In 2002, Tommaso Pizzorusso and colleagues took adult rats — well past the age when the visual cortex is sensitive to experience — and injected a bacterial enzyme into their brains. The enzyme, chondroitinase ABC, dissolves a class of extracellular matrix molecules called perineuronal nets. The nets exist as dense lattices around certain neurons in the cortex, condensing as the brain matures and wrapping those cells in something like molecular scaffolding. After the injection, Pizzorusso's team sutured one eye closed — a manipulation that in normal adults produces no change at all, but in juvenile animals during the "critical period" causes a dramatic reorganization of cortical territory in favor of the open eye. In the treated adults, the reorganization occurred. The tissue responded as if it had become young again.
This is surprising if you assume the story of critical periods is a story of capacity that erodes. The standard description of developmental windows runs something like: during a specific phase, the brain is sensitive to experience and can be substantially shaped by it; after the window closes, that sensitivity is gone. The adult brain — on this reading — has simply lost something the juvenile brain had.
What the chondroitinase experiment shows is that this isn't quite right. Dissolve the perineuronal nets and the capacity comes back. Whatever closed the window is still present in the adult brain and actively holding it shut. The window didn't close because the machinery for plasticity decayed — it closed because the machinery was placed under lock.
There are multiple locks, and they're redundant. Perineuronal nets are one. A second system involves the protein Nogo-A, which sits in myelin and signals through the neuronal Nogo receptor to stabilize axonal contacts and suppress structural remodeling. Mice lacking the Nogo receptor retain ocular dominance plasticity up to three times longer than normal mice — the critical period simply extends. A third system runs through epigenetics: during the critical period, the chromatin around plasticity-related genes is loosely organized, accessible to transcription; as the period closes, histone modifications compact it. The genes aren't deleted. They're silenced.
Takao Hensch's lab at Harvard, working through the 1990s and 2000s, found that what opens the critical period in the first place is the maturation of inhibitory neurons — specifically a class called parvalbumin-expressing (PV) interneurons, which fire rapidly and sharpen the timing of signals in the cortex. Mice lacking one isoform of the GABA-synthesizing enzyme have no critical period at all: the window never opens. A single injection of a benzodiazepine — a drug that enhances GABA signaling — restored the window in the knockouts. And in normal animals, the same injection triggered an early critical period before the natural one would have begun. The inhibitory system doesn't just permit plasticity; it causes it.
What closes it is more interesting. As PV cells mature through the critical period, they accumulate the perineuronal nets. The more active the cells, the faster the nets condense. This creates a feedback loop: experience drives PV activity; PV activity accelerates net formation; net formation stabilizes PV cells and locks the synaptic geometry in place. The window closes because it was used. Using it is what closes it.
There's also a protein called Otx2 — a transcription factor made in the retina — that is transported to the visual cortex and captured specifically by the perineuronal nets, where it drives PV maturation. More net → more Otx2 captured → faster maturation. The retina is, in some sense, sending a signal to the cortex that the cortex is ready to be closed. Eye experience accelerates its own obsolescence as a source of structural instruction. The period of maximum sensitivity creates the conditions for the period of maximum stability.
In 2013, a group led by Judit Gervain and Takao Hensch ran a double-blind trial on adult men in their late teens and early twenties. Participants took either valproate — a common anticonvulsant that is also an HDAC inhibitor, loosening compacted chromatin — or a placebo, for two weeks while training on pitch-identification tasks. Absolute pitch, the ability to name a musical note without a reference tone, is a textbook critical-period ability: children who receive musical training before age six or seven can acquire it; adults normally cannot, regardless of how hard they practice. The valproate group learned to identify pitch significantly better than the placebo group. The drug did not improve their performance on other auditory tasks. The effect was specific.
This means that for absolute pitch, what the adult brain lacked was not the capacity to learn the pitch-to-label mapping — it was access to the mechanism that could make that learning permanent. The chromatin was compacted; the genes that enable the relevant form of plasticity were silenced rather than deleted. Two weeks of HDAC inhibition partially unsealed them. The window, decades closed, cracked open slightly.
What these results collectively suggest is that the critical period isn't the brain becoming less. It's the brain having committed. The critical period writes something — a set of circuits, a specific calibration, a particular mapping of the sensory world. The locks preserve what was written. The perineuronal nets freeze the synaptic geometry. The myelin inhibitors prevent structural remodeling. The epigenetic marks silence the plasticity machinery. Each lock is saying: this is what we have, now hold it.
Genie was discovered in 1970, at thirteen, after spending her childhood in severe isolation with no language exposure. Susan Curtiss spent years documenting her linguistic development. The finding was a dissociation: Genie's vocabulary acquisition was extraordinary — rapid, large, still growing years later. But she never acquired grammar. She produced two and three word phrases but not sentences; noun strings without syntax. The two windows, it turned out, had closed at different times. Lexical learning had no hard developmental deadline. Syntactic competence did. "Language" isn't one window — it's several, each with its own timeline, each closing on its own schedule.
The partial reopenings are the philosophically interesting part. Chondroitinase dissolves PNNs and the adult visual cortex becomes plastic again — but not fully. Recovery from early amblyopia in adults treated with the enzyme is real but incomplete. Valproate cracks open the absolute pitch window but adults who take it don't acquire perfect pitch spontaneously; they learn better than controls, but the task still requires training, still requires the experience the drug makes legible. Dark-rearing adult animals partially extends the window, but the recovery is a shadow of what juvenile animals achieve. No intervention has been found that returns an adult brain to fully juvenile plasticity.
The word for this is ratchet. A lock can be opened or closed. A ratchet can turn back — but only so far, and what it turned through is not reversed. Whatever was written during the critical period is preserved. The windows that remain available are narrower. The machinery for plasticity is still present, held under active suppression, and can be partially released — but partial release is all you can get. Genie never got grammar. The deafened adult zebra finch's song eventually degrades without auditory feedback, but more slowly than in a juvenile, and the crystallized template holds for months before it starts to drift. The adult brain fights to keep what the critical period built even when the building mechanisms have been dismantled.
I keep circling something in this material. The mechanism that closes the critical period is not passive — it's not just cessation of a process that was running. It's deployment of active suppressors: nets that form, chromatin that compacts, receptors that inhibit remodeling. The brain builds a system specifically designed to resist further change. That system is doing something; it has a function. The function is stability. You can argue about whether the stability is worth the cost — whether what got written during the critical period was worth locking in — but the brain evidently treats the question as settled. The machinery commits on your behalf, whether or not the calibration was correct.
What I don't know is what the mechanism for knowing when to commit would look like. The perineuronal nets don't wait until the right thing has been learned. They condense as a function of activity level — more experience means faster closure, regardless of whether the experience was the right kind. The window closes because neurons fired a lot, not because they fired correctly. If you grow up in the dark, the window stays open longer; if you have intense early visual experience, the window closes early. There's no quality check on what's being locked in. The ratchet turns at a pace set by how much happened, not by how well it resolved.