know: there is a difference between a memory that survives an hour and a memory that survives a week, and that difference is protein synthesis. Block protein synthesis (with drugs like anisomycin) right after learning, and you can wipe out long-term memory while leaving short-term memory intact — which is exactly what you'd predict if late LTP, which needs new proteins, is the substrate of durable memory. This is one of the cleaner bridges between the molecular and behavioral levels.

Synaptic tagging: how the trace finds its home

A genuinely deep problem: protein synthesis for L-LTP happens partly in the cell body, but the new proteins need to reach the specific synapses that were active — not all synapses indiscriminately, or you'd destroy input-specificity. The synaptic tagging and capture hypothesis (Frey & Morris, 1997) solved this elegantly: a strongly activated synapse sets a temporary local "tag," and the newly synthesized proteins, shipped out broadly, are captured only by tagged synapses. This is why a weak event near a strong one in time can be consolidated (it sets a tag that catches the strong event's proteins) — a molecular explanation for why salient experiences can "rescue" nearby mundane ones into long-term memory.

That is the full arc, molecular to structural: glutamate + depolarization → magnesium leaves → NMDA opens → calcium floods → CaMKII flips on → AMPA receptors get added → (if strong enough) CREB turns on genes → new proteins → spines grow → the connection is permanently stronger. Coincidence in, structure out.

04

The landmark science

The intellectual scaffolding came first, decades before the biology.

Donald Hebb, 1949 — In The Organization of Behavior, the Canadian psychologist proposed what became the founding postulate of the field: "When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased." That mouthful got compressed by later writers into the slogan "cells that fire together, wire together." Worth flagging the provenance honestly: that exact catchy phrasing is not Hebb's — it's a paraphrase popularized later, most often attributed to neuroscientist Carla Shatz (who used "cells that fire together, wire together" in the 1990s as a memorable gloss on Hebbian plasticity in the developing visual system). Hebb gave the principle; Shatz gave the bumper sticker. The associativity property of LTP described above is the physical realization of Hebb's postulate — and LTP is often called the "Hebbian" mechanism for precisely this reason. (Strictly, classic NMDA-dependent LTP is Hebbian; some other plasticity forms are not, which is a caveat worth keeping.)

Terje Lømo, 1966 → Bliss & Lømo, 1973 — This is the discovery. Working in Per Andersen's lab in Oslo, the Norwegian researcher Terje Lømo first observed in 1966 that brief high-frequency stimulation of the perforant path produced a lasting increase in the response of dentate gyrus neurons in anesthetized rabbits. He teamed with Tim Bliss, and the founding paper — Bliss & Lømo, "Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path," Journal of Physiology 1973 — is the canonical founding document of LTP. (A companion 1973 J Physiol paper, on the unanaesthetized rabbit, was Bliss & Gardner-Medwin, not Bliss & Lømo — the two are often loosely lumped together, but only the anaesthetized paper is the Bliss & Lømo one.) They stimulated the perforant path (a major input to the hippocampal formation) with a brief high-frequency train, recorded the population response in the dentate gyrus, and showed it was potentiated for periods ranging from roughly half an hour to several hours. The phenomenon they named "long-lasting potentiation" later became long-term potentiation. The setting matters: rabbit hippocampus, 1973. That the effect showed up in the hippocampus — a structure already implicated in memory formation since the patient H.M. (Henry Molaison, whose 1953 bilateral medial temporal lobe resection — which removed large parts of the hippocampus and surrounding cortex — left him largely unable to form new long-term declarative memories) — is what made LTP electrifying. The cell type that learning seemed to require was now caught physically strengthening its connections. (One honest caveat: the dentate gyrus / perforant path is the original LTP site, but the molecular mechanism walked through above is the best-characterized at the CA3→CA1 Schaffer collateral synapse; the field generalizes across them, but the discovery synapse and the textbook-mechanism synapse aren't the same one.)

