Pedro Domingos

But if you applied the brain’s physics inference engine to the full range of observations, it would blow Einstein away.

But only for the type of physical phenomena which were relevant for the evolution of brains. For other phenomena (at non-human scales, or for electromagnetism, molecular, atomic, quantum, ...) math is our only window to reality. Still created by human brains, though.

Our current math is about as sophisticated as retina + V1 + V2 (which is why that’s all we can - barely - understand). Your full visual cortex etc. would leave QFT in the dust if it was evolutionarily useful.

Yes, but it wasn't. And, nevertheless, we developed QFT, QED, QCD, ..., which are certainly not based on intuition. Other parts of our (collective) brain opened up windows to reality way beyond our perception and intuition.

My point is that our brains running on microscopic data would leave QFT in the dust.

Not *our* brains, which evolution hardwired to work at certain space-time scales and with certain innate physicas (e.g. locality, "objectiness", some space-time equivariance,...). For that to work, we would have to re-evolve different brains and (very) different senses.

Our brains are much more general than you assume. E.g., we easily understand scenes that are vastly different from anything we evolved with. (And QFT still assumes all of that.)

Scenes with vastly different contents, but with a common underlying physics. Shrink a human to 1mm and have him interact with water or wind, and he totally fails in doing anything, including moving.