Fallibilism — Transcript

Hello YouTube. Today we're going to talk about fallibilism in epistemology. Now, there is a general consensus among epistemologists that knowledge can be fallibilistic, which is to say that a subject can know that p even if the justification for p is not infallible. Or another way to put it, a subject can know that p even if there is no conclusive justification for p. The justification for p does not guarantee the truth of p.

infallible or another way to put it a subject can know that p even if there is no conclusive justification for p the justification for p does not guarantee the truth of p uh for example i currently see a cup

For example, I currently see a cup on the table. I can reach out and touch it. I can check my general environment to see that I'm not dreaming, and so on. Many epistemologists would accept that I know on this basis that there is a cup on the table. Nevertheless, my evidence for the proposition that there is a cup on the table does not guarantee the truth of that proposition. Maybe I'm being deceived by an evil Cartesian demon. Maybe I actually am dreaming. It's just that the usual indications of dreaming are absent.

table nevertheless my evidence for the proposition that there is a cup on the table does not guarantee the truth of that proposition maybe i'm being deceived by an evil cartesian demon maybe i actually am dreaming it's just that the usual indications of

So if I know that there is a cup on the table, it must be fallibilistic knowledge. This same belief could have been held with the same grounding even in a situation where it isn't true and thus wouldn't be knowledge. The level of justification required for knowledge is compatible with error.

wouldn't be knowledge the level of justification required for knowledge is compatible with error so let's try to define fallibalism a bit more precisely baron reed in his philosophy compass article on fallibilism presents several characterizations of the position the first approach to fallibilism is as

So let's try to define fallibilism a bit more precisely. Baron Reed, in his Philosophy Compass article on fallibilism, presents several characterizations of the position. The first approach to fallibilism is as follows: S knows that p in a fallibilistic way just in case S knows that p on the basis of some justification j, and yet S's belief that p on the basis of j could have been false.

1940 how do i know this because i read it on wikipedia and i know that wikipedia is usually reliable about the birth dates of famous people i i know that wikipedia is checked by thousands of people that frank zappa is famous enough that many

So I know that Frank Zappa was born in 1940. How do I know this? Because I read it on Wikipedia, and I know that Wikipedia is usually reliable about the birth dates of famous people. I know that Wikipedia is checked by thousands of people, that Frank Zappa is famous enough that many people will have checked the birth date, and so on. But suppose that Wikipedia had got it wrong. Maybe Frank Zappa's birth date was widely misreported for some reason.

about his birth date on his wikipedia page in that case i would have held the same belief with the same justification but the belief would have been false so it wouldn't have been knowledge but of course because he was in fact

Well, in that case, I would still have believed what I read about his birth date on his Wikipedia page. In that case, I would have held the same belief with the same justification, but the belief would have been false, so it wouldn't have been knowledge. But of course, because he was in fact born in 1940, it is knowledge, so says the fallibilist.

does not entail j does not logically entail that p is true what justifies my belief that frank zappa was born in 1940 is the testimony from wikipedia along with my background assumption that wikipedia is generally reliable about this sort of

A second way to characterize fallibilism is: S knows that p in a fallibilist way just in case S knows that p on the basis of j, and yet j does not entail—j does not logically entail—that p is true. What justifies my belief that Frank Zappa was born in 1940 is the testimony from Wikipedia, along with my background assumption that Wikipedia is generally reliable about this sort of thing.

belief that he was born in 1940 is compatible logically compatible with many other claims about his birth date now as reader notes one of the drawbacks of both of these formulations is that neither of them allows for fallibilistic knowledge of necessary

But obviously, the fact that Wikipedia has reported that Frank Zappa was born in 1940, and the fact that Wikipedia is generally reliable about birth dates, does not entail—does not logically entail—that Frank Zappa was born in 1940. The justification for my belief that he was born in 1940 is compatible, logically compatible, with many other claims about his birth date.

necessary truth then it's not the case that it could have been false so our knowledge of this will not be fallibilistic per the first formulation of fallibilism furthermore necessary truths are logically entailed by everything this is because one

Now, as Reed notes, one of the drawbacks of both of these formulations is that neither of them allows for fallibilistic knowledge of necessary truths, such as the truths of mathematics. Suppose it's necessarily true that p. So take Fermat's Last Theorem. We take it that this is true, and being a mathematical theorem, if it's true, it's arguably necessarily true.

proposition so every justification will entail a necessary truth so our knowledge of fermat's last theorem is not fallibilistic for the second formulation of fallibilism and the problem with all this is that many people want to say that our knowledge of fermat's last

