WRIT 5.04: What is Knowledge?

Uniqueness vs. Permissivism

Sometimes it is helpful to think about our cognition like a machine. Perhaps we are engineers making a machine that can tell if an object is round. We put a shape in; we get a roundness-rating out. Maybe the machine gives us a coarse identification: “this is definitively round,” or: “this is not round at all.” Or, perhaps, we get a confidence rating from 0 to 1, with a score of 0.5 indicating something not quite round but not quite square. If we put different objects in, it’s fair to assume that we will get different results from our machine. But what really makes an object “round”? Perhaps we encounter a group of competing engineers who have made another roundness-identifier. If we put the same object through both machines, it seems reasonable to expect different results from each.

In other words: is every disagreement irrational?

We can translate this case into the language of epistemology, the study of reasoning and knowledge. We “put into” the roundness-machine evidence regarding a particular proposition (“this object is round”); we “get out” the doxastic attitude we have toward that proposition (this doxastic attitude may be a coarse belief/disbelief, or variable credence). As engineers, we might subscribe to the thesis of uniqueness for the scope of our machine—that one body of evidence, one input object, should permit only one rational doxastic attitude towards whether it is round (Jackson, Turnbull forthcoming, 1). If uniqueness were true, our two teams could never rationally disagree—at least one machine would have to be interpreting the evidence wrong, or be biased in some irrational way. Does uniqueness hold true for every proposition we reason about in our day-to-day lives? In other words: is every disagreement irrational?

The denial of uniqueness is permissivism, the thesis that some bodies of evidence permit more than one rational doxastic attitude toward a particular proposition (Jackson, Turnbull forthcoming, 1). Permissivism allows for situations like our roundness disagreement to be considered epistemically acceptable. There is an important logical distinction between these two theses. Uniqueness is a universal claim, meaning that it applies to all propositions and all bodies of evidence (Jackson, Turnbull forthcoming, 1). Under permissivism, which is instead an existential claim, there still may exist some bodies of evidence which permit only one doxastic attitude (Jackson, Turnbull forthcoming, 2). Because of this, arguing permissivism is an offensive process: even if you rule that most cases of disagreement are irrational, all we must do is find a single rational case to disprove uniqueness.

​​In this paper, I will explain the presuppositions, strategies, and stakes of reasoning that vary between agents or across time. I will then defend the rationality of these factors’ influence on permissible doxastic attitudes to defend permissivism.

Uniqueness and Evidentialism

Another way to view the distinction between permissivism and uniqueness is in terms of “slack” (Kelly 2013, 1). Under uniqueness, there is no leeway between what evidence suggests and how reasonable it is to believe that suggestion, but the permissivist believes that there may be some wiggle room (Kelly 2013, 1). At first glance, permissivism seems like an argument against evidentialism, which is the thesis that justification is a direct function of evidence. Uniqueness does imply evidentialism; by uniqueness, your evidence makes rational only one doxastic attitude towards a proposition (Ballantyne, Coffman 2011, 2). We could argue against uniqueness by denying evidentialism, asserting that there are some other factors to reasoning other than our evidence. However, while uniqueness implies evidentialism, evidentialism does not necessarily entail uniqueness (Ballantyne, Coffman 2011, 2). If uniqueness is false, then it is still possible for us to have two agents who have the same evidence but reason differently. This is the definition of relative fit, the thesis that a proposition may fit evidence more or less for different agents (or, Jackson and myself would add, for different times) (Ballantyne, Coffman 2011, 3). So we can also argue against uniqueness by defending rationally different strategies of reasoning. This is what we will focus on going forward.

Time and/or Place

Now that we have established how we will be arguing for permissivism, we should also establish the scope of our argument in terms of time (synchronic vs. diachronic) and place (interpersonal vs. intrapersonal). We would consider our roundness case to be one of “synchronic interpersonal permissivism,” by which different agents reach differing doxastic attitudes at the same time. The one configuration of these two factors we will not concern ourselves with is synchronic intrapersonal permissivism—the thesis that it is rational for an individual to reach different doxastic attitudes toward a proposition from the same body of evidence at the same time (Jackson, Turnbull forthcoming, 2). This thesis is controversial because it allows for situations like “arbitrary toggling,” a case in which our roundness-machine could spit out different rational credences seemingly at random. Therefore, the factors to reasoning beyond evidence that I will defend in this paper are those that vary between agents or across time.

Evaluating Possibilities

The moment we begin to evaluate a possible world where our evidence does not fit our proposition, we lose confidence in the proposition.

