A Skeptics' Epistemology

We live in an amazing age of discovery and science, and it's astonishing to think of all the things that we know today that people of even a few centuries ago didn't know. Regrettably, much of what we consider to be known today is not accepted by everyone, and established scientific theories are constantly under attack by those who don't believe they are true. I am often asked questions like "do you believe in evolution?", or "do you believe in the Big Bang?” My response is usually that it's not a matter of belief, but of high probability, given that there is ample evidence for it. Similar questions are asked about belief or disbelief in God, and I try to explain that a deistic creator (or a god, with a lower case g) is largely unknowable, while the more specific depictions of God (capital G), are almost certainly myths. I'm careful when using the word knowledge, as it implies absolute certainty when that's not always the case.

But what exactly is knowledge, and how can one say that it is knowledge that one holds, rather than belief? Is it possible to say that one knows something, and yet be wrong about it?

What follows is my own simplistic version of a skeptics’ epistemology, which is my attempt to answer these concerns. This is by no means intended to be a complete explanation of skeptical theories of knowledge (that would take a whole book), but a simplistic explanation of a methodology we can use to determine what is reasonable to believe (although as you read further, you will see that my use of the word 'belief' is used for want of a better term). It is how I gauge what is acceptable to call belief or knowledge, while remaining rational.

My Previous Simplistic Scales of Belief and Knowledge

In my previous article, Agnostics, Get off that Fence!, I gave an explanation of the meaning of the words atheism and agnosticism and how they differ. Part of that explanation was a diagram comparing scales of varying degrees of both belief and knowledge. It was a simplified explanation, and I mentioned varying degrees of certainty. Now, I'm going to go deeper into that, but what I'll provide here will actually partially override what I wrote before. This might seem like a contradiction, but the reason for it is that although the concepts of knowledge and belief are distinct, I will argue that the two concepts must be united in order to maintain a rational assessment of certainty. In everyday usage, belief is a word used to describe a person's positive acceptance of a principle, or their point of view, but for the more specific usage - to refer to a position accepted as absolute truth - I will demonstrate why this is redundant.

The Need for an Honest Assessment of Evidence

Most of us would agree that the truth is important. Even people with a strong faith-based belief would have some kind of reason for their belief, whether it be what they think is common sense, or their interpretation of evidence, or their trust in what they have been taught by their parents. Very few people would say that what they believe is more important than what is really true, and the reason for this should be clear. People believe what they do because they assign a truth value to it. The belief doesn't have precedence over truth, the belief is seen to be the truth. But with so many different opinions on what the truth really is, how can we be certain?

The purpose of the system that I will propose is just that - to align, as closely as possible, our beliefs with reality. Where there is insufficient evidence to make a determination, the "I don't know" answer is the only correct answer. But there are varying degrees of certainty. If we do have some evidence to support a proposition, but not enough to say that it is known, we would say that it's probable, based on the evidence. Only once a high threshold of evidence has been passed can we then call it knowledge, end even then it is not absolute knowledge, only practical knowledge.

To summarise the idea - it is that for any proposition that could be true or false, rather than coming to an absolute determination, we need to build a supporting body of justification that would be proportional to the level of certainty assigned to the proposition. Only once surpassing a high level of certainty would we claim to know it, and even then would fall short of absolute knowledge - rather our use of the word knowledge would refer to a practical knowledge, or things proven beyond reasonable doubt.

Why not Absolute Knowledge?

In this article, I will write as though I am assuming that absolute knowledge is not possible, at least for most propositions. I'll go into more detail on this in a later article, and that will support my arguments used here, but for now it's not entirely necessary. To briefly introduce the argument, it would be to say that there can always be some possibility of being wrong. The human mind is fallible, and what we know of the world can only be known through our senses, and from our own limited perspective on the world. We can never attribute absolute certainty to a proposition, as there is always the possibility that evidence may arise that proves us wrong. It is simply an honest statement of fallibility.

This seems as though it might pose a problem - If nothing can really be certain, how can we have knowledge at all? This depends on what is meant by "knowledge", and I'll explain that from here on.

What is Knowledge?

Plato defined knowledge as a "justified, true belief", that is, in order for something to be called knowledge, it must be all three things: justified, true, and believed. Some problems with this simple definition have been pointed out, such as the Gettier problems, which demonstrate situations in which these conditions are met but which couldn’t really be considered knowledge. I won't explain all the details, but I would recommend reading up on them.

What I'm going to take from this is the idea of justification, which means any supporting evidence or reasoned argument that supports the proposition. This could be empirical evidence or data, observations, and philosophical extrapolations using rational arguments, such as logic and probability theory. Care needs to be taken not to use faulty reasoning caused by bias, and I'll also discuss that in the next article.

