During my time in the counter-apologetics crowd on Twitter I’ve noticed that several of the arguments deployed by our side rest on a lay-understanding of what constitutes ‘knowledge’. As most people haven’t studied philosophy formally, this isn’t surprising. Their motivation for taking the stance they do isn’t surprising either; it’s easier to deny your opponent’s position than it is to build a positive case of our own (and rebut the criticisms). Yet I believe it can be demonstrated that we can build such a positive case, and this post will be the first part in my attempt to detail how.
This first post will address how you can have knowledge without certainty, and who justifiably has the burden of proof for a claim.
Before I begin, a word of caution: words like ‘know’, ‘prove’, ‘belief’, ‘certainty’, have a much narrower meaning than their lay-use would suggest, and can (and do) cause confusion when discussing this matter. I’ll attempt to signpost those differences as I go.
What is Knowledge?
The branch of philosophy that studies knowledge is called epistemology. Its goal is to settle the ‘What is it?’ and the ‘How do I get it?’ questions. Now there are many kinds of knowledge, such as ‘know-how’ (skills knowledge), knowing a person (being acquainted with), and knowing a place (knowing of it but not strictly anything about it). But the kind most relevant to our discussion is called ‘propositional knowledge’. This is the ‘know-that’ variety you recall from memory in sentence (proposition) form, and it ranges from everyday ‘I know that my desk is messy‘ to the scientific ‘I know that Mars is the fourth planet from the Sun‘.
So what do we need in order to say with confidence that we have ‘know-that’ knowledge? In philosopher-speak: what are the necessary and sufficient conditions for knowledge? According to the traditional account first described by Plato, knowledge is (at least) a ‘justified, true, belief’ (JTB):
- Knowledge isn’t something you have accidentally. If I correctly guess how many coins are in my pocket at random, I can’t be said to have known it before checking my answer. There needs to be a proper connection between what you know and how you came to know it. Therefore, knowledge requires ‘justification’.
- I cannot know something false (I can only know that something is false). Therefore knowledge requires a statement to be ‘true’.
- If I don’t have a single belief about cats, I can’t know anything about cats. Therefore knowledge requires ‘belief’.
(Note: philosophical definition of ‘belief‘: ‘a mental state that represents a state of affairs which is accepted as true by the believer‘. There is no connotation of it being religious in nature, or association with a low confidence in the truth of its content. Example: I have several separate beliefs that Obama is the president of the United States; that my mechanic is an honest man; and that avocado seems to taste bad to me. Notice I can be wrong about the first two, but not the third. I can’t be wrong about how something seems to be to me.)
There is a problem with this account of knowledge (see Gettier Problem), and the traditional/non-traditional accounts differ on whether a fourth condition should be added to solve it. It’s an interesting discussion but not one that concerns our purposes here.
To put this definition into practice, consider a discussion between two cavemen concerning the shape of the Earth. Uugh-the-Firestarter thinks the Earth is round. When pressed, he admits he has no evidence to back this up, but he just thinks it is because roundness is an appealing shape. Gurk-the-Spearmaker says the Earth is flat, and points to the relative flatness of the savanna to persuade Uugh. He also points out how ridiculous it would be if it were round, because people on the bottom would fall off.
Uugh has a true belief, but it is not justified. He has no evidence to account for it beyond his fondness for round shapes. Gurk has a justified belief, which from his available evidence is true. From where we stand, he’s missing crucial information that falsifies his claim. But from his point of view, all the evidence fits his hypothesis. From his point of view, he has knowledge. We only know better than he does because we have access to more information.
I can imagine some readers will object right now and say ‘No! Gurk doesn’t have knowledge because he’s wrong!’ But notice, we are in the exact same boat as Gurk and Uugh. Without the benefit of hindsight, we are unable to know which of our theories about the world are true and which will ultimately be shown to be incorrect in the future. Relativity is one of the most well-tested theories in physics, yet for all that it could be superseded tomorrow. Do we want to say in light of this that we don’t know relativity theory is true (within its domain of application) in spite of all the evidence in its favour?
Not so fast. There are perhaps three things we can say here. We might say that we probably have knowledge, but we’re forever ignorant about when we do and don’t have it. We might say that knowledge is impossible. And finally, we might say that knowledge is the kind of thing that is always provisional, and is contingent on our ever-updating evidence.
I contend that the first two responses are confused about the nature of knowledge. They confuse ‘what there is’ with ‘what we know about what there is’. In philosopher speak, they confuse ontology with epistemology. This error is so common it deserves its own unique name, but I’m unaware of it having one. Solipsists make this error in totalis by asserting that since they can’t know if anything else exists, they are the only thing that exists.
The Irrelevance of Certainty
At this point I imagine my twitter opponent will object. Knowledge, they say, requires certainty that the statement is true, but the definition of belief I’ve just given is one in which I could potentially be mistaken, or two people could make opposite knowledge claims about the same object/event (and they can’t both be right). If I can’t prove it, if I can be wrong about it, then I can’t claim to say that I know it. Or so the objection goes.
