Sunday, 5 September 2010

K-scepticism and c-scepticism (part 1 of, probably, 3)

This post started of as a brief post on why I didn't like spelling "scepticism" with a "k". I thought, in fact I still think, that those who call themselves skeptics pay far too much (that is to say "some") deference to authoritarian views of epistemology. To say why, though, I found I had to spell out x and in spelling out x I needed to address y and so on. The post grew into a summary of all that I currently think in this area. The post will, to bring it back down to manageable proportions, be split into three. This first adresses the traditional view of knowledge. The second further criticises the traditional view of knowledge. The third develops Popper's ideas of conjectural knowledge and seeks to promote its advantages.


Since the time of the Ancient Greeks it has been held that knowledge must have a warranting basis. The basis differs depending on the analyst. There are claims of knowledge from authority, by divine revelation, by derivation from self-evident ideas, clear and distinct ideas, a reliable methodology, evidence, proper function and others. Some are held to stand alone, some in combination with others but all are held by someone to be necessary for knowledge and, in combination with truth and belief, sufficient for knowledge.

“Scepticism” has been used to denote many attitudes of denial. Sometimes it is used to denote denial of specific things, denial of the external world, denial of logic or morals or rationality. More globally it denotes denial of knowledge which, in its milder form, denotes denial of knowledge unless certain conditions are met. Here the sceptic takes a position of being resistant to belief. The sceptic will not allow as known, and resolves as a consequence not to believe, propositions with insufficient evidence or that have not been derived from a reliable methodology, received from an authority, or the like. This is the scepticism currently popular. In particular there is a desire for science and evidence, the sceptic will reject beliefs not derived from a scientific investigation or not backed by sufficient evidence. Once the proposition is adequately supported by scientific evidence, this sceptic will likely invert his attitude to the proposition and insist that others put aside their doubts. As the spelling of sceptic as “sketpic”, with a “k”, is also fashionable, I will use the term k-sceptic.

Another scepticism is a little more extreme. The other scepticism demands neither warrant nor warranting basis, this scepticism denies the possibility of a warranting basis. I will use the term c-sceptic. It follows from the denial of warranting basis that the c-sceptic must deny knowledge, as traditionally thought of. The c-sceptic uses the word “knowledge” all the time, though. The c-sceptic is quite happy to claim knowledge, allow that others know, even to claim that he knows that others know. If he is not to be ridiculously contradictory in his beliefs the c-sceptic must mean something different from warranted, true, belief when talking of knowledge. I will refer to this knowledge as c-knowledge. I will refer to the traditional concept of knowledge as k-knowledge.

I hope to convince the reader that k-skepticism is wrong, c-scepticism is right, that k-knowledge is unobtainable and its pursuit counter-productive. It draws, heavily, on the works of Karl Popper, his collaborators and critics. David Miller, Popper's research assistant and, later, collaborator and friend, may find the denial of authority of science and the treatment of the usefulness of warrant eerily familiar, should he chance upon the post.


One of the Ancient Greeks’ concerns was to draw a distinction between episteme (knowledge) and doxa (belief). Yes, yes, you believe that the earth was created around 6,000 to 10,000 years ago, that the stars are a guide to our destiny and that Liverpool are a good football team but do you know?

Things that are rubbish are not known. You cannot know that the earth was created around 6,000 to 10,000 years ago in six days of twenty four hours because the earth was not created then. You cannot know that the stars are a guide to our destiny because the stars give no clue as to what awaits us. You cannot know that Liverpool are a good football team because they are a bunch of red, koppite, gobshites. The Greeks (and almost everybody) would say something has to be true to be known.

It is important to note that this “truth” is not truth, full and precise, but “reasonable truth”. “The earth is round”, for example, is not true. The earth is slightly fatter around the equator making it an “oblate spheroid”. It is reasonably true. The differences between a round earth and an oblate spheroid earth are so small and unimportant that the two can be said to be very similar, or “much the same”. One would have to be either extremely pedantic or working in a very specialised area to object to anyone saying that the earth is round. However, if we take it seriously that both are true we can end up in serious difficulties. There will be a point, call it “a”, which “round earth” says is on the face of the earth but “oblate spheroid earth” says is not on the face of the earth. Call the assertion that point a is on the face of the earth “p”. If we take “round earth” as fully and precisely true we must assent to p. If we hold “oblate spheroid earth” as fully and precisely true we must assent to the negation of p: ¬p (pronounced "not p"). If we hold both “round earth” and “oblate spheroid earth” to be fully and precisely true we must assent both to p and ¬p.