Evaluating possibilities is the first of these variable strategies of reasoning that we will defend. If we are thinking of doxastic attitudes as an indication of how relatively well our evidence fits a given proposition, we must also be inherently considering how the evidence might not fit the proposition. There are usually, if not always, competing hypotheses that explain our evidence. For instance: when we put an object into our roundness-detector, our confidence might be undermined by the possibility that our machine is defective. One model of reasoning involves weighing your evidence against every plausible possibility (Lewis 1996, 551). Suppose we put an object into our roundness-machine, and we get a 1.0 confidence rating from the machine. The proposition “this object is round” seems to be true. But a member of the competing team is with us, and they introduce some skeptical hypotheses into our consideration (we call this “priming with a particular possibility”) (Jackson 2021, 320). Maybe our machine is faulty! What if the object only appears round when it is observed? The moment we begin to evaluate a possible world where our evidence does not fit our proposition, we lose confidence in the proposition, since we can no longer know that it always holds true (Lewis 1996, 562).

This objection makes it seem as though we should never be confident in anything; this is the skeptical thesis. After all, there is always the chance we are all being deceived in a grand simulation, and “roundness” is but a construct. But remember back to our definition of possibility-based reasoning. We are not weighing all possibilities, only plausible ones. When we form our doxastic attitudes, we are making an inference—a guess—to the most likely possibilities. We eliminate, or ignore, possibilities which require more work to explain than fit our evidence (it would be very hard to explain how roundness does not exist when we experience it constantly). We call this process inference to the best explanation, and it is the basis for the following argument for epistemic supererogation.

Epistemic Supererogation

Is making inferences in this manner—assuming certain possibilities and ignoring others—rational behavior? The skeptical view is that it is epistemically wrong to not consider every possibility. But reflecting upon every possibility of every doxastic attitude you hold is not a practical affair. Jackson (2021, 317) uses the language of supererogation in her defense of permissivism:

Supererogation: Performance of more than duty or necessity requires.

Jackson 2021, 317

We usually identify supererogation in the context of ethics—heroically going above and beyond the line of duty, i.e. running back into a burning building to save a dog (Jackson 2021, 317). But Jackson (2021, 317) applies it to reasoning in an epistemic sense. When we reason, we employ various strategies, including the weighing of possibilities. Many strategies are epistemically good, if not required—for instance, considering other possibilities helps us reveal blind spots. But there is a level to which they are unnecessary, and further reasoning is supererogatory (Jackson 2021, 317). Jackson paraphrases Tidman: “even if… low-stakes, everyday beliefs would change upon reflection, that does not mean one has an epistemic duty to reflect on them; maybe right after forming the beliefs, one is distracted by an automobile accident and never considers the belief again” (Jackson 2021, 317).

The significance of epistemic supererogation in the context of inference to the best explanation is that we cannot consider every possibility.

The significance of epistemic supererogation in the context of inference to the best explanation is that we cannot consider every possibility. Consequently, it must be epistemically acceptable to ignore implausible possibilities. So: if you and I are given the same evidence but consider different possibilities, it is possible for us to rationally arrive at different doxastic attitudes. By this, uniqueness must be false.

The plot of the 1985 black comedy mystery film Clue is a perfect demonstration of how inference to the best explanation plays into rational disagreement. The film’s initial release featured three different endings, in which different audiences were primed with different “conclusive” possibilities of how one of the suspects would have killed their victims. Viewers spend the majority of their time deliberating which suspect best fits the crime given a barrage of inconclusive evidence. Midway through watching, it is possible that you are struck with a possibility that leads you to another conclusion, a “sudden flash of insight [in which] you see that your evidence is actually better explained by the [revelatory] hypothesis” than any others you were considering (Jackson 2021, 318). The insight was critical to your reasoning, helping you eliminate other possibilities, but you could have reached the same (or a different!) conclusion as to the identity of the murderer without it. However, as Jackson explains, “seeing that the evidence better explains this [shown] hypothesis is not required by rationality, as rationality does not require brilliant insights”; it went above and beyond your usual considerations, and was supererogatory (Jackson 2021, 318).

An objection to the Clue scenario is that having such a revelation should be considered obtaining new evidence. Jackson (2021, 318) identifies a distinction between “evidence” in this sense, and “evidence” as used in the literature of inference to the best explanation. In this epistemological context, hypotheses are explained by evidence, not a component of it.

Pragmatic Encroachment: Stakes and Doubt

Doubt is a critical factor of belief to the argument for permissivism.

Most of the time, we use our roundness machine on everyday objects—it’s mostly a way to confirm our suspicions as to whether objects are round. But one day we are approached by members of the International Avogadro Project, who present us with a nearly perfect silicon sphere. They ask us to confirm their measurements to the sphere; we run it through the machine and it gives us a rating of 0.99—nearly certain it is round. We are confident in the machine, and we initially adopt this attitude. Later, we overhear the significance of the sphere’s precision: its mass will be used to define the kilogram unit used around the world. Suddenly, we are struck with doubt. What if our machine was wrong? The consequences might be dire for anyone who uses the metric system of measurement going forward. When we are asked for our final credence regarding the proposition “the silicon sphere is round,” we only give 0.9 instead of our initial 0.99, even though our evidence—the machine’s result—has not changed. Is this change in doxastic attitude rational?

