Sceptical arguments are the best place to start in presenting the Middle Way, because they enable a philosophical argument about its justification to be built up. To start by facing up to all our uncertainties, and then consider what positions we can justify in spite of them, is a pattern of presentation that has often been used in Western philosophy (for example, by Descartes and Hume, who both attempted to confront scepticism after it had re-arisen in Western civilisation). What this approach reflects is a concern with justification which I share with Descartes and Hume, even though of course I disagree with them in other ways.

A concern with justification is ultimately a practical concern. If we do not face up to the challenges of justification, we may remain deluded in ways that could have been avoided, and those delusions may well catch us out with practical consequences in the future. I want to argue that much Western thought has turned its back on this concern with justification, because of a set of interrelated unnecessary assumptions about it, and that this has had negative practical consequences.

So, I am going to begin with an account of a range of sceptical arguments. Many of these arguments are well known and are often taught on introductory philosophy courses. The first of them were introduced to Western Philosophy by Pyrrho of Elis, the founder of the Pyrrhonian school of scepticism in Hellenistic philosophy, but they are also found in Indian philosophy, which may have influenced Pyrrho. These arguments set up a basic challenge in Western philosophy that the most prominent philosophers have been struggling to address ever since.  What all of these arguments have in common is the casting of doubt on all claims of knowledge.

1. The ten modes of Pyrrhonism (first given by Aenesidemus[1]) give a range of reasons why our senses do not necessarily give us correct information about objects. These are
   1. that different animals have different sense organs, so therefore animals perceive objects differently from humans
   2. that different humans have different sense abilities (e.g. some have visual impairments) and thus perceive objects differently
   3. that different senses perceive objects differently (e.g. I may be able to hear something I cannot see)
   4. that differences in circumstances lead to different perceptions (e.g. a hand put in hot water and then cold will find the cold colder through contrast)
   5. that differences in spatial position relative to an observed object (e.g. a distant landmark) lead to different perceptions and to perceptions that may be mistaken
   6. that our perceptions of an object will be altered by what we see it with or near, which may lead us to see it differently (e.g. camouflage)
   7. that the same object will vary in the way it is perceived when in different quantities or when composed differently, making it impossible to identify the object with certainty (e.g. wheat grains look different from flour, but are composed of the same substance)
   8. that if objects are claimed to be absolutely existent this claim is still only understood relative to other claims
   9. that the constancy or rarity with which something appears changes our perception of it (e.g. comets are rarer, and thus seeing one is more significant to us, than stars)
   10. that moral claims also differ between people (one person’s good child is another’s bad).

In general, then, these arguments point out that all our perceptions are relative, because influenced both by the specific circumstances of our perception and of the object we are (or may be) perceiving. This means that any perception may be in error.

• The dream argument considered by Descartes[2] and others, suggests that we cannot tell with certainty that we are not dreaming (or that our whole experience is not otherwise illusory) at a given moment, and therefore that our perceptions are not erroneous. This argument is problematic if applied to all our experience through time, as it then deprives us of any contrast between ‘dream’ and ‘reality’, but we could consistently maintain this distinction to assert that at least some of our past experience must not have been a dream, and yet not be certain that our current experience is not.

• The error argument points out that even if our whole experience at a given time is not erroneous, particular objects that we think we perceive may still be so. Past mistakes in perception show that mistakes are possible, and we were not aware of those mistakes at the time we made them, so we may not be aware of our current mistakes. This argument can be applied to current perceptions and also to memory, to point out that with a past perception we may have made a mistake in the original perception or in our memory of that perception.

• The time lapse argument used by Bertrand Russell[3] suggests that we cannot be certain of the object of perception because the conditions of that object may have changed by the time we receive the perception (e.g. the sun may have ceased to exist 7 minutes ago, but due to the distance from the sun and the time it takes light to traverse that distance, we wouldn’t know about it yet).

• The relativity of cultural background. Earlier sceptical arguments acknowledged all the physical reasons for the relativity of perception, but more recent psychological and linguistic research tells us more about the mental reasons. Our cultural background may lead us to perceive objects differently: for example, perceivers of the Müller-Lyer Illusion (see figure) make a bigger misjudgement about the relative lengths of the lines if they are accustomed to environments with rectilinear architecture[4].

