Donald Hoffman and Chetan Prakash asked a precise question using the mathematics of evolutionary game theory: when a creature that perceives some part of the world accurately competes against a creature that perceives only what helps it survive, which one wins? They ran the payoff structures across a wide range of environments and the answer barely wavered. The truth-seeing strategy was driven to extinction almost every time.
The reason is economy. Seeing the world as it is takes in far more information than survival requires, and gathering, storing and processing that information is slow and costly. A creature that ignores the truth and tracks only fitness payoffs, what to eat, what to flee, what to mate with, spends less and acts faster. In a race measured entirely by offspring, cheap and fast beats accurate and complete.
It is worth being careful about the size of the claim. The theorem does not say our perceptions are random or useless. A good interface has to be reliably tied to fitness or the creature dies. It says our perceptions are tuned to payoffs rather than to truth, which is a different thing. The beetle's read on "female" was extremely reliable and completely untrue at the same time.
Like any formal result, it inherits the assumptions built into its models, and critics question whether real evolution is as tidy as the game-theoretic setup. What is not in dispute is the direction of the pressure. Even Hoffman's opponents tend to grant that natural selection rewards fitness first and truth only when the two happen to coincide. How often they coincide is the whole argument.