Well, if it's a necessary truth, then it's not the case that it could have been false. So our knowledge of this will not be fallibilistic per the first formulation of fallibilism. Furthermore, necessary truths are logically entailed by everything. This is because one proposition logically entails another just in case it's not possible for the first to be true and the second to be false.

truths s knows that p in a fallible way just in case s knows that p on the basis of j and yet s's belief that p on the basis of j could have failed to be knowledge a belief

Well, if p is a necessary truth, then it's not possible for p to be false no matter what we substitute for the first proposition. So every justification will entail a necessary truth. So our knowledge of Fermat's Last Theorem is not fallibilistic for the second formulation of fallibilism.

mathematical truth by applying tools that they are very familiar with so they use reliably but now suppose that when working out a particular problem the student makes two mistakes and both of these mistakes just so happen by chance

And the problem with all this is that many people want to say that our knowledge of Fermat's Last Theorem is fallibilistic. After all, maybe there's some undetected mistake in the proof. We can't be absolutely sure, absolutely certain that we haven't made an error somewhere.

but arguably it's it's not knowledge and since this kind of thing is always logically possible um so you know it's we we can never be conclusively sure that this kind of thing hasn't happened right um the knowledge of mathematics

Reed suggests a third formulation of fallibilism which will allow for fallibilistic knowledge of necessary truths. S knows that p in a fallible way just in case S knows that p on the basis of j, and yet S's belief that p on the basis of j could have failed to be knowledge.

that this could have failed to be knowledge because she could have made mistakes um so these are just some different ways of defining fallibalism i mean the general idea is that we want to say that you know so knowledge is fallablistic

A belief could have failed to be knowledge in two ways. First, the belief could have been false. Or second, the belief might have been true but only by accident, or at least only accepted by accident.

evidence it's just that conclusive evidence is not required for knowledge so yeah maybe sometimes we do have conclusive evidence um but a fallibilist will say in many cases we do not and we can have knowledge either way right it's the

So suppose that a student learns a mathematical truth by applying tools that they are very familiar with, so they use reliably. But now suppose that when working out a particular problem, the student makes two mistakes, and both of these mistakes just so happen by chance to cancel each other out. So the equation that she mistakenly writes down is in fact correct, even though she got the calculation wrong.

there's about as universal a consensus on this as you will ever find in philosophy uh the most sophisticated forms of many different epistemological theories you know internalism versus externalism foundationalism versus coherentism and so on all of these are forms of fallibilism um

Well, arguably, this is a case of a true belief, right? But arguably, it's not knowledge. And since this kind of thing is always logically possible, so we can never be conclusively sure that this kind of thing hasn't happened, right?

mentioned baron reed identifies the primary motivation as the problem of skepticism and the need that many epistemologists have felt to give a more realistic account of knowledge infallibilism seems to lead to very radical forms of skepticism it perhaps entails that nobody has any

The knowledge of mathematics that we have is fallible knowledge because, for all you know, it may have been the case that we've made some mistake when working it out. The student accepts the theorem on the basis of our calculations, and this could have failed to be knowledge because she could have made mistakes.

never conclusively rule out deception by the evil cartesian demon even very basic forms of reasoning you know when we when we perform simple mathematical calculations three plus three equals six we can't be sure we can't be absolutely certain that we haven't made some sort

So these are just some different ways of defining fallibilism. I mean, the general idea is that we want to say that knowledge is fallibilistic just in case the justification required for knowledge does not guarantee the truth of the belief.

have very very little knowledge however our cognitive faculties do still seem to be quite good and they can be improved we can identify better and worse circumstances for belief formation we have developed processes such as controlled experiments and peer review

It's worth noting that, on our usage, a fallibilist might say that some beliefs have infallible justification, that some beliefs are supported by conclusive evidence. It's just that conclusive evidence is not required for knowledge.

so many philosophers think we're referring to this kind of more modest success so most philosophers feel that infallibilism just doesn't model very well our intuitive use of like our intuitive concept of knowledge we often attribute knowledge to people

So yeah, maybe sometimes we do have conclusive evidence, but a fallibilist will say in many cases we do not, and we can have knowledge either way. It's the core claim of fallibilism that conclusive justification is not required.

one important point here is that even if all beliefs are open to doubt we still want to distinguish between those beliefs that are well supported and those that are not there is a difference between on the one hand forming beliefs by

Having stated how fallibilism is to be characterized, let's consider the position in a bit more detail. Now, as I said, fallibilism is almost universally accepted in contemporary epistemology. I think there's about as universal a consensus on this as you will ever find in philosophy.