• Problem of Induction. All generalisations based on specific observations lack certainty, because the observations do not provide enough evidence to cover the possibly infinite number of instances referred to in the generalisation. For example, if I claim that all physical objects have mass and are subject to gravity, I have not checked all the physical objects in the universe to ensure this.

• The infinite regress of justification. There is no possible claim for which one could not ask for further justification (i.e. there are no self-evident claims). However, if a justification is offered, one could then ask for a justification of the justification, and so on ad infinitum.This argument works not only for empirical claims but for a priori ones. For consideration of possible self-evident claims which might be claimed to undermine this sceptical argument, see 1.b

• The relativity of linguistic categories. Even if we were able to overcome the above sceptical arguments in other respects, the language out of which we represent claims about objects in the universe does not have an absolute relationship with the objects themselves, either as they may exist in themselves or even as we experience them. We cannot be certain either that another person understands the same as we do by a particular proposition about the world, or even that we mean the same ourselves  when we return to our previous utterances after an interval of time[5]. Even if we were to weaken the requirement to one of identical representation of our experiences to ourselves after a few seconds, we cannot be sure that our mental representation of that experience has not changed, and thus that the language does not mean something different from what it meant to us beforehand. Claims of certainty depend on the absolute consistency of language used to represent those claims, otherwise any certainty that might apply to a statement at one instant will immediately be lost at the next instant, even for the person who made the statement.

• The vagueness of linguistic categories. Any possible representational term out of which a claim of certainty might be made is also inadequate for the representation of any reality (or even any experience) because of its vague relationship to that reality or experience. The terms used for representing objects (even abstract ones) are nouns, and any given noun is vague in terms of the scope of what it represents either in experience or the object of experience.  For example, if I use the word ‘pen’ to describe an object, and even if I give a precise and unique description of that pen, giving measurements and physical co-ordinates, what I am referring to is vague both in terms of space (some molecules or even smaller particles may not be clearly defined as part of the pen or not) and time (any interval of time I may specify for my statement about the pen will have duration, and during that duration the pen may change). If, on the other hand, I make no claims for the object which take up any space or time, my claims will be uninformative. It might be claimed that a priori claims such as those about numbers avoid this vagueness, but when applied to any claim about the universe these numbers depend on counting and measurement, which are unavoidably vague (see 1.f for more discussion of a priori claims).

Together these sceptical arguments provide a huge over-determination of the sceptical case. We do not need them all. Only one of them has to be successful to prove that there can be no certainty attached to any claim of knowledge. It is not surprising that philosophers have often been concerned with questioning the assumptions behind these sceptical argument rather than refuting them directly in their own terms. In their own terms they are unanswerable. However, as I will argue in the rest of section 1, it is the assumptions of those who attempt to undermine the sceptical approach that are unnecessary.   

Before I go on to defend the basis of these sceptical arguments further, and then develop an account of their implications, I should re-iterate that my purpose in doing this is ultimately practical. I am not defending scepticism in order to assert the relativity of all judgements. In fact, I think that sceptical arguments of the kinds listed above offer us the key to avoid relativism and assert that some judgements are better than others – but only if sceptical arguments are consistently and unflinchingly applied, and we do not try to dodge scepticism nor ever cease to take it seriously.

---

[1] Sextus Empiricus (1996)

[2] Descartes (1912)

[3] Russell (1940) p.13

[4] Segall et al (1963) pp. 769-771

[5] This is the scenario considered by Wittgenstein in his ‘Private Language Argument’ (Wittgenstein 1967 §258), which I discuss in Ellis 2001 pp.258-62. I argue that although Wittgenstein is correct in pointing out the lack of standards of correctness based on defeasibility in private language, this is no different from the situation with public language. Standards of correctness are not absolute in either case because meaning is not purely representational, but this does not deprive us of a degree of meaning (see volume 3 for much fuller discussion of meaning).