Full Idea
S knows P if S's evidence eliminates every alternative. But the nature of the alternatives depends on context. So for Lewis, the context sensitivity of 'knows' is a function of contextual restrictions ln the domain of quantification.
Gist of Idea
We have knowledge if alternatives are eliminated, but appropriate alternatives depend on context
Source
report of David Lewis (Elusive Knowledge [1996]) by Stewart Cohen - Contextualism Defended (and reply) 1
Book Reference
'Contemporary Debates in Epistemology (2nd ed)', ed/tr. Steup/Turri/Sosa [Wiley Blackwell 2014], p.80
A Reaction
A typical modern attempt to 'regiment' a loose term like 'context'. That said, I like the idea. I'm struck by how the domain varies during a conversation (as in 'what we are talking about'). Domains standardly contain 'objects', though.