CASES
by
JIM STONE
Abstract: This article argues that justified true beliefs in Gettier cases often are
not true due to luck. I offer two unlucky Gettier cases, and its easy enough
to generate more. Hence even attaching a broad anti-luck codicil to the
tripartite account of knowledge leaves the Gettier problem intact. Also, two
related questions are addressed. First, if epistemic luck isnt distinctive of
Gettier cases, what is? Second, what do Gettier cases reveal about knowledge?
1.
We add a wrinkle to SSR 1 (call this new example SSR 2). Unbeknownst
to anyone, the county through which Im hiking is infested by a breed of
shy sheep whose keen senses detect approaching human observers. These
sheep hide behind anything big enough to hide them, but they favor rocks.
As sheep in this county have always vastly outnumbered rocks, from time
immemorial sheep are lined up several deep behind every rock big enough
to hide a sheep. Also, rocks in this area occasionally shift position due to
earth tremors. Especially they tend to roll down hills. Call this world A
(for actual).
As the closest possible world (w) where the hill is sheepless resembles
A in the above respects, in w the sheep-shaped rock I see in A has rolled
down the hill. Otherwise sheep would be hiding behind it on the hill. So
in w I see neither sheep nor rock. Hence the sensitivity condition is satisfied: I wouldnt have believed a sheep was on the hill if none had been
there.
Nozicks adherence condition is unsatisfied, however. He writes of
this condition that not only does S actually truly believe p, but in the
close worlds where p is true he also believes it; the subject believes
that p for some distance out in the p neighborhood of the actual world
(1981, p. 176) Now moving some distance out, but remaining close to
the actual world, there is a world where the rock behind which the sheep
hides isnt sheep-shaped. I apply the same epistemic method in that
world looking at that very hilltop and do not believe a sheep is on it.4
As the adherence condition is unsatisfied, so is our strong anti-veritic
luck constraint.
A new feature must be added to our example (SSR 2 will henceforth
denote this amended version). An ancient tribe of sheep-idolaters secretly
peoples the unexplored region through which Im trekking. They chisel all
appropriately sized rocks on hilltops into perfect facsimiles of sheep. Consequently the hilltops are peppered with sheep-shaped rocks with sheep
hiding behind them. These sheep are paranoid a breed-wide neural
artefact of their genetically-determined morbid shyness. Consequently
they hide only behind sheep-shaped rocks, since they distrust other rocks.
Moving some distance out, but remaining in the neighborhood of the
actual world, there is no world where a sheep (perhaps of another color or
gender) is on yonder hill but the rock I see on the hilltop isnt sheepshaped. (Note that neither the rock nor its shape is caused by the sheeps
presence, nor are the rocks representations of particular sheep.) Is there a
close-in world where the sheep-shaped rock is obscured by trees or other
rocks? A basic tenet of the idolaters religion is that nothing should
obscure the view of a sacred rock; long ago they removed all obstructions.
Consequently the adherence condition is satisfied.
2013 The Authors
Pacific Philosophical Quarterly 2013 John Wiley & Sons Ltd & University of Southern California.
What about the safety condition? My belief that a sheep is on the hill is
true in the closest worlds to A where I arrive at it in the way I actually did,
namely, by looking at that hilltop. For in nearly all nearby worlds where
I arrive at my belief in this way, I see a sheep-shaped rock that hides a
sheep. As my justified true belief satisfies the safety, adherence and sensitivity conditions, it satisfies our strong anti-veritic luck constraint.
Nonetheless SSR 2 is a Gettier case. For I know nothing about any of
this. I believe a sheep is on the hill because Im making a knowledgenullifying mistake, namely, mistaking a rock for a sheep. If I learned I was
looking at a rock, not a sheep, I would judge that I didnt know what I
thought I did. Since the falsehood that I see a sheep on the hill is my whole
reason for believing one is there, as the first belief plainly isnt knowledge,
neither is the second.
2.
Lets say I make a mistake when I take one thing for another, where
thing covers objects, like rocks, and particular situations, like particular
parallel lines being equal in length. Paradigmatic mistakes flow from
immediate sensory inputs, but this can be extended to include false testimony and inductive evidence about what will soon happen. It is required
of a mistake, as I will use the term, that it consists at least partly in a
cognitive change in the makers mind. When I mistake a rock for a sheep,
for instance, I come to believe that I see a sheep. When I reach into a box
of Sure Fire matches, convinced by strong inductive evidence that striking
Sure Fire matches always causes them to light, the cognitive change is
coming to believe that striking this particular match (the one I now grasp
in my hand) will cause it to light.
