If you want a gentler introduction to these topics than the two survey articles (SEP and Phil Compass), have a look at this:
I’m going to talk about how Cohen set things up, but it’s not necessary to read that article if you felt comfortable with the survey articles.
Epistemologists fall into four camps:
Cohen introduces the third view on p. 22 and the fourth view on pp. 22-23. Some philosophers use the word “context” to talk about view 3 (David Annis and Michael Williams for example), and if you do so, you should clarify by saying “the subject’s context” as opposed to the “attributor context” invoked in view 4. But epistemologists have mostly gravitated towards using the word “context” and the label “Contextualism” only in connection with view 4. (What the SEP survey calls “Attributor Contextualism,” and focuses on after its introductory pages.) For view 3, we’ll talk about the subject’s “circumstances” rather than “context,” in the same way that Cohen does.
Views 1-3 are called Invariantists, as opposed to Contextualist views like view 4.
View 3 (the Invariantist who introduces novel kinds of dependence on facts about the subject besides justified true belief, and some Gettier factors), has come to be called “Subject-sensitive invariantism.” (Cohen doesn’t use this label.)
The kinds of extra dependency factors invoked for views like 3 include:
As Cohen points out, there are elements of Dretske’s work in epistemology that suggest each of view 3 and view 4. Cohen thinks on balance view 4 is stronger. In a paper responding to this piece by Cohen, Dretske thanked him for disentangling views 3 and 4, and said that he wanted to be understood as proposing view 3.
One kind of novel extra-evidential factor that could be invoked under view 3 are the practical stakes or risks the subject faces. (What Brian Kim in his Phil Compass article calls “nontruth-relevant factors.”) If it’s more important to the subject whether P, then they’d need more evidence for P, or need to be able to rule out more alternatives to P, to count as knowing P. Cohen discusses this idea around pp. 29-31. He observes that Dretske himself resists bringing practical stakes into the extra-evidential circumstances that knowledge is relative to; but other subject-sensitive invariantists welcome this. Cohen himself thinks that the Contextualist views 4 account for practical considerations in a better way.
When a Subject-Sensitive Invariantist brings practical considerations into the story in this way, we call that Pragmatic Encroachment. (Or sometimes “Interest-Relative Invariantism.” Kim also includes Contextualists as a kind of practical encroacher, and he motivates that way of talking, but I think the dominant way of talking is to count only this subgroup of view 3 that I’m describing as “pragmatic encroachers.”) The main proponents of these views are: Fantl & McGrath, Hawthorne, Stanley, and Weatherson.
For Contextualist views 4, the features of the attributor’s and audience’s context that can matter to whether a subject counts as “knowing” include:
As the SEP article explains, some Contextualists combine these variations with “externalist” stories about how reliable the subject’s belief is or what explains their having that belief (DeRose, Heller, Rieber, Greco). Many instead combine them with “internalist” stories about how much evidence the subject needs in order to count as “knowing,” or what possibilities their evidence needs to rule out, or what counts as evidence for them.
There are views we’re not exploring that say we need to separate the attributor’s context from that of the audience (think for example of cases where I write something down and then several of you read my message at different times and places, and with different expectations and so on). These views are called Relativist, and are defended by John MacFarlane and others. As I said, we’re not going to expore these views.
Most of the discussion around Contextualism and Pragmatic Encroachment has focused either on knowledge or on whether a belief counts as justified. That includes the views Kim calls “justification encroachment.”
We’ll be thinking this term about parallel views that say analogous things about whether a subject counts as believing something, in the first place. Maybe what an attributor means by “belief” partly depends on that attributor’s context. Or maybe there are facts about the subject beyond how confident they are, and so on — for example, facts about the subject’s interests or stakes — that make a difference to whether they count as believing. This includes the views Kim calls “belief encroachment,” and cites a few examples of.