If scientific knowledge is based upon first principles, how can we come to know first principles? If we claim that 'all cows are ruminants' is a first principle, in that it described the essence of a cow, then we need to be able to develop it into a full explication of the essence of a cow (e.g. 'cows are herbivorous, hoofed animals...'). Dialectic allows us to do this.
The process of dialectic is:
1) Begin from existing and conflicting opinions of reputable sources
Since the opinions taken from (1) have been formed from some reasoning, it is plausible that each of them will have some truth in them, and that therefore the conclusion (4) will contain some truth.
Through taking step (2), we gain a clearer understanding both of the problems we must resolve, without which we cannot begin the process (akin, Aristotle says, to trying to untie yourself without first studying the knot). And through taking step (3) we can evaluate each opinion with this clearer understanding and so be in the best possible position to recognise and deduce the elements of truth, and weed out false aspects of the beliefs.
Objection: Dialectic undoubtedly gives us a coherent set of beliefs, but why should we assume that these accurately represent the objective truth? Particularly when we use dialectic to establish first principles, are we not just allowing our methodology to deceive us, when in fact we may deduce entirely false first principles and so build all of our knowledge on false foundations?
Reply: Though one may be able to make quite a large web of coherent beliefs that are entirely false, at some point you will inevitable find anomalies, and so will be forced to evaluate the entire web against those anomalies.
Objection reply: But if we are evaluating these anomalies in the context of their coherence with the web, and in particular with endoxa, they will simply discounted. To allow for radical new ideas to upset the web, and so to assure us that dialectic is able to get us closer to an objective truth, we must ignore the endoxa or be open to the idea that the endoxa may be entirely wrong, e.g. evolution upsetting ideas of creationism, anthrocentrism, etc.
Objection 2: Through dialectic, Aristotle wants to establish first principles. In Metaphysics, Aristotle says dialectic will also establish universal truths. But necessary truths are objective, and coherent truths can only be contingent, since we have Little reason, except that it hasn't yet happened, to suppose those belief will always be true.
Possible replies: That we have no better way of knowing of objective truth; that, as in (c) we have no choice in accepting certain beliefs in discourse
Can, then, metaphysics have any claim on truth? Or is it simply experiential deduction?
And my essay on this subject:
In his enquiries in metaphysics, the natural and social sciences and mathematics, Aristotle employed a method of acquiring knowledge called dialectic; he thought that dialectic was the only method available that could be said to establish knowledge as opposed to belief. This view predominated for over one and a half millennia, and has in the past few hundred years been further developed and extended in various different ways to provide many quite different understandings of what constitutes knowledge, and by what methods we can acquire knowledge, and even if we acquire knowledge at all. Given the epistemic ambiguity of dialectic itself, it is perhaps best to begin by studying its roots in Aristotle's writings, and to analyse its key problems there.
In Topics, Aristotle set himself the task of explaining and justifying dialectic as the only method of acquiring knowledge because he wanted to "find a line of enquiry whereby we shall be able to reason from opinions that are generally accepted" (Topics, 100a 19-20). In other words, he wanted to avoid situations where critics could doubt his premises and so undermine his arguments, and so sought to establish, in most of his works, first principles upon which his other fields of enquiry, from philosophy to botany, could be soundly and unobjectionably based. Once established, these principles could be employed in further dialectic, expanding our field of knowledge in a reasonable manner.
So what is dialectic? As a form of learning, it involves four stages: One begins with two or more conflicting opinions that come from reputable sources (that is to say from those whom we deem to be learned already); then, one deduces the problems posed by each opinion and the implications of these problems, and the problems and implications of the conflict between the two opinions; thirdly, one evaluates each opinion according to a set of agreed beliefs; and finally one can reconcile the differences between the two opinions, finding, according to Aristotle, the truth within the resultant unifying opinion.
Why take this approach? To begin with, Aristotle suggests, we are more justified in our certainty if we have good reason to believe the opinions we begin with, as opposed to taking two radical and unpopular views, since they will be based upon good rational or commonsense arguments. Though this may be the case, it does present a significant problem, since it doesn't allow for radical changes in our belief system; we might instinctively ask whether or not we would have taken Darwin or Einstein at all seriously if we were to only allow dialectic as a route to knowledge. This however slightly misses the point, since both Darwin and Einstein did base their work upon the ideas of others, and so developed their radical theories in the context of an acknowledged new belief system that itself was worked upon by reputable sources. In fact, it is common sense that we don't take a proposition seriously unless we have reason to believe that the proposer is at all likely to be correct. Quite how far we take this is unclear in Aristotle's writing, and could range from only accepting the work of already-eminent theorists (which begs the question as to how anyone could ever ascend to that position unless they were apprentices to eminent theorists) to accepting any opinions from anyone with a reasonable formal education.
Once we have the two conflicting reputable opinions, Aristotle says that there must be some truth in each opinion, else the sources would not hold them to be true, and so we must begin to find what truths can be found in each. This is best done by first isolating the key problems with each; Aristotle compared this to untying a knot, and pointed out that one could not begin to untangle the knot until one first appreciated that the knot exists, and further by appreciating the nature of the knot. By analysing the endoxa we not only find that seems unlikely about them, but also points in common between the differing views. From here, it is a matter of employing rational arguments to find the truth within the statements, almost as if in a dialogue between two people.
