The Gettier Problem and the
Conditions for Knowledge,
by Anthony Peter Iannini, 1999 |
|
|
Essay Overview: This short essay poses a few novel examples of the Gettier problem and proposes why it is not a problem but a matter of misunderstanding luck for knowledge as some beliefs rely not on a proper causal chain of reasoning.
In a now famous article (1), Edmund L. Gettier presents an argument that challenges the
traditional account of knowledge as true justified belief. Before presenting a formal
account of the problem, I will present a Gettier-type example that demonstrates why problems have arisen for such traditional definitions of knowledge.
CASE 1:
________________
Let’s suppose that Smith is home one evening and learns,
from several reputable news organizations, that leading astronomers have agreed that a
planet-sized asteroid will hit the Earth the next day.
Smith, who is justified in
believing this, infers that he will die the next day. Now, as Smith believes (A)
that a planet-sized asteroid will hit the Earth the next day and (B) that he will
die the next day, he decides to do all of the crazy, dangerous things he had never done.
As Smith’s body is not used to such excitement, it turns out that Smith ends up dying
of a heart attack around noon the next day. It also turns out that some strange phenomenon
in space had caused the astronomers of the world to falsely conclude that an asteroid was
heading for Earth.
It is apparent (assuming some basic understanding of what happens when two planet sized
objects collide) that A does entail B. Smith, then, believed B, was
justified in believing B, and it turned out that B was true.
However, it
turned out that A was false, and that knowledge of B was inferred from a
false proposition. The problem, then, is that we do not want to say that Smith knew B.
Therefore, the traditional account of knowledge must be revised or else we must allow that
Smith did know B.
Before continuing, it is notable that there are other Gettier problems that are of a
different logical form, though they create the same problem.
CASE 2:
_________________
If S,
for example, is justified in believing p, and S infers that either p or q (which is a logically valid inference), and it turns out that p is
false and q is true, then, according to the traditional account of knowledge, S would be said to have known q. CASE 3: Also, perceptual Gettier
examples are relevant.
Suppose S sees some object x, and states that
"There is an x in front of me". It turns out, however, that the x that S saw was, in fact, merely a holographic representation of an x.
However, it also turns out, that there really was a hidden x behind the hologram.
Therefore, according to the traditional account of knowledge, it seems as
though S knew that there was an x in front of him, even though he never saw
the actual x.
A number of approaches have been taken to overcome the difficulties that Gettier-type
illustrations present to the traditional account of knowledge. One such attempt is that of
Alvin Goldman (2), who puts forth a theory of causal connection between knowledge of
propositions.
Goldman states that S knows that p if and only if the fact p is causally connected in an "appropriate" way with S’s believing
that p.
Goldman defines "appropriate knowledge-producing causal
processes" as including perception, memory, and a causal chain that is correctly
constructed through warranted inferences or any combination of these. In both case one and
case two (previously defined by CASE), Goldman’s theory, though
complex, does seem to at least have the capacity to solve the Gettier problem.
Goldman’s theory, stated naïvely, is that for someone to know something, it must be
related to something else that such a person knew in a certain way. In the case of the man
who believes he is going to die as a result of a catastrophic collision with an asteroid,
his belief is causally connected to the belief that there is an asteroid coming. In
other words, if the asteroid does not come, then his belief that he is going to die has no
causal connection with any facts about the world.
However, Goldman’s condition does not appear to solve the perceptual
Gettier-type problem illustrated above in case three. In this case, someone sees an
illusory representation of something, makes a claim about the actual existence of the
thing in question, and it turns out that both the representation of the thing and the
thing itself are present in the manner that the proposition was made.
For example, cases
in which someone states "There is a dog is within ten feet of me" and, in which,
there is a ‘fake’ or ‘illusory’ dog and a real dog within ten
feet of the speaker. We could imagine that the fake dog is a hologram, while a real dog is
laying behind a piece of furniture so as to be out of sight.
If perception is one of the
"appropriate" ways in which propositions are causally connected, then
Goldman’s analysis fails because for someone to believe and be justified in believing
that a dog is within ten feet, there must be some actual perception of a dog.
