I think he's mischaracterizing the issue.
Beliefs serve multiple functions. One is modeling accuracy, another is signaling. It's not whether the environment is harsh or easy, it's which function you need. There are many harsh environments where what you need is the signaling function, and not the modeling function.
I think the quote reflects reality (humans aren't naturally rational so their beliefs are conditioned by circumstance), but is better seen as an observation than a recommendation. The best approach should always be to hold maximally accurate beliefs yourself, even if you choose to signal different ones as the situation demands. That way you can gain the social benefits of professing a false belief without letting it warp or distort your predictions.
Here's the new thread for posting quotes, with the usual rules: