Why I'll keep on flossing, thank-you.

You should floss your teeth everyday.

Knowing how much importance I place on evidence, and given the recent media hype spawned by a story by the Associated Press about the paucity of good quality evidence supporting that recommendation, you might be tempted to call me a hypocrite.

But you’d be wrong.

If you’ve been reading this blog, you’d also know that I fully recognize that we most often have to make decisions in the absence of complete or even good information (see here). And besides, as nice as it would be, we don’t need the highest quality evidence - multiple, consistent, well-conducted randomized controlled trials (RCTs) -  to provide a reasonable answer to every question.

So what is the state of the evidence regarding daily flossing? You could find that out for yourself – no media craze required – at the Cochrane Collaboration (which I’ve written about before here) who wrote:

"Twelve trials were included in this review which reported data on two outcomes (dental plaque and gum disease). Trials were of poor quality and conclusions must be viewed as unreliable. The review showed that people who brush and floss regularly have less gum bleeding compared to toothbrushing alone. There was weak, very unreliable evidence of a possible small reduction in plaque. There was no information on other measurements such as tooth decay because the trials were not long enough and detecting early stage decay between teeth is difficult."

And here's what they concluded:

"There is some evidence from twelve studies that flossing in addition to toothbrushing reduces gingivitis compared to toothbrushing alone. There is weak, very unreliable evidence from 10 studies that flossing plus toothbrushing may be associated with a small reduction in plaque at 1 and 3 months. No studies reported the effectiveness of flossing plus toothbrushing for preventing dental caries."

It should be no surprise that there are no randomized trials reporting a reduction in cavities (dental carries) among flossers compared to non-flossers. Such a study would require that large numbers of subjects be randomized and followed for years (since it takes years for carries to develop). Large, long-term studies are costly and labor intensive, and that's why they haven't been done. Who's going to pay millions of dollars to do those studies? Are you willing to cough up some money for the cause?

Perhaps the industry that profits from the manufacture of dental floss should cover the costs, but why would they when the available evidence is enough to reasonably conclude that regular flossing prevents gum disease, which is really enough to recommend it? Even if that's all that it does, you should be flossing. Never mind that there isn't evidence showing that it prevents cavities.

This is an especially important question among people like me with valvular heart disease (VHD). Gum bleeding that permits oral microbes access to the bloodstream is an important cause of infectious endocarditis: a life threatening infection on heart valve tissue. Among the most common causative microbes are oral bacteria that get into the bloodstream by way of gum injury and micro bleeding that occurs throughout regular daily life (1,2).

When I was diagnosed with VHD about 4 years ago, I started flossing every day. Prior to that, I was like Margaret Wente (whose inflammatory piece in the Globe and Mail sparked my desire to write this post), which is to say that I only flossed a day or two before seeing my dentist, and when I did, my gums always bled. But since flossing daily, my gums hardly ever bleed, and haven’t bled at all at my last 2 dental visits. I’m living proof that flossing prevents gingivitis and there is plenty of other anecdotal evidence like mine to add to the already existent and reasonably good higher-quality evidence that it does.

Another widely reported anecdote that I can speak to: despite brushing twice a day, that string used to have quite a smell when I was finished, but after a few weeks in the habit, the foul scent was gone. Preventing bad breath is another terrific reason to floss your teeth daily, don't you think?

Now, one caveat: all of my comments refer to regular (ie. daily) flossing. Intermittently flossing and stopping will likely lead to more gum bleeding because the gingivitis that causes the underlying bleeding just returns in between flossing stints. That could actually be harmful, since everybody is at risk of endocarditis (though not as at high a risk as people with VHD). But other than that, the potential risk of daily flossing is almost certainly astronomically low.

Just as we don’t need a RCT of parachutes for jumping out of airplanes, we also don’t need anymore RCTs of flossing. As tempting as it may be - because I know how unpopular the idea of regular flossing is to most people - you should not mistake an absence of evidence (or a paucity of evidence, in this case) for evidence of absence. I'm sufficiently convinced that daily flossing is having a positive impact on my oral hygiene, and perhaps even to my overall health, and you should be too.

The stakes may not be as high for you as for me, and you may reasonably decide that a reduction in gingivitis is not worth daily flossing for you, which would be fine because you'd be making a decision based upon a reasonable and well-informed interpretation of the evidence. But please, do not make the mistake that Margaret Wente and others in the media have made, which is to conclude, because it's the conclusion that they like and/or that they think their readers will like, that we should all stop flossing because there is evidence that it doesn't do any good.

That would not be fine at all.


1. Lockhart PB, Brennan MT, Sasser HC, Fox PC, Paster BJ, Bahrani-Mougeot FK. Bacteremia associated with toothbrushing and dental extraction. Circulation 2008;117:3118–3125

2. Veloso TR, Amiguet M, Rousson V, Giddey M, Vouillamoz J, Moreillon P, Entenza JM. Induction of experimental endocarditis by continuous low-grade bacteremia mimicking spontaneous bacteremia in humans. Infect Immun2011;79:2006–2011.

On Disagreement: An Index

Ezra Klein has a great video explaining why wine ranking and pricing is complete bullshit. You can watch it here:

https://www.facebook.com/ezraklein/videos/10153775718848410/

Implicit in his case is the significance of agreement if wine expertise is to be considered a real thing.

Here are links to my five-parter on disagreement and it's important epistemic implications:

Part 1
Part 2
Part 3
Part 4
Part 5

On Disagreement, Part 5: Objections and Conclusion

"Honest disagreement is often a sign of progress"
- Mahatma Gandhi

So far, I’ve discussed three cases (which I’ll call clocks, dress, and cardiology) that illustrate what I consider to be a more widely applicable principle (p) indicating that the right thing to do upon discovering that an epistemic peer disagrees with one after full disclosure is to suspend belief. Today, I’d like to begin examining two lines of criticism of – or disagreement with – this proposition. I'll address them in reverse order. The first is that my proposition is self defeating, and I will bring this up at the end of the post. The second, which I'll deal with first, comes in the form of several other cases where epistemic peers disagree but apparently, “nearly everyone supposes that it is perfectly acceptable for one to (nevertheless) hold fast - i.e. to fail to adjust credences to bring them closer to the credences of another”(1). So let's begin by examining one of several such counter examples as described by philosopher, Graham Oppy, which he calls elementary math:

“Two people who have been colleagues for the past decade are drinking coffee at a café while trying to determine how many other people from their department will be attending an upcoming conference. One, reasoning aloud, says: ‘Well, John and Monima are going on Wednesday, and Karen and Jakob are going on Thursday, and, since 2 + 2 = 4, there will be four other people from our department at the conference’. In response, the other says: ‘But 2+2 does not equal 4’. Prior to the disagreement, neither party to the conversation has any reason to suppose that the other is evidentially or cognitively deficient in any way; and, we may suppose, each knows that none of the speech acts involved is insincere. Moreover, we may suppose, the one is feeling perfectly fine: the one has no reason to think that she is depressed, or delusional, or drugged, or drunk, and so forth. In this case, it seems plausible to suppose that the one should conclude that something has just now gone evidentially or cognitively awry with the second, and that the one should not even slightly adjust the credence she gives to the claim that 2 + 2 = 4.”

Well, of course it seems plausible to us, readers whose independent opinions agree with the one that 2 + 2 = 4, that something has gone cognitively awry with the second! Our opinions serve as additional evidence that alters the 0.5 probability that each disagreeing party is correct. Watch how re-wording the alleged counter example illustrates this reason for its failure:

“You and your colleague for the past decade are drinking coffee at a deserted café while trying to determine how many other people from your department will be attending an upcoming conference. You, reasoning aloud, say: ‘Well, John and Monima are going on Wednesday, and Karen and Jakob are going on Thursday, and, since 2 + 2 = 4, there will be four other people from our department at the conference’. In response, your colleague says: ‘But 2+2 does not equal 4’. Prior to the disagreement, neither you nor your colleague has any reason to suppose that the other is evidentially or cognitively deficient in any way; and, we may suppose, you each know that none of the speech acts involved is insincere. Moreover, we may suppose, you are both feeling perfectly fine with no reason to think that you or the other is depressed, or delusional, or drugged, or drunk, and so forth.”

I hope that you agree that it no longer seems plausible to suppose that you should conclude that something has just now gone evidentially or cognitively awry with your colleague, and that you should not even slightly adjust the credence you give to the claim that 2 + 2 = 4. One of you has a problem, but as bizarre and implausible as this scenario seems, if we are to really consider it a relevant counter example to the principle I have proposed , then neither you nor your colleague can have any reason independent of the disagreement itself to think that the other is the one with that problem. Without a reason independent of your disagreement to justify maintaining your belief, no matter how confident each of you is in being correct, it seems clear to me that you must both suspend belief until further evidence (which is, in this example, easy to acquire) sorts it out. Why? Because, as I have discussed earlier, from an epistemic perspective, neither of you has any reason to think that you are more likely to be correct than the other, so the probability at that point in time that either of you is correct – again, from each of your epistemic perspectives - is 0.5. Since assent to a belief that seems no more likely to be true than false is irrational, suspension of belief is required. Oppy provides several other examples of disagreements concerning “cognitively basic judgments” (those immediately grounded in memory or perception or elementary mathematics), but I think that they all fail for similar reasons. Essentially, if you found dress convincing of my proposition (an example of a cognitively basic judgment), then Oppy’s other similar counter examples should seem pretty unconvincing.

Imagine that you wake up and all of your memories and all of your intuitions tell you that 2 + 2 does not equal 4, while every other person on Earth disagrees. You hold your belief as confidently, sincerely, and intuitively as everyone else. Every time you take two oranges and put them in a box with two other oranges, you count the total number of oranges and you never get 4, while everybody else around you always does. This would be very strange, indeed, but no matter how confident you feel that you know better, I hope you can see that you simply cannot insist that you are right; you must suspend your belief or risk suffering from a delusion. On the other hand, everybody else can draw epistemic confidence in the otherwise perfect agreement that 2+2 does equal 4. Agreement really does matter since it serves to identify what sort of criteria we can use to determine what's normal and what isn't. Here's another example: if you think that killing others for your own pleasure is fine and dandy, and you can't see any problem with that, you're not a lone champion of an obscure moral truth, you're a psychopath.

Alright: I've saved the best for last. The final challenge posed to my proposition is that there appear to exist not just my own epistemic peers, but my own epistemic superiors who disagree with (p), including Dr. Oppy and Dr. Alvin Plantinga, among others. (There are, of course, other philosophers who agree with (p) (or something like it), but the fact that there are those who disagree is the very problem.) Lest I have some way of saving (p) from self-referential defeat, even I must suspend my own belief in it.

But I do have a way of saving (p) from this challenge, at least for now. I have argued logically for (p), and the mere disagreement of epistemic peers or superiors is not enough for me to dismiss it. Recall that (p) requires disagreement despite full disclosure. The latter requires that those who disagree explain which of my premises they disagree with and why. If after such a process, I am left with no reason independent of our disagreement to think that (p) is correct, then it seems that I will have to become agnostic regarding (p) because that's precisely what (p) requires that I do. That hasn't yet happened.

One of my interlocutors on this subject rejected the notion that when n epistemic peers disagree after full disclosure, the epistemic probability that any of them are correct is 1/n, for that is precisely what the disagreement calls into question. I am sympathetic to this criticism, and I interpret it as indicating that it is sometimes very difficult to determine if those disagreeing really are epistemic peers. However, there are times when this really isn't difficult at all, such as, when large groups of people disagree about, for example, cognitively basic judgments, as was the case in dress. Cardiology is a good example of a similar case involving a cognitively non-basic judgment. Relevant epistemic differences will tend to even out among large groups, so if you saw the stripes on the dress as white and gold, and you knew that your spouse saw them as blue and black, you might wonder whether there was something wrong with your spouse visuo-neurologically, but when you realize that there are  literally thousands of people agreeing with you, and thousands of others agreeing with your spouse, it's much more clear that there is something about the situation - about the picture itself - and not with either of the two individuals or camps that is preventing rational assent to a belief about the stripe colors. But this is just to say that the more reasons one has to think that the person disagreeing really is an epistemic peer, the more one must reduce the confidence in one's belief. Interestingly enough, it seems that applying Bayesian math to disagreements among perfectly rational cognizers leads to just this conclusion.

