# What is Falsification, and Who is This Popper Fellow, Anyways? -Part 1- The Problem of Induction

What distinguishes science from pseudoscience? In my experience, if you ask a practicing scientist this question they will likely answer:

“We have logic on our side! *Proper* scientific theories can be *falsified*, while pseudoscientific theories cannot.”

You might remember Dr. Ng bringing this up in the first lecture on climate change in January. At the same time, I was reading an essay written by Karl Popper, the father “falsification” as we now know it, for a grad course I’m currently enrolled in.

So, what *is* falsification?

Here’s the gist*. There are inductive and deductive logical assertions. Inductive assertions involve some sort of non-guarantee, while deductive assertions provide a guarantee. The most intuitive form of inductive logic takes the form of enumerative induction, or:

Every piece of paper I have seen has been white.

- Therefore,
the nextpiece of paper I see will be whiteor:

- Therefore,
allpieces of paper are white

Deductive logic takes a different form, and follows a number of deductive rules. Here is the rule, *modus ponens*:

- If p, then q
- p
- therefore, q

Of course, q must only be so insomuch as the prior requirements of q are met fully (i.e. that p was so).

I think the difference between deductive and inductive logic is fairly intuitive. The natural question that arises, then, is, “Does science deal with inductive, deductive, or both forms of logic?”

Lets think about this question for a moment: does science concern itself with non-guarantees? I think the answer is obvious. Yes. The scientific method is *founded* on non-guarantees, after all. Hypotheses are by their very nature not guaranteed – or are they?

David Hume, a renown 18th century British skeptic and philosopher first formulated the Problem of Induction (POI) into how we know it today. The POI can be set up and described as follows:

- Scientific theories are not only concerned with describing past events, but also with predicting future occurrences of events
- In order to move beyond particular observations (i.e. this piece of paper is white) to
*generalized*assertions (all pieces of paper are white), one must use inductive logic, since it it*irrational*to assume the universe possess any notion of regularity (that is, it is irrational without having observed the entire universe and all occurrences of paper throughout time). - Therefore, one must somehow be guided by a principle, whose tenets successfully allow one to rationally make use of inductive assertions (i.e. hypotheses)
- But what form of logic must this principle take? It cannot be deductive, as it would then cease to be inductive.
- It cannot be inductive, as that would require one of two things: an infinite regress of higher order principles to rationally justify the next lower principle (Popper); or the use of circular logic (one cannot use an inductive rule to justify inductive logic)
- Since any justification must use either inductive of deductive logic, a principle cannot be rationally justified. This is the POI.

Popper thought he had a nice and tidy remedy for the problem of induction: get rid of it entirely, and adopt deduction as the *sole* basis for scientific inquiry and progress (i.e. since induction cannot be used rationally, get rid of it and figure out a way to deal with the fall out). Popper then utilized falsification: the notion that there must always exist the possibility that a true counter example for every hypothesis exists. Therefore, if one were to test a hypothesis and observe the counter example, one would have *deductively* falsified the hypothesis of interest.

Naturally, one may then ask, “What happens when you fail to falsify a hypothesis? Is the hypothesis then confirmed?” According to Popper, no, since doing so would implicitly make use of inductive assertions (since one can never prove a theory to be true without having observed all possible cases where it is being tested).

“Popper, you’re a mad man!” you might say, “How can you possibly confirm a hypothesis without using inductive logic?”

The “Popperian science” consists of the following: “*bold conjectures*” and concomitant *refutations*. Popper thought he could do way with the POI by requiring science to make use of some serious fire power when it came to coming up with hypotheses, nothing that could be knocked down by a philosophical breeze. After rigorous testing and verification, via internal (is it internally consistent? is it consistent with the rest of science?), and external tests (can you formulate something meaningful with the theory, and then test it in the real world).

If a theory held up to the most scrupulous eye, it was deemed *corroborated*. This is the equivalent to a time log of all the instances where the theory was not falsified and, as Popper maintained, contains no shred of confirmation, nor implications regarding the likeliness a given hypothesis will be true in the future.

*Sounds strangely like induction, doesn’t it? *

*Coming up*:**Part 2: Objections to Popperian Falsification**

***Note**: *I am neither a philosopher, nor a philosophy graduate student. I simply find this stuff fascinating. If you take offense to anything in the above article and feel obliged to correct me in the comments field, please do so with the most honorable intentions in mind. Crushing my spirit will do neither of us good (unless you sustain on spirit juice, perhaps muddled with some lime and fresh mint. In that case, I can’t really say no, can I?). *

## 4 Responses to “What is Falsification, and Who is This Popper Fellow, Anyways? -Part 1- The Problem of Induction”

I expect of my science no proven theories. I think, for the most part, that science is concerned with creating necessarily false abstractions of the real world. These abstractions are useful to us because they are simplified cases of a complex system, in which we can make predictions that come out true a lot of the time.

Most theories have counter examples if you look hard enough. Try going really fast and you will break most physics done earlier than Einstein. This is not to say that Newtonian mechanics is ‘wrong’ any more than relativistic mechanics is ‘right.’ Rather, Newtonian mechanics looses it’s usefulness for speeds close to that of light. It is no longer an appropriate abstraction. We always throw out information when we do science – or else we would just be concerned with tracking the state and behaviour of the most reduced aspects of the system we can find.

I suspect I know much less about the philosophers who concerned themselves with such things than you, but hopefully the following idea helps:

The real world is not a formal system. In a formal system we rigorously define truth values, and operations on those truth values. We operate by deductive logic, given a set of axioms. The real world is not amenable to such a treatment because the list of axioms that we take for granted is subjective. Whereas in a formal system, we can define axioms to be whatever we like and we get some interesting mathematical structure as a result; the axioms we choose for the real world have to do double duty – at once they need to spawn the things we want to believe about the world, and they have to be accurate in an endless number of ways with regard to the real world themselves. You can thus argue, subjectively, about what axioms are ‘right’ in the real world, whereas in a formal system you accept some things as right and observe the consequences – different axioms give rise to interesting and diferent logical structure.

Many apologies if this wandered. Maybe one should refrain from philosophy of science at 2 am?

I liked the article, not many want to communicate logic and science to the general audience. kudos.

Josh

[…] Part 1: The Problem of Induction […]

[…] Part 1: The Problem of Induction […]

[…] Part 1: The Problem of Induction […]