Part of what is required for my belief that p to be based on a mistake,
as I will use the term, is that the cognitive change (c) in which the mistake
partly consists is part of my beliefs cause. Also required is that the
causation involves a cognitive process by which c is part of my reason for
my belief.5 In standard Gettier cases, the mistake closely precedes the
beliefs formation, perceptual and testimonial contexts being the natural
habitat of Gettier cases (as mistaking a rock on the hill for a sheep is at
least a near-proximate cause of my believing that a sheep is on the hill).
By way of definition, to say a mistake upon which my belief is based is
knowledge-nullifying is to say that, if Im rational and I come to think my
belief is based on my making the mistake, I would for that reason seriously
doubt my beliefs truth. Seriously signals the doubt is stronger than mere
skeptical doubt; note that knowledge doesnt occur in the definiens. In
SSR 2, if I came to believe that I had mistaken a rock for a sheep, I
wouldnt even believe a sheep is on the hill. Suppose, on the other hand, I
2013 The Authors
Pacific Philosophical Quarterly 2013 John Wiley & Sons Ltd & University of Southern California.
know that sheep are hiding behind every sheep-shaped rock. Then the
mistake would not be knowledge-nullifying: I know that, even if somehow
Im seeing a rock, a sheep is behind it. So, if I learned my belief is based on
that mistake, I would not doubt a sheep is on the hill. Mistakes neednt
nullify knowledge. Note that I make the same knowledge-nullifying
mistake in SSR 1, which is sufficient to explain why it isnt knowledge.
3.
4.
My friend, Sally, takes me for a drive in her new Porsche, shows me the
bill of sale, and so on.7 Sally has always proven honest in the past, and Im
aware of no reason to disbelieve her. I conclude that she is presenting me
with what she says is her new Porsche because she owns one, and I
whimsically infer D: Sally is presenting me with what she says is her new
Porsche because she owns one, or Venusians have bribed her to deceive
me. (I routinely add the second disjunct to suitable propositions, a private
joke to which Im addicted.) In fact a Venusian research team has bribed
Sally to deceive me, and lent her the automobile. Suppose, too, that
Venusians alone want to bribe Sally, she doesnt play practical jokes
without a substantial bribe, shes broke (though I dont know it), and
nobody is about to give her a car.
Note that my belief that D tracks the truth. If D had been false, it would
have been because the Venusians didnt bribe Sally to deceive me. As
stipulated, no one is waiting in the wings to bribe her, and Sally, unbribed,
isnt going to come up with a Porsche. So if the Venusians dont bribe her,
shes not going to show me her new Porsche or even tell me she owns one.
What if D had been true? Note that the second disjunct is true in the close
possible non-actual worlds where D is true; hence I believe D in those
worlds, e.g. the Venusians lend Sally a different color or model Porsche.8
Further, D is true in the closest relevant worlds where I arrive at my belief
that D in the way I actually did, so the safety condition also is satisfied. So
my belief isnt veritically lucky.
Nor is it subject to reflective luck. After all, I know that Sally, who has
always proven honest in the past, has taken me for a drive in what she says
is her new Porsche, shown me what she says is the bill of sale (which looks
fine to me, an expert on such documents), and that Im aware of no reason
to disbelieve her. D is true in virtually all nearby possible worlds so
ordered. So my belief that D is not true due to reflective luck. In short, my
belief that D is justified by good reflectively assessable reasons.
Nonetheless its strongly counterintuitive to insist I know D. I inferred
D from the false disjunct and I have no warrant for the true one; indeed,
I attached it because I disbelieve it. My true belief fails to be knowledge for
2013 The Authors
Pacific Philosophical Quarterly 2013 John Wiley & Sons Ltd & University of Southern California.
standard Gettier reasons. Note that the example mimics the logical structure of the second of Edmund Gettiers original cases (Gettier, 1963).9
Here we have a second Gettier case where my justified true belief isnt
true due to epistemic luck. Why isnt it knowledge? On account of Sallys
deception, I mistake a situation where Sally doesnt own a Porsche for one
where she does. If I learned Id made it, I wouldnt even believe D.
5.
Two questions
6.
Conclusion
10