Now, think of something utterly ridiculous, say “lager is better than real ale”, “Led Zeppelin weren’t very good” or “Liverpool are a not a bunch of red, koppite, gobshites”, and call that q. Since p is true “p or q” must be true (“or” is taken to be “not both of these are false”). Now we have “p or q”, one of which cannot be false and, from ¬p, p is false. So q, anything you like, must be true. (This demonstration is taken from Popper, Conjectures and Refutations, p. 426 -427)

Of course the non-earth-shape-specialist and non-pedant is happy to accept and use the reasonably true proposition that the world is round. For our purposes the there is no need to make a distinction between the reasonably true proposition and the actually true proposition. More, for our purposes, there is no need to maintain the qualification of “reasonable”, no harm will come if we just call “the earth is round” “true”. Difficulties will only arise if we enter the specialist area where greater precision is needed, succumb to pedantry or look in depth at epistemology. Many of the arguments against k-knowledge, counter-intuitively, are bolstered by a slackening of the truth condition. In the philosophical literature the fine distinction between “reasonable truth” and “truth” is often blurred. As, counter-intuitively, many of the arguments against k-knowledge are bolstered by a slackening of the truth condition it is important in what follows not to blur the distinction.


So, to be known a statement has to be a “reasonable” approximation to the truth. But that is not enough for knowledge. Each team in a knockout competition may believe that they will win that competition. For one of them that belief will be true, yet they cannot be said to know. True (or reasonable approximation to true) belief does not make knowledge. We move from “what turns belief into knowledge” to “what turns reasonably true belief into knowledge”. This is warrant.

What warrant is fills many pages of many philosophical works. Its general character, however, is pretty clear; it acts kind of “stamp of approval” on the belief. As we are interested in the truth its primary function is to approvingly stamp that truth.

Consider telling the time and when it would be right to move from saying that one believed that the time was 8:55 to saying that one knew that the time was 8:55. A belief that the time is 8:55 is just that, a belief. A belief that the time is 8:55 when in fact it is 8:55 is a true belief. If one looked at a reliable clock, saw that it indicated 8:55, it was 8:55 and one believed that it was 8:55 then one may be said to know that it was 8:55.

The same cannot be said of the sofa. The sofa would look the same were it 8:50 or 9:00 and is thus not reliable. If it were 8:55 but one guessed 8:50 a look at the sofa would confirm 8:50. A look at the clock would deny 8:50 and simultaneously confirm 8:55. Providing the clock is reliable if it says 8:55 then 8:55 is true. It warrants the truth of 8:55 and thus, added to true belief, creates knowledge.

Thus a warrant for p has to be, at some level of reliability, of the form:

If warrant then p
If ¬p then not warrant
(Most logicians would consider “if w then p” to be identical to “if not p then not-w”)

So there we have a three-part elucidation of knowledge. For someone to know p:
1. One must believe p
2. p must be true
3. One must have a method of deciding whether p or ¬p such that, to a reasonable level of reliability, if p then you believe it and if not p then you do not
That is not quite the end of the matter. If your method is reliable then you would be foolish to not to follow it. This means that if your method says p you are going to believe p and p is going to be true. If your method says p then you know p. If your method says ¬p then you are going to believe ¬p and ¬p is going to be true. If your method says ¬p then you know ¬p. The first two conditions for knowledge are a consequence of the third and so knowledge boils down to the level of warrant you have for any assertion.

A basis

At a minimum the relationship between warrant and a proposition that it warrants is that the proposition can be inferred from the warrant (“if warrant then proposition”). This can be represented by a simple argument schema:


For example, we can illustrate the “knowing what time it is” example:

The clock says that it is 8:55
It is 8:55

Or take the example of whatever navigation mechanism a migratory bird uses to “calculate” its route:

Navigation mechanism

The bird is said to “know” the route to take by virtue of a navigation mechanism. The bird infers, all be it non-consciously, the route from the navigation mechanism. The route is inferable from the navigation mechanism, if the route were materially incorrect the navigation mechanism would not produce it and the navigation mechanism tends to produce routes that are correct.