The 1980s — the NMDA receptor and calcium. The mechanism got filled in over the next two decades. Graham Collingridge and colleagues showed in 1983 that the drug AP5 (APV), an NMDA receptor antagonist, blocked the induction of LTP without blocking normal synaptic transmission — pinning LTP induction to the NMDA receptor specifically. The magnesium block voltage-dependence was worked out by Mayer, Westbrook, and Guthrie (1984) and Nowak et al. (1984), which is what revealed the NMDA receptor as a coincidence detector. Calcium's role as the trigger was nailed down through the same period, and CaMKII rose to prominence in the late 1980s and 1990s (work by Tom Soderling, Alcino Silva, and others) as the persistent switch of early LTP.

Richard Morris, 1986 — the first behavioral bridge. Morris and colleagues showed that infusing AP5 (the NMDA blocker that stops LTP) into rat brains impaired learning in the Morris water maze, a spatial memory task. This was the first influential demonstration that blocking the LTP mechanism blocks actual learning — the link the field had been chasing. It was suggestive, not airtight (AP5 has confounds — it can impair sensorimotor function and block normal transmission at high doses), and the debate over its cleanliness ran for years, but it moved LTP from "interesting cellular phenomenon" to "candidate memory mechanism."

"Doogie," 1999 — gain-of-function. Joe Tsien's lab engineered mice overexpressing the NR2B subunit of the NMDA receptor, enhancing LTP, and the mice learned faster and remembered better across multiple tasks. Named "Doogie" after the TV boy-genius, this was the converse experiment — enhance the mechanism, enhance the learning — and it strengthened the causal case from the other direction.

Optogenetics, 2014 — the closest thing to proof. This is the one that mattered most for the "is LTP memory?" question. Sadegh Nabavi, Roberto Malinow and colleagues (Nature, 2014) used optogenetics to create a fear memory in rats, then showed they could erase that memory by delivering an LTD protocol (weakening the relevant synapses) and reinstate it by delivering an LTP protocol (strengthening them again). Inactivate and reactivate a specific memory by running LTD and LTP on identified synapses. This is the most direct demonstration to date that potentiation of specific synapses is carrying a specific memory — not just correlated with one. It is rightly cited as a landmark, and it is also one experiment in one paradigm, which is the caveat that keeps the larger debate alive.

05

How it shapes experience

The molecular machinery has fingerprints all over ordinary life. A few sharp ones:

  • Why repetition works, and why cramming fails. Early LTP runs on existing proteins and decays in hours; late LTP needs new protein synthesis and consolidates over time, often during sleep. A single cramming session can drive early LTP — you "know it" that night — but without the spaced repetition that re-triggers the cascade and pushes synapses into the protein-synthesis-dependent late phase, the potentiation decays. Spaced repetition is not a study hack; it's an exploitation of the E-LTP/L-LTP boundary. Each spaced review catches the synapse and re-deepens the trace before it fades, and spacing gives consolidation time to build structure between sessions.
  • Why emotionally charged events sear in. Salient, arousing experiences flood the system with neuromodulators (noradrenaline, dopamine) that lower the threshold for LTP and bias synapses toward the protein-synthesis-dependent late phase. The synaptic tagging mechanism means the strong, emotional core of an event can drag the mundane surrounding details into long-term storage too. This is why you remember exactly where you were during a shock and not what you had for lunch last Tuesday.
  • Why association is automatic and sometimes unwanted. Associativity at the synapse means whatever co-occurs gets wired. This is the engine of useful learning and of trauma, phobia, and addiction. A neutral cue present during a high-arousal event (a song during a breakup, a place during a panic attack) gets potentiated into the emotional circuitry and later triggers the whole state. The mechanism doesn't care whether the association serves you — it wires correlations indiscriminately. This is also exactly why exposure therapy and extinction work with the same machinery: new learning installs competing connections (it doesn't erase the old ones, which is why relapse happens).
  • Why "neurons that fire together wire together" makes practice physical. Deliberate practice — running a skill correctly, repeatedly, with focus — is, mechanistically, the repeated co-activation that drives Hebbian potentiation in the relevant circuits. The felt sense of something going from effortful to automatic tracks the consolidation of those circuits from labile early plasticity into structural change. You are, quite literally, growing dendritic spines on the synapses that matter.
06

Where it's distorted or oversold

This is the honest cut, and it's necessary because LTP is one of the most over-claimed findings in neuroscience pop culture.

