One of the three major views of the nature of epistemic justification, the coherence theory (or "coherentism") experienced a revival during the 1970s and 1980s after its near total eclipse earlier in the twentieth century. Although its origins can be traced to idealists, including Francis Bradley, Bernard Bosanquet, and Brand Blanshard, the coherence theory has more recently been espoused by empiricist-minded contemporary philosophers such as Wilfrid Sellars, Nicholas Rescher, Keith Lehrer, Gilbert Harman, and Laurence Bonjour. The coherence theory of justification stands as an alternative to both the more traditional foundations theory and the view called reliabilism. It should not be confused with a coherence theory of truth. A coherence theorist about justification can acknowledge a fact that cripples the coherence theory of truth, namely, that there are instances of coherent, hence justified, beliefs in falsehoods.
Although the details of different versions of the coherence theory vary widely, all versions share a positive thesis and a resulting negative claim. The coherence theory's positive thesis is that a belief is justified or warranted for a person to the degree that that belief coheres with the rest of that person's belief system. As a fabric derives its strength from the reciprocal ties and interconnections among its constitutive threads, so, for the coherentist, beliefs derive their justification from their interconnectedness with one's other beliefs. The negative claim endorsed by all coherentists is that foundationalism is in error when it asserts that some of our justified beliefs are privileged or basic—that is, their justification is at least partly independent of their connectedness with other held beliefs.
The coherentist's picture of mutual support or fit among our beliefs departs (to varying degrees) from the strictly linear image of justification that classical foundationalism endorses. For the foundationalist epistemic justification is transmitted to nonbasic beliefs, from those that are basic or foundational, along lines of inference and explanation. Inferred beliefs are justified by those from which they are inferred. For the coherentist the belief's justificatory status has less to do with the grounds on which a belief is based and more to do with the whole cluster of relations (of consistency, implication, probability, explanation, and the like) that more or less strongly fix that belief within the network of other held beliefs.
The exact nature of epistemic coherence, however, is very difficult to clarify, and disagreements occur even among coherentists. Some have argued that coherence is always and ultimately explanatory coherence, a question of whether a belief is a member of the best overall explanatory account accessible to an individual. Others claim that there are justificatory relations of comparative reasonableness of competing beliefs that reflect concerns wider than explanation alone, including measures of subjective probability and the relative informativeness of the proposition believed. Logical consistency seems to be a minimal necessary condition for maximal coherence, but some have argued that at least certain inconsistencies are unavoidable but do not so undermine coherence as to prevent beliefs from being justified. Speaking generally, coherence is a property of a belief system that is determined by the (various) connections of intelligibility among the elements of the system. Most agree that these include deductive, inductive, and abductive relations, as well as other explanatory and probabilistic connections. Some writers, especially pragmatists, are prepared to add relations such as the relative simplicity or the power of the explanations contained in one's belief system as contributors to overall coherence.
Motivation for the coherence theory comes most directly from finding foundationalism unworkable and believing as a consequence that some version of coherence must be correct. Another motivation comes from the observation that it seems apt and possible to ask about any belief what a person's reasons are for holding it. The theory also appears particularly compatible with the realization that all instances of epistemic justification are defeasible—that is, the justification of a given belief is always liable to undermining by other held beliefs, no matter how strong the initial grounds or evidential basis of the belief might be. Since undermining can come from any element of one's system that might be negatively relevant to a specific belief, it appears that complete epistemic justification, the kind necessary to support claims of knowledge, is sensitive to all of the connections among our beliefs, precisely as the coherence theorist urges. This argument for the coherence theory is not decisive, however, since foundationalists can freely admit that warrant is undermined by a lack of coherence while still rejecting the coherentist's positive claim that coherence is the source of all epistemic justification.
In addition to the unclarities surrounding measuring degrees of coherence, numerous objections have been offered to coherentism. Four have been particularly prominent.
The Circularity Objection
If there are no foundational beliefs that act as the ultimate source of epistemic justification, and if the lines of justification transmission are not infinitely long (which appears absurd given the finitude of our mental capacities), then the coherence theory seems forced to claim that justification can be ultimately but not viciously circular. It is not immediately clear how circularity of this sort is anything but vicious, no matter how wide the circle may be, even though some have argued that wideness of a justificatory circle immunizes against viciousness. But if A is the source of justification for B, how can B be the source of justification for A ? The coherentist can reply that the "source" of justification is the entire belief system. The linear model of justification on which the circularity objection is based may not be forceful against a more holistic construal of the relation. Taken as a holistic and higher-order relation constituted by lower-order reciprocal relations (at least some of which are asymmetric, such as "explaining" and "being explained by"), coherence might be able to avoid the problem of vicious circularity.
