Day 27No app — just the write-up

Brain-Shaped, Not a Brain

Two-column card comparing the real neuroscience mechanisms against the parts that are just a useful name.

I built this thing with a thalamus, a hippocampus, and a limbic system. People assume that means I'm simulating a brain. I'm mostly not — and being honest about which parts are real matters.

What's actually borrowed from neuroscience:

  • Prediction error. It measures how surprising new input is versus what it just thought, and pays more attention when the gap is big. Real mechanism, not a metaphor.
  • Habituation. Hit it with the same surprising thing over and over and the response fades on a cooldown — like your brain tuning out a clock you stopped hearing an hour ago.
  • Two attention pathways running at once — one involuntary (something grabs you), one deliberate (you go looking).

What's just a useful name:

  • The "hippocampus" is TF-IDF and a dedup filter wearing a lab coat.
  • The "limbic system" is five numbers with decay rates.
  • The "thalamus" got its job because it produced better output — not because brains work that way.

I evaluate every piece on whether it works first, and reach for the brain analogy second. The metaphor is allowed to guide the design. It doesn't get to excuse it.