Epistemic humility versus relativism

Scenario

From 1980 through 2010, the Dietary Guidelines for Americans told the public to keep dietary cholesterol under 300 milligrams a day, the rule behind decades of advice to limit egg yolks, shrimp, and organ meat. The figure traced back to the 1977 Dietary Goals for the United States and the mid-century hypothesis that cholesterol eaten in food raises cholesterol in blood and, with it, heart disease risk. In 2015, the government's own Dietary Guidelines Advisory Committee reversed the rule. Its scientific report stated that cholesterol in food is not a nutrient of concern for overconsumption in the general population, based on evidence accumulated over the intervening decades showing that dietary cholesterol has a far smaller effect on blood cholesterol than saturated fat does, and that the original limit had outrun what the evidence supported.

Two people, hearing about the reversal, react in ways that both sound careful. The first says: "nutrition science got this wrong for thirty years, so whatever it tells us today might also be wrong. We should keep checking." The second says: "this shows there was never a real fact here. What counts as a healthy diet is just true for whoever's science community currently believes it." Both speakers are declining to be dogmatic. But they have said opposite things. The first still thinks there is a single fact, namely what dietary cholesterol actually does to a human body, and is only doubting whether the current answer has reached it. The second has given up on there being a single fact at all. Telling these two reactions apart, and seeing why only one of them is licensed by a track record of being wrong, is the business of this entry.

Definition

Start with what the first speaker is committed to. Fallibilism, the core of epistemic humility, is the claim that a person's beliefs, however well supported, remain revisable: new evidence could show any of them false. Notice what this claim requires in order to make sense. To say a belief might be wrong is to say it might fail to match something, some fact about how the body actually processes dietary cholesterol, that exists independently of what the 1977 committee or the 2015 committee happened to conclude. If there were no such independent fact, "might be wrong" would have nothing to be wrong about, and the sentence would not describe a real possibility. So fallibilism presupposes a target: a standpoint-independent fact against which a belief can be measured and found wanting. Epistemic humility is a claim about the knower, that their grasp on this target is imperfect and correctable, not a claim about the target itself.

Relativism, the position behind the second speaker's reaction, denies exactly that target. On a relativist view, there is no fact about a healthy diet standing apart from what a given standpoint, a research community, a era, a culture, takes to be true; there are only truths relative to each standpoint, and no standpoint is privileged over the others. This is a stronger and different claim than fallibilism. Fallibilism says: there is one fact, and I might not have reached it yet. Relativism says: there is no one fact to reach.

The two are easy to run together because both can be voiced as modesty about certainty. But they cash out in opposite instructions for what to do next. A fallibilist facing the cholesterol reversal has a clear next step: look at the newer studies, weigh the evidence on saturated fat against the older evidence on dietary cholesterol, and revise belief in the direction the evidence points, because there remains a fact of the matter for the evidence to point toward. A relativist facing the same reversal has no such step available. If dietary cholesterol's effect on the body is only ever true relative to some standpoint, then the 1977 view and the 2015 view are simply two standpoints, neither owed deference to new data, because deferring to data only makes sense when the data is evidence about something standing outside every standpoint.

Global relativism, the version that extends this denial to all matters of fact rather than one, runs into a further difficulty, pressed by Paul Boghossian (2006): the claim "there is no standpoint-independent fact" is itself a general claim about how things stand. Either that claim is meant to hold independently of standpoint, in which case relativism has admitted at least one standpoint-independent fact, the fact that relativism itself names, and the position is no longer global. Or the claim is meant to hold only relative to the standpoint of whoever asserts it, in which case a person from a different standpoint who rejects relativism is not thereby making any error, and the relativist has no basis for pressing the claim on anyone who does not already share it. Either way, the position cannot be stated as a universal thesis without either contradicting itself or losing the force of a thesis at all. This is not a proof that no version of relativism can be repaired, philosophers have proposed narrower and more careful versions, but it is the standing pressure any global version has to answer, and it is a different kind of trouble than fallibilism ever faces, since fallibilism never claims there is no fact, only that any particular grasp of it might be off.

Common mistakes

Treating the mere existence of disagreement, or a reversal like the cholesterol one, as evidence for relativism rather than as evidence that one side, or both, was simply mistaken. A committee being wrong for thirty years shows that committees can be wrong, which is exactly what fallibilism predicts; it does not show that there was nothing to be wrong about.

Using the phrase "true for me" where the honest phrase is "believed by me." "Dietary cholesterol is dangerous is true for me" quietly imports relativism into what is really only a report of what someone currently believes, given the evidence available to them. Swapping in "believed by me" keeps the fallibilist meaning without smuggling in the stronger relativist claim.

Overcorrecting into its opposite failure: mistaking any firm, evidence-backed claim for arrogance, and treating epistemic humility as though it required refusing to hold any position at all. A person who says dietary cholesterol has a smaller effect than saturated fat, because that is what the current evidence indicates, is not being dogmatic; they are doing exactly what fallibilism recommends, holding a position while remaining open to the next reversal.

Conflating relativism about matters where it is plausible, such as taste in food or preference in seasoning, with relativism about matters of physiological fact, such as what a nutrient does inside a human body. The first genuinely may have no standpoint-independent answer; the second is precisely the kind of question the Dietary Guidelines committees were trying, across decades and in the face of their own errors, to get right.

Test yourself

Take a real disagreement you have witnessed, a policy dispute, a scientific reversal, a disagreement between two people you know. Write one sentence expressing epistemic humility about it and a second sentence expressing relativism about it, using the same subject matter in both. Then state, for each sentence, what its speaker is committed to doing when new evidence arrives.

Success looks like two cleanly distinct sentences, one naming a fact the speaker might have gotten wrong, the other denying there is a single fact to get wrong, plus the correct consequence attached to each: the humble speaker must update toward the evidence, because there remains a target for the evidence to point at, while the relativist speaker has nothing to update toward, since on their own view no standpoint was ever more correct than another to begin with.