Let’s see where this leads. Near as I can tell, there are two things that ‘certainty’ could mean in this context:
- (1) Subjective Certainty: the felt experience that something is definitely the case, and
- (2) Absolute/Irrefutable Proof: exactly representative of ‘How Things Are’; a 0% probability of being wrong.
Now I assume it is clear to all readers that (1) is not at all helpful, and I am including it only for completeness. The universe doesn’t care how much you feel it in your bones that X is true, if it’s the case that X is actually false.
Certainty (2) on the other hand is an extremely bold claim. If this is what is meant by certainty, it seems there aren’t any beliefs we possess that counts as knowledge, except ‘I exist’. Your consciousness, locked away in the prison of your brain, can only access the outside world through your bodily senses, and these are fallible channels of information. We misrepresent the world all the time, from mishearing lyrics, to mistaking a stick for a snake (or worse, a snake for a stick). We hallucinate on a daily basis. Recall the last time you couldn’t find something even though you were staring right at it; a negative hallucination. So our senses are at least sometimes unreliable, and that is enough to undermine Certainty (2).
In logical form, the argument looks like this:
- Premise 1: Certainty (2) is a necessary condition for knowledge to be attained.
- Premise 2: Certainty (2) is a practical impossibility due to our fallible senses.
- Conclusion: Knowledge is impossible to attain in practice.
In an effort to salvage their position, the earlier person responds, “That may be true, but the scientific method allows us to overcome our biological shortcomings and achieve knowledge by weeding out human error and cognitive biases”.
While I can see how this is might be an attractive ledge to jump to, it’s a trap for at least three reasons:
- The scientific method, while better at deriving true statements about the world, is also a fallible method. This fallibility undermines certainty.
- It begs the question against the previous argument. One would imagine that in order to carry out proper science, I’d have to observe technical readouts, measure quantities precisely, etc. You’re employing your fallible senses to carry out the experiment, and that ‘contaminates’ the experiment with fallibility. Even if the scientific method was infallible, your fallible senses are ‘downstream’ of it.
- Even if 1 and 2 weren’t an issue, It entails that nearly nothing you believe about your life counts as knowledge. You can’t claim to know you ate breakfast this morning, or what you’re wearing. The only knowledge you could possess are the facts attained from scientific experiment. This seems, to me at least, to be an absurd thing to say. If I know anything, I know I’m writing these words right now.
So it seems the scientific method can’t save certainty here either. Yet another reply I’ve encountered: “Science improves the definition of knowledge. Plato couldn’t have foreseen the power of the scientific method when he laid down the JTB theory of knowledge. Our definition of knowledge has changed in response to the invention of science since Plato.”
I’ll be honest, I’m not sure what to make of this response. It seems to be implying that our capacity to be confident in the truth of our statements has increased as a result of the invention of science. I take this to be indisputable. But it doesn’t get you to the absolute proof desired by Certainty (2), for all the reasons outlined above. Or perhaps it is appealing to the criterion of ‘reliability’, the purported fourth condition of knowledge that non-traditionalists add to the definition. If so, that has its own problems, and still doesn’t get you to Certainty (2).
So it seems certainty is a dead end. Perhaps if we look into why such people want certainty, we can see that we don’t need the extreme they are asking for, and that something else will do in its place.
‘Proof’ Doesn’t Exist
Already the use of the word ‘proof’ here gets my hackles up. Outside the world of mathematics, proof is a yeti. It’s a myth. We should all get used to speaking in terms of ‘evidence’ rather than ‘proof’, as the word ‘proof’ denotes a finality of inquiry that we’re not entitled to. Knowledge isn’t a synonym for Truth with a capital T.
What we’re butting up against here is the difference between the scientific and lay definitions of ‘proof’. I find it strange that for words such as ‘theory’ we skeptics are the first to correct creationist misuse of the word, but then go on to commit the same error ourselves with ‘proof’. Scientists don’t speak in terms of ‘proof’ – in fact it’s bad scientific practice to do so. When your ‘proof’ is overturned, you’ve got egg on your face. You can have evidence that supports or negates your hypothesis, but proof is not on the cards. The reason for this is that any data set can fit an infinity of explanatory hypotheses. To see how this is possible, consider three data plots:
We have three identical data sets here, but three competing interpretations: linear, polynomial, and logarithmic. If three weren’t bad enough, there are an infinity of lines that I can draw through those six points of data. Now from the available evidence we can narrow down the list of candidates to something more manageable and arrive at an agreeable answer by applying rules like parsimony. But notice the key phrase: ‘available evidence’. There will always be more data points to add, which will rule out further hypotheses, and we can’t know in advance what that data will be, whether it will support or negate our current favoured hypothesis.
