The slides detail the process by which Goldman arrives at what he hopes will be a non-circular and explanatory substantive condition for knowledge.
He replaces the traditional view according to which:
Knowledge = (Justified True Belief, or) true belief which is justified and where the subject herself ‘possesses’ and perhaps ‘knows’ the justification with. The latter idea is called ‘internalism’.
with:
Knowledge = true belief arrived at by a reliable process. Crucially as long as the process is reliable, the subject does not need to know this to have knowledge. Externalism.
There’s a slight complication because in this paper Goldman says that by offering this analysis he is saying what justification is. In other words, he does not actually reject the claim that Knowledge = Justified True Belief. Still it is convenient to take reliabilism to be a rejection of the JTB analysis (because. eg. he is an externalist and most JTB theorists are internalists).
We should ask two preliminary questions:
What is a process?
How reliable?
What is a process?
A process is general: a general method for gaining beliefs. This implies that for any particular instance of belief formation, there will be different ways of describing it. It will instantiate different processes described at different levels or in different contexts. This is not a problem for reliabilism but suggests it owes an account of how the general process should be individuated.
In the meantime, however, reliabilism does seem to be able to deal with Barn Facade County. Intuitively Henry does not have knowledge because it is a matter of luck that he is presented with the one real barn in Barn Facade County. Had he stopped his car half a mile earlier he would have a false belief.
For the traditional JTB approach:
Henry has a belief (that that’s a barn) which is true and he has a justification for it (it looks like a barn). But since, intuitively, Henry does not have knowledge, that’s a probel,
For the reliabilist:
Henry has a belief which is true. But – whether or not he knows it – in Barn Facade County, his method is NOT a reliable process. So according to Reliabilism he does NOT have knowledge and that fits intuition.
Henry’ justification (according to traditional and internalist analysis) is the same in Barn Facade County and in the UK and thus cannot be used by the traditional internalist JTB analyst to distinguish between the UK (where Henry, intuitively, has knowledge) and Barn Facade County (where, intuitively he does not).
But Henry’s belief forming process is reliable in the UK but unreliable in Barn Facade County.
But note that this suggests a complication.
The same method of arriving at beliefs is reliable in one context and unreliable in another. So a method is not a reliable or an unreliable process in itself. Processes are reliable in contexts. So Henry’s method is EITHER: having a look in Barn Facade County, OR: having a look in the UK. How the world is – whether or not Henry knows it – plays a part in whether his method is a reliable process or not.
How reliable?
Two initial options. EITHER 100% reliable. OR Less than 100% reliable. Let us pick a figure for the second: 80% but it doesn’t matter which figure.
100% reliable
Is any general method ever 100% reliable? It seems that there will not be much knowledge possible if we pick 100%.
80% reliable
Suppose Henry can spot a barn 80% of the time. Then for any belief he forms there’s:
an 80% chance it will be true and hence (since it is arrived at by what we are calling a reliable process) knowledge; and
a 20% chance it is not a barn and hence false and hence not knowledge.
But even if Henry knows that his method is 80% reliable (not that he needs to know this, according to reliabilism) he does NOT know which case applies. There is a 4/5 chance he will be right and 1/5 chance wrong. If he is right, then, according to this version of reliabilism he has knowledge. If not, not. But then does this not seem to be mere luck?
Compare:
Henry always bets of dice rolls that they will not score 6. He has a 5/6 chance of being right and 1/6 chance of being right. So his method is 5/6 reliable (83%). But we would never say of his true beliefs that they were knowledge.
So if 100% is implausible and anything else is not enough for knowledge, then that’s a problem.
Two further comments
1) In the UK, there are no barn facades so Henry’s method might be 100% reliable here. Had he been in Barn Facade County it would not have been. But he is here, so it is. But should we worry about the mere possibility of barn facades? Does the fact that “had there been barn facades, the process would not have been reliable” make Henry’s actual process unreliable?
We might want to say that a method has to be 100% reliable in the actual world even if they are unreliable in other possible worlds. It is contingent that they are reliable. But...
2) Consider a case where Henry looks at a real barn at time T1 and forms a true belief that it is barn. Call this particular event a process (“Barn-Process-T1”). Is it reliable? Well it only ever happens once. Any later applications of what we might think of as the same method – that is Henry having a look with his skilled eyes from 50 yards, say – will not be at time T1 and so will not count as this process (“Barn-Process-T1”). So this process is, in the actual world, 100% reliable.
But, I said earlier, processes are general. That is what cases the problems. Even in this case, had the method been applied at time T1 under other counter-factual circumstances - in Barn Facade County or if the UK had had a facade building custom – then it would not be reliable. So we might say, the fact that “Barn-Process-T1” was 100% reliable in the actual world was just a matter of luck. And if so, then mere 100% reliability in the actual world is not enough.
So we might need to say 100% reliable in the actual world and in nearby possible worlds. (In every possible world I which the method were applied, it would be successful).
Bonjour's objection
This is how it is summed up by Goldman himself on the Stanford Enyclopaedia page:
The second objection is that reliability isn't sufficient for justification. The principal example of this kind is due to Laurence BonJour (1980). BonJour presented four variants of a case in which a subject has a perfectly reliable clairvoyant faculty, but either has no evidence for believing he has such a faculty, or has evidence against this proposition, etc. In each of the cases BonJour argues that the subject isn't justified in believing the output of the faculty, namely, that the President is in New York City. Nonetheless, this is what the subject does believe. So BonJour concludes that reliabilism is wrong to say that being the output of a reliable process suffices for being justified. Of course, “What Is Justified Belief?” added a further condition to (try to) handle a similar case, as explained above. BonJour didn't address that condition, but he did formulate a similar supplement for Armstrong's reliability analysis of knowledge. As he points out, the supplementary condition would handle his cases of Casper and Maud, who believe (correctly) that they have powers of clairvoyance despite having substantial contrary evidence. BonJour also offers the case of Norman, however, which he claims cannot be handled by the supplementary condition (nor by the similar condition in “What Is Justified Belief?”). Norman is described as possessing no evidence or reasons of any kind for or against the general possibility of a clairvoyant power, or for or against the thesis that he himself possesses one. But he holds the belief that results from his clairvoyance power, viz., the belief that the President is in New York City. BonJour argues that, intuitively, he isn't justified in holding this belief. (He is said to be “subjectively irrational” in holding it.) So reliability isn't sufficient for justification.