The coherence theory of knowledge asserts that epistemic justification can be realized by determining how well a new belief fits within the existing system. The theory relies on the inferability of one belief to another, along with how one belief some aspect of the external world, and the ability to gain access, based on the Doxastic Presumption, to a belief system that is approximately like the one that the believer actually possesses. There are three major objections to the coherence theory: the alternative coherent systems objection, the input objection, and the truth objection. Each of these objections will be answered, but I would first like to give attention to an addition to the theory, one that allows for observation to factor in to the ability for one’s coherent system to grow and change based on changes in the external world. This addition will help us answer the input objection, which I will explain a little later.
It would seem to be a contradiction to allow for observational input from the external world since observational data is necessarily non-inferential and the coherence theory requires that all justification be based on how well a belief coheres with the overall system (which is necessarily inferential in nature); however, there are two different ways in which a belief can be considered inferential or non-inferential. The first way is how the belief was arrived at, or how I came to believe such-and-such. In this sense, it seems fairly obvious that not all beliefs can be inferential, unless there is a vast system of a priori beliefs that we are born with, or we somehow have access to an infinite number of beliefs that we can reference in order to acquire any other belief. The second way a belief can be considered inferential or non-inferential is by how it is justified. A belief can (in theory) be inferentially justified in virtue of its relation with the other beliefs in the system, or it might be justified in some other, non-inferential way, such as some data from the external world. It is in the latter way that a belief must be considered inferential in order to remain a part of the framework for coherence theory. It is important to understand that the coherentist is not arguing that every belief must be inferential in every way. It would be rather absurd to say that a belief, especially an observational belief, must be inferential in origin in order for it to become a part of a coherent belief system. What the coherentist is pointing out is that, while it is quite possible that a belief is non-inferential in origin, it must have some sort of inferability with other beliefs in the system in order to be justified.
It is entirely conceivable for a proponent of the coherence theory to insist that observational beliefs be non-inferential in the manner in which it was arrived at, for this would allow for a belief justified by an appeal to coherence to act as an independent check on the system. Instead of relying on current beliefs to cause other beliefs, observation from the external world can be brought into the system to “check out” the trustworthiness of existing beliefs. This would actually be the ideal situation for the coherentist!
Imagine for a moment that I have come to hold a belief (that is not yet justified) that there is a cookie in my hand. I did not infer this belief from anything, rather it was cognitively spontaneous. This belief is a prime example of observational knowledge: I see the cookie in my hand, I feel it crumble in my palm, I smell the chocolate chips. This is the sort of belief that strikes me based on my senses. What we now need to consider is how this belief is to be justified. Lawrence BonJour offers a justificatory argument for cognitively spontaneous beliefs that I have modified slightly to fit our example:
(1) I have a cognitively spontaneous belief, let’s call it K, that there is a cookie in my hand.
(2) Conditions C obtain (this is to say that my current situation is one in which it is possible that there is a cookie in my hand).
(3) “Cognitively spontaneous visual [sensory] beliefs of kind K in conditions C are very likely to be true.”
Therefore my belief that there is a cookie in my hand is very likely to be true.
Therefore, there is probably a cookie in my hand.
Of course, as knowledgeable coherentists, we know that in order for my belief to be completely justified, each of the premises must be justified in turn, but we are well on our way to allowing input into our coherence framework without opposing our fundamental assertions. The interesting point of this justificatory argument is that my belief does not have its justification intrinsically (which is dangerously foundationalist). Instead its justification is dependent on the context in which the belief came into existence, how well it coheres with the other beliefs in my system, and its overall explanatory value. This is how the coherentist must account for all types of observational knowledge.
So far we have discussed instances of positive observational belief, but we must also consider negative observational belief. As I sit in my most comfortable and stylish antique chair, I come to have the belief that there is no beer in my hand. It’s not that I see the absence of beer in my hand, it’s simply a matter of failing to see its presence; I have no spontaneous sensory belief that there is a beer in my hand, even though the conditions are such that it is possible for there to be a beer in my hand. This is an interesting feature of spontaneous observational beliefs: not only are they most likely really there when I observe them, but they are also likely to be observed in situations where they can be observed. BonJour refers to this quality as converse reliability, or the fact that an object that causes some sort of observational belief can be held reliable in the negative sense, as well as the positive.
So, my observational beliefs can be considered epistemically reliable based on prior epistemic qualifications (my education on the reliability of such beliefs) that qualify such beliefs as ones that tend to be dependable. We now have a coherentist account for observation that has three major requirements: (1) There must be some grouping of observations that are discernible to the person that has them, (2) The grouping of observations must be epistemically reliable in both the positive and negative aspects, and (3) The believer must have access to the premises of the justification for the observations in question.
There is much more that could be said about this additional piece of the coherence theory, but we may proceed now with an understanding that it is possible in the aforementioned way, to account for observation in a coherence system of beliefs, and we are now able to move on to answering the three major objections mentioned earlier.
The input objection alleges that the coherence system, as a theory that seeks to avoid externalism, cannot allow for input from the external world. This suggests that any correspondence between my belief system and the external world is the result of either a miracle or an accident. We are, however, after our discussion of observation, now in a position to refute a large part of this objection. If the coherence account of observation is accurate, then there is no longer any lack of input from the external world with a system that relies only on the internal system to justify knowledge. If our account is true, and there are recognizable groups of observations that are likely to occur, and indeed do occur, then we have justifiable knowledge as a result of information gathered outside the internal system that allows it to grow and change based upon new information gathered from the external world.
