But that’s what gives away the game. The problem of induction offers the conclusion that our generalizations are unreliable, and yet we are to accept that conclusion as reliably applicable to all generalizing. It is as though one stated, “All generalizations are unreliable, and my generalizations prove that!” And yet, theists who deploy the problem of induction as an apologetic device apparently do not see how it falls on its own sword.
According to what David Hume’s argument supposedly demonstrates, no instance of induction can be accepted as reliable because the reliability of induction as such cannot be justified in some non-circular manner, induction essentially being the epistemological process of drawing general conclusions from a handful of specific samples. But isn’t that essentially an inductive conclusion? It’s a conclusion about an entire category – namely induction as such, and in fact the premises supporting that categorical conclusion must themselves have been arrived at by means of inductive inference, the very process the argument is said to call into question, in order to repudiate it wholesale. To say that inductive reasoning is unreliable is to say that all instances of inductive reasoning are unreliable, to which one should ask: How did you draw this general, all-encompassing conclusion without making use of inductive reasoning?
Typically the problem of induction is positioned in such a way as to conceal this obviously self-defeating flaw. Often we see the problem of induction cast in terms of making future estimates based on past experience. For example, in his paper Secular responses to the Problem of Induction, James Anderson defines the “inductive principle” as “the principle that future unobserved instances will resemble past observed instances.” But even in these terms, the problem of induction so-conceived overlooks its own propensity to take something critically fundamental entirely for granted.
Suppose you drop an uncooked egg on your hard kitchen floor, and upon impact with the well-weathered flooring the egg shell breaks and its loose and gooey contents spill out onto the floor. Now you have a mess to clean up. Now my cat will never infer from this specific instance that the same result will happen if I drop another uncooked egg onto the floor. But you and I can! However, according to David Hume a la his problem of induction, I’m wrong to take any such inference as reliable. How could I possibly know that the next egg will behave similarly? He would say that I am illicitly jumping to a conclusion about future unobserved instances on the basis of past observed instances. Just because I dropped one egg and it splattered all over my kitchen floor, how can I thereby draw the general conclusion “eggs will break when they’re dropped on a hard surface”? The skeptic will ask: “How do you know that eggs will break when they’re dropped on a hard surface?”
Here the Objectivist can turn the tables on the would-be skeptic and ask: “Eggs? Drop? Break? Hard Surface? What do these words mean?” When it is acknowledged that these words in fact denote concepts, we can ask the skeptic: Where did you get the concept ‘egg’? Where did you get the concept ‘break’? Where did you get the concept ‘surface’? Etc. These concepts are clearly necessary constituents to the question “How do you know that eggs will break when they’re dropped on a hard surface?” and therefore more fundamental to any questions they are used to inform, such as the skeptic’s. Which means: One would first have to know what eggs are, what dropping is, what surfaces are, etc., just to raise the stated “how do you know?” question to begin with. I’m simply pulling the skeptic’s own question back a step: How does he know what eggs are? In fact, his question not only assumes his own knowledge of what eggs are, it also assumes that we know what they are as well. So let’s call him to account for the knowledge his own question assumes.
Now here’s what’s important to note about the concept ‘egg’: the concept ‘egg’ does not restrict its range of reference only to those specific eggs one has come in contact with. On the contrary, it includes every egg that exists, every egg that has existed, and every egg that will exist, without any quantifiable restriction (i.e., it does not refer to, say, only 2000 eggs, and thereafter a new concept needs to be formed for the next 2000 eggs). The concept ‘egg’ is open-ended and quantitatively limitless. Which means: it includes not only those eggs one has personally come in contact with in the range of his limited experience, but in fact all eggs - regardless of time and place. Time and place are thus omitted measurements and are thus not essential to the meaning of the concept ‘egg’ or to the units which it subsumes. Considerations about past versus future become as irrelevant as considerations about this side of the Atlantic versus the other. Indeed, we make inferences about what must have happened in the past all the time, such as when I go into my kitchen and find an eggshell on the floor: someone didn’t clean up thoroughly!
Since knowledge is essentially those concepts which we have formed and integrated into sum total retained in consciousness, this means that by means of the concept ‘egg’ we have knowledge - not just of those eggs which we have seen and touched firsthand – but of all eggs, in a very generalized manner. Moreover, our concept ‘egg’ includes only eggs, not furry critters, alphabet blocks, glass bottles, mailboxes, steam shovels, hydro-electric dams or pontiff’s caps. It also includes everything we learn about eggs as an ever-growing sum of knowledge in its own right. It’s not a broad leap, then, to go from observing that one eggs breaks when it’s dropped on a hard surface to supposing that other eggs will do the same. This is simply an application of the data we integrate in our concept ‘egg’ to specific instances under consideration, thus providing a model for deduction. We then refine such generalized instances by discovering those conditions which apply and those which do not. I have yet to find, for example, an egg which breaks when I sing to it.
So induction, then, not only presupposes the validity of those concepts which we bring to bear in our inferences, the validity of those concepts in fact constitutes the epistemological basis of those inferences. So I think the question of how the skeptic knows what eggs are is fair game. Indeed, any question about what we know generally about eggs goes right back to how we formed our concept ‘egg’ in the first place. But we don’t find apologists who raise the problem of induction discussing the nature and formation of concepts in their treatments of the problem of induction. Indeed, their worldview has no theory of concepts to begin with. And that’s why they are so readily captivated by the problem of induction, for their worldview does not equip them to understand the conceptual preconditions which make inductive generalization possible.
by Dawson Bethrick