and i mean this kind of thing extends to our interactions with others much of our cognitive activity in modern society is highly collaborative so obviously science is an essentially collaborative institution but you'll find that for pretty much every area of

The most sophisticated forms of many different epistemological theories—internalism versus externalism, foundationalism versus coherentism, and so on—all of these are forms of fallibilism in the kind, in the way that they are developed by contemporary epistemologists.

plumbing expertise when you have a plumbing problem there is a very high division of cognitive labor in society in modern society so we have to distinguish reliable sources of information the legitimate experts from the unreliable sources again the concept of knowledge

Historically, it was more popular to suppose that knowledge must be grounded on infallible foundations. So why has this changed? Well, in the Philosophy Compass article I mentioned, Baron Reed identifies the primary motivation as the problem of skepticism and the need that many epistemologists have felt to give a more realistic account of knowledge.

this kind of approach to epistemology it's also worth noting that in the background of all this there's been a change in the attitude of epistemologists concerning uh the appropriate method for for doing epistemology uh so there's a famous problem of the

Infallibilism seems to lead to very radical forms of skepticism. It perhaps entails that nobody has any knowledge whatsoever. At the least, it seems to entail that, for instance, no one has any knowledge of the external world.

of the criterion is is a vicious circle arising from these two options so we could start by identifying the instances of knowledge we start by identifying particular propositions that particular subjects know and then we use these instances of

This is because all of our cognitive faculties are imperfect. Even in the best circumstances, perception sometimes misleads us. We can never conclusively rule out deception by the evil Cartesian demon. Even very basic forms of reasoning, you know, when we...

are well um particularism as it is known will begin by identifying instances of knowledge and seeing what they have in common and so you start with the instances of knowledge and you figure out the general criteria of knowledge from those

that's particularism alternatively methodism will start by specifying the general criteria for knowledge and then applying these criteria to specific cases to determine whether or not a person has knowledge per those criteria in that specific case now this is a problem right there's a problem

here because it looks like we have a kind of vicious circle the the trouble is that we can identify the instances of knowledge only if we have well accepted criteria for knowledge right like if you don't already know what knowledge is then you

won't know whether or not you've correctly identified the instances of knowledge but equally how can we make any claim about what the criteria for knowledge are if we do not first have various instances of knowledge to work with and

examine so basically particularism seems to depend on methodism but methodism seems to depend on particularism and we have a vicious circle and that's a whole other debate i'm not going to get into this debate here suffice it to say that methodism i think

used to be fairly popular but there's now a general consensus in favor of particularism particularists again think we start by taking common sense instances of knowledge and using those to figure out the criteria for knowledge and we can improve those criteria by

testing them against our intuitions about particular cases we can apply these criteria to help decide vague cases particularism tends to be more in line with common sense indeed that's where particularism starts there are a number of instances that are

paradigmatic cases of knowledge uh at least common from from a sort of just common sense point of view there are a bunch of propositions that various subjects intuitively no for instance when i visually perceive a cup on the table and i'm in a sound

state of mind and so on then i know that the cup is on the table um this is like that's just on the basis of intuition common sense we would attribute knowledge to me in that situation there are many

scenarios where it's perfectly intuitive to almost everybody to attribute knowledge to subjects but this won't be infallible knowledge certainly my knowledge that the cup is on the table is not infallible knowledge so one reason for the turn to fallabalism

is that most contemporary epistemology contemporary epistemologists favor particularism in epistemological method and that seems to favor fallabalism um with all this said then i think we can state a general argument for fallibilism it's rarely explicitly stated but i do

think it's implicit in much contemporary epistemology it's simply this some beliefs are knowledge and this is supported by our everyday use of the concept knowledge and by the particularist methodology premise two no belief is supported by infallible justification

so conclusion knowledge is fallibilistic i guess we might state more restricted versions of this argument if you think that some beliefs are infallible maybe we infallibly know facts about our present mental states for example so an alternative version of this

argument might be some beliefs about the external world of knowledge no belief about the external world is infallible so knowledge is fallibilistic whatever um so as i say i think this argument is the primary reason why fallibilism is so widely accepted um to be clear i

mean it's it's i don't think a a a a watertight argument for palliative fallabalism for all that's been said here it may not be so implausible to suppose that ordinary ascriptions of knowledge are usually false that in fact people know very little if

anything contemporary philosophers place a lot of weight on preserving intuitions preserving common sense but you know the very fact that skeptical arguments strikers is so simple and powerful um is itself evidence that the infallibleist model of knowledge is