So at least for now, and at least on those occasions where one seems compelled to conclude that the disagreement really is among epistemic peers, (p) still stands. If you know where I might encounter objections to the premises leading to (p), please link or provide references in the comments below. Or even better, explain what they are in your own words. I'm keen to hear all about your disagreement . . .

(1) Oppy, G. Disagreement. Int J Philos Relig 2010 68:183-199.

On Disagreement. Part 4

"Who shall decide, when doctors disagree, and soundest casuists doubt, like you & me?"
-Alexander Pope

Before Angioplasty and After

Duncan was 67 years old when he had his first heart attack. Squeezing chest tightness came on Friday evening while watching ‘Wheel of Fortune’ and didn’t go away when the puzzle was solved. His wife called 9-1-1, and though the discomfort eased off a bit when the paramedics gave him a few puffs of nitroglycerine, they insisted on bringing him to the hospital. His EKG showed signs of a heart attack and an emergency angiogram showed that one of his three main heart arteries was abruptly and completely blocked, depriving valuable heart muscle downstream from much needed blood flow. The discomfort in his chest finally went away when the small balloon his cardiologist had positioned at the blockage was deflated, revealing that the clot that had been blocking the artery had broken up, restoring the flow of blood. A second inflation of a balloon was required to place a metal stent in the artery to help keep it open over time.

Duncan was lucky: the clot that formed in his heart artery and blocked it off had only been present for a few hours, so not much heart muscle damage had occurred. Untreated, a large territory of heart muscle would have been damaged, leaving him with impaired pumping function, a large scar, and a significant risk of life threatening heart rhythm problems. But the prompt opening of the blocked vessel had averted all of that. In a heart attack, time is muscle.

But Duncan wasn’t completely out of the woods.

His angiogram also showed that one of the other two main heart arteries had an 80-90% narrowing in it. Unlike the clot that blocked off his artery in a few minutes when the contestant bought a vowel, this narrowing was the result of cholesterol deposits in the artery wall that had been slowly building up over years. Interestingly enough, it hadn’t caused him any obvious problems, though in retrospect, he had had some of that chest discomfort before when playing with his grand children, which he’d attributed to indigestion.

Cardiologist #1 admitted Duncan to the coronary care unit (CCU) and told him that some time the following week, this other narrowing should also be treated with a balloon and stent, and that made good sense to Duncan, so he approved.

On Monday morning, cardiologist #2 took over the care of the CCU patients for that week, and after reviewing Duncan’s chart, informed him that the plan would be to continue treating him with a variety of new (to him) medications that had all been shown in randomized trials to reduce his risk of subsequent heart attacks and death. Only if he had problems with chest pain that these medications couldn’t prevent, would he undergo angioplasty and stenting of the remaining 80-90% narrowing. Cardiologist #2 explained to Duncan that other randomized trials comparing angioplasty to treatment with medication hadn’t shown an improvement in survival or reductions in heart attacks when stable patients with one narrowed artery were examined. Why undergo the small but real risks of having a second angioplasty procedure if no obvious benefit seemed likely? Besides, if the artery continued to cause problems despite medication in the future, it could always be treated with angioplasty then.

Duncan regrettably agreed with the new plan, and as soon as cardiologist #2 left the room, he called his nurse with a few questions.

“Do these doctors know what the hell they’re doing? ... How come the first doc said that I should have an angioplasty and the second doc said that I should just take pills? ... What are these people’s credentials?"

