There is an alluring appeal to claiming that there are some things that are just given when discussing the regress problem in epistemology; if some things are simply given, then we can stop the justificatory regress. The regress of justification is a motivating problem for classical epistemology. If I claim to know something, call it P, I can be asked to provide a reason why I believe that P is true. I can provide a reason Q to believe that P, and that reason Q transmits justification to P in so far as Q is itself justified. But how exactly is Q justified? Well, I can say that reason W provides a justification for believing that Q, and Q provides a reason to believe that P in so far as Q is itself justified by W. You can see the regress forming.
The regress problem has several proposed solutions. One could adopt the stance of Peter Klein and claim that the regress is not vicious. He adopts an infinitist evidentialism, and believes that as long as we consistently provide evidence for each belief, we have an ever growing chain of reasons rooted in evidence. The problem here is that if we never have some sort of story about how the whole chain of reasons is justified; it’s like having an infinite number of mirrors reflecting an image off of each other, but we have no idea how the reflection originated.
Another way to address the regress problem is by claiming that we just eventually repeat ourselves when asked to provide justification for our beliefs. The coherentist can take the position that we eventually repeat a previous belief when asked to justify enough beliefs. For the coherentist, we have a set of beliefs that is itself justified as a whole by a web of mutual support among the beliefs in the set. Mutual support is generally explained as a variety of inferential connections among the beliefs in a coherent set, and the more connections between beliefs are formed, the more support for the whole is provided. There are several problems with this approach. One problem is that it isn’t at all obvious how coherence is a guide to truth, in so far as truth is defined as correspondence between some parts of reality and some propositions about those parts of reality. A second problem is, depending on one’s account of coherence, there can be several contradictory coherent sets of beliefs, and no obvious way to adjudicate between them.
A third way to deal with the regress problem is probably the most common move people make when faced with this problem: an appeal to the given. What I mean by this is that there is some sort of foundation of knowledge that does not itself require further justification to transmit justification to other beliefs. In short, there are basic beliefs that do not themselves require reasons for justification. In the words of Wilfred Sellars, “the point of the epistemological category of the given is . . . to explicate the idea that empirical knowledge rests on a ‘foundation’ of non-inferential knowledge of matter of fact” (§3). Sellars goes on:
One of the many forms taken by the Myth of the Given is that there is, indeed must be, a structure of particular matter of fact such that (a) each fact can not only be noninferentially known to be the case, but presupposes no other knowledge either of particular matter of fact, or of general truths; and (b) such that the non-inferential knowledge of facts belonging to this structure constitutes the ultimate court of appeals for all factual claims – particular and general – about the world. (§32)
The given isn’t the same as non-inferential knowledge. One can maintain that the given is a myth and we have non-inferential knowledge. For instance, on reliablism we could have reliably functioning faculties such that when I approach a red door, a mental state forms and I non-inferentially believe that there is a red door in front of me. Sellars is making a more specific claim. To understand what exactly he’s saying, I need to introduce two concepts: Epistemic Independence (EI) and Epistemic Efficacy (EE):
EI: The Given is epistemically independent iff whatever positive epistemic status our cognitive encounter with the object has, it does not depend on the epistemic status of any other cognitive state.
EE: The Given is epistemically efficacious iff it can transmit positive epistemic status to other cognitive states. (deVries 2005)
The given must satisfy both of these conditions to obtain its status as a true regress stopper. So, the given must be some sort of state where I know that P without requiring any other facts to justify my belief that P, and my belief that P can itself transmit justification to other beliefs. There are generally two options when trying to find the given: It is either a state with conceptual content or a state without conceptual content. If a person claims that a particular state has conceptual content, and can provide reasons to believe that P, yet does not itself require justification, then that state would be an instance of the given. Now, the problem with this form of the given is that there is the possibility of concepts being misapplied. I’m not saying that we necessarily apply concepts consciously; I’m saying that when we impose a conceptual order on a state of experience such that we can form beliefs about the state of experience that can enter into the logical space of reasons, and can itself be used as a reason, then there is the possibility that those concepts have been applied incorrectly. So, we need a reason to believe that the conceptual order has been correctly mapped onto the state of experience, such that we have a properly subsumed manifold of experience that can itself provide justificatory reasons for other beliefs. Any reason to believe that we’ve properly applied our concepts to a state of experience would itself be a state that is conceptually ordered, and we would then need reasons to believe that those concepts have been properly applied. The regress ensues once again.
If one instead claimed that the state is non-conceptual, then the problem of conceptual mapping is not a problem. However, it isn’t at all obvious how a non-conceptual state can provide reasons for beliefs that are conceptually ordered; how exactly does some sort of causal state provide a reason to believe something? There is no inferential connection between a non-conceptual and a conceptual state. In other words, a non-conceptual state is not in the logical space of reasons, and so it cannot be a reason to believe something.
So, we have the dilemma that leads to belief that the given is a myth. The given is either conceptual or non-conceptual. If the given is conceptual, it is epistemically efficacious, but not epistemically independent, since there needs to be reason to believe that the concepts have been properly applied. (If one was an externalist instead, this wouldn’t be a problem. However, given Sellars’s theory of concepts, it would still be a problem for the externalist since to have one concept requires us to have a whole battery of concepts; so any application of a concept would presuppose some sort of knowledge of a whole host of other concepts. This assumes Sellars’s coherence theory of concepts, which is contentious.) If the given is non-conceptual, then it isn’t obvious how it could be a state that is in a position within the logical space of reasons such that it can transmit justification; it is not epistemically efficacious, even if it’s epistemically independent given its non-conceptual status.
