Stuff like map() is generic iteration, over any structure that exposes iteration. When it's a member function, it means that every collection has to implement map itself basically. When it's separate, the collections only need to provide the interface needed to iterate over it, and the generic map() will just use that.
Taking OOP more seriously, this kind of thing should be implemented through inheritance, interfaces, mixins, etc. Even though I've got used to it, Python has these inconsistencies that sometimes it wants to be more OOP, sometimes it wants to be more FP.
Ruby has chosen OOP as its size, and implements nicely those functional operations as methods (same for Kotlin, for example). That makes easy for composing them:
# Sum of the square of even elements of a list, in Ruby
my_list.filter{|x| x % 2 == 0}.map{|x| x * 2}.sum
Python could do something like that, but we quickly fall in a parentheses hell:
# The same code, in Python
sum(map(lambda x: x * 2, filter(lambda x: x % 2 == 0, my_list)))
Languages that have chosen the FP path more seriously implement function composition. They could do the same way as Python, but composition makes it more readable:
# The same code, in Haskell
sum . map (^ 2) . filter (\x -> x `mod` 2 == 0) $ my_list
PS: I know that I it would be better to use comprehensions in both Python and Haskell, but these are general examples
So semantically it's actually closer to Python, with the only difference that, since Python doesn't have extension methods, it has to use global functions for this, while Kotlin lets you pretend that those methods are actually members. But this is pure syntactic sugar, not a semantic difference.
FWIW you can have both in this case; you just need to make dotted method calls syntactic sugar for global function invocations.