Note: This of course doesn't go into a full examination of what amounts to justification, as that would take a lot more time than I can put into this article. The details are not entirely necessary for my purpose here - the point is that there are forms of justification based on evidence and inquiry, and part of the skeptical process is to assess the forms of justification for their validity. This of course assumes a rational assessment of empirical evidence and reasoned inquiry, rather than an appeal to subjective experiences. Again, this is something I'll write about in a follow-up article.

The Difference between Absolute Knowledge and Relative Knowledge

If we consider that we can't have absolute knowledge, we can make a distinction between 100% certainty and a degree of certainty that we can consider to be "proven beyond reasonable doubt". If we have loads and loads of justification for a proposition, there would come a point where we could realistically say that something was "known" (in an everyday practical sense rather than an absolute sense) even though there is a slim chance that it could turn out to be wrong. In this way, we could say we know that the Earth orbits the sun, because there are massive amounts of scientific evidence to prove it. There is still a chance, however slim, that the Sun, the other planets, and all the other stars and galaxies in the universe orbit the Earth – there would have to be some massive and intentional manipulation, orchestrated by a prankster god – but from the evidence that we have, it's completely reasonable to say that we know that the established models are extremely accurate. This is an extreme example as it's something that very few people would contest (some people do believe the whole universe orbits the Earth, sadly), but the same principle can be applied to any other proposition that may be either true or false.

I'll summarise this point by saying that for questions concerning objective truths, the reality is independent of people's opinion of what is true. People can believe whatever they want regarding the existence of a god, but there is either a god or there isn't one; it's not a matter of belief. In fact, it's possible that all are wrong, and that there is a god of a kind that has not yet been imagined by human beings.

We can think of true or absolute knowledge as an unattainable ideal, and what we would call knowledge in a practical sense is the attribution of a high degree of certainty to a proposition (often called relative knowledge). The level of certainty that we have for a proposition would only become knowledge once it surpasses a theoretical high threshold of certainty.

Justification, Certainty, and the Rational Ideal

To put this idea into a more easily understandable graphical form, we can think about any proposition that could be objectively true or false, and set up a scale that shows the variable degrees of certainty that we can assign to the proposition. This would range from -100% (absolutely not true) to 100% (absolutely true), with 0% in the middle (completely uncertain).

This would be very similar to what I used in my previous article. On this scale, we can plot any proposition given the amount of certainty that we assign to it. For example, the proposition "Does god exist?" could have a high degree of certainty (if you believe in God), or a low degree of certainty (if you are fairly sure that there's no god). Or, it could be 0%, meaning that you have no belief either way. So how does justification come into it?

We can plot justification horizontally on the chart to show the relationship.

A rational “belief” would have a higher degree of certainty coinciding with a higher amount of justification. In other words, we wouldn't assign a high degree of certainty to a proposition if there wasn't a high amount of justification for it. Similarly, if we are looking at the inverse, we wouldn't say there was a high degree of certainty that a proposition was not true, unless there was a good amount of justification that it was not true.

This is not to say it's the absolute objective truth - it's our rational assessment of certainty based on the justification.

Thus, a diagonal line can be plotted on the graph to show this relationship. The line increases as justification increases, and decreases into negative certainty as justification becomes negative (it's not always possible to provide evidence of the negative proposition, and I will get to this later).

Given this relationship, this is what the diagram would look like:

Rational Ideal

Approaching the Unattainable Absolute

As I mentioned earlier, we shouldn’t claim absolute knowledge, so our line should always fall short of absolute. This means the line representing a rational relationship between justification and certainty would taper off and never reach 100%. In other words, if something is justified enough to call knowledge, and if we were to find yet more evidence, it would become even more certain but would still never reach 100% certainty.

For propositions that have inadequate justification, the only rational position to take is "I don't know". A lack of evidence disproving a position is not the same as evidence supporting it, and a lack of evidence supporting a proposition is not the same as evidence against it. Often, particularly among the religious, a false dichotomy is assumed – they say a person either believes there's a god or believes there isn't a god. This is clearly a false generalisation, as it's possible to take the "I don't know" position, or any other point in between.


When the amount of justification is high enough, we can say that we have knowledge, in a practical sense, that a thing is true. A proposition without enough justification to meet these thresholds might simply be something likely or probable, but not known. These thresholds can only ever be theoretical, and not specifically defined, but we need to make sure they are of a high enough standard.

This could be a problem – different people have different ideas about what constitutes justification or evidence, so these arbitrary thresholds can be disproportionate – for example, a person might accept a much lower standard of evidence for a particular preconceived belief, and a higher standard for anything that contradicts that belief.

How then can we determine what is suitable justification, or what amount is sufficient? Because of this difficulty, it would be necessary to establish a methodology that would define certain high standards of evidence, testing procedures and methods, error checking procedures, and assessment of outcomes. Fortunately, hundreds of years of painstaking work have contributed to creating such a methodology, which we now call the Scientific Method. As yet, no other methodology has been created that can achieve the same high standard of results.

