Imagine someone lives beneath a volcano and knows that at some point an eruption will destroy his house such that he no longer extracts economic benefit from it. Do you think that the way he builds his house (e.g. using straw, or stone) will differ based on whether the eruptions come every 10 years or every 100?
If artists are motivated by the economic incentive afforded by copyright, then the level of that motivation is variable based on the length of copyright.
While that's undeniably true, I think the objection is to your leap from observing that motivation significantly increases with term length at term = 1 day, to concluding it's still significantly increasing at term = life+70yrs.
Economic incentive can be measured and modelled, and while there isn't that much research into this, there is some. One Cambridge study (Pollock [2007]) found that the optimum length of copyright term is ~14 years (any longer and the economic benefits of works going into the public domain outweigh further increases in incentive to produce). Now, yes, economics models should be taken with a large grain of salt (especially ones when it's unfeasible to test their predictions). But it's a start, and certainly there don't seem to be any econ studies supporting a length anywhere near life+70 or longer.