隐喻性的解释:

我有一个包裹要寄给一个朋友，我也想知道我的朋友什么时候收到。

所以我把包裹带到邮局，让他们把它送到。如果我想知道我的朋友什么时候收到包裹，我有两个选择:

我可以在邮局等邮件送到。

(b)邮件送达后我会收到电子邮件。

选项(b)类似于回调。

假设你需要一个函数返回10的平方，那么你写一个函数:

function tenSquared() {return 10*10;}

之后你需要9的平方，所以你写了另一个函数:

function nineSquared() {return 9*9;}

最终你将用一个泛型函数替换所有这些:

function square(x) {return x*x;}

同样的想法也适用于回调。你有一个函数，它做一些事情，当完成时调用doA:

function computeA(){
    ...
    doA(result);
}

之后你想要完全相同的函数调用doB，而不是你可以复制整个函数:

function computeB(){
    ...
    doB(result);
}

或者你可以将回调函数作为变量传递，并且只需要使用该函数一次:

function compute(callback){
    ...
    callback(result);
}

然后你只需要调用compute(doA)和compute(doB)。

除了简化代码之外，它还让异步代码通过在完成时调用任意函数来让您知道它已经完成，这与打电话给某人并留下回调号码类似。

为了教授回调，你必须先教授指针。一旦学生理解了指向变量的指针的概念，回调的概念就会变得更容易。假设您使用的是C/ c++，可以遵循这些步骤。

First show your students how to use and manipulate variables using pointers alongside using the normal variable identifiers. Then teach them there are things that can be done only with pointers(like passing a variable by reference). Then tell them how executable code or functions are just like some other data(or variables) in the memory. So, functions also have addresses or pointers. Then show them how functions can be called with function pointers and tell these are called callbacks. Now, the question is, why all these hassle for calling some functions? What is the benefit? Like data pointers, function pointer aka callbacks has some advantages over using normal identifiers. The first one is, function identifiers or function names cannot be used as normal data. I mean, you cannot make a data structure with functions(like an array or a linked list of functions). But with callbacks, you can make an array, a linked list or use them with other data like in dictionary of key-value pairs or trees, or any other things. This is a powerful benefit. And other benefits are actually child of this one. The most common use of callbacks is seen in event driver programming. Where one or more functions are executed based on some incoming signal. With callbacks, a dictionary can be maintained to map signals with callbacks. Then the input signal resolution and execution of corresponding code become much easier. The second use of callbacks coming in my mind is higher order functions. The functions which takes other functions as input arguments. And to send functions as arguments, we need callbacks. An example can be a function which take an array and a callback. Then it performs the callback on each of the item of the array and return the results in another array. If we pass the function a doubling callback, we get a doubled valued array. If we pass a squaring callback, we get squares. For square roots, just send appropriate callback. This cannot be done with normal functions.

可能还有更多的事情。让学生参与进来，他们就会发现。希望这能有所帮助。

应用程序通常需要根据其上下文/状态执行不同的功能。为此，我们使用一个变量来存储关于要调用的函数的信息。根据需要，应用程序将使用要调用的函数的信息来设置这个变量，并使用相同的变量来调用该函数。

在javascript中，示例如下。在这里，我们使用方法参数作为变量，我们存储关于函数的信息。

function processArray(arr, callback) {
    var resultArr = new Array(); 
    for (var i = arr.length-1; i >= 0; i--)
        resultArr[i] = callback(arr[i]);
    return resultArr;
}

var arr = [1, 2, 3, 4];
var arrReturned = processArray(arr, function(arg) {return arg * -1;});
// arrReturned would be [-1, -2, -3, -4]

回呼是一个贴了邮票的回邮信封。当你调用一个函数时，就像发送一封信一样。如果您希望该函数调用另一个函数，则以引用或地址的形式提供该信息。

A callback is a function that will be called by a second function. This second function doesn't know in advance what function it will call. So the identity of the callback function is stored somewhere, or passed to the second function as a parameter. This "identity," depending on the programming language, might be the address of the callback, or some other sort of pointer, or it might be the name of the function. The principal is the same, we store or pass some information that unambiguously identifies the function.

当时间到来时，第二个函数可以调用回调，根据当时的情况提供参数。它甚至可以从一组可能的回调中选择回调。编程语言必须提供某种语法，允许第二个函数调用回调函数，并知道它的“标识”。

这种机制有许多可能的用途。通过回调，函数的设计者可以通过调用所提供的任何回调来定制函数。例如，排序函数可能将回调函数作为参数，而这个回调函数可能是用于比较两个元素以决定哪个元素先出现的函数。

顺便说一下，根据编程语言的不同，上面讨论中的“函数”一词可能会被“块”、“闭包”、“lambda”等取代。