just the model that we've always been using which is to say that knowledge requires conclusive justification i mean consider how people usually respond to the radical skeptical arguments like the cartesian evil demon or the brain into that argument if our implicit criteria

for knowledge are fallibilistic if knowledge does not require conclusive justification why is anybody remotely concerned about these kinds of skeptical challenges like the very fact that we all recognize these as presenting serious challenges maybe suggests that our concept of

knowledge is infallibleist the infallible list might also suggest that what's really going on is that in everyday contexts people regard their justification as conclusive even though reflection reveals that it isn't people just might not realize all of the

ways in which their beliefs are open to doubt or they might realize this but ignore it and this would be the case because there's often no good practical reason to consider all of the possible sources of doubt so yes in an everyday context it's

perfectly intuitive to say i know that the cup is on the table but this isn't because i'm implicitly assuming that knowledge is fallibilistic rather i'm implicitly assuming incorrectly that i have conclusive justification because i'm ignoring the ways that this

might be false so yeah i mean it's certainly not a watertight argument for fallablism but i think this is the main reason why uh philosophers have adopted fallabalism uh okay i'll just know a couple of other arguments for

fallibilism first fallibilism is supported by a plausible view of the function of knowledge arguably knowledge is the aim of inquiry that is inquiry ends when knowledge is achieved so the idea here is that when i'm inquiring into whether

say frank is at home i may appropriately end my inquiry when i know that he's at home or when i know that he's not at home so inquiry is one of the things that humans do for various reasons and

knowledge is the product of inquiry this is a kind of you know functional definition of knowledge it sees knowledge as the the aim of a particular practice or what is achieved by a particular practice now this view of inquiry can only work

if knowledge requires less than conclusive justification if conclusive justification is required then inquiry would um never end right a given inquiry would never come to an end we would never come to any conclusion because the inquiry would never achieve

knowledge we can always conceive of possibilities of error that we haven't ruled out it would be indeed pointless to begin an inquiry in the first place because it would be guaranteed to fail um at least in the vast majority of

cases like i i'm never going to have conclusive justification that frank is at home um so i just have to continue inquiring into that forever another argument for fallibilism appeals to the epistemic goals of inquiry in general there are two goals that we

have with respect to belief formation first of all we want to avoid believing falsehoods and second we want to believe truths we want the minimum number of false beliefs and the maximum number of true beliefs now these two goals pull us in different

directions whatever you use as a source for forming beliefs will likely permit some false beliefs so maybe you can form some true beliefs on the basis of perception but you'll inevitably have occasional false perceptual beliefs too because perception isn't perfect

we could avoid falsehood entirely by simply suspending judgment about everything by refusing to adopt any beliefs at all that would be guaranteed to minimise the number of false beliefs but this leaves the goal of acquiring true beliefs unfulfilled

a good inquirer will balance the goals of minimizing false beliefs and maximizing true ones to go too far to either extreme would be unreasonable now the key point here is that infallibilist models of knowledge are often considered to place too much

weight on the goal of minimizing false beliefs requiring conclusive justification would allow you to believe very little if anything we need to form beliefs about things for which conclusive justification isn't available in circumstances where our cognitive capacities are imperfect or where

there's pressure on time and resources so we have to make do with inconclusive justification basically the epistemic goal of believing truths puts downward pressure on the level of justification required for knowledge it makes it less than conclusive um so yeah there's a couple of arguments

that appeal to the nature of inquiry and the relation between inquiry and knowledge one response to both of these arguments would just be to deny that knowledge is the aim of inquiry so we might say that knowledge does require

conclusive justification but that when we engage in inquiry we're not actually aiming for knowledge we're aiming for something less than knowledge the problem with this kind of move is that it makes it unclear why anybody should care about knowledge

um you know so like you're defining knowledge in a way that makes it impossible for us to achieve it in uh most or even all circumstances and in a way that disconnects it from any of our practices and from anything we care about

and so yeah it just becomes unclear what the value of knowledge even is if the aim of inquiry is something less than knowledge then it looks like epistemologists should be concerned about that thing about the thing that's less than knowledge and it's unclear why

anyone should be concerned about achieving knowledge um but anyway those are some uh arguments for fallibilism as i've noted fallibalism has achieved an almost universal consensus but like any philosophical view it does face difficulties some of them quite serious so we will

now consider the case against fallablism one argument against fallibilism appeals to linguistic intuitions consider the statement i know that my car is parked outside but i might be wrong about this or maybe the statement scientists know that humans and spiders have a common

