dynamic binding

In object oriented programming, dynamic binding refers to determining the exact implementation of a request based on both the request (operation) name and the receiving object at the run-time. It often happens when invoking a derived class's member function using a pointer to its super class. The implementation of the derived class will be invoked instead of that of the super class. It allows substituting a particular implementation using the same interface and enables polymorphism.
See more at Wikipedia.org...

The property of object-oriented programming languages where the code executed to perform a given operation is determined at run time from the class of the operand(s) (the receiver of the message). There may be several different classes of objects which can receive a given message. An expression may denote an object which may have more than one possible class and that class can only be determined at run time. New classes may be created that can receive a particular message, without changing (or recompiling) the code which sends the message. An class may be created that can receive any set of existing messages.
C++ implements dynamic binding using "virtual member functions".
One important reason for having dynamic binding is that it provides a mechanism for selecting between alternatives which is arguably more robust than explicit selection by conditionals or pattern matching. When a new subclass is added, or an existing subclass changes, the necessary modifications are localised: you don't have incomplete conditionals and broken patterns scattered all over the program.
See overloading.