Justin Horky

Epistemology

Paper 1

Vogel and a Critique of Externalism In his article "Reliabilism Leveled," Jonathan Vogel provides a critique of the externalist position of knowledge by arguing that different interpretations of this position are either to lax or too stringent when it comes to higher order levels of knowledge. Specifically, with regards to his Roxanne scenario, he argues that an externalist position which relies upon the safety condition leads to the absurd outcome of legitimizing bootstrapping, an illegitimate means by which we can obtain higher-order knowledge. The Roxanne example is directed specifically toward an account of knowledge which Vogel refers to as 'neighborhood reliabilism.' 'Neighborhood reliabilism' is defined by Vogel as "the view that you can know X, if and only if you have a true belief that X which satisfies some version of [the safety condition]" (p. 305). The version of the safety condition which Vogel's Roxanne example is directed towards focuses upon processes and is formalized by Vogel as "In N, all or nearly all beliefs B(Xi) formed by process P are such that B(Xi) implies Xi" (p. 604). What this means is that the Roxanne example is directed towards an account of knowledge which makes the following claim: one knows X if and only if (1) one believes that X and (2) X is true and (3) the belief in X was formed by a process which is safe. A belief formation process which is used in the actual world is safe if it is the case that in all or almost all nearby possible worlds (i.e. our 'neighborhood'-possible worlds which are not dramatically different from ours), this same process yields beliefs which are true. For example, let us suppose that I form the belief that it is currently raining, that it is in fact raining, and that the process by which I came to form this belief is my perception that it is currently raining. According to the view

under review here, I do in fact know that it is raining because I have a true belief which was formed via a reliable method. The method is reliable because in all or almost all nearby possible worlds in which I use my perceptive faculties to form a belief about the weather, that belief is true. The intuition here is that it would take an outlandish set of phenomena to trick my perceptive faculties, a set of phenomena which are not present in any nearby possible world (i.e. a possible world that is significantly or mostly like our actual world). My perceptive faculties can indeed be tricked under the extreme circumstances which exist in worlds that are dramatically different from this one but that is all right; in order to be reliable, a belief formation process does not have to be infallible. It is unreasonable to expect something to work under any and all conditions; rather, it is enough for it to be able to work under reasonable conditions. And if such a process does in fact work reasonably well, then that reliability should be enough to rule out capriciousness, whimsicality and other features of a belief formation process which we intuitively believe cannot yield knowledge, even if they occasionally happen to yield true beliefs. This Neighborhood Reliabilism is an externalist view of knowledge. Internalism about knowledge is the view that necessary conditions for knowledge other than truth are always accessible to the agent. By contrast, externalism about knowledge is the view that not all of the necessary conditions for knowledge, besides truth, are always accessible to the agent. Let us further explicate this. It is widely agreed that truth is a necessary condition for knowledge (you do not have knowledge if the proposition which you believe in is false) but that truth is not accessible to the agent in question in the sense that, although that agent can know a proposition, he cannot know that that proposition is true

(with the exception of certain propositions like "I think, therefore I am"). It is also widely accepted that belief is a necessary condition for knowledge (you do not have knowledge of something if you do not even believe it) and that belief is accessible to an agent in the sense that an agent should be able to tell whether or not he has a certain belief. Internalists and externalists about knowledge divide over the question of whether or not other necessary conditions for knowledge are always accessible to the agent. For example, if we take a simple JTB theory of knowledge, because, as has already been said, virtually everyone agrees that truth is not (usually) accessible and belief is, the crucial question concerns justification. An internalist who advocates a JTB theory will argue that in order for a person to know something, he must have a justified true belief and that that justification must be accessible to him, i.e. he must be capable of knowing that his belief is justified. An externalist will argue that this is not necessary; it is enough merely that the agent has a belief which is true and which is justified in some way. The justification does not have to be accessible to the agent; the agent can still know a proposition even if he does not know or is not justified in believing that his belief is justified. To use our rain example again, an externalist JTB theorist would believe that one has knowledge that it is raining so long as one believes that it is, it actually is raining, and this belief is justified by the senses or perception; and the agent in question, according to this theorist, has this knowledge even though he does not know that perception justifies beliefs. Neighborhood Reliabilism, as we have construed it, is thus clearly an externalist position because, according to the advocate of this view, if you have a belief and that belief is true and you formed it using a safe process, then you have knowledge, even if you don't know that the process is safe. What matters is the belief formation process and

whether or not it is reliable, i.e. whether or not it yields true beliefs in nearby possible worlds, not that you know that this is the case. Vogel's Roxanne scenario is meant to demonstrate that Neighborhood Reliabilism cannot be the correct theory of knowledge because it leads to an unacceptable conclusion. In this scenario, a person, Roxanne, is looking at her gas gauge. Crucially, Roxanne does not know that her gauge is reliable; she has not tested, has no background information about it that would make her think that it is reliable, etc. However, according to Neighborhood Reliabilism, she does know that, on this occasion, the gauge currently reads 'F.' She knows this because it is true, she believes that it is true and the belief was formed via perception, which is a safe belief formation process. Therefore, she meets all three of the necessary conditions which Neighborhood Reliabilism lays down for knowledge. Furthermore, although she does not know that the gauge is reliable, she nevertheless believes what the gas gauge tells her. Hence, she believes that the tank is full, a belief which is also true. This belief was formed via the process of using the gas gauge and, because the gauge does in fact read reliably, this belief formation process is also safe. Hence, once again, Roxanne meets Neighborhood Reliabilism's three criteria for knowledge and she thus knows that, on this occasion, the tank is full. Now, via deduction, Roxanne is able to combine her two previous beliefs to arrive at the belief that "on this occasion, the gauge says 'F' and the tank is full." Roxanne also knows this proposition as it is true, she believes it, and the belief was formed via deduction, another safe process. It follows s from this that she also knows that 'on this occasion, the gauge is reading accurately' as this is yet another true belief arrived via deduction. At this point, Roxanne begins to repeat this entire process over and over again

with this gas gauge, such that when the gauge reads '3/4' she knows that the gauge reads '3/4,' knows that the tank is three quarters full and knows that on that occasion, the gauge is reading accurately, etc. She does this enough that, via induction, she is able to arrive at the belief that the gauge reads accurately all of the time. This proposition is true, Roxanne believes and she arrived at it via induction which (Vogel thinks) is also a safe process. Hence, Roxanne again meets all of Neighborhood Reliabilism's criteria and thus also knows this proposition. And, via deduction from this proposition, Roxanne finally arrives at the knowledge that 'the gauge is reliable.' Thus, she can come to know that her belief forming process of reading the gauge is reliable and thus she can also come to know that she knows the tank is full (or whatever level the gauge reads after the induction process). Vogel's point here is that this astonishing process by which someone can "promote many, if not all, of one's beliefs that were formed by reliable processes into knowledge that those beliefs were formed by reliable processes" is illegitimate. I can no more come to know that my gas gauge reads accurately merely by looking at it repeatedly and believing what it tells me than Descartes could defeat his evil demon merely by repeatedly noting what he was perceiving and repeatedly believing what his senses were telling him. And yet, there does not seem to be any stage at which one or more of the three criteria that Neighborhood Reliabilism has for knowledge is violated. This being the case, Neighborhood Reliabilism improperly ratifies an illegitimate means of gaining knowledge and, indeed, seems to undercut the Neighborhood Reliabilist's own externalism. For this former reason at least, Vogel argues, it should be rejected.