The Problem of Perceptual Beliefs
Certain simple and apparently immediate perceptual beliefs seem to be justified for us on the basis of the perceptual experience we currently are having rather than on any considerations about how that belief coheres with the rest of our belief system. Experience often seems to warrant beliefs that are anomalous—that is, do not cohere with already-held beliefs. In such cases we do not think that we are justified in rejecting the new belief on grounds of incoherence but often concede that revision of some previously held beliefs is appropriate. Coherentists have replied to this objection by arguing that the justification of even the most immediate perceptual belief requires that that belief cohere with our metabeliefs regarding how reliable or trustworthy we take our perceptual processes to be in the particular conditions. It is such metabeliefs that make it more reasonable to accept the anomalous perceptual experience than it is for us to conclude that we are hallucinating or have been deceived in some fashion. The introduction of metabeliefs into the explanation why immediate perceptual beliefs are often justified for us has struck many, however, as overintellectualizing our epistemic situation, as well as possibly reintroducing foundational principles into the theory of justification.
The Isolation Objection
This objection, closely related to the problem of perceptual beliefs, begins with the observation that coherence is a cognitively internal relation, relating belief to belief. But might not a thoroughly coherent system of beliefs nonetheless fail to be justified because they are not properly linked to the external perceptual circumstances? Would acceptance of a coherent fiction be justified if it were entirely the product of wishful thinking? The continual perceptual input we receive from the world must be assimilated into our belief system or else the justification for those beliefs will often suffer from undermining. The coherence theory seems too internalist to be a complete theory of epistemic justification, the objection concludes. Since coherence does not necessarily serve the epistemic goals of pursuing truth and avoiding error in our belief system, further constraints seem necessary if our notion of justification is to relate appropriately to knowledge. Coherentists respond in a number of ways to the isolation objection.
One alternative is to admit the objection's force and add a requirement that all justified systems include the belief that certain kinds of spontaneously occurring beliefs such as perceptual and memory beliefs are reliable or likely to be true. Demonstrating that this constraint is not an ad hoc amendment to coherentism is a difficult matter. A similar requirement applied to acceptances based on spontaneous wishful thinking would be obviously ad hoc and unacceptable. Some have suggested that metabeliefs about the trustworthiness of our perceptual beliefs in certain circumstances are not ad hoc and are important and legitimate members of our belief system, justified, as all beliefs are, through their coherence with our other beliefs. Whether such beliefs can be noncircularly defended, whether they constitute a sort of foundational belief, and whether they are realistically necessary for epistemic justification are each open matters.
The Inferential-Structure Objection
The foundationalist's traditional view—that whether one is epistemically justified in believing some proposition depends crucially upon the actual course of inference taken in arriving at a belief—is not easily relinquished. Coherence, however, is a relation determined only by the contents of beliefs and not by the order in which they have been inferred. Consequently, it appears possible that a series of beliefs inferred one from the other in a wholly fallacious manner might nevertheless cohere maximally with a background system of beliefs as long as there is another valid (but unused) course of inference that does connect them. This leads to the conclusion that, even if the coherence theory adequately captures the concept of epistemically justifiable beliefs relative to a system, it fails to explicate the notion of being justified in believing a proposition. Coherentists have responded to this challenge by relying once more on metabeliefs, claiming that when we infer A from B and B from C we also accept or believe that A follows from B, and not, for example, that C follows from A. Incorrect metabeliefs will, on some versions of coherentism, cause incoherence and loss of justification, keeping blatantly fallacious reasoning from ending in justified beliefs. This response, however, may generate an infinite regress of metabeliefs. Not all uses of inference schemes contain premises stating that the scheme is valid. One can infer B from A without first having to infer that B follows from A. Some coherentists answer this and other objections by admitting that their proposed conditions for coherence constitute ideals to which human knowers should aspire but seldom in actuality achieve. Debate over the merits of the coherence theory promises to continue unabated.
See also Blanshard, Brand; Bosanquet, Bernard; Bradley, Francis Herbert; Classical Foundationalism; Coherence Theory of Truth; Epistemology; Epistemology, History of; Harman, Gilbert; Lehrer, Keith; Reliabilism; Rescher, Nicholas; Sellars, Wilfrid.
Audi, R. "Foundationalism, Coherentism, and Epistemological Dogmatism." In Philosophical Perspectives, edited by J. Tomberlin, Vol. 2. Atascadero, CA: Ridgeview, 1988.
Bender, J. W., ed. The Current State of the Coherence Theory. Philosophical Studies, Vol. 44. Dordrecht: Kluwer Academic, 1989.
Bonjour, L. The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press, 1985.
Lehrer, K. Theory of Knowledge. Boulder, CO: Westview Press, 1990.
Plantinga, A. Warrant: The Current Debate. Oxford: Oxford University Press, 1993.
Pollock, J. Contemporary Theories of Knowledge. Totowa, NJ: Rowman and Littlefield, 1986.
John W. Bender (1996)