Now apply that thinking to human-based neural nets...