Alexandra Zinke has been appointed JProf. at the University of Tübingen where she has successfully applied for a DFG-grant to further pursue this reasearch project.

Abstract

We all know that there are many things we don’t know. But even more: with regard to many questions, we have not even formed an opinion. Some questions we haven’t considered yet, others are still under investigation, and yet others seem to allow for no definite answer. We often suspend belief and, more importantly, we do so rationally. Suspension of belief is no non-committal fallback position, but one of the driving forces of science. As long as we have not settled on an opinion, as long as we suspend belief, we have an open mind and keep on investigating.

Despite its core role in our doxastic everyday life, in scientific inquiry and in philosophical theorizing, suspension of belief has not received much attention in the epistemological literature. Current research focuses on the attitudes of belief and disbelief, while the third attitude, suspension of belief, is often totally eclipsed or referred to merely negatively as the absence of both belief and disbelief. As yet there is no systematic treatment of suspension of belief. This is surprising as we actually suspend belief in regard to most questions, and, furthermore, suspension often seems to be the only rational doxastic attitude one can adopt. The project sets out to fill this lacuna and provide an account of suspension of belief. It aims to distinguish and classify different forms of suspending belief, tries to provide formal representations of the different forms of suspension, discusses the impact of pluralism about suspension of belief on decision theory, and applies the findings to various philosophical discussions, such as peer disagreement, religious agnosticism, and Pyrrhonian skepticism.

Outline

Within the doxastic taxonomy of traditional epistemology suspension of judgment is analyzed as the absence of both belief and disbelief: S suspends belief in p if and only if S does neither believe that p nor believe that non-p. Traditional epistemology thus only distinguishes three doxastic attitudes: belief, disbelief, and suspension. However, belief is no yes-or-no affair. Some things we (dis-)believe stronger than others. This idea prominently figures in formal epistemology, which comes with a much more fine-grained terminology by introducing the notion of degrees of belief. There are competing formal representations of degrees of belief, e.g. subjective probability theory and ranking theory, coming with different rationality constraints on the doxastic states of the agent. Consider exemplarily the most prominent proposal to model our doxastic states, viz. subjective probability theory, which distinguishes continuum-many different attitudes a subject may have towards a proposition p: the degree of belief, or credence (cr), a subject has in a proposition p is given by some member of [0,1]. The degrees of belief of a rational agent are then supposed to obey the laws of probability.

Standard subjective probability theory models complete ignorance regarding p with a credence of 0.5 for p. This is required by the Principle of Indifference, which says that the ideal doxastic agent has to equally distribute her subjective probabilities over the possibility space, i.e. over p and non-p. If a subject fully suspends judgment on p, both her degree of belief in p and her degree of belief in non-p equal 0.5. This condition on suspension is relaxed in the middling-credence approach, which says that S suspends judgment on p iff cr_S(p) 9[0+t, 1-t] (for some 0<t≤0,5 ). It can easily be seen that the classical and the more general middling-credence approach are not very promising. Firstly, they inherit a problem akin to the Lottery Paradox, the most pressing problem for the classical suggestion, viz. the Lockean Thesis, of translating credences into full belief. Secondly, it has been plausibly argued that this approach is unable to distinguish ignorance from knowledge about chances. Another attempt tries to model suspensions of belief with interval-valued credences, i.e. not with a single probability function, but with a set of probability functions. The size of the interval may then reflect grades of suspension. However, also this account seems to falter: it struggles with the so-called dilation problem, and, so I will argue, it is still unable to distinguish different kinds of suspension of judgment. It seems that so far no convincing formal representation of suspension of judgment has been put forward.

I do not want to set out for yet another proposal how to represent suspension of judgment within subjective probability theory. Rather, my working hypothesis is that there are several forms of suspension of judgment for which we cannot give a uniform treatment. There is not one correct formal representation of suspending belief, but there are different ‘kinds of’ suspending belief, or different grounds for suspension, each governed by its own rationality norms. Perhaps, therefore, we need not even fully reject the above-mentioned proposals of formally representing suspension of judgment, but only restrict their respective scopes. They may adequately model one ‘kind’ of suspension.

This hypothesis is based on the observation that different subjects who suspend judgment on p may have relevantly different doxastic attitudes towards p, demanding different formal encodings. To give just some examples: A subject may suspend judgment on p …

… in virtue of not having considered p, or even lacking the conceptual resources to consider p.
… in virtue of tending to believe p, i.e. expecting p, but not yet being convinced that p.
… in virtue of being in a state of committed neutrality, e.g. because its reasons for and against p are in balance.
… in virtue of fully believing that the objective chance of p is 0.5.
… in virtue of fully believing that p is a genuine borderline case (e.g. if p contains vague concepts).

In all these different cases, the subject neither believes nor disbelieves p and thus qualifies as suspending judgment on p. I further argue that different kinds of suspension will give rise to different rational actions. Which action is rational in a given situation does not only depend on whether a subject suspends belief or not, but on how exactly she suspends. Thus, if classical decision theory based on a belief/desire model is to deliver the correct results also for cases of suspension of judgment, different kinds of suspension must come with different formal representations.

The first part of the project will consist in finely distinguishing different forms of suspension of belief and their classification within the epistemic landscape. This will then prepare the second part in which I try to find adequate formal representations of the different notions of suspending belief. I think that this can only be successful if we fully embrace the plurality of suspension of belief, i.e. accept that there are different attitudes qualifying as suspension of belief, and do no longer opt for one correct formal representation of suspension.

A proper analysis of the attitude of suspension of belief must be part of any complete description of our doxastic states. Furthermore, it will prove invaluable also beyond the confines of theoretical formal epistemology. Let me give three examples for the fruitful application of accounts of suspension of belief. In the vastly growing literature on peer disagreement one proposed solution is the so-called Equal Weight View, saying that in case of peer disagreement, you should give equal weight to your opinion and the opinion of your peer. In one interpretation of the Equal Weight View, this means that you and your peer should both suspend judgment on the debated question. Obviously, the plausibility of this approach depends on the employed notion of suspension of judgment. Another field which will benefit from a fine-grained analysis of suspension of judgment is the discussion of religious agnosticism. With different notions of suspension of belief, we can distinguish, and discuss, different kinds of religious agnosticism. One strong form of religious agnosticism involves a higher-order belief that it is impossible to have a justified belief in the subject matter. It will be illuminating to see how this kind of suspension relates to the other, first-order kinds. As a last example, let me mention Ancient or Pyrrhonian scepticism, which seems to say that we should suspend judgment on any matter. Although there is an extensive debate whether the Pyrrhonian sceptic really seeks suspension of all matters, the question which kind of suspension is involved has received almost no attention. An account of different forms of suspending judgment, using the tools of formal epistemology, would increase our understanding of the Pyrrhonian position.