Many epistemologists accept the mentalist version of internalism about rationality. In a slogan, this is the view that rationality supervenes on the mental states that the relevant thinker has at the relevant time.
Coherentism, as I shall understand it here, is the view that rationality requires nothing except that one’s mental states should cohere in certain ways: for it to be rational for you to have a certain belief (or other attitude) at a certain time is just for your having that belief (or attitude) at that time to be part of a system of mental states that meet all these rational requirements of coherence.
I shall argue here that given the right understanding of what “coherence” is, the only plausible form of internalism is equivalent to coherentism.
First let me give a slightly more precise statement of this mentalist-internalist view. According to this view, whether or not a belief or attitude of some other kind counts as rational or as irrational depends, not on the attitude’s relation to the external world, but on how the attitude relates to the mental states that are present in the relevant thinker’s mind at that time, or just before that time. (Strictly, a complete account would have to take account of mental events, including events in which we form or revise our attitudes in various ways, as well as enduring mental states and attitudes; but for simplicity’s sake, I shall forget this complication here.)
I shall assume here that in addition to assessing the rationality or irrationality of individual beliefs or attitudes, we can also assess whole systems of beliefs and attitudes as rational or as irrational; and the assessment of individual beliefs or attitudes must harmonize with the assessment of systems of beliefs and attitudes – in the sense that every individual belief in any rational system of beliefs is itself a rational belief, and every individual rational belief is part of a possible rational system of beliefs.
What exactly is “coherence”? There are certainly narrow understandings of “coherence” on which coherentism is not plausible. For example, some forms of coherentism in epistemology imply that the only requirements of rational belief are that one’s outright or full beliefs should cohere with each other – where whether or not these beliefs count as “cohering” with each other is determined purely by the propositional contents of those beliefs.
This narrow notion of coherence would not even include what formal epistemologists call “probabilistic coherence”, which is a feature of systems of partial degrees of belief, not just of sets of outright or full beliefs. To determine whether a system of partial degrees of belief is probabilistically coherent, we need to look not just at the propositional contents, but also at the degree of belief that the believer attaches to each of these propositional contents.
Moreover, it seems clear that it is not just full and partial beliefs that can cohere or fail to cohere. Many philosophers have suggested that to be rational, one’s choices or intentions must cohere both with one’s other choices or intentions, and also with one’s beliefs, and perhaps with one’s desires and preferences as well. (For example, Michael Bratman argues that rational intentions must be “means-end coherent, relative to one’s beliefs”; and many forms of decision theory require that one’s preferences must cohere in a certain way both with each other and with one’s partial beliefs or credences.)
If the requirements of rational choice can include requirements about how one’s choices or intentions should cohere with one’s beliefs, why shouldn’t the requirements of rational belief include requirements that one’s beliefs should cohere with other mental states besides beliefs? For example, the “foundherentist” view of Susan Haack, according to which rational beliefs must cohere with one’s experiences as well as with each other, should also be counted as a form of coherentism.
Admittedly, as Haack in effect pointed out, forms of coherentism that include experiences among the mental states and events that must cohere, if one’s current beliefs and other attitudes are to be rational, will have an affinity with foundationalism. While we can in principle assess merely possible systems of mental states as rational or irrational, we are particularly interested in assessing the possible systems that are available to a particular thinker at a particular time. Since it is not up to the thinker what experiences she now has, all the systems of mental states that are now available to her will involve her actual experiences – which are thus in a way being “held fixed”, like a kind of “foundation” for the different systems of beliefs and other attitudes that count as coherent to a greater or lesser degree. (Similarly, it is not now up to the thinker what mental states she had in the past; and so all the systems of mental states that are now available to the thinker will be systems that succeed the system of mental states that the thinker had in the past; in this way, the facts about the thinker’s past mental states are also being “held fixed” in a similar way.)
Still, even if this “foundherentist” view has an affinity with foundationalism, it still seems to be a kind of coherentism – since its fundamental claim is that to be rational, the thinker’s mental states (including her beliefs and experiences) must cohere in a certain way, and for a belief to be rational is for it to be part of an optimally coherent system of mental states of this kind.
Some other philosophers might insist that coherence is a strictly synchronic relation, which can only hold between mental states that are present in the thinker’s mind at the same time. But on the face of it, it seems that we can also make sense of one’s present mental states’ “cohering with” some of one’s past mental states. Perhaps a radical conversion event, in which one makes a radical shift from one set of beliefs to another completely different set of beliefs, could count as a failure of diachronic coherence. Intuitively, one’s current set of beliefs does not “fit together” with one’s earlier beliefs. So it seems that we can make sense of diachronic coherence just as much as synchronic coherence.
So what is distinctive of coherence, then, if it is not defined as a relation that holds only among beliefs, or only synchronically among the mental states that the thinker has at a single time?
I propose then that we should interpret “coherence” as follows: coherence is a relation that holds between mental states purely in virtue of the content and nature of those mental states – specifically, a relation that helps to make that set of those mental states rational.
Suppose that this is all that it means for a system of mental states to cohere. Now internalism about rationality implies that whether or not a thinker’s system of beliefs and other attitudes is rational depends on the mental states that are present in the relevant thinker’s mind at that time, or just before that time. It seems clear then that internalism implies all the requirements of rationality are requirements of coherence in the large sense of the term that I have just explained.
If the requirements of rationality supervene on mental states, and coherence is just a rationality-enhancing relation among mental states, then all requirements of rationality are requirements of coherence.