Now it might seem that I’ve switched sides here and have begun arguing for the position I argued against in the last section, that we can never really know anything. This would be true if knowledge required absolute certainty, the finality of investigation. I imagine the concern is that if we shift away from requiring 100% ‘proof’, everything suddenly becomes relative because any other number is arbitrary. It may be arbitrary, but it is not entirely so. Let’s see why this is the case.
Different disciplines have different thresholds for discovery, and these are tailored to the relative capabilities of the field. Physics sets the threshold for discovery at sigma-5 (99.9999994% probability of an accurate result) because it’s possible for physicists to collect tens of millions (or even trillions) of data points in a short amount of time. Medical trials set it far lower – at perhaps sigma-3 (99.7%) – due to both the nature of what they study (people) and the expenses involved. Far from these probability thresholds expressing a lack of certainty in their findings, they express sheer confidence! A probability of 0.0000006% equates to a 1 in nearly 2 million chance of being wrong.
This is ‘practical’ certainty. To insist on absolute certainty is to confuse ‘beyond reasonable doubt’ with ‘beyond doubt’, and jettison your reason in the process. We don’t need absolute certainty, as reason allows us to justifiably bridge the gap. The proponent of absolute certainty is defining knowledge to mean ‘the truth’ as opposed to our potentially fallible assessment of the truth.
The Burden of Proof
One popular meme I see floating around the Twitterverse is the assertion that the burden of proof is always on the positive claim, or that ‘you can’t prove a negative’ (we’ve all seen or said this plenty of times). As a rule of thumb, this does quite nicely. We more often than not run into positive claims than negative claims, so it rarely comes up. But nearly nobody on Twitter who uses the burden of proof in debate understands when it properly applies. This leads them to believe that proponents of negative claims are intellectually dishonest, as it allegedly saddles them with an impossible burden of proof.
The assigning of the burden of proof relies on where the claim sits in relation to known background information. If I make a claim that is consistent with everything we know about the world, there is little to no burden on me. The work has already been done. The likelihood of what I’ve said being true is so high that we don’t even talk about burdens at all. In Darwin’s day, the prevailing explanation for the complexity of life was intelligent design. It was his burden of proof to argue this was not the case. Yet today the burden is reversed, because intelligent design is inconsistent with what is known about the origin of biological complexity.
But it might be objected that this is just “one positive claim (evolution is true) against another positive claim (design is true). What we’re concerned about is the impracticality of proving negative claims, claims to non-existence”. Superficially, this is the case. But notice that the claim ‘evolution is true’ is a denial of the claim that ‘design is true’. One can rewrite the positive claim ‘evolution is true’ as the negative claim ‘non-evolutionary explanations are false’. The same evidence supports both claims. I sense many will be unsatisfied with this answer, so let’s use another example.
If I claim to have $20 in my pocket, you’ve really got no reason to disbelieve me on that information alone since it’s trivial. A mere $20 isn’t an unusual amount of money to have in your pocket, so the burden is minor or nil. But if I make the (positive) claim that I have $20 in my pocket and a leprechaun loaned it to me then you have cause to raise an eyebrow, and ask further questions. The existence of a money-lending leprechaun is so out of whack with everything else we know about the universe that it demands special explanation, and the burden now applies.
Now consider the converse. I make the (negative) claim that money-lending leprechauns don’t exist. This is consistent with known background information. As such, there is little to no burden to demonstrate it. It’s a trivially true claim against the background of metaphysical naturalism, the philosophical position that there are no supernatural entities of any kind. We are justified in provisionally accepting this metaphysical position because we’ve never seen evidence to its contrary, and all the evidence we do have supports it. If supernatural entities like leprechauns existed it would require a wholesale revision of everything we think we know about the universe, and the probability of this occurring is infinitesimal given the wealth of evidence we have for the universe operating how we currently believe it does.
In this way, the burden of proof is related to the informal fallacy of ‘argument from ignorance’. It’s important to understand when this charge is actually a fallacy, and when it is not. Whether it sticks will depend on who has the burden of proof, which will depend on known background information. When intelligent design proponents claim that everything in the genome has a designed function and we just haven’t found it yet, that’s a fallacy. When the evolutionary biologist claims that there has never been found an instance of an intelligently designed biological organism, therefore ID is false, it is not a fallacy. The evidence is clearly on one side.
Now a hard-nosed skeptic will perhaps reply that science isn’t complete, and we don’t know everything, so how can we categorically rule out the existence of things? But this is again to confuse ontology with epistemology, to treat knowledge as absolute truth rather than a provisional assessment of truth.
This is a bad habit that needs to be broken.
The next post in the series will apply the current discussion to the existence of gods, both theistic and deistic. This ought to influence how we ought to use words like ‘agnostic’ and ‘atheist’, and which definitions of these terms are more useful than others.