Theists claim that there are things that can't be answered by science, and only theology can provide answers, however there is no methodology for testing the accuracy of their claims, and in reality, more often than not, their claims are shown to be false. An example would be the efficacy of prayer. Claims that intercessory prayer works are common, based on anecdotes, however studies show prayer to have no measurable benefit.1 Of course, the theist might argue that God could be deliberately influencing the results so as not to give verifiable evidence of his intervention (to the detriment of the unfortunate patients), and so begins the descent into unfalsifiable hypotheses. We could just as easily say that the results of any inconclusive study might have been tampered with by pixies, but we are looking for an honest assessment of facts, not speculative fantasy. Results are what matters.

What Lies Outside the Ideal Line - Irrational Beliefs and Delusions

If a person has a belief that doesn't lie on, or at least very close to, the rational ideal line, it is an irrational belief. In other words, it would be a belief that doesn’t correspond to the required amount of justification.

Those beliefs that completely contradict the available justification are highly irrational beliefs, or delusions.

These terms often cause offense, as people sometimes think it means they are crazy, hallucinating, or unable to grasp reality. But their definitions are the best descriptions of the person's belief. An irrational belief is one that is not sufficiently justified; a delusion is a high level of belief in a proposition of which the opposite is justified. As shown on the diagram, a delusion is a belief of a high certainty where the evidence is contradictory.

Some of the best examples of delusional beliefs are the belief in creationism in the face of overwhelming contradictory evidence, and the rejection of the Theory of Evolution where there is overwhelming evidence to support it.

Belief or Knowledge?

In Agnostics, Get Off That Fence!, I used two scales comparing the different degrees of belief and knowledge - there were two because they are not mutually exclusive, even though they usually overlap. The diagram above shows the rational relationship between justification and certainty, with high certainty being knowledge. So where does belief come into it?

We might use a similar second chart to plot the relationship between justification and belief. However, the chart for belief becomes redundant, because if we were to look at a proposition for which we would attribute a low degree of certainty, why would we attribute a higher degree of belief? The degree of belief should have the same relationship to justification that knowledge has.

In other words, a thing wouldn't be believed unless it was also known, and for those with less certainty, we would say that it’s probable, but we neither know nor believe it.

Any point on the chart could be considered a belief (in a non-absolute sense), but because of the ideal relationship between certainty and justification, if a thing is known, there is no need to say that it's a belief. If this is the case, "belief" would only refer to those things that are believed but not known - this would lie outside the rational line, so would be irrational. Thus anything now referred to merely as a belief would be an irrational belief.

Burden of Proof

The diagram gives us an idea of what can be called knowledge, but what about a proposition for which it seems that there is no justification? As long as we make no knowledge claim (thus taking the "I don't know for sure" position), there is no burden of proof. Take for example the proposition that there is a bigfoot. Since there is no credible evidence of its existence, we can't claim knowledge that it does, but nor can we claim knowledge that it doesn't exist. But are we completely without justification? Since we know that most of the world has been completely explored and there is very little chance that it could exist without being discovered, and we understand human fallibility, gullibility and propensity for hallucination, we do at least have some justification to suggest that the bigfoot is a myth. So, even though we can't claim knowledge that it doesn't exist, we can still lean towards its inexistence, rather than being completely neutral. As soon as we claim knowledge either way, we bear the burden of proof, i.e. we require significant justification.

The crucial point is that the amount of justification must correspond to the degree of certainty claimed.

Proving the Negative

One of the arguments often heard, particularly in regard to religious belief, is "prove there's no God", and the atheist will point out that the burden of proof is on the person making the claim that God exists. This is a reasonable request - for if the person is making a claim of certainty that a god exists, it would be necessary to have justification for the claim. If a person was saying with certainty that there's no god of any kind, they would certainly need to provide justification for this claim. However, most atheists are partially agnostic, that is they are not claiming with absolute certainty that there's no god. There are also different depictions of possible gods that might be considered, as I explained in my previous article.

Rational Ideal AB

It would be easiest to show this as examples on the chart (right). If we look at position A - This represents a non-specific deity; a creative force; a first cause - a being of which characteristics are unknown, and is (and perhaps always will be) a philosophical speculation. This kind of god cannot be proven or disproven, and while there may be some philosophical arguments to justify it, there are still some good criticisms of those arguments. I would consider this kind of god possible, although unlikely based on what we know, so I would place its level of certainty slightly below zero. Now look at position B - this would represent a specific god with determined characteristics, whose existence depends on the credibility of the source material which presupposes it - the bible. The source material is flawed: it contradicts scientific evidence, it’s immoral, it contains contradictions, and these are just a few of its problems – all very good reasons to believe it's a myth. So I would say this God is extremely unlikely to be possible, and I would place its certainty at a very low negative. I wouldn't say it's absolutely certain to be false, but certain enough to call knowledge in the practical sense.