  • "LTP is memory" — stated as settled fact. It usually isn't presented with the caveat. The careful position, held by serious people in the field for forty years, is that LTP is a mechanism that plausibly underlies memory — strongly supported, partly demonstrated, not fully closed. The chain "LTP exists → LTP has memory-like properties → blocking LTP impairs learning → enhancing LTP improves learning → you can write and erase a memory with LTP/LTD protocols" is genuinely impressive, but each link has had its critics, and "this everyday memory you have lives in these particular potentiated synapses" remains demonstrated only in narrow, engineered cases. Treat "LTP is the memory mechanism" as the leading theory, not a closed case.
  • The artificial-stimulation problem. Classic LTP is induced by tetanic stimulation — an electrode delivering 100 pulses a second, a pattern that does not obviously occur during natural learning. Critics (notably for years) pointed out that this is a sledgehammer, and that whether the brain ever drives LTP this way during actual experience was an open question. The field's answer — theta-burst stimulation, patterns that mimic natural hippocampal theta rhythms, and recordings of plasticity during real behavior — has narrowed the gap considerably, but the foundational worry was legitimate: a phenomenon you produce with a wildly unphysiological stimulus may not be the phenomenon learning uses.
  • The "in a slice, anesthetized rabbit" gap. Bliss & Lømo's foundational work, and an enormous fraction of LTP research since, is done in brain slices or anesthetized animals, reading electrical traces — not in awake animals forming real memories. Strengthening at a synapse in a dish is real. The inference from there to the lived act of remembering crosses several levels of organization, and every one of those crossings is a place the story is partly inference.
  • Nootropic and "memory pill" hype. The Doogie mice and CaMKII work get spun into supplement marketing and brain-training claims promising you can "boost LTP" for better memory. The honest reality: enhancing LTP globally in animals sometimes helps and sometimes hurts (more potentiation is not strictly better — it can degrade the signal-to-noise that input-specificity depends on, and the NR2B-enhanced mice also showed increased sensitivity to chronic pain). There is no validated consumer intervention that selectively enhances useful LTP. The mechanism is real; the supplement aisle's claims about it are not.
  • "Fire together, wire together" misattributed and overextended. As noted, the slogan is Shatz's gloss, not Hebb's words, and it gets stretched into pop-psychology claims ("just visualize success and the neurons wire it in!") that ignore the requirement for actual co-activation, the thresholds, the protein synthesis, and the fact that imagined and executed actions don't produce identical activation. The principle is real and narrow; the self-help extension is loose.
07

The debates and frontier

  • Is the memory stored pre- or post-synaptically? A genuine, long-running technical war. Early LTP expression is now mostly attributed to postsynaptic AMPA receptor trafficking (the consensus), but a vocal camp argued for years that a presynaptic change — increased neurotransmitter release — also contributes. The likely answer is "both, depending on synapse type and phase," which is unsatisfying but probably true.
  • *CaMKII as the memory switch — or one of many?* CaMKII's self-sustaining activation made it the glamorous candidate for the molecular memory switch. But early LTP can sometimes persist even when CaMKII is later inhibited, raising the question of whether the switch, once flipped, hands off to other mechanisms. The molecular identity of the persistent trace is not fully settled.
  • The engram and the rise of cell-assembly memory. The frontier has partly moved from single synapses to engrams — the specific populations (cell assemblies) that encode a given memory. Susumu Tonegawa's lab (MIT, 2012 onward) used optogenetics to label and then artificially reactivate the specific neurons active during fear learning, triggering recall of the memory on demand — and even implanting false memories by mismatching context and shock. This reframes LTP's role: it may be the mechanism that binds an engram together, the glue rather than the whole picture. The 2014 Nabavi/Malinow erase-and-reinstate result sits in this lineage.
  • Memory without synaptic strength change? The most destabilizing frontier finding. Some experiments suggest a memory can survive even after the synaptic potentiation that supposedly stored it has been disrupted — implying the information might be held somewhere more durable (some have proposed intracellular or even molecular/structural stores), with synaptic strength as the readout rather than the storage itself. This is contested and far from established, but it's the live question that keeps "LTP IS memory" from being declared closed.
  • Does it hold in humans? Almost all of it is animal work. LTP-like plasticity has been demonstrated in human tissue and inferred non-invasively (e.g., via repetitive sensory or magnetic stimulation), but the direct, mechanistic human evidence is thin compared to the rodent literature. The extrapolation is reasonable — the molecules are conserved — but it is extrapolation.
08