The nurse explained that cardiologist #1 was the director of the hospital’s angioplasty program and was recognized as a researcher and leader in the field both nationally and internationally. Cardiologist #2 was the Chief of the Cardiology Department, and the Director of the CCU. She was a co-author of the National Guidelines for the treatment of heart attack victims. He explained that both had many years experience looking after patients like Duncan, and that it wasn’t uncommon for experienced and thoughtful cardiologists to disagree about the best treatment for a given patient. He advised Duncan to make his decision about whether to undergo angioplasty or medical treatment on the basis of his personal values, not the current state of the evidence. Does he prefer the idea of taking medicine, which is simple to do? Or does he prefer taking fewer medicines without minding too much about the risk of another invasive cardiac procedure?

But Duncan couldn’t accept that advice. He wanted to do what was best, not what he seemed to prefer for other reasons.

Cardiologists #1 & #2 represent the larger cardiology community on the question raised by Duncan’s situation. On the basis of their interpretation of the available evidence and experience, some recommend opening the remaining narrowings during the initial stay after a heart attack, while others, on the basis of their experience and interpretation of the same evidence, recommend treatment with medicine and opening the narrowing only if further problems arise down the road. What should Duncan do? What should Cardiologists #1 & #2 do? What should the Cardiology community do?

Should Cardiologists #1 & #2 just continue offering their advice to every patient like Duncan that they see? Should Duncan just flip a coin? Should he get a third opinion?

Isn't it obvious? It certainly was obvious to Duncan! They don’t know the answer and further evidence is required to sort the problem out. In this case, Duncan was lucky, because the cardiology community had recognized that there was, regarding the question posed by his circumstances, a condition known as clinical equipoise. This means that the community had come to the conclusion that they ought to suspend their belief because they just don't know. In fact, a randomized clinical trial (RCT) was developed and was enrolling patients just like Duncan to either medical treatment or angioplasty of the remaining narrowing and following subjects closely for the next 4 years to determine which strategy better improved survival, reduced heart attacks, and improved quality of life. The experiment aimed to recruit almost 4,000 patients.

Some philosophers disagree with the approach to disagreement that I have been arguing for so far. They conclude that it’s perfectly rational for Cardiologst #1 & #2 to disagree. But if that’s true, then it’s perfectly acceptable for each to continue treating patients as they rationally believe. If, say, Cardiologist #1 is rational to believe that the best treatment for Duncan is angioplasty, then it’s unethical for her to enroll Duncan in the trial and expose him to a 50% chance of not getting the treatment that she rationally believes is best for him, and vice versa for Cardiologist #2. If every member of the cardiology community maintained their belief in this fashion, none of them would be able to ethically enroll patients in the trial, the trial would never be completed, and the question would never be answered. It is only by recognizing and accepting that they don’t know the answer that it becomes ethical for disagreeing doctors to enroll their patients in the trial and make progress. Not only has much progress already been made this way, but time and time again, what “thoughtful and reasonable doctors” thought was the best treatment has been shown to do more harm than good when properly tested.

Doctors should be accurate with their patients, and that often means being humble about the community's state of knowledge and their own. They should recognize the limits of their personal assessments based on experience. They should tell patients when there is significant reasonable disagreement, and how confident they are of their advice and why. They should fairly often be saying things like “probably”, “possibly” “we really don’t know”, “our best guess at the moment is”, etc., and patients shouldn't get upset with their doctors when they honestly just don't know.

Cardiologists #1&2 should both tell Duncan that they really don't know what should be done about his remaining 80-90% artery narrowing. They should be free to tell Duncan that each has a hunch about what course of action would be best, but that that's all that they have: a hunch. And this is how the rest of us should behave when faced with the reasonable disagreement of our epistemic peers. Admitting that there is a problem with our belief- a problem big enough to justify suspending a previously held belief-  is the first step towards making sure that our beliefs and the strengths with which we hold them accurately map onto reality.

Notice that Cardiologists #1 & #2 should suspend their belief whether or not a randomized trial addressing the question exists, for the existence of the trial in no way affects their inadequate justification. They should suspend belief before the trial exists, while it is being planned, and until the results are published and shown to warrant one approach over the other.

Next time, I'll be looking at a few objections to the approach to peer disagreement that I have been advocating. Have you got any? Do you disagree?