There are however, some aspects of this objection that remain unanswered, even with an account for observation. First, some may wonder if cognitively spontaneous beliefs qualify as input in any more than a minimal sense, or, put another way, as input that can really count as a contribution from the external world. On the empirical level, operating from within the system (as a coherentist insists), this data does count as “real”, as it is input that is caused in a significant way by the external world.
The second, thus far unanswered, element of the input objection is the suggestion that it would be possible for one to construct what might be considered to be an epistemically justified system of beliefs without sufficient input from the external world. We can, however, stipulate that any justifiable coherent belief system must have sufficient input from the external world, thus eliminating the possibility for one to leave this important feature out of their belief system and still be considered epistemically responsible.
The alternative coherent systems objection suggests that there is a very distinct possibility that there are equally coherent belief systems that exist that would prevent us from accepting one belief system over the other. But with the ability to allow input from the external world through observation, this objection loses some of its strength. It simply doesn’t seem like a major problem as long as we are eventually able to figure out which belief system is the “right” one. As a matter of fact, this is exactly what occurs in the study of science when there are competing, seemingly equal, hypotheses; the scientist must figure out which one is correct through experimentation and observation. It is reasonable to take our seemingly equally coherent belief systems and test them through observation to see which one is correct, and that we can expect that the most accurate belief system will win out.
The final objection, the truth objection, may be the most difficult to answer to the satisfaction of a skeptic or foundationalist. This objection states that there is no reason to conclude that the coherence theory truth-conducive; that is, it offers system that gives us only true beliefs. Because one of the most important tasks of any epistemological theory is to lead us to truth, the inability to answer this objection would be devastating to the coherence theory.
A satisfactory answer to this objection is one that can show that, at the very least, the coherence theory of justification results in correspondence with the truth. The correspondence theory of truth holds that a belief is true if it agrees with the equivalent independent reality. For example, my belief that there is a cookie in my hand is true if, in fact, there is a cookie in my hand and my belief that there is not a beer in my hand is true if there really is no beer in my hand.
One thing that we must assume is that there actually is a real external world that our beliefs can match-up with. So we must subscribe to some form of realism.
Correspondence is the relationship between two things: the truth-bearer (the thing that is true) and the portion of reality that makes the truth-bearer true. There is some dispute over what tat truth-bearer actually is, but here we will go ahead and assume that the truth-bearer is a proposition. The second term has been said to be a “fact”, but it better serves us to consider that term the actual independent reality that makes the truth-bearer true; that is, the events, situations, etc. that make up the reality in which the proposition in question is true.
We can now move on to a standard skeptical objection that can be raised for almost any theory of knowledge. There is the possibility that although my belief system is coherent, corresponds with the world, and this thus an accurate measure of what I “know”, that my knowledge is not really knowledge at all, but an illusion. In this case, my justificatory system has failed and I do not know all that I thought I knew. This could theoretically occur as a result of a mental illness, an evil scientist who has fooled me, a brain in a vat, into thinking that I am a real person, or an evil demon who is causing the world to (falsely) correspond to my belief system. The Cartesian “evil demon” is probably the objection used most often, so we will answer to it specifically.
The evil demon in question is one whose mission is to cause the world to correspond to my belief system for the sole purpose of confusing me now and forevermore. This demon is steadfast, never wavering, always making sure that things are not as they should be, but are rather the way my belief system would have them be. In this situation, I would never know that I was being tricked; nevertheless, while my beliefs may be justified and entirely coherent, what I believe to be the truth is not the truth and my belief system is not actually truth-conducive.
Because there are so many ways that cognitively spontaneous beliefs could be caused by the world, it seems rather unlikely that they would be caused in such a way that they adhere to the strict principles of the correspondence theory, and as a result very a priori unlikely that the correspondence theory is true. The interesting point here is whether reasons can be found to believe that the correspondence theory is a priori less likely to be true than the evil demon hypothesis, or vice versa. There appear to be more evidence that supports the correspondence theory than the evil demon, not the least of which is biological evolution, which shows how cognitive beings with spontaneous beliefs could (and did) come to exist).
The whole point of a skeptical hypothesis such as the evil demon is to be equally compatible with any other system so that it is neither able to be refuted nor confirmed. This is not entirely possible with a theory that relies on the spatio-temporal world, which has a definite and orderly character, unlike the evil demon’s, whose causes and effects can be both defined and predicted. The world is not a neutral producer of beliefs like the evil demon, but instead causes beliefs based on its nature – in a definite and orderly fashion. This can be seen by looking at our beliefs, determining where they originated, and showing that they came about in an orderly way. In this case, the correspondence theory, which is framed by distinct patterns of causation, would be more plausible.
With solid answers for both the input and alternative coherent systems objections, and the makings of a reasonable answer for the skeptic on the truth objection, it seems that the coherence theory is quite an acceptable account of knowledge. The coherentist’s version of epistemic justification maintains an internalist framework while allowing for input from the external world, and seems to be the most truth-conducive of any epistemological theory offered thus far.