However this is not enough for the k-sceptic. The k-sceptic wants our choice of propositions to be as a result of the warrant. Whether the navigation mechanism does reliably inform the bird or no the bird has no choice in the matter, it will follow the route. The “sofa” example also fits the schema above or, perhaps better, the example of an unreliable clock:

The unreliable clock says that it is 8:55
It is 8:55

“It is 8:55” is clearly not known, the unreliable clock does not give warrant for “it is 8:55”. We should, according to the k-sceptic, reject “it is 8:55”. Where a reliable clock says 8:55 we do have warrant and, according to the k-scpetic we should accept “it is 8:55”.

If we are to follow the k-skeptic’s stipulation some way of distinguishing between the non-warrant and the warrant must be available. To do the job this way must, at some level of security, give assent to “this is warrant” when “this is warrant” is true and reject “this is warrant” when “this is warrant” is false. That is to say it would be warrant for our warrant:

The clock is accurate
“The clock says it is 8:55” is warrant for “it is 8:55”

Or we could show it as a stepped schema, with each line warranting the one below it.

The clock is accurate
The clock says it is 8:55
It is 8:55

But of course “warrant for warrant” is, itself, warrant. If warrant is required to be warranted then “the clock is accurate” must be warranted. Perhaps we have, or someone else has, tested the clock and concluded that the clock is accurate:

The clock has been tested
The clock is accurate
The clock says it is 8:55
It is 8:55

It is pretty clear where this is going: each warrant for warrant itself requires warrant, resulting in an infinite series of warrants that can never be completed.

Perhaps you think that I am being too strict. Certainty requires an infinite regress but, just as we will accept something that is reasonably true rather than true simpliciter, we will accept a certain level of uncertainty. Unfortunately any regress combined with any level of uncertainty makes the situation worse. If we a 0.95 confident that a working clock will tell the correct time and we are 0.95 confident that this is a working clock then the chances of us having a working clock that has just told the right time is 0.95 x 0.95 = 0.9025. If we are 0.95 sure that the test we use does confer a 0.95 chance of the clock working then the chances slip back still, 0.95 x 0.9025 = 0.87375. The more levels of warrant that we add the more the chances of the end result reduce until, fairly soon, the difference between a warranted proposition and a complete guess are negligible. Warrant that fails to raise the probability that a proposition is true is no warrant at all.

Thus if we are to secure warrant a basis is needed to stop the regress. A set of secure propositions is needed to provide the initial warrant which can be carried to all further propositions. Some contenders for this basis are listed above: authority, divine revelation, self evidence, clear and distinct ideas, a reliable methodology, evidence, proper function. But which of these bases, or combinations of these bases, are to form the basis?

The prospective bases often differ markedly in the propositions they certify. The pronouncements of authority, for example, differ so radically from some of the pronouncements of evidence and the scientific method that if one were true the other, far from being reasonably true, could not be remotely true. We must, therefore, choose a basis and this creates a dilemma. If we try and justify our choice of basis we do not treat it as a basis, we claim some warrant for the choice which creates a new recess. On the other hand if we do not offer any justification then our choice is entirely uninformed and arbitrary.

Alternatively, we could follow the fundies in their (perhaps not favourite, but extremely annoying) argument that “the Bible is a revelation from God because the Bible says it is a revelation from God”. What do we say to that? Really that they have given us no argument at all, just restated a conclusion as a premise. A third way fails and we find ourselves in Hans Albert’s “Münchhaussen trilemma”. All searches for an ultimate basis of warrant must “end” in:

1. infinite regress
2. circularity, or
3. an arbitrary exemption from the need for warrant
(Hans Albert. (trans. Mary Rorty) 1985. Treatise on Critical Reason. Princeton, New Jersey. Princeton University Press p. 18 or Wikipedia)

The k-sceptic, who insists on a basis for propositions, insists on something he will never have. There is no basis of knowledge, there is no primeval warrant and thus no warrant. There is no k-knowledge.

I sense a face-palm. “If we cannot know anything then we are all doomed” is the objection. Another is the enquiry as to whether, if I do not know that exiting my house from the front door is safe whilst exiting from the bedroom window is not, I do not jump out of the bedroom window every now and again. I will attempt an answer to the second objection later. First we must consider “if we
cannot know then we are all doomed”.

The first response is that, from the above, we cannot know anything. So, if we cannot know anything then we are all doomed, then we are doomed. There is little point in bewailing the fact and even less pretending we know just because we do not like the consequences of not knowing. We are better off finding ways to cope with our damnation.

There is another response, though, one a little more palatable but rather counterintuitive. “If we cannot know anything then we are all doomed” is false and a prime reason why it is false is because knowledge is useless.

That is the topic for the next part.