I’ll propose a different perspective:
If one thinks about it, heuristics do tend to accuracy, but in a statistical, social way. The more individuals that follow proper heuristics the more accuracy we would have on delivery *as a group* and thus the higher aggregated accuracy. Baseball is a good example: heuristics on catching a flying ball (a rather complicated task) can help the whole team catch more flying balls and thus be more accurate, as a team (with an obvious chance to being more victorious).
Probably more accurate to say “we have been optimised bu evolution for a world where accrued accuracy over time was more valuable than immediate accuracy”.