I often get into disputes over epistemology. I don’t think I can hope to convince anyone of my position in a blog post, but I can try to clarify it a bit for further discussion. This is a hastily-written, rough first attempt to make it slightly clearer “where I’m coming from”.
My attempted clarification has two steps, which correspond roughly to two stages in my own epistemological development. The first involves the rejection of foundationalism.
Traditionally, from the time of Plato’s dialogue Thaeatetus to the twentieth century, philosophers understood knowledge as justified true belief. Different schools of thought differed on how to understand justification.
Typically, “rationalists” held that beliefs were justified by being based on a priori truths such as Descartes’ “clear and distinct ideas”, while “empiricists” held that beliefs were justified by being based on the deliverances of experience such as “sense data”. What these supposedly opposed schools of thought have in common is foundationalism.
By “foundationalism”, I mean the assumption that beliefs are justified by being “based on” a foundation of basic beliefs. According to this picture, knowledge is like an edifice: it rests on its firmest or most certain parts in much the same way as the conclusion of a valid deductive argument can be derived from its premises. So basic beliefs function like axioms in mathematics. As theorems can be derived from axioms by following the rules of inference, empirical knowledge is supposed to be derivable from basic beliefs following analogous rules of deduction or induction.
I reject foundationalism for many reasons. Perhaps the most straightforward of them is that I’m inclined to take natural science rather than mathematics as my “model” of human knowledge. In science, theories do not rest on foundations – rather, they are tested against observations, and indeed they are tested alongside one another. According to this alternative picture, knowledge is more like a spider’s web than an edifice: it does not rest on foundations, but instead is “anchored to the world” at various points where observations test it.
To develop this metaphor a little further, an individual’s “web of belief” is most abstract and uncertain near its centre, furthest away from where it is anchored by – i.e. checked against – observation. Its most concrete and secure parts are closest to these anchor points. This outer area consists of inter-subjectively verifiable observations and modest but widely shared theory. Because it is mostly directly observed or “sensed”, and because it is almost universally shared by users of the same language or “common”, we may as well call this area “common sense”. According to this understanding of common sense, observations consist of everyday truths we learn to recognize as we acquire our native language, such as “it’s raining”, “there’s a banana in this brown paper bag”, and “the needle points to the letter N”. Because these are mostly truths that are obvious to anyone present whose sense organs are working properly, practically everyone who understands the language agrees to them. In addition to these directly observable truths, common sense consists of some other very widely-held truisms that are somewhat less directly observed, such as “the Earth is round”, “the poles are unusually cold”, and “many diseases are spread via infectious material”.
So much for this non-foundational image of human knowledge, which we might call “pragmatism”. What about the justification of beliefs in this picture?
The answer to this question is rather subtle, and I will try to answer it in two separate passes. The first pass is a quick-and-dirty answer: justification is coherence. A belief is justified if it is implied by and/or implies other beliefs in the web of belief. So the more smoothly a belief meshes with the rest of the system, the more justification the believer has in retaining it as part of that system.
Now you might wonder how could mere coherence with the rest of a belief system justify a belief? – The answer is that most of our beliefs are true. That’s hardly surprising, because most of them are about utterly mundane, everyday, obvious facts. We acquired them via innate biological processes whose reliability is more or less guaranteed – at least in normal use – by evolution. By cohering with a largely true belief system, any belief gains something in credibility. Truth being a matter of correspondence with the facts, we might say that most of the time coherence yields correspondence.
That’s the first pass, the quick-and-dirty answer to the traditional question of justification. But strictly speaking, it’s a mistaken or misleading answer. Why? – This traditional question of justification is a mistaken question. No matter who you are, or how misguided your opinions may be, your beliefs form a system, and any one of your beliefs coheres with the rest of the system. So all beliefs are justified, at least according to the lights of the believer whose system they belong to.
If we are forced to embrace the traditional epistemological language of justification, pragmatists like myself will be forced to say that beliefs are justified by coherence. But we have to abandon that language. This brings me to the second stage of my own epistemological development, and the second pass at answering the question above.
The second stage goes further than rejecting foundationalism – we have to reject epistemological internalism as well. By “internalism” I mean the assumption that knowledge consists of beliefs that – in addition to actually being true – enjoy a “guarantee” or assurance of truth that is internal to the mind of the believer. Instead, we have to consider the reliability of the external processes that give rise to the belief – in other words, we have to consider the links that connect the belief and the purported fact (i.e. state of affairs) that makes it true (if it is indeed true).
One “end” of such a link is usually outside the head of the believer. So the link bridges a gap between the external world and the mind of the believer. Such links can be causal, of course, but more generally they must be “informational” so that they connect items that co-vary in a lawlike way. In a word, the link must be reliable.
Turning our attention from the mind’s own “grounds” for belief to the reliability of the links between mind and world involves a sort of paradigm shift. Our attention switches from the “doctrinal” matter of “how to refute the (radical) sceptic”, to the “conceptual” matter of how best to understand belief, information, cause, hypothesis, test, probability, truth, etc.
This paradigm shift opens a gulf between (A) the traditional epistemology of Plato and Descartes, and (B) a new “naturalized” epistemology, which often looks more like a branch of philosophy of science and mind. Although the question of internal justification is largely left behind, there remain questions about what we ought to believe or disbelieve. How can we distinguish denial and scepticism? How does pseudo-science differ from the real thing?
Obviously, although much has changed, some beliefs remain more belief-worthy than others. But just appealing to the contents of the rest of the system any given belief belongs to does not provide any obvious ways of deciding which is better than which. The traditional appeal to the reasonableness of a belief would usually trace it back to its supposed foundations via arguments, including “inductive arguments”. The new approach sees far less “unreasonableness” in belief systems than the old approach, but accepts quite a lot of plain falsity. Arguments are overrated; the real poison is the effect of falsehoods on a belief system, and the damage done by unreliable processes of belief-formation. The unfortunate people who suffer from such unreliable processes tend to have many false beliefs, and the contagion spreads. New beliefs may be internally justified according to the lights of the believer, but they remain embedded in a bad system.
To the new paradigm of externalism and naturalized epistemology, questions about what ought to be believed are generally not settled through navel-gazing or introspection, but rather through discussion, experiment, observation, etc. – in other words, by social activities between believers.