Another approach and condition for knowledge that attempts to solve the Gettier problem
is put forth by Keith Lehrer. In summary, Lehrer’s fourth condition (where the
previous three conditions are those of the traditional account of knowledge) is that: if S is completely justified in believing any false statement p which entails (but is
not entailed by) q, then S would be completely justified in believing q even if S were to suppose that p is false (3). Lehrer reaches this
condition through a rather complicated analysis in which he examines and rejects
proto-conditions that he finds problematic.
For clarity, I will present case one in terms
of Lehrer’s final condition. If Smith is completely justified in believing any false
statement "a planet-sized asteroid will hit the Earth tomorrow" which entails
(but is not entailed by) "he will die tomorrow", then Smith would be completely
justified in believing "he will die tomorrow" even if Smith were to suppose that
"a planet-sized asteroid will hit the Earth tomorrow" is false.
Though
Lehrer’s condition does pose a viable solution the Gettier problem, it seems to be so
overly strong that many cases in which we have extremely good evidence for inferring a
proposition would fail to meet its requirements.
In cases such as the one in which Smith
believes he will die the next day, Smith would require, in order to meet Lehrer’s
fourth condition, additional evidence for the knowledge that he will die.
In my own analysis of the Gettier-type problems, I have attempted to formulate a number
of conditions that may preserve the traditional account of knowledge.
Firstly, any
proposition that makes any claim about a future event can not meet the truth
condition until it actually happens, because truth is dependent on correspondence with
what actually is.
|
|
|
This takes care of "Smith believes that p will happen"
type examples, such as in case one. Examples such as in case two, where "Smith
believes p and infers p or q", where there is no evidence for q but it turns out that p is false and q is true can be handled by relying on
Goldman type requirements for a causal relationship.
Similarly, if, as in G.E.
Moore’s example, a man is dreaming that he is speaking in the House of Lords and he
wakes up and finds that he actually is speaking in the House of Lords (4) then I
think it is appropriate to invoke the words of Barry Stroud "At best you have a
thought or belief which just happens to be true, but that is no more than coincidence and
not knowledge." (5)
In case three, where there is the perception of some object x,
and the object referred to turns out to be fake or illusory and there actually is another object that can be included in the proposition (as it is similar enough to object x)
we can again say that this, too, is coincidence rather than knowledge.
The fact that Gettier-type examples require such specific and unlikely circumstances
points to the notion that they do not undermine the foundations of the traditional account
of knowledge, but rather that they point to interesting uses of language that seem to undermine the justified true belief conditions of knowledge.
When someone forms a
justified belief, it is based upon another justified belief, as in the case of
Smith who thinks he is going to die the next day because he also thinks an asteroid
is going to destroy the Earth.
Also, when someone says that object x is physically
present, they are making such a claim based upon the ‘fake’ or
‘illusory’ object x. Because these grounds for inferring other
generalized beliefs turn out to be false, the inferred beliefs themselves become baseless
and unwarranted. In essence, if it turns out that my inferential belief based upon a false
belief turns out to be true, I was, in such a case, lucky— not knowledgeable.
Endnotes:
________________
(1) Gettier, Edmund L. "Is Justified True Belief Knowledge?" Analysis, Vol.
23, 1963, 121-123.
(2) Goldman, Alvin I. "A Causal Theory of Knowing," Journal of
Philosophy, LXIV, 12, 1967, pp. 357-372.
(3) Lehrer, Keith. "Knowledge, Truth and Evidence," Analysis,
Vol. 25, 1965, 168-175.
(4) Moore, G.E. Philosophical Papers (London, 1959), p. 245.
(5) Stroud, Barry. The Significance of Philosophical Scepticism. 1984,
Oxford UP p. 15.
Reflections on this essay as of 06/11/2010:
__________________________________________________
If knowledge is a true, justified, belief there will always be a problem because to define justification is nearly impossible and how can we ever, without absolute prior knowledge of reality, know what is, in fact, true?
We know nothing other than there is something happening here. We can know, more weakly, lots of things about what I call proximal reality- which is the closest reality we find ourselves in.
Our proximal reality, for instance, could be a dream. But, we need not bother to consider that we are dreaming when we know, for instance, there is a table in the room we are in. It just so happens, however, there is no room and we are in it only in our minds. But, that does not mean, at least in some sense, that the table and the room are not "there" in our minds.