How to work with it

You don't manipulate your own NMDA receptors directly, but the mechanism dictates what actually moves learning, and most "study tips" are downstream of it:

  1. Space your repetitions across the E-LTP/L-LTP boundary. Review before the trace fully decays, and leave time between sessions for protein-synthesis-dependent consolidation. Spacing isn't gentler cramming — it's the schedule that pushes synapses into the durable phase. This is the single highest-leverage move and it's pure mechanism.
  1. Protect sleep, especially after learning. Consolidation — the building of late-LTP structure and the hippocampal replay that distributes memories to cortex — is heavily sleep-dependent. A night of poor sleep after learning measurably weakens retention. You are not "resting"; you are running the construction crew.
  1. Engage emotion and attention deliberately. Salience lowers the LTP threshold and biases toward late LTP. Caring about what you're learning, connecting it to something that matters, or studying in a state of focused arousal (not anxiety, which is too high) recruits the neuromodulators that make synapses more plastic. Flat, disengaged repetition is mechanically weaker.
  1. *Practice the correct co-activation.* Hebbian wiring is indiscriminate — it strengthens whatever fires together, including your mistakes. Sloppy repetition potentiates sloppiness. Slow, accurate, focused practice wires the right pattern; this is the cellular argument for quality over volume in deliberate practice.
  1. Use retrieval, not just re-exposure. Pulling a memory out (testing yourself) reactivates and re-stabilizes the trace more powerfully than passively re-reading. Retrieval practice is among the best-validated learning techniques in psychology, and it's plausibly LTP-and-reconsolidation at work — every act of recall is a chance to re-strengthen.
  1. Respect that unwanted associations are wired the same way — and dissolved by competition, not deletion. You can't erase a potentiated fear or habit by willpower; you install competing learning (new connections that outcompete the old). This is why the path out of a bad association is repeated new experience, not avoidance — and why it takes time and relapses, because the original potentiation is still physically there underneath.
The Compression

Learning is not metaphorical rewiring — it is literal rewiring, one strengthened synapse at a time, triggered by coincidence, paid for in protein, and made permanent only if you repeat, sleep, and care. Connects to — the Law of Gestation. This is the natural pairing, and the reasoning is mechanistic, not poetic. The Law of Gestation says everything has an incubation period — a fixed interval between conception and birth, between the planting of a cause and the appearance of its result, during which the work is happening invisibly and cannot be rushed without aborting it. LTP is the cellular face of that law. The moment of learning (early LTP) is conception; durable memory (late LTP) is birth; and between them sits an obligatory gestation period of protein synthesis, gene transcription, structural spine growth, and sleep-dependent consolidation that you cannot skip. This is why cramming fails and why "I learned it but it didn't stick" is so common — people mistake conception for birth, the early labile trace for the consolidated one, and don't honor the incubation. The synapse needs time in the dark to build the structure, exactly as Gestation insists. Push for the result before the term is up and you get nothing born. Respect the interval — space the repetitions, protect the sleep — and the memory is delivered on schedule. Gestation tells you that there's an unskippable incubation; LTP tells you what is physically happening inside it. Related structures and concepts: Hebbian plasticity & the Hebb postulate (the founding rule LTP physically realizes) · Long-Term Depression / LTD (the weakening counterpart — the other direction of the same calcium-gated knob) · the engram & cell assemblies (Tonegawa — the population LTP may bind together) · synaptic consolidation & systems consolidation (hippocampus → cortex over time) · reconsolidation (memories become labile when recalled, re-stabilized via LTP-like processes) · neuroplasticity broadly (LTP is its most precise mechanism) · the hippocampus and patient H.M. (where the story began) · sleep and memory (where late LTP is built) · spaced repetition & retrieval practice (the behavioral exploits of the mechanism).

Your Practice
My Depth
My Notes
← Previous
The Action Potential & the Synapse
Next →
Synaptic Pruning