Contrast Classes

If we don't take anything for granted, it's possible to doubt just about everything. Descartes tried to discard all assumptions and found the only thing he couldn't doubt was that he, the doubter, must exist.¹ This fact would hold even if all of his memories and sense experience were somehow being faked (as happens in some science fiction).

Normally, we don't question such basic assumptions about the world. If I ask you whether you drive a white car, I don't expect you to answer, "I'm not sure. I might drive a white car or I might be accessing an artificial memory of a white car." Fake memories were outside the scope of the question.

Walter Sinnott-Armstrong suggests that all beliefs are justified (or not) relative to some set of alternative beliefs.² For example, the belief that you are wearing a cotton shirt may be justified out of the following contrast class:

{ cotton shirt, silk shirt, no shirt }

...but not this wider contrast class:

{ cotton shirt, silk shirt, perfectly faked sensory experience of wearing a cotton shirt }

...and maybe not this contrast class:

{ cotton shirt, imitation cotton shirt }

Similarly, you are easily justified in believing you hold a $5 bill in your hand as opposed to a $1 bill or a $20 bill, but you probably aren't justified in believing you hold a genuine $5 bill as opposed to a very expert counterfeit of a $5 bill. According to Sinnott-Armstrong:

"Someone, S, is justified in believing a proposition, P, out of a contrast class, C, when and only when S is able to rule out all other members of C but is not able to rule out P."³

See how that applies to the paper money example? You would be justified in believing you hold a $5 bill if — by simple inspection — you rule out the possibility that it's a $1 bill or a $20 bill, leaving only the possibility it is a $5 bill. However, you wouldn't be justified if counterfeits are also under consideration and you lack the skill to rule that possibility out.

Note: Other theories of what makes one belief justified (or more justified) compared to another can still benefit from limiting the scope to a particular contrast class without sticking to Sinnott-Armstrong's definition of "ruling out" everything else.

How This Helps

The notion of contrast classes captures what we intuitively mean by saying a person is 'justified' in a belief, even when we're not counting scenarios of extreme deception. We can acknowledge that Descartes' thought experiment has merit when we're considering things at that level, but explain why it isn't usually relevant.

Some disagreements about whether a belief is justified can be explained by differences in contrast classes. If Claude boards a train which has two stops left (A and B), his friend Edward may be justified in believing Claude will leave the train at Stop B after noticing Claude remained on board at Stop A. However, Claude's other friend, Jaquelin, may disagree with Edward because she knows Claude is prone to jumping off trains between stops. Edward had the class { Stop A, Stop B } in mind while Jaquelin had the class { Stop A, Stop B, between stops } in mind. It's important to note that Edward's belief was, in fact, justified relative to the first contrast class...just not the second.

Appropriately Weak

This understanding of the meaning of 'justified' is negative. Alternatives are ruled out, but the last remaining possibility — the proposition that is justified relative to the class — is not necessarily ruled in as correct. Something outside the class could be correct instead.

This shouldn't come as a surprise. Philosophers often talk of "justified, true beliefs" which would be redundant if all justified beliefs were true. I think what we want from 'justified' is not a guide to truth so much as an account of due diligence. Contrast classes are a way to specify which beliefs were discarded (or found to be less justified) on the way to a justified belief.

Justified Actions

Finally, contrast classes make sense of calling a particular action 'justified' when the other available options are worse, even if the same action would not be justified if a better alternative were available.

1. Meditations II.1-3
2. Sinnott-Armstrong, W. (2006). Moral Skepticisms. New York: Oxford University Press, p. 84.
3. Ibid. p. 86.