The following post is my summary of a paper by Linda Zagzebski titled The Inescapability of Gettier Problems, and is posted for the benefit of my colleagues at the Theory of Knowledge class at Monash. I quote from the paper gratuitously without further acknowledgement, and my attempt was to summarize the paper as accurately as possible.
The inescapability of Gettier problems
We have seen that Gettier problems arise in cases when it is only by chance that justified true belief is knowledge. The moral we draw from that is that either justified true belief is not sufficient for knowledge, and that we need some extra component, or otherwise, the justification part of justified true belief must be rethought in a way such that it would become sufficient.
Zagzebski argues that it is impossible to avoid Gettier problems, given the assumption that the relation between justification and truth is close but not inviolable. Secondly, that it makes no difference if the extra component is identified as anything other than justification. She concludes that Gettier problems are unavoidable whenever knowledge is defined as true belief plus something extra.
She goes through a number of different theories of knowledge and shows how Gettier problems can apply to each. She starts with internalism and externalism, though she only explicates internalism.
In internalism, the grounds for justification are accessible to the consciousness of the believer. Gettier problems then arise when there is something wrong with the aspect of the situation which is inaccessible to the believer. That’s the case in the ‘Smith owns a Ford or Brown is in Barcelona’ scenario. Suppose that Smith tells you he owns a Ford and gives you all of the possible justification you might need to believe that he owns a Ford, but then it turns out that he is actually lying, and also it just happens that Brown is in Barcelona. The original statement ‘Smith owns a Ford or Brown is in Barcelona’ just happens to be true and justified, however it does not seem the case that you ‘know’ it.
The problem arises because the believer does everything that is reasonably expected of him or her to reach the truth, however those efforts do not lead to the truth. Zagzebski here talks about luck. It is a stroke of bad luck that Smith lied, and it is cancelled out by a stroke of good luck since Brown is in Barcelona.
Next she moves on to Reliabilism, which she defines as ‘a group of theories in which believers are justified when their beliefs are formed in a reliable, or truth-conducive, manner’. The ‘fake barn’ scenario is used to illustrate how Gettier problems can arise.
Suppose you are driving through the country and unknown to you, the inhabitants of the particular region you are in at the time, to make themselves look more prosperous, have erected 3 fake barn facades. You have good eyes, and they are reliable enough in ordinary circumstances to know a barn from a certain distance. These are ordinary circumstances, however, in this case the fake is not distinguishable at this distance from the real thing. Now suppose you look at a real barn and form the belief ‘That’s a jolly fine barn’. The belief is true, and justified. Again though, it does not amount to knowledge. Again, it is a case of bad luck followed by a case of good luck. It is bad luck that your eyes are not reliable enough in this case, good luck that you looked at a real barn.
Next she moves on to a theory postulated by Alvin Platinga, in which he calls the quality, which in sufficient amount converts true belief into knowledge, ‘warrant’ rather than ‘justification’. He proposes that ‘warrant is the property a belief B has for believer S when B is produced in S by S’s faculties forking properly in the appropriate environment, according to a design plan successfully aimed at truth’. Again, Zagzebski builds a Gettier problem as a counterexample to this theory.
So suppose that a woman named Mary has very good eyesight. Not perfect, but very good. Good enough to allow her to identify her husband sitting in his favourite spot in their lounge room from a distance of 15 feet in dim light. She’s done so many times before, and always her faculties have been working well enough for the environment. There is nothing unusual about the environment or her faculties in those cases. Her faculties in fact, are functioning well enough, for her to form the belief ‘my husband is sitting in the living room’ with enough warrant to constitute knowledge when true, and be true most times.
Suppose that she simply misidentifies the person in the chair. It might be her husband’s brother. Furthermore, the husband is in the living room, just out of sight. Now in this case, there is nothing unusual about the environment (like in the barn scenario), nobody wants to fool her (like Smith did), her degree of warrant is as high as it is every other time.
Again, the belief ‘my husband is in the living room’ is true, but again there is a stroke of bad luck and a stroke of good luck make it that it is accidental that the statement is true.
Platinga concludes that ‘what is essential to Gettier situations is the production of a true belief despite a relatively minor failure of the cognitive situation to match its design’. Zagzebski argues that in Platinga’s theory admits that for the degree of warrant to be sufficient, the faculties need not be working perfectly, just well enough. In Gettier cases, either there is enough warrant for knowledge or not. If there is not, then a large number of our beliefs are unwarranted. If there is, then it faces the same problems as justification based theories.
The three scenarios that have been looked at so far show that if there is even a small degree of independence between justification, or whatever the third element in addition to true and belief is, and the truth, then Gettier style scenarios can be created for that definition of knowledge.
Next Zagzebski discusses the addition of so-called ‘defeasibility conditions’ to the analysis of knowledge, as a strategy to counter Gettier problems. This adds to the requirement of justified true belief the restriction that the belief must also be justified in certain counterfactual conditions. She quotes a definition by Steven Levy which is ‘that a requirement to the effect that for S to know that p, there must be no other evidence against p strong enough to undermine S’s belief that p, should this evidence come to S’s attention’. So if this was applied to the three cases discussed previously, if S was to be advised of the falsity of the underlying belief, S would retract the belief under discussion.
A different definition of the defeasibility conditions makes an even greater problem to Gettier cases. ‘The evidence e must be sufficiently complete that no further additions to e would result in a loss of justification and hence a loss of knowledge’. If the belief is false, further evidence would result in the loss of justification, and thus knowledge.
Strong defeasibility conditions then appear to threaten the assumption of independence between the justification and truth conditions of knowledge. Weaker defeasibility conditions though can be shown false with a Gettier style example.
Suppose Doctor Jones has very good evidence for her patient, Smith, suffering from virus X. Smith has all of the common symptoms, and let us suppose that they are not compatible with any other known virus. All the evidence the doctor bases her diagnosis on is true, and there is no other evidence available to her. There is nothing defective in the justification of her belief.
Now let us suppose that the belief is false, that in fact, Smith has virus y, which is unknown to science, but has similar (or even the same) symptoms. The belief, though false, is still undefeated and is still justified.
Now what if Smith actually does have the x virus? Suppose he contracted it the day before the visit at the doctors, and as such, the virus has caused no symptoms of its own yet. Now the belief ‘Smith has virus x’ is true and justified, yet it is not knowledge.
This shows that so long as there is a small degree of independence between truth and the other conditions of knowledge, Gettier scenarios are going to be possible.
As becomes clear, one way to solve this is to eliminate this independence. So the cases where the belief is true, but could have equally well been false, are eliminated from our view of knowledge.
A second possibility is to go to the other extreme, and make the justification condition completely independent from the truth condition. It could be that the best justification puts the subject in the best position to get the truth, but perhaps simply, the best position isn’t that great. In this case, most justified beliefs will be false. Still, some might be willing to call them knowledge, if they just turn out to be true.
In this case, Gettier scenarios would simply be accepted as knowledge.
Thirdly, one could say that true belief + whatever is never going to be sufficient. Here it will always be necessary to count luck into the analysis. This makes knowledge justified true belief + luck.