On Disagreement. Part 3

So far in this series, I’ve considered two straightforward instances of disagreement and argued that in each instance, the rational thing to do because of the disagreement is suspend belief (see here and here). Today, I’d like to summarize what I think are the circumstances where disagreement requires suspension of belief.

Quite simply, one should suspend belief whenever, as far as one can know (from an epistemic perspective), the probability that the belief is true is roughly equal to the probability that it is false.

Not all disagreement presents such a situation. For example, Dr. Rik Willems is an expert in the treatment of slow heart rhythm disorders with cardiac pacemakers. If a first-year medical student on her first clinical cardiology rotation thinks that a patient should have a pacemaker implanted, and Dr. Willems disagrees, the probability that Dr. Willems is right is considerably greater than the probability that the medical student is. After all, medical students are supposed to get their plans for patients vetted by attending physicians, not the other way around!

Dr. Willems and the medical student are not epistemic peers. That is, they are not in equally good positions to make judgments upon pacemaker therapy. This is not to say that just because Dr. Willems is in a superior position to make such judgments, that his opinion must be right. The rational thing for him and the student to do is explain to each other the reasons for their opinions. Maybe Dr. Willems has contracted viral encephalitis and evidence of his cognitive dysfunction will be disclosed in the conversation. More likely, however, the medical student has missed an important detail of the patient’s situation, or misinterpreted the available evidence addressing pacing in that situation. This conversation comprises a process known as “full disclosure”; it represents the best possible attempt for disagreeing parties to consider and share the reasons for their own belief and the reasons for the opposing belief. In many such instances, the reasons on one side of the disagreement will really be better and the disagreement will be resolved. We can all, medical students included, learn a great deal this way, even though not all disagreements end so educationally and amicably.

The disagreeing clocks left little to no room for consideration of which time reading was more likely to be correct. Electronic quartz clocks these days are all remarkably accurate, so these two machines are “epistemic peers”. Maybe one had suffered a power loss that the other had not. Maybe somebody spilled a Coke into the one on the night table and caused a malfunction. Or maybe steam and humidity from the adjacent shower caused a malfunction in the bathroom clock. Since the clocks can’t speak and arrive at full disclosure, it seems quite clear that the weight that one must put on the reading of each clock is about equal, and so one must suspend belief about what the time actually is.

The disagreement about the dress also leaves little to no room for consideration of which opinion is more likely to be correct. If just two individuals disagreed, they’d have at least a few things to discuss. Is one looking at the monitor from a particular angle, or in a room with a particular reflection that is affecting her perception? Is one color blind? Is one deceiving the other? But since the disagreement occurred on a global scale, all of these possibilities even out among the two disagreeing camps. Upon becoming aware of the scale of the disagreement, one really is left with no good reason to think that one perception is more likely to be correct than the other, and the rational thing to do is suspend belief. Since the weight of one perception is, as far as anyone can tell, equal to the weight of the other, the circumstances are not unlike considering a coin flip, and this is true even when both parties are disagreeing on the very private evidence of perception.

Why can’t the parties agree to disagree? For the simple reason that both parties have, in the genuine opposition of the other, a good reason to believe that their own perception is, as far as either can tell, the wrong one. Had the opposing belief resided in your own mind – a situation people sometimes find themselves in when they are torn between 2 equally strong but opposing beliefs – you’d be perfectly agnostic. The fact that the opposing belief resides in another mind is, as far as either can tell, arbitrary, and therefore not sufficient to render one belief more likely to be true.

So there we have it.  If epistemic peers disagree after full disclosure, and there remains no good reason independent of the disagreement itself to consider one belief more likely to be correct than the other, the rational thing to do is to suspend belief and try to find other information that will settle the question. If further deciding information is unavailable, either in principle or in practice, then the question will have to just remain open, and cognizers will just have to remain agnostic, at least until such new reasons are available. 

