New AI Theory Explains Why Some Moments Feel Vivid and Others Blurry

A researcher named David Balduzzi posted a paper on arXiv, a public site where scientists share drafts before review. He doesn't ask whether AI is conscious. He asks a narrower question: could the way signals ripple through a network explain what experience feels like from the inside? To test it, he built a make-believe world called Gradland, populated by simple neural networks, where every rule is known and the math behaves.

His main tool is the gradient, which measures how much a tiny nudge in one part of a system changes another — like noticing that pressing one piano key makes a nearby string hum. He turns that into two scores: effective rank (roughly, how many separate "dials" are active at once) and cohesion (how tightly those dials move together). Together, they describe the shape of a moment.

Using those scores, the paper claims to account for a surprising list of everyday feelings: why an experience can stretch over hundreds of milliseconds instead of flashing by; why some things feel vivid and others hazy; why surfaces have texture; the "blooming, buzzing confusion" newborns likely experience; the difference between a clear idea and a muddled one; what learning feels like; and why rich, dense moments might be useful rather than wasteful.

The catch: Gradland is imaginary. No brains were scanned, no AI was asked how it feels, and the paper is a mathematical argument, not evidence. A high score on these measures does not mean a chatbot is having a bad day. Still, it's the kind of idea that could eventually be tested — and if it holds up, it gives researchers a practical ruler for something philosophers have argued about for centuries.