From Tim McGrew, The Foundations of Knowledge, chapter 1
A Rival Account of Justification and Truth
There is an alternative way to weave together a theory of truth and a concept of justification, and on this account certain arguments we will deal with in later chapters might well be forceless. Although I do not think such theories have much of a claim either to historical primacy or to intuitive appeal, it is worthwhile here to take a brief polemical detour into the realm of idealism, with its reliance on "coherence" for a theory of truth and justification, in order to prepare the ground for the ensuing discussion.
The clearest contemporary explication of a coherence concept of justification is due to Jonathan Dancy. The notion of coherence of a set of propositions, which figures so largely in this theory, involves not only mutual consistency but also the extent to which each proposition in the set is explained by the others. As explanations may be better or worse, the coherence of such a set may be increased by adding a new proposition which does something to strengthen the internal explanatory powers of the set as a whole.
On the coherence account, a proposition is true if and only if it is a member of a coherent set. This is criterial rather than definitional: it indicates that we are justified (perhaps) in counting a proposition true if adding it to our belief set increases the coherence of that set. But if this is the criterion, the coherentist seems to be faced with a severe problem in finding an adequate definition of ‘truth'. For may there not be rival coherent sets of propositions? And if so, which one is to count as the touchstone of truth?
Dancy's answer is that idealists have always insisted that their theory of truth be tied down to the real world by means of the deliverances of perception. "There is only one coherent set," says Dancy in explicating the idealist's case, "and . . . this set is distinguished from all rivals by being empirically grounded." Presumably he intends this to mean that there is only one relevant coherent set, only one that matters for the definition of truth. Dancy continues with a quotation from F. H. Bradley to support his exposition:
My experience is solid . . . so far as in short it is a system. My object is to have a world as comprehensive and coherent as possible, and, in order to attain this object, I have not only to reflect but perpetually to have recourse to the materials of sense. I must go to this source both to verify the matter which is old and also to increase it by what is new. And in this way I must depend upon the judgements of perception.
Much might be said regarding this theory of truth, but our present concern is with what happens when we combine it with the coherence notion of justification which, as Dancy points out, bears a close affinity to the coherence definition of truth. As Dancy explains,
each belief is to be evaluated by appeal to the role it plays in the belief‑set. If the coherence of the set would be increased by abandoning the belief and perhaps by replacing it by its opposite, the belief is not justified. If the set is more coherent with this belief as a member rather than with any alternative, the belief is justified.
The link between justification and truth here is that the increasing coherence of a belief‑set is, by definition, an increase of the justification of its members, and by definition the unique coherent set of propositions which account for the data of experience is the sum of all truth. As one's belief set increases in coherence, one has better and better reason to think one's beliefs true. Indeed, as Dancy points out, it is difficult to distinguish between the coherence criterion of truth and a coherence definition of justification.
The reason this is so appealing is that without the coherence account of truth the coherence theory of justification is not obviously truth‑directed. Why should an increase in the mutual explanatory relations of my beliefs be taken as a sign, even a probable sign, that my beliefs are true? On the minimal correspondence theory of truth sketched earlier, there is no obvious answer to this question.
But Dancy's marriage of coherence theories of truth and justification is not without problems. In the first place, the introduction of observational control on the relevant coherent set of beliefs seems to turn this into a version of moderate foundationalism. The aim of thought and inquiry on this view is
to start from the data of experience and to construct a set of beliefs around those data which will order the data in the most systematic (coherent) way. To do this we may need to reject some of the data, but we cannot reject them all because our very aim is to make sense of what we have as data. So the set of beliefs which we do construct must be empirically grounded, and this grounding in the data of experience guarantees that there will be only one set which constitutes 'the most systematic ordering.'
Again, waiving questions as to whether this really does define ‘truth' in a way that is unique, and waiving as well worries as to whether any definition which links truth so tightly to subjective perceptual data can retain the notion of truth as non‑epistemic, the description here of the human cognitive enterprise should warm the heart of any moderate foundationalist. The data of experience form a privileged set, and we cannot reject them all, or apparently even a very high proportion of them, since it is our goal to give an account of them. Other beliefs which help us to give a better explanatory account of them (inter alia) are justified in virtue of this fact. But in this case, the privileged status of the observational beliefs requires some account independent of their explanatory powers, an account of their favorable epistemic status which is separate from the notion of justification Dancy is wielding. The definition of justification as explanatory coherence has apparently led him to a foundationalist theory of epistemic justification, and that in turn requires a notion of justification that goes beyond the bounds of the coherence concept.
This is a consequence which Dancy rejects. It is an advantage of the coherence theory of justification, he claims, that it dispenses with the evidential asymmetries of foundationalism.
Each belief is assessed in the same way, by considering the effect of its presence on the coherence of the whole. So there are no restrictions on what can be appealed to in support of what. The test, as Bradley says, is system and not any one‑directional criterion of fitting the evidence.
But seizing now upon Bradley's use of the term 'system' and ignoring his emphasis on the primacy of observation is rather opportunistic. Dancy continues to stress the systematic aspect of Bradley's position in his account of how a coherentist views a conflict or inconsistency within the web of belief:
Equally, in the event of a difficulty there are no antecedent requirements about where revision should be made. We have no independent reason to prefer to retain highly observational beliefs in preference to theoretical ones. The right revision is the one that results in the most coherent new whole, but we cannot tell in advance what sort of revision is most likely to achieve this. . . . In practice there are no taboos on what can be appealed to in support of what and no requirements about which sorts of statements should be retained in preference to others if there is a clash. . . . In the absence of fixed points and the lack of any clues about where revision should start, we know that at any time our belief‑set is merely provisional. Revisions will be called for, and the need to revise may occur anywhere.
This will indeed remove the danger of a collapse into foundationalism and the attendant demise of the coherence notion of justification, but the price is that the coherentist no longer has a right to appeal to observational grounding as a means for distinguishing the one relevant coherent set of propositions, membership in which constitutes truth. At best, selection of that set reflects an arbitrary preference for which no argument can be given. Once again, the coherentist is left with a set of beliefs which may very well, for all he can tell, be in splendid isolation from the way things really are.
Dancy attempts to cope with this criticism by characterizing sensory beliefs as having "antecedent security" which is not justification and which consists simply in our determination to retain them in the absence of any reason to reject them. Thus, although our taking the deliverances of the senses as data gives a genetic asymmetry to our belief set, all justification is still a matter of "subsequent security" and foundationalism is avoided. But either this determination is a mere attitude, or it reflects nothing more than the additional belief that such sensory beliefs are often true, or it expresses an independent justificatory status possessed by sensory beliefs apart from their coherence relations to other beliefs. The third option again amounts to a form of foundationalism; the first and second both leave the idealist without an answer to the charge of arbitrariness.
This discussion, with its prolific use of the term 'coherence,' may foster a terminological confusion. What is principally at stake here is a concept of justification, and having some such concept is a precondition for formulating a theory as to how justification accrues to particular beliefs. The coherence theory of epistemic justification, though historically related to what we have been discussing, is concerned with the latter question; it is compatible with the traditional notions of justification and truth and will be considered later. None of the foregoing considerations amounts to an argument for foundationalism over coherentism as a theory of epistemic justification; what the argument does show is that the idealist concept of justification as a coherence relation seems to require a coherence theory of truth, that a coherence theory of truth is only tenable if there is an empirical control on the relevant coherent set which is definitive of truth, and that introducing such a control introduces a foundationalistic structure to one's theory of knowledge which is incompatible with a coherence concept of justification.