ancestor but they might be mistaken in general any statement similar to s knows that p but it might be the case that not p some people feel that statements like this are just bizarre it makes no sense to assert that you

know something even though you might be wrong about it but of course that's exactly what fallibilism entails because on fallibilism knowledge does not require conclusive justification and if i don't have conclusive justification i can't be absolutely sure that i'm not mistaken

um now of course we might question whether statements like this are in fact all that odd um i should say just from my own point of view these kinds of statements don't strike me as being odd at all perhaps this just reflects my

philosophical background right i've existed in a community where basically everybody is a fallibilist so these kinds of statements no longer strike me as unusual you know i'm happy to say that my car is parked outside and also that i recognize the ways in which

this could be false like maybe i've misremembered where i parked it maybe somebody stole it in the period between parking it and recording this video whatever um even so i i'll say yeah i know that it's parked outside but this might just

be because as i say i've existed in this community of falliblists um so so there are some people certainly who will claim that these kinds of statements are very bizarre um and that maybe lends support to infallibilism um so one response here is to say well

one reason why these kinds of statements strike people as odd is that they might be misinterpreting the might claim so again take i know that my car is parked outside but i might be mistaken about this now it's agreed by all parties that

knowledge requires truth right like a belief can only count as knowledge if that belief is true so if i do know that my car is parked outside if we hold that fixed if we hold fixed the claim that i

know that my car is parked outside then it can't be the case that i'm mistaken that my car is parked outside if i know that p it must be true that p so the point here is well look

we shouldn't interpret uh the statement i know my car is parked outside but i might be wrong about this we shouldn't interpret this statement as asserting that i could know that my car is parked outside even if it actually isn't parked outside

rather it's an expression of epistemic uncertainty i can assert that i know my car is parked outside i can assert it's true that my car is parked outside even if i'm not conclusively sure that my car is parked outside

i can have doubts about my beliefs even if those beliefs are in fact true so you know even if my beliefs are in fact true and justified and they meet the conditions that count as knowledge i can still have doubts

about them and when i say i might be wrong about this i'm expressing those doubts that's what we're that's what we're doing when we say i know my car is parked outside but i might be wrong about this i'm

asserting knowledge but also recognizing there are certain sources of doubt another reason why these statements might seem odd is that in everyday contexts we assert knowledge only when we are very confident by contrast we tend to draw attention to

the possibility of mistakes only when we're being especially cautious now given the universal fallibility of our senses and of our reasoning capacities given that you always might have made a mistake somewhere there's no need to point this out every

time you state a belief about the world right because otherwise everything we say would have to be qualified by the phrase but i might be wrong and that would just get annoying right so usually we just state our beliefs

without saying that we might be wrong so the point is if a person does explicitly draw attention to the fact that they might be wrong then we tend to interpret them as having relatively little confidence we tend to interpret them as having by

their own lights not enough justification to count as knowledge so when somebody says i know that my car is parked outside but i might be wrong about this this say this statement seems odd because the first part is naturally

interpreted as expressing a very high confidence in the belief that their car is parked outside whereas the second is naturally interpreted as expressing caution um usually we only explicitly state that we might be wrong when we take ourselves not to have good

enough justification so for the the oddness of statements like statement o uh is pragmatic uh not semantic i i guess is one way to put this um right a second objection noted in baron reed's philosophy compass article is that fallibilism gives rise to getty

a problems these are of course one of the major topics in contemporary epistemology i'm assuming you know how gettier problems work if not i recently uploaded a video with cole on gettier so you can check that out anyway gettier challenged the

justified true belief analysis of knowledge but gettier cases apply much more broadly than that indeed no responses to gettier have been widely accepted and it seems that attempts to fix the definition of knowledge by adding new conditions just tends to give rise to new gettier

cases um so suppose we try to fix the justified true belief analysis by adding some new condition x so we say that knowledge is justified true belief plus x where x is independent from justification truth and belief well we have a general method for

generating gettier cases first we specify a situation where the subject has a justified false belief but also meets condition x and then we tweak the case so that the subject's belief is true but merely by accident linda zagsebski in

her article the inescapability of getting a problems discusses this point in more detail the fundamental issue here is is that of accidental success um what happens in gettier cases is that there's a case of bad luck that gets

cancelled by good luck the subject has bad luck in so far as she meets the appropriate standards of justification or she meets condition x or whatever but she nevertheless has a false belief but then the good luck is that some independent element of the

situation unknown to the subject ends up rendering her belief true and the issue here is that it seems that fallibilism will always permit some sort of accidental success some accidental knowledge the reason is that for fallibilists justification and truth can come apart

