Why I'm writing this.

Every so often the same word gets used for a brain and for a machine, and nobody stops to ask why. We say a neural network "learns." We say it "pays attention." We say a person's memory is "corrupted," the same word we'd use for a bad file. Either that's a coincidence of language, or it's a clue that these two things — the wet kind of thinking and the silicon kind — are closer relatives than they look. I don't know which yet. That's the whole reason this exists.

Computation and cognition is the name for that question, dressed up as a field. On one side: how a machine takes information in, transforms it, and produces a decision. On the other: how a brain does something that looks a lot like the same thing, except nobody fully agrees on what "the same thing" even means here. Most people learn about one side or the other. Fewer people spend time in the gap between them, which is exactly where the interesting disagreements live.

I ended up caring about this less through theory and more through building. I've spent time writing code that has to make decisions — a camera that has to decide whether the thing in frame is a bird, a program that has to guess what someone means from a rough draft. Every time, I hit the same wall: the machine is doing something that resembles judgment, but it isn't judgment, and I couldn't articulate the difference cleanly. That gap between "resembles" and "is" turned out to be worth more of my attention than the code itself.

So this blog, Decoding Intelligence, is where I chase that gap in public. I'm going to read the papers, get things wrong, revise, and write it all down as it happens — not polished after the fact, not from a position of already knowing. Some posts will be closer to explainers, for anyone who wants a clear way into a confusing topic. Some will be closer to open questions I haven't resolved. Both are the point.

There's no syllabus for this and no finish line. I'll add posts as the thinking earns them, not on a schedule. Consider this the first entry in a conversation I plan on having with myself, in public, for a long time.

— Vedant