What the data actually shows
The strongest evidence for an age effect comes from language. The idea of a 'critical period' for language traces to Eric Lenneberg's work in the 1960s, and later research refined it into a 'sensitive period' — a stretch, mostly in childhood, when language is acquired with unusual ease. Accent shows the steepest age effect of all: acquiring a fully native-like accent in a new language after childhood is uncommon, even for people who become highly fluent.
A large-scale study by Hartshorne, Tenenbaum and Pinker (2018) used data from hundreds of thousands of English speakers to estimate when the ability to reach native-like grammar declines. Their analysis suggested learners can still reach very high proficiency if they start in childhood, with a notable drop in the ceiling for native-like mastery around the late teens. Crucially, this is about reaching a native-like ceiling — adults can and do become highly proficient; what gets harder is the last increment to indistinguishable-from-native.
Outside language, the evidence for hard age windows is much weaker, and adults often have advantages. Adult learners typically acquire the early stages of a new skill faster than children because they can use existing knowledge, follow instructions, and apply learning strategies. Children's apparent edge is often a matter of time and immersion rather than a closing biological window. For most adult goals, the data does not support the idea that you have missed your chance.
Why this feels different from how it actually is
The 'best age' idea feels true partly because the most famous example — children soaking up language and accent effortlessly — is genuinely real and very visible. We generalise from that vivid case to all learning, even though language acquisition is unusual in how strongly it depends on early exposure. Most skills don't work like accent.
It also feels true because of comparison and self-consciousness. Adults learning something new are often surrounded by people who started young, learn in public, and judge their early clumsiness harshly — so the experience feels like fighting biology when it is often just the normal awkward beginning that children get to have privately and without embarrassment.
And adults usually have less time, not less capacity. A child learning piano may practise daily for years inside a structured routine; an adult squeezes in occasional sessions around work. When the adult progresses more slowly, it is easy to attribute it to age when the real difference is hours of focused practice — the variable that actually drives most skill.
'Too old to learn this' is true for very few skills — the age you start matters far less than how, how much, and how deliberately you practise.
What the research says to do about it
Match your expectations to the specific skill. If your goal is a native-like accent in a new language, the evidence says that particular bar is genuinely harder after childhood — so aim for clear, confident, highly intelligible speech rather than indistinguishable-from-native, which is an unusual outcome at any adult age. For high fluency short of native accent, age is a much smaller obstacle.
For most other skills, treat age as roughly irrelevant and focus on the variables that actually drive learning: the quality and quantity of practice, useful feedback, and prior knowledge you can build on. Adults can deliberately exploit their advantages — strategy, self-direction, the ability to connect new material to what they already know — to learn efficiently.
Lower the cost of starting. Because much of the adult disadvantage is self-consciousness and limited time rather than capacity, the practical moves are to practise where early clumsiness feels safe, to protect regular practice slots, and to expect a slow, normal beginning. The research consistently points to consistent practice over starting age as the thing that determines how far you get.
What the research says does not help
Deciding you are 'too old to learn this' is, for most skills, an inaccurate belief that becomes self-fulfilling by stopping you before you start. The evidence of hard age windows is concentrated in a few domains — chiefly native-like language and accent — not in the broad swath of skills people assume are closed to them.
Trying to 'learn like a child' — through pure immersion with no strategy — usually wastes the adult's real advantage. Children rely on immersion because they have little else; adults learn faster when they use explicit instruction, structure, and their existing knowledge rather than abandoning those tools to imitate a child's approach.
Chasing a native-like accent as the measure of success in a new language sets up a goal that even highly fluent adult learners rarely reach, and that the research suggests is unusually age-dependent. Treating it as the bar can make genuine, impressive progress feel like failure when it is nothing of the kind.
Adults usually have less time, not less capacity. The real difference is often hours of focused practice, not age.
What this looks like in real life
Starting an instrument at 40
Taking up piano in mid-life feels like fighting biology, but there's little evidence of a closing window for music. The slower progress usually comes from having less time and more self-consciousness than a child practising daily in a structured routine — not from a lost capacity. Protecting regular practice slots and expecting a normal, awkward beginning does more than worrying about age.
Learning a language as an adult
An adult learner can become highly fluent — reading, working, holding real conversations — while never sounding indistinguishable from a native speaker. That last increment, especially a native-like accent, is the one thing the sensitive-period research says is genuinely age-dependent. Aiming for clear, confident, highly intelligible speech turns impressive progress into success rather than a felt failure against an unusual bar.
Real numbers in context
The clearest age effect is specific to language. Research on the sensitive period, including the large Hartshorne, Tenenbaum and Pinker (2018) analysis, suggests the ceiling for reaching native-like grammar starts to drop notably around the late teens, while accent shows an even earlier and steeper decline. But these findings are about reaching a native-like level — adults routinely become highly proficient; what gets rarer is the final step to indistinguishable-from-native.
For most non-language skills, there is no comparable evidence of a closing window, and adults often learn the early stages faster than children by leaning on prior knowledge and strategy. The honest summary is that 'best age' is a real but narrow phenomenon: meaningful for native-like language and accent, largely overstated everywhere else. Starting age usually matters far less than how much and how well you practise.