you can have a justified false belief and this is what allows you to set up gettier cases you give the subject a justified false belief and then tweak it so the belief is true by chance and the same can be said for whatever

other condition the fallibilist adds to knowledge so suppose we say that knowledge is true beliefs formed by reliable processes well a reliable process does not guarantee true belief beliefs formed with a reliable process can occasionally be false so um there's a there's a disconnect

between truth and the other conditions for knowledge and this is an essential feature of fallibilism um because you know for the fallibilist right you can have knowledge right you can hold a true belief with the right level of justification

even though that level of justification is compatible with the belief being false so that disconnect between truth and the other conditions for knowledge is an essential feature of valuablism and that seems that you're it's always going to be enough to generate gettier

cases so it has proven very very difficult to insulate fallabalism from this kind of problem in a rigorous way um but for many people this kind of consequence is just absurd right it's absurd to think that gettier cases could

be instances of knowledge and since every fallibilistic account of knowledge seems to entail gettier cases fallibilists are committed to an absurd conclusion on the other hand consider the infallible list the infallibilist says that belief a belief counts as justified or

at least a belief has enough justification to constitute knowledge only when it is guaranteed to be true so the justification condition and the truth condition are no longer independent for the infallible list which means it would no longer be possible to set up

gettier cases for any proposed gettier case the infallibilist will simply object that the justification is not strong enough because the justification does not guarantee the truth of the belief so one benefit of infallibilism is that it seems to avoid getting a

problems another argument raised by reed is that fallibilism is vulnerable to lottery paradoxes so so the standard version of the lottery paradox um depends on two principles uh principle a is that if p is highly probable then it is rational to believe

that p and principle b if it is rational to believe that p and it is rational to believe that q then it is rational to believe that p and q the paradox here can be seen by considering a lottery with 1 000 tickets

only one of which will win if you take ticket number one well it's highly probable that it will lose so by principle a rational to believe that it will lose similarly for ticket number two it's highly probable that it will lose so

rational to believe that it will lose and so on for every individual ticket but now if we can join these beliefs we get the result that it is rational to believe that every ticket will lose and that's a

contradiction given that we've just stipulated that there will be one winning ticket now this in itself isn't a problem for fallibilism this doesn't touch on fallibilism but we can consider slightly modified versions of those two principles that do raise trouble for

fallibilism so reed considers these two principles principle a1 says that if it is both highly probable with p and true that p then one knows that p and principle b1 if one knows the p and one knows the q then one knows that

p and q any fallibilist will need to accept a1 or something like it we can substitute justified or warranted or based on good evidence or whatever for highly probable if you'd prefer not to frame it directly in terms of probability

now with these two principles we can't generate a paradox in quite the same way as before so suppose that the winning ticket will be ticket 1000.

well then it will be highly probable that ticket 1000 will lose but false the ticket 1000 will lose so you won't know that ticket 1000 will lose and this means you're not going to know the contradictory propositions that

one ticket will win and all the tickets will lose there is however another way to get to the paradox suppose again that ticket 1000 will be the winning ticket i then consider the tickets methodically taking them one by one

to decide whether they will win or lose ticket one has a 0.999 probability of losing it is also true that it will lose so by principle a1 i know that it will lose for the same reason i know the ticket 2 will lose

by principle b1 i can conjoin these claims ticket 1 will lose and ticket 2 will lose i know this i know that ticket one will lose and ticket two will lose i can apply the same reasoning to every

ticket up to ticket 999 then before considering ticket 1000 i remember that one of the tickets will win well i know that tickets 1 to 999 will lose so ticket 1000 must be the winner my belief the ticket 1000 will win

is true it's also straightforwardly entailed by other things that i know so do i know that ticket 1000 will win um that would seem like a very strange conclusion uh arguably this is actually a kind of getty a problem uh it's it's a case

where there's a justified true belief that intuitively seems not to be knowledge um however i must say that this argument doesn't strike me as being quite so problematic for the fall of list the obvious response here is to deny principle b1

so that is deny that if you know the p and you know the q then you know the p and q um similarly it strikes me as quite plausible to reject quincy principle b in the standard version of the lottery

paradox the reason why this can be fairly easily rejected is that conjoining probabilities increases the risk of error so if the probability of p is 0.999 and the probability of q is 0.999 and p and q are independent then the

probability of p and q is 0.998 or thereabouts if the justification for belief in any given proposition is probabilistic then you will have less justification for belief in the conjunction of those propositions so even if p and q taken independently