Doubt is a critical factor of belief to the argument for permissivism. My evidence may point me to believe a proposition, but if that belief can be affected by doubt without a change in evidence, then permissivism is true (Jackson 2021, 319). This argument sounds very similar to the argument for pragmatic encroachment, the thesis that practical circumstances (stakes) rationally influence epistemic affairs.

Should varying practical affairs count as evidence? They might, in the sense that they are an input to reasoning about a proposition. But rarely do they interfere with the evidence relevant to the permissivism debate—whether the proposition is true (Jackson 2021, 320). When there are higher stakes for an object being round, there is a change in how I act regarding the proposition; I go back and double-check, or discuss the matter with my fellow engineers. But my evidence as to whether the object is round remains the same—it almost certainly is. Therefore, since practical stakes may vary across time or between agents, and pragmatic encroachment rationally influences doxastic attitudes without a change in evidence, it is possible to reach different rational doxastic attitudes as a consequence of doubt or stakes. Uniqueness is false.

Doubt and pragmatic encroachment often come into play in justification for a simple reason: no one likes to be wrong. Kelly (2013, 9) uses the term “cognitive goals” to distinguish between the priorities of different agents in determining doxastic attitudes. For instance, I might focus on what the evidence at hand suggests, and believe or disbelieve even on a very slim margin in the interest of possessing a fact I can work with. On the other hand, you might more heavily consider the potential for error in the evidence and suspend judgment. Agents weigh evidence differently depending on their context—certain factors, such as error, may be more important than others even if both agents are fully aware of these factors. This is an alternative strategy for relative fit from evaluating possibilities, but it implies the same permissive effect.

Objections to Permissivism

It seems as though permissivism is a slippery slope.

Permissivism addresses the question of rational disagreement, but it raises other concerns. While our context may rationally affect our judgment, the factors that go into it can be arbitrary. White (2013, 318) emphasizes this situation with the case of a belief-switching pill, which arbitrarily toggles the coarse doxastic attitude of an agent—I believe a proposition, take the not-p pill, and no longer believe it. This seems like irrational behavior. But what makes our situational factors better than arbitrary switching? Some epistemologists, including Kelly (2013), argue that your epistemic goals or standards dictate what attitude is permissible you to hold; the how or why you reason is unique to you, so the output of that reasoning will differ as well. But, generally, the way in which you reason is static; this refutation works to defend interpersonal but not intrapersonal permissivism. However, as Jackson (2021, 322) identifies, White’s example is narrow. It suggests that if an agent can arbitrarily adjust doxastic attitudes, those attitudes are acceptable at any given moment—his example is one of synchronic intrapersonal permissivism. If we consider our reasoning to be like a machine with evidence as an input and doxastic attitudes as outputs, it would be wrong (irrational) for the machine to produce inconsistent outputs. But, over time, we may adjust the machine to favor certain inputs and ignore others so that our results are more useful. This is analogous to diachronic intrapersonal permissivism, which White’s toggling objection doesn’t necessarily account for.

It seems as though permissivism is a slippery slope. After all, there can always be higher stakes or more skeptical possibilities to consider when evaluating evidence. But if uniqueness is true, then if we are not absolutely accountable in our reasoning we are reasoning irrationally. If epistemic supererogation is false, then we are forced to consider an infinite number of worlds in order to come to a conclusion regarding our beliefs. Without pragmatic encroachment, we must assume the highest stakes in our justification and suspend judgment out of doubt. There is an extent to which each of these subjective practices irrationally interfere with justification. But it is irrational to rule them out.

Works Cited

Ballantyne, N. and Coffman, E. J. (2011). “Uniqueness, Evidence, and Rationality.” Philosophers’ Imprint 11.

Jackson, E. (2021). “A Defense of Intrapersonal Belief Permissivism.” Episteme 18 (2):313–327.

Jackson, E. and Turnbull, M. G. (forthcoming). “Permissivism, Underdetermination, and Evidence.” In Clayton Littlejohn and Maria Lasonen-Aarnio (eds.), The Routledge Handbook of the Philosophy of Evidence. New York: Routledge. pp. 1-13.

Kelly, T. (2013). “Evidence Can Be Permissive.” In Matthias Steup and John Turri (eds.), Contemporary Debates in Epistemology. Blackwell. pp. 298.

Lewis, D. K. (1996). “Elusive knowledge.” Australasian Journal of Philosophy 74 (4):549 – 567.White, R. (2013). “Evidence Cannot Be Permissive.” In Matthias Steup and John Turri (eds.), Contemporary Debates in Epistemology. Blackwell. pp. 312.