it's not just the human brain. certain g-loads at certain vectors can cause a situation where 2 different scenarios are indistinguishable without outside reference. Illustrated most famously in Einstein's elevator thought experiments.
Could you explain what you mean? My understanding was that Attitude indicators solved this problem using a gyroscope a long time ago. Maybe this falls into what you meant by an "outside reference", but it seems to me autopilots would clearly take these measurements into account.