are highly probable taken together they may not be um which means you can't just conjoin probabilities in the way that the principles b and principle b1 suggest now of course this does raise another problem for the fallible list because i mean rejecting

the conjunction of knowledge claims is deeply counter-intuitive um it really seems obvious that if you know the p and you know the q then you know the prob the conjunction p and q or at least you can come to know this by

applying a very simple logical rule of conjunction introduction um now i mean okay that's counterintuitive but then probabilities are often counterintuitive so maybe that's not a big deal but we might wonder whether it really makes sense to suppose

that there is a difference between believing that p and believing that q and believing the conjunction p and q i mean do beliefs even work this way so what's the difference between these states right so state number one

is the belief that frank zappa was a musician and the belief that frank zappa was born in 1940 and state number two is the belief that frank zeppel was a musician and frank zappa was born in 1940.

what's the difference between those two that that really isn't isn't so obvious and if beliefs don't work in such a way that allows this kind of distinction to be drawn then the falliblist is going to have some trouble denying

principle b1 um or at least it's not not so obvious what the fallabilist should say here okay let's turn to a fourth problem this is the threshold problem and it's discussed by michael hannon in his article why purists

should be should be infalliblists um it's one of the most serious difficulties for fallibilism as we have seen the basic claim of fallibalism is that a subject can know that p even though the justification for p is not conclusive

now this raises a question exactly how much justification is required for knowledge where do we draw the line we might frame this in probabilistic terms how probable must a true belief be in order to qualify as knowledge it

seems there are there are two options that are non-arbitrary first infallibilism infallibilists will require that a true belief counts as knowledge only if it has probability one only if it's certain to be true so there can be no doubt about it uh second

there's what we might call minimalism which says that true beliefs count as knowledge just in case they have a probability of over 0.5 so they're more likely to be true than not the obvious problem with minimalism is that

that just doesn't seem like a high enough standard for knowledge anything over 0.5 probability means that the probability might be barely higher than just flipping a coin even the weakest possible justification will be enough um and and that doesn't seem like enough

for knowledge now suppose we we choose something else suppose we say that knowledge requires 95 reliability well then the question will be why settle for this level of reliability rather than 92 percent or 98 this is going to be

arbitrary it's not clear why achieving the 95 level would make a relevant difference by contrast it's easy to see why the infallibleist's conclusive justification makes a difference once you have conclusive justification you can be certain that the belief in question is

true you have ruled out the possibility of falsehoods um i mean there's a sense in which the difference between 99 reliability and 100 reliability is a difference in kind rather than a mere difference in degree you know we're moving from degree of probability

to certainty to ruling out all falsehood and it's easy to see why that would make a difference now of course all of this is exacerbated by the fact that in many cases we can't even put a number on the level of justification in this

kind of way so suppose we had some good non-arbitrary reason to say that the knowledge requires 95 percent reliability well now consider my belief that my car is parked outside what kinds of evidence and how much of this evidence would allow me to say that

i'm 95 sure that this belief is true is it enough that i remember parking it outside memory is in general fairly unreliable but then i also know that parking my car outside is part of my usual routine so maybe my memory for that kind of

thing is is better but you know is that enough for justification or do i need to actually look outside my window and perceive my car there maybe even direct perceptual evidence isn't enough for 95 reliability maybe you take

seriously the skeptical possibilities like the cartesian evil demon um and you know like you think that there's a more than five percent chance that actually there just is no external world at all um i mean obviously this whole line of

inquiry seems rather ridiculous uh not only then is there no non-arbitrary way to fix the level of probability there is often no non-arbitrary way to link kinds of justification to probability figures hanan suggests that one way to begin

dealing with the threshold problem is to appeal to imperialism about knowledge purism and imperism concern the relation between knowledge and practical interests according to purism what counts as knowledge is independent of practical concerns whether or not a true belief counts as

knowledge is dependent purely on epistemic factors so whether the belief is supported by good evidence whether the belief is formed in a reliable way and so on and then what counts as evidence what counts as justification all of this is going to be independent

of what we care about independent of our goals and concerns epistemic concepts such as justification and knowledge are independent independent from practical concepts what exactly do we mean by practical concerns well things like our need to act the stakes involved in the situation the

time and resources available for inquiry and so on according to interests by contrast whether a belief counts as knowledge whether a belief has sufficient justification to be knowledge depends on practical factors such as the cost of being wrong

so here's a classic example clara believes that her bank will be open on saturday her justification for this is that she remembers that the bank was open on a saturday two weeks ago when she deposited a check there and she

