NSCoder: how does the class know which object we pass?

NSCoder: how does the class know which object we pass?

There is already a similar question regarding decoding, but I wanted to
try summarize and see if I got the full picture right.
Basically Apple provides the NSCoder class that allows to encode and
decode an object if it is sublcass of the NSCoding class.
Inheriting from that class allows to use the NSCoder encode and decode
functions provided by Apple that allow to "serialize" NSDataTypes (and all
primitive C structs and data types).
My understanding of the NSCoder class reference is that the encodeObject
function is able to detect which object type we pass and hence serialize
it in the proper way.
Instaed, when using specif types of int (e.g. int 64) Apple decided to
provide a specific function as understanding the "signature" of the data
wasn't that trivial (I guess that in low levelel memory an int64 is
relatively similar to an int as is a primitive data type and hence there
is no NSObject "signature" from which Apple could reverse engineer the
data type).
Is my understanding correct? Anyone has a different explanation?
I would also like to ask you some more explanation on the usage of
"encodeRootObject" function. There isn't much on the reference and I have
seen from some code examples. I understand that archiveRootObject is used
to actually save an object, but when shall I use encodeRootObject? Is this
used when I try to encode an object of a class that I defined?