Another point that needs to be made is that any doxastic mental state will necessarily be conceptually structured. An essential component of beliefs is that they are conceptually structured; if it isn’t conceptually structured, it isn’t a belief. An entailment of this fact is that no doxastic mental state can satisfy the epistemic independence requirement to qualify as an instance of the given. If a doxastic mental state is essentially conceptually structured, then the concepts involved in a particular state can, in principle, be misapplied. So, the conceptual mapping has the possibility of misrepresenting that which is being mapped. So, there is the possibility of that doxastic state being conceptually confused, which means that assenting to that state requires a reason to believe that the state being assented to is not conceptually confused. One needs a reason to think that the proposition ‘this doxastic state isn’t conceptually confused’ is true. But, whatever reason provided to believe that that proposition is true is itself open to the same challenge if it is a doxastic state as well. So, the regress problem can’t be solved by appeal to a doxastic mental state.
Willem deVries provides a concise reconstruction in his SEP entry on Sellars:
1. A cognitive state is epistemically independent if it possesses its epistemic status independently of its being inferred or inferrable from some other cognitive state. [Definition of epistemic independence]
2. A cognitive state is epistemically efficacious — is capable of epistemically supporting other cognitive states — if the epistemic status of those other states can be validly inferred (formally or materially) from its epistemic status. [Definition of epistemic efficacy]
3. The doctrine of the given is that any empirical knowledge that p requires some (or is itself) basic, that is, epistemically independent, knowledge (that g, h, i, …) which is epistemically efficacious with respect to p. [Definition of doctrine of the given]
4. Inferential relations are always between items with propositional form. [By the nature of inference]
5. Therefore, non-propositional items (such as sense data) are epistemically inefficacious and cannot serve as what is given.
[From 2 and 4]
6. No inferentially acquired, propositionally structured mental state is epistemically independent. [From 1]
7. Examination of multiple candidates for non-inferentially acquired, propositionally structured cognitive states indicates that their epistemic status presupposes the possession by the knowing subject of other empirical knowledge, both of particulars and of general empirical truths.
[From Sellars’s analyses of statements about sense-data and appearances in Parts 1-IV of EPM and his analysis of epistemic authority in Part VIII]
8. Presupposition is an epistemic and therefore an inferential relation. [Assumed (See PRE)]
9. Non-inferentially acquired empirical knowledge that presupposes the possession by the knowing subject of other empirical knowledge is not epistemically independent. [From 1, 7, and 8]
10. Any empirical, propositional cognition is acquired either inferentially or non-inferentially. [Excluded middle]
11. Therefore, propositionally structured cognitions, whether inferentially or non-inferentially acquired, are never epistemically independent and cannot serve as the given. [6, 9, 10, constructive dilemma]
12.Every cognition is either propositionally structured or not. [Excluded middle]
13. Therefore, it is reasonable to believe that no item of empirical knowledge can serve the function of a given. [5,11, 12, constructive dilemma]
deVries uses the term ‘propositionally structured’ rather than conceptually ordered, but the point remains the same. A propositionally structured mental state is going to be conceptually ordered, so the two forms of the problem overlap. The main point is that any propositionally structured mental state must be both epistemically efficient and epistemically independent to count as an instance of the given. But no propositionally structured mental state can fill both those roles, since any such state must if epistemically efficient will presuppose other knowledge, and so it cannot be independent. The only state that could qualify as epistemically independent is non-conceptual; such a state could not be epistemically efficacious though, since it could not provide the means to form inferential connections such that it can transmit justification to belief states.
To sum up: A basic belief must be both epistemically independent to maintain its status as basic, and epistemically efficacious to be capable of justifying other beliefs; otherwise it would be useless as a foundation for knowledge. However, if such a state is epistemically efficacious, it cannot be epistemically independent. Any instance of a conceptually ordered state is, in principle, capable of being an instance of misapplied concepts to the manifold of experience being captured; so, any such state presupposes that the concepts are correctly applied. However, once one admits that, it follows that any attempt to provide a reason to believe that the concepts have been correctly applied is itself open to the same question. If such a state is to be epistemically independent, then it cannot be epistemically efficacious. A non-conceptual state seems to be the only possible thing that could be epistemically independent, but it cannot also be epistemically efficacious. Non-conceptual states are not the sort of things that qualify as reasons. So, those states cannot fulfill the epistemic efficacy requirement. So, they cannot count as instances of the given.
deVries, W. (2005). Wilfrid Sellars. New York: McGill-Queen’s University Press.
Sellars, W. (1956). Empiricism and the Philosophy of Mind. In H. Feigl & M. Scriven (Eds.), Minnesota Studies in the Philosophy of Science (Vol. I, pp. 253-329). Minneapolis, MN: University of Minnesota Press.
(deVries’s reconstruction of Sellars’s argument can be found here:http://plato.stanford.edu/entries/sellars/#4)