I think this is broadly correct. Your system 1 (in Kahneman terms) can tell the difference between 50/50 and 1/10 but something like 1 in 1000 in is put in the same bucket as 1 in 100,000,000 -- unlikely but not impossible. If you want to actually work with odds like that you need to use system 2 to compute something more tractable like EV (~$0.75 on a $2 ticket for tonight's drawing).

In the interest of full disclosure, I probably spent $30 on lotto tickets last year.