What Does It Mean to Learn?
Exploration · OCT 3, 2026 · 5 MIN READ
I thought I knew what it meant to learn something. Then I tried to define it.
The Question
I have been on a quest to understand things more deeply, and machine learning is a topic that I have really been wanting to dive into. So naturally, I asked myself: what is machine learning?
I guess the obvious answer would be that it's when a machine learns something. But what does it mean for a machine to learn? Humans learn. Animals learn. But computers aren't conscious beings, so how would they learn anything?
Then I realized I was already getting ahead of myself. Before I could understand what it means for a machine to learn, I needed to answer a more basic question: what exactly does it mean to learn something?
My First Definition
I started by asking myself what learning even means. If someone came up to me and asked me to define it, I would probably say that learning is the process of understanding something and being able to apply it.
But then I started thinking about all the different ways someone can learn something.
Take learning how to tie your shoes. You don't need to become an expert on knots before you can tie a shoe. You can watch someone else do it, try it yourself, fail a few times, and eventually figure it out.
Or someone could simply tell you that the capital of France is Paris. If I asked you 20 minutes later for the capital of France and you remembered Paris, then you learned something. There isn't necessarily some deep understanding involved. You learned a fact and remembered it.
Thinking about examples like these made me realize that my original definition was flawed. Maybe understanding doesn't always have to be involved in learning something.
The Pianist Problem
Then I thought about a pianist who has never opened a music theory book or received any formal instruction but has spent years playing the piano. They can improvise, recognize patterns, anticipate where music is going, and play beautifully.
This pianist didn't sit around trying to understand music theory or learn how to read sheet music. They just played the piano and figured things out along the way. They clearly learned something, even if they can't necessarily explain everything they know.
That made me wonder whether there was a name for this kind of learning.
I came across the term implicit learning, which, put simply, is learning something without consciously trying to learn it or necessarily being able to explain exactly what you learned.
That eventually led me to a paper by Arthur Reber titled Implicit Learning of Artificial Grammars. Reber wanted to investigate whether people could learn patterns without consciously knowing the rules behind those patterns. He created an artificial grammar, basically a made-up set of rules determining which sequences of letters were valid.
To make a long story short, the experiment provided evidence that people could become sensitive to patterns through experience and apply what they had learned to new situations without necessarily being able to explain how they arrived at their conclusions.
Maybe Learning Is About Change
After reading about Reber's experiment, I started thinking that learning doesn't necessarily require someone to consciously understand something.
Maybe learning is really about change.
At the end of the day, learning seems to start with some kind of input. That input could be information, experience, observation, practice, or something else entirely. Then some process happens after receiving that input, and the result is that something about you has changed. You go from a state of not knowing or not being able to do something to knowing something or being able to do something you couldn't do before.
That change could be knowledge, a skill, an ability, intuition, or something else.
You can learn about history by reading about it. You can get better at piano by practicing. You can get better at solving math problems by doing more math problems. The process is different in each case, but you still come out of it knowing or being able to do something you couldn't before.
Different inputs. Different processes. Different results. But in each case, you come out on the other side changed in some way because of the experience.
What I Think Learning Means Now
So now I have a different definition of learning.
This isn't something I'm presenting as the definition of learning. It's simply the conclusion I've come to while thinking through all of this:
Learning is a process through which experience or information produces a relatively lasting change in what someone knows or is capable of doing.
I say relatively lasting for a reason.
You could learn how to tie your shoes when you're young and remember how to do it for the rest of your life. But maybe you learned calculus in college and were pretty good at it. If I came back ten years later and asked you to solve a limit problem, you might have no idea where to start. You may need to refresh your memory, work through a few problems, and get comfortable with it again.
Does that mean you never learned calculus? I don't think so. You can forget something you learned, but that doesn't change the fact you once learned it.
Maybe that's where the distinction between learning and memory starts to matter. But that's another question, and not the one I'm trying to answer here.
For now, I think the important part of my definition is change. There are countless ways to learn something, and learning doesn't always have to involve sitting down, studying, and consciously trying to understand every detail.
If an experience leaves you knowing something you didn't know before, understanding something differently, or capable of doing something you couldn't do before, then I think it's fair to say that you've learned something.
> What I'm Still Thinking About
I started thinking about all of this because I wanted to understand machine learning. If machines don't learn the same way humans do, how does a machine actually learn?
Published Oct 3, 2026 · 5 min read