This does not sound like an accurate description of "rational agents". A rational agent may value something less because of a perceived second-order effect (e.g. "If I accept a price hike, there will be more price hikes in the future"). There's nothing non-rational about that.

In general, the "rational agent" abstraction is so flexible it can represent almost any real-world behaviour (which is why it's a bad model, not that humans are fundamentally irrational in some way).

It's not the core idea of "fatigue of paying for value" itself but something more akin to "fatigue of keeping track of which things were actually worth paying for, which starts to take more value than you can create with the same amount of time or could get with a few well thought out larger payments". It's usually brought up around subscriptions precisely because they are a lot of work to track but easy to forget.

What if what I value is that something is not corrupted by monetary incentives?