If you think about that for a moment, you should realize that if you accept it, you're going to have to suspend belief about a whole lot of things. This approach to disagreement leads to a significant amount of skepticism, though not, at least as far as I can see, the kind of sweeping philosophical skepticism that is intellectually crippling. We can still believe, for example, that a computer screen is in front of us, that Kennedy was assassinated in the sixties, that OJ was probably guilty (even if that belief isn't beyond all reasonable doubt) and that the gene is the unit of inheritance. But what should minimum wage be? What should be done about income inequality, anthropogenic global warming, and ISIS? Is Allah or Jesus God? These kinds of questions would seem to require the humble approach of agnosticism, and further argumentation, experimentation, and evidence. Sometimes, we are forced to act despite being agnostic, but notice that there's nothing wrong with taking a "best guess" when that's all that is available.

In part 4, I’ll apply this reasoning to a case of disagreement in the Cardiology community and explain how it is being addressed. Chime in now with your own disagreement and you just might find me addressing it in part 5, when I will consider some criticisms of approaching disagreement in the logical fashion I have been describing.

On Disagreement. Part 2

Here's a pretty dull picture of a dress with gold and white stripes, right? As you probably know, this past February, a tumblr user posted this photo and it went viral. Why? Because of disagreement.

If, like me, you saw gold/white stripes, then you were rational to believe that the stripes actually are gold/white. But what are you rational to believe when you realize that a huge population of people disagree with you? While looking at the same picture, they see black and blue stripes. How does your awareness of that disagreement influence the rationality of your original belief?

It turns out that the dress in the photo has been identified, so evidence that will settle the question exists. However, while one is aware of the genuine disagreement and before one is aware of what that definitive evidence shows, we can and should ask the following questions:

- Are those who see gold/white stripes rational in continuing to believe that the stripes are gold/white?

- Are those who see black/blue stripes rational in continuing to believe that the stripes are black/blue?

- Or should both camps suspend belief and conclude that there is something fishy about this situation - something that's preventing either group from rationally forming a belief about the actual colors of the stripes?

It seems to me that just as the disagreeing clocks in my previous post prevent rational belief regarding the actual time, so does the disagreement that captivated the world-wide-web prevent rational belief regarding the actual colors of the stripes on the dress. Until further evidence is available to settle the question, anybody who insists that the dress stripes actually are as they appear to them in the face of that disagreement is just special pleading.

The definition of arrogance is displaying a sense of superiority, self-importance, or entitlement. Without a reason for one group to think that their perception of the dress colors is more likely to be correct than the other, any member of each group who is aware of the genuine disagreement that exists, yet who insists that the colors actually are as they appear, is being arrogant. The humble thing to do here is the epistemically right thing to do, and that is to recognize that one simply can't rationally believe that the dress colors are as they appear. Not, at least, until further evidence settles the issue. The rational thing to do here is to remain agnostic on the question, despite the deliverance of your senses.

Let me explain. Assent to a belief is only rational when it is more likely that the belief is true than false. Since there is no reason to think that one group is more likely than the other to have true beliefs about the dress stripes, the principle of insufficient reason (also known as the principle of indifference) suggests that the probability that either group is correct is no better than 0.5 (after all, both groups could be wrong). Accordingly, the genuine disagreement in this case prevents rational belief. Again, the rational thing to do is to remain agnostic about the dress. One could humbly say that one's best guess is that the colors are as they appear to them, but one would not be rational to say that they believe that the dress colors actually are as they appear.

What if someone perceiving the colors as white/gold were to think to themselves something like this: "Maybe those people seeing black/blue stripes have something wrong with their visual systems? Maybe they are falling prey to an illusion? Accordingly, I can rationally continue to believe that I'm right and they are wrong." Would this kind of argumentation provide a good reason for rationally maintaining the belief that the dress colors actually are white/gold?

Well, if those seeing black/blue stripes are falling prey to a visual illusion, then those seeing white/gold stripes are rational to continue to hold their belief that the stripes are white/gold, but the disagreement calls that very conditional into question! Assuming that the other group is the one falling prey to a visual illusion is a classic case of begging the question (also known as circular reasoning). To avoid this fallacy, one would have to not be assuming that the other group is likely to be wrong; that is, one would need to have reason(s) independent of the disagreement itself to believe that the other group is likely to be wrong.