knows that most banking institutions will keep the same timetable each week let's also suppose that her belief is true the bank will in fact be open on saturday so she has a true belief and she has some justification

for it but does she have sufficient justification for the belief to count as knowledge well an impurist would say that it depends on the situation take a low stakes case suppose we want to deposit a check but there is no real

urgency in doing so we know that we'll be visiting the town for shopping on saturday and we wonder if the bank will be open in which case we can get two jobs done at once but if when we arrive it turns out the

bank is closed that's no big deal we can deposit the check some other time well in this situation the claim is that intuitively clara's justification is sufficient clara knows the bank will be open on saturday given that like we want to go to a bank

but it doesn't really matter when we do it yeah clara's got enough justification um i think so she knows she knows she has a true belief and it's supported by adequate justification she knows the bank will be open on saturday

now compare this with a high stakes case in this case the check really does need to be deposited by the end of the week if not we won't have enough money in the bank to pay some important bills or whatever like

for one reason or another it's crucial to us that we get the money in the bank in this case the claim then is that clara's justification no longer seems to be sufficient um like if we ask does clara know that

the bank will be open on saturday intuitively the answer is no and the reason for this is that also the imperialist will say the reason for this is that doubts about her justification will be more salient to us

we will worry well what if clara's memory is wrong and she actually wasn't there exactly two weeks ago you know what if the bank has changed its timetable for some reason um intuitively we'll say that clara does not know that the bank will be open on

saturday even though the belief is true and even though she has some justification for it the justification isn't enough in the in the low stakes case then clara has done enough for the purposes of our inquiry since it doesn't really matter

if she's wrong relatively little justification is required for knowledge this is not the case so in the high stakes case in that case her evidence isn't good enough and what makes it not good enough is our practical concerns

if this kind of thinking is right then practical factors can partly determine the level of justification required for knowledge in in these two cases the low stakes and high stakes cases the epistemic factors are held constant while the practical factors differ

but it seems also the imperialist thinks it's intuitive that clara has knowledge in one case but not the other so what does all of this have to do with the threshold problem well the idea is that if you're an impurist

then you can say that the level of justification required for knowledge is that which is enough for the subject to rely on her belief in her reasoning and action given the particular circumstances that she's in the level of justification is set by the

highest stakes situation that the subject faces at the time to which her belief in question is relevant practical constraints explain why knowledge requires less than conclusive justification and they can set in a non-arbitrary way the kind of justification that knowledge

requires again take the example of depositing a check at the bank in the high stakes case this requires a high degree of justification we don't want to rely on clara's single memory and on the assumption that the bank will be

following the same timetable because if we get it wrong there's a high cost to us but we don't require a conclusive justification either why not well because there are first of all potential sources of error that are kind of irrelevant given the

circumstances we don't worry about the fact that we might all be being deceived by a cartesian demon for example we all assume that there really is a bank that we really do owe the money and so on i mean we have to assume that in order

to describe our circumstances in the first place and second there are constraints of time and resources so maybe the bank is some sort of fake or scam maybe when we deposit money there we'll just lose that money that's logically

possible and no matter how much evidence we have that the bank is genuine we all will always be able to imagine logically support possible scenarios where it is a scam but you know if we have to deposit the

money by saturday and if we have other things we want to do with our time we have no choice but to stop exploring these doubts at some point probably quite soon right like that you're not generally going to put in

a lot of effort to prove that some bank isn't a scam like maybe you'll ring them up you'll look them up online and and so on but like you're not going to really go out of your way to

rule out scenarios in which the bank is a scam um so the the thought is then an impurist actually can begin answering the threshold problem and the the impurist can appeal to uh practical factors as setting the level

of justification required for knowledge now there are two things to say about this maneuver first of all even if we accept impurism we're not going to be able to say exactly how much justification is required in any given case

we can specify different practical factors and we can say how they affect the kind of justification required but there's still going to be vagueness here and i mean still arbitrariness too so impurism can reduce the force of the

threshold problem but it doesn't solve it entirely a second thing to bear in mind is that imperialism is extremely controversial many perhaps even most epistemologists are purists there are ways to account for our different intuitions about the low stakes high stakes cases

without supposing that there is literally a difference in knowledge um so if the fallibilist takes this option in dealing with the threshold problem it's going to raise its own difficulties um right well that was fallablism that's all i have to say about this

i just wanted to yeah introduce the position because as i say it's a position that is almost universally accepted um and uh it's not so so common to see people explicitly discuss the reasons for and against it so

hopefully that was helpful and that's all for today goodbye