So it seems that we have some reason to trust that dialectic might provide us with relatively unobjectionable conclusions, but why employ this method at all, since no other academic disciplines use it? 'Philosophy', according to J. D. Evans, can be 'distinguished by its methods and focus of interest' (Evans, p.2). If its primary method is dialectic, then it is because philosophy is not just 'occupied with purely conceptual problems rather than substantive issues of fact' (Evans, p.3). Philosophers must appreciate ideas from a diverse range of disciplines that may not have a single method acceptable to each discipline that could unite them. For example, one might want to use substantive facts of biology to influence an argument over the nature of the mind; since these two discussions are methodologically incompatible, one needs a method that can unite them and cross over axiomatic bases. That method is dialectic.
But if this is true, then we must accept the assumption that all knowledge needn't reduce to the same axiomatic bases, which seems like a fairly troubling conclusion, especially given Aristotle's commitment to first principles. If we are to accept the idea of first principles, we must consider the possibility that there is at least a substantial number of principles that form more than a collection of priori and logical truths upon which all disciplines are based. We must do this if we accept the epistemological position known as foundationalism, according to which every belief is based upon some others, and those others are based on yet more fundamental beliefs, and that this pattern continues until one finds foundational beliefs that are based upon no other, i.e. they are self-evident or a priori truths. If we accept foundationalism, then we must accept that every truth can reduce to the foundational truths, and can be explained purely in terms of these foundational truths, in the same way that all of mathematics can be proved in terms of a few logical and observed truths. So in this case, why should we think of dialectic as a useful way of discovering foundational beliefs, let alone other less fundamental beliefs, if it doesn't work from a single axiomatic base, but rather covers all it encounters? Should we not, as Descartes proposed, begin from the foundations and work our way up the 'knowledge tree'?
Aristotle provides a clear answer to this, but one that confuses his epistemological position and brings into doubt his claims for first principles. He emphasises, through his justifications for dialectic, that what is important is that the beliefs we acquire cohere, hence dialectic is such a good method, since it finds incoherence in conflicting beliefs and resolves it, until, conceivably, we have one coherent set of knowledge that is the truth about everything. According to this view, there simple aren't any foundational beliefs, they're an illusion or misconception; those truths that are self-evident or a priori are simply not 'knowledge' at all, at least not in the sense that 'John is wearing a green hat' is knowledge. To say: 'a bachelor is an unmarried man' is not to express any knowledge about those two terms, but rather to express an appreciation of their nature, their definition. To say: '1+1=2', meanwhile, is simply to show that one appreciates the way in which language conceptualises logical truths about reality. To then say: 'John has one green hat and has bought a new green hat, and so has two green hats' is to express knowledge, based upon its coherence with one's observations of John's hat collection according to logical truths. Other supposedly 'foundational' beliefs may simply have very small 'webs' of coherence upon which they are based, and so appear foundational when compared to more complex beliefs, e.g. how trees grow.
So, if we take a coherentist epistemological stand, the role of dialectic is to work on beliefs until we eventually find entirely self-contained webs of truths. Of course, Aristotle still holds that we can use strong dialectic to find first principles, but in a sense strong dialectic only helps us to discover the logical, a priori and linguistic truths upon which all of our arguments are based.
There remains, however, a serious worry about this stand, for 'coherence within common beliefs does not seem to be a ground for claiming to have found objective principles' (Irwin, p.8). Aristotle can't ever be immune from scepticism, from the doubt that his whole web of knowledge is in fact wrong and that, if he were to reassess his first principles or find a major new coherent web that conflicted with his current web, he might find that all that he had once believed was untrue. This criticism is especially telling of Aristotle's dialectic, because he placed such a lot of value on intuition, upon which many endoxa are founded. Aristotle himself believed the world to be spherical, but had he not, one wonders what he might have said to Copernicus when he challenged one of the most fundamental endoxa of his time. Again this susceptibility to doubt is made worse by the possibility of radical and unpopular opinions being closer to the objective truth than received and accepted opinions. Aristotle might have given this more thought had he considered his example of the knot more closely, for what use is it knowing that there is a knot if one refuses to take seriously the suggestion that one has completely missed the most important threads in the tangle? To work with a coherntist epistemology, one must surely scrap Aristotle's first step in dialectic and allow for all views, endoxa or not, to be considered. This really oughtn't be a big problem for Aristotle, since if a radical view is really completely false and baseless, then it won't take long under rational analysis and in comparison to endoxa to determine that it is in fact totally false. If, on the other hand, it is true, or closer to the truth than any endoxa, then it ought to animate in a dialectical investigation sufficient problems and questions to make it worth serious consideration, and so the truth ought to come out.
In this way also we can forget the worry of epistemological reduction, because even if they do all reduce to certain axiomatic or logical bases, we needn't know those bases nor the relationship of all knowledge to those bases in order to make a claim to knowledge, although being able to do such a thing would certainly strengthen a case for coherence.
Furthermore, because dialectic concentrates more on conserving endoxa where appropriate and developing them to resolve incoherence, and is interdisciplinary, we are more likely to acquire knowledge of reality, since any incoherence with reality will immediately be considered as a problem. Other methods of knowledge acquisition, on the other hand, may take the inquirer down one route ignoring other evidence, and so may ignore apparent facts of reality altogether in order to develop a particular theory or paradigm. In this sense, though we cannot say that dialectic gives us absolutely objective knowledge of reality, it does seem to give us more knowledge of reality, and a better knowledge of reality, than other possible methods and other epistemological systems.
Aristotle, Metaphysics, in The Basic Works of Aristotle, ed. R. McKeon, 2001
Aristotle, Topics, in The Basic Works of Aristotle, ed. R. McKeon, 2001
J.D. Evans, Aristotle (Philosophers in Context), The Harvester Press, 1987
T.H. Irwin, Aristotle's First Principles, Clarendon Press, 1998