I've now considered two instances of disagreement: (a) quartz clocks displaying different times, and (b) two large groups of people disagreeing about the colors of a dress in a photo. In both instances, there was no good reason to think that one clock, or one group, was more likely to be correct than the other, and in both instances, assent to belief was irrational, or so I have argued.

Next time, I'll try to summarize what I think are the logical principles involved in considering how disagreement should affect the rationality of one's belief(s). This is the time to chime in if you think that I've made some mistake in my reasoning so far. This is your time to disagree.

By the way, here's a picture of the actual dress in the photo:

And here's a link to a great discussion of the explanation for this disagreement by Canadian experimental psychologist and cognitive scientist, Dr. Stephen Pinker.

On Disagreement. Part 1

Suppose you’ve just completed an over 20-hour series of flights to an exotic location. You’re exhausted. When you get to your hotel, you close the curtains tight, curl up in a cool, crisply made bed, and finally fall into a delicious sleep. After what seems like an eternity, when you stir again, you crack open one eye and see that the bedside alarm clock reads 07:00 am. Refreshed and remembering that you have a busy day ahead, you pop out of bed, planning your day.

When you reach the bathroom to take a shower though, something strange catches your eye. The clock on the bathroom wall says 09:45 am.

Hmmm.

Up until you walked into the bathroom, it was quite reasonable (ie. rational) for you to believe that it was 7 am. What’s it now rational for you to believe upon seeing the virtually simultaneous reading of 09:45 am on the second clock?

I don’t give a flying fruit what the actual time is, and I don’t mean for a second to suggest that if the scenario I just posed were to actually happen in real life, one ought to go back to bed and deliberate at length about the question I asked. There’s no doubt that you can just pick up the phone and ask the front desk what time it is, or check your smart phone that synchronizes automatically over WiFi. 

Boring. 

I care about what it’s rational to believe before sorting the problem out. Why? Because many disagreements that we regularly face are not so easily resolved and it is precisely those that are the most interesting and challenging disagreements to handle. For example:  “I thought we should pay down our mortgage, but my sister said it’s better to save for retirement.” “I really think I should marry him but my parents think otherwise.” “My cardiologist thinks I should put off having my valve replacement surgery, but the cardiac surgeon said that the operation is called for now.” I suggest that there might be something for us to learn from simple cases of disagreement that we might - no, we should - apply to the more complicated and important disagreements with which life is brimming.

So please stop and consider for a moment what impact the disagreement between the two clocks has on what one can rationally believe. Remember, you were rational to believe that it was 7 am right up until you saw the second clock. Should you (a) continue to believe that it’s 7 am, concluding that the second clock must be wrong? Should you (b) believe that it’s 9:45 and conclude that the first clock must be wrong? Should you (c) think that it’s probably half way between the two times (8:22:30)?  Should you (d) believe that you have no idea at all what time it is? Should you (e) believe that it’s probably morning?

You probably felt compelled to seek out further information as you contemplated the situation I posed, and that should be a good indication that (a) and (b) are not reasonable. After all, there’s no reason to think that one clock is more likely to be correct than the other. Perhaps it’s reasonable to believe that it’s morning, but notice that had the second clock read 7 pm, you’d be completely lost and you’d have to conclude with (d).

It seems obvious to me – a fact of rationality itself – that the awareness of the disagreement of the second clock must dramatically reduced the confidence that one rationally had in initially believing that it was 7 am. 

In Part 2, we’ll explore some more complicated disagreements, but this is an important time to chime in if you think that my conclusion is mistaken. I’ll repeat it one more time: the instant you become aware of the significant and mutually exclusive disagreement of the second clock, you have a very good reason to drop your belief that it’s 7 am. You suddenly have a very good reason to doubt that you can tell anything reasonable about the time, except that maybe, it’s morning, and that'll just have to do until you gather information that will settle the question. I think that if you agree with me here, you’ll have to admit that disagreement ought to have a much greater impact upon the confidence we have in our beliefs than it seems to have. Join me in the rest of this series on disagreement to see if I’m right, or if you disagree!