Great question, I used to wonder the same thing until I learned how C++ was implemented using vtables and whatnot. Basically, code only needs to be in memory once, and objects are just pointers to data structures that the interpreter/compiler knows how to deal with. Having a bunch of objects is basically the same thing as having a bunch of structures; the effective difference being that the the order of the function/method name and it's first parameter is swapped. Consider how in perl, the following two are almost identical:

$obj->meth(@args);

and

meth($obj, @args);
[download]

The relevent difference is that perl can work some magic knowing what $obj is blessed as, such as finding it's "meth" method in superclasses.

So basically, don't worry about memory overhead: it's not an issue. Code performance is the more serious problem, because of the added indirection in OO architectures and added abstraction inherent in OO designs.