What I meant is the following: You have a type A, and B inherits from A. If you have a variable with type constraint A, but it contains a B, then C++ dispatches with the type constraint (A), while Perl 6 dispatches with the actual type (B). My attempt to make it more concrete:

class A { };
class B { };

multi sub f(A $x) { say 'A' }
multi sub f(B $x) { say 'B' }

my A $x = B.new;
f(b)   # says B

And in C++:

#include <iostream> 
using namespace std;

class A { };
class B : public A { };

void f(A x) { cout << "A\n"; }
void f(B x) { cout << "B\n"; }

int main (int argc, char** argv) {
    A obj = B();
    f(obj); // prints A
    return 0;
}