As I explained in the previous article, the terms agnostic and atheist are umbrella terms, covering any and all possibilities, so to assign a higher degree of certainty to god A would mean the term agnostic atheist is appropriate, as it would be conceding the possibility of this kind of god, even though considering type B to be almost impossible.

At no point have I claimed absolute certainty that there is no god of any kind, and justification for the certainty that I have assigned to these different types can be provided. This is not a matter of belief, or faith, but a rational assessment of certainty based on available justification.

Theologian William Lane Craig has argued the following in his blog2

it is incumbent on the atheist to prove that if God existed, He would provide more evidence of His existence than what we have.

This might be what is happening here – God could be deliberately hiding. Of course, the same argument applies to the existence of fairies or leprechauns. The burden of proof cannot lie with the person who rejects a claim because of insufficient evidence, but lies with the person who claims that a thing is true, regardless of whether or not the attempt to prove or disprove it is futile.

Shorthand Expressions

Even though a skeptic wouldn't make the claim of knowledge, there are still times when we say "I believe" or "I know" a thing to be true. The reason for this is that there is a lack of words for conveying the various degrees of certainty, at least not without resorting to longer sentences, so we tend to deal only in absolutes. A rational belief is one that we could also call knowledge, so the use of either word doesn't convey the meaning of the other, i.e. it gives the false impression of being either belief, or knowledge, without being both.

But brevity is needed, and because of this lack of terminology, we use the absolute terms belief or knowledge to convey something that could fall short of absolute. We can't always say, in a conversation about the solar system, "I don't technically believe with absolute certainty that the Earth orbits the Sun, but assign around 99.9% certainty to the proposition, based on the evidence"! We would simply say "We know the Earth orbits the Sun". This would be acceptable as a shorthand expression, as it means we assign a high certainty to it, not necessarily meaning that we have absolute knowledge of it. Another shorthand expression would be “God doesn’t exist” (again, capital G), which might mean a claim of absolute certainty, but is not always intended to be. Because of the limited choice of words, confusion is quite common.

We might propose new words to refer to these skeptical versions of knowledge or belief. However, as I have shown, a rational mind must unite both belief and knowledge as a "working" concept - the absolutes of belief and knowledge remaining unattainable ideals, but overlapping in the sense that a person only believes and knows when a high degree of certainty is justified. In other words, belief and knowledge would ideally be one and the same, so “belief” can almost be discarded completely.

The word "belief", however, is also used to refer to a person's own assessment, for example, one might say that a thing is known because of empirical evidence objectively proves it, and it is believed because one can make an assessment of the evidence and come to the conclusion. This is a subtle difference, but exists nonetheless. These nuances are not obvious though, and difficult to convey in every day conversation.

I have used the term "knowledge in a practical sense" (relative knowledge is also used), but I welcome any suggestions for a better term, as well as for the two different senses of belief.

Appealing to Intuition, or Subjective Experience

One more thing that I haven't gone into a lot of detail here is the theists' flip side of the coin – subjective experience. Many theists would argue that there is more than just empirical evidence that proves their faith. The system that I have described is only a rational, objective system of gauging certainty. I will go into more detail in a later article about the flaws of subjective experience.


There is a good deal of explanation that I have had to leave out to keep this concise, and I hope to expand on those soon in follow-up articles. This should still give the basis for my skeptics’ epistemology, and it shows why many arguments used by theists (for example, "a disbelief in god = a belief there's no god") are plainly false.

Many of us don't claim with absolute certainty that there's no god, or that we have to sit on the fence, or that we have to reject the possibility of a first cause, and I think my diagram should give anyone good grounds for refuting any of these generalisations. What we are trying to achieve is as honest an assessment of evidence as is possible, and there is nothing wrong with withdrawing certainty when it is not justified - it would be the truly honest position to take. We may say that something is known in a practical sense when the evidence supports a high degree of certainty, and if we fall short of this mark, we need only have a relevant amount of evidence proportional to the degree of certainty that is representative of the truth.

The truth, after all, is what is most important.

Add comment


Andrew Graham's picture

Edit - I decided to make a couple of small changes in the graphics - I had on the horizontal axis more justification and less justification, but this is incorrect. It should have been positive justification and negative justification, because "less" is towards to middle, the centre being zero justification.

The reason this is important is that "less" to the left would imply that less justification means "proving the negative", which is not correct. To prove the negative would require justification itself, not the lack of justification of the positive. I think people were understanding the point well enough, but it was bugging me! So it has now been fixed.