Contemporary Epistemology: Positism

History

There is very little that has been said about the positist theory of the structure of epistemic justification, probably because there are so few adherents. It has been explicitly acknowledged as a distinct position only recently. James van Cleve’s contribution to Contemporary Debates in Epistemology (2005, link) is the first instance in which I encountered the term, and I have not found anything prior to that essay which mentions “positism” in an epistemic context. He defines it there as the belief that “chains of justifying reasons can terminate in reasons that are not justified themselves, but are simply individual or societal posits” (pg. 168). In a footnote to this essay, van Cleve writes:

The distinction between positism and foundationalism is lost on those who cannot hear the word “justified” as anything but a past participle, implying that some act or relation of justifying has occurred whereby a belief is justified by something else that serves as a reason for it. For foundationalists, “justified” simply connotes a favorable epistemic status, which a belief may have even though the subject has no reason for it. In this connection, another term, such as “evident” or “credible,” might be less misleading than “justified.” (pg. 178)

In other words, van Cleve believes a positist would admit his basic beliefs do not have a favorable epistemic status. It's just a belief one has, and on the basis of this belief other beliefs of his can be “justified,” or one can be “justified” in coming to believe other things. Ryan Herbert (link) defines positism similarly, stating positists hold that basic beliefs “can serve to justify other beliefs” (pg. 15), although these basic beliefs “are not autonomously warranted and… are neither epistemically justified nor unjustified” (pg. 14). [Herbert slightly misstates positism when he says this, as positists do view basic beliefs as epistemically unjustified. He also does not seem to distinguish, as Engel does, between doxastic and personal justification (see below).] Herbert adds that this encompasses both doxastic and propositional justification (pg. 15), which is significant in comparison to some who are infinitists about propositional but not doxastic justification. 

Both van Cleve and Herbert credit the origin of positism to what Ludwig Wittgenstein called “hinge” propositions: 

…the questions that we raise and our doubts depend upon the fact that some propositions are exempt from doubt, are as it were like hinges on which those turn.  

That is to say, it belongs to the logic of our scientific investigations that certain things are in deed not doubted.  

But it isn’t that the situation is like this: We just can’t investigate everything, and for that reason we are forced to rest content with assumption. If I want the door to turn, the hinges must stay put. (On Certainty, §§341-3)

Any act of questioning or doubting presupposes we believe certain propositions. In a skeptical context, question or doubt has to relate to some (or even all) propositions, but that is to already acknowledge, in some sense, the existence of this or these propositions, or at least that which gives this or these propositions meaning. In short, one can't deny, doubt, or question all propositions without rendering those denials, doubts, or questions unintelligible.

Wittgenstein goes on to say hinge propositions can’t be known, but that may just be due to his definition of knowledge: “One says “I know” when one is ready to give compelling grounds. “I know” relates to a possibility of demonstrating the truth” (On Certainty, §243). John Robbins had a similar definition, and I very much doubt he was a positist. In any case, it is obvious that on this definition, only something believed by inference can be “known.” If one were to have confronted Wittgenstein with a different definition of knowledge that would not have automatically ruled out basic beliefs – perhaps this happened – I am unsure of how Wittgenstein would have replied. Positism does remind me of the emotivist position taken up by the logical positivists (link), and Wittgenstein was involved with this latter group in the early part of his life. I still think it would be too ambitious to call Wittgenstein a positist on the foregoing evidence alone, although in an undeveloped or logically implicit sense it could be true.

Introduction to Engel's Paper

Surprising as it may seem, this is all that I could find that had been said about positism prior to last year. Perhaps it has been elsewhere discussed, but aside from these few scant references in papers by non-positist philosophers, I haven't encountered a self-aware defense of this position besides Mylan Engel Jr.'s “Positism: The Unexplored Solution to the Epistemic Regress Problem” (link). The reason is, I think, simple: philosophers have for a long time assumed epistemic justification is either intrinsic or transferred. Thus, there is no logical room for a position in which a belief is justified by an unjustified belief (link).

This assumption that epistemic justification is transferred has been recently challenged by a rising number of infinitist epistemologists, so it is not unexpected that at least one defendant of positism has seen and taken the opportunity to piggy-back on this objection and also claim “reasoning itself can be justification generating” (Metaphilosophy, Volume 45, Issue 2, 2014, pg. 152). In his abstract, Engel defines the transmittance assumption - which leads to an epistemic regress, as positists don't believe in justified basic beliefs - as follows: “Person S is mediately justified in believing p iff (1) S has a doxastic reason q for p and (2) S is justified in believing q.” Two defenders of infinitism, Jeanne Peijnenburg and Scott Aikin, also highlight this aspect of Engel's defense of positism:

Mylan Engel Jr.'s “Positism: The Unexplored Solution to the Epistemic Regress Problem” develops and defends the view that a justification-conferring chain of reasons may legitimately begin with an unjustified belief. Engel holds that, under certain circumstances, reasoning itself can be justification generating, and not just justification transmitting. He argues that a person S can be justified in coming to believe a proposition p on the basis of an unjustified posit R, provided S does not realize she is unjustified in believing R and she has no defeaters that defeat R's status as a reason for p. Engel maintains that when one believes a proposition p, one is rationally committed to what is knowingly entailed by p, unless one is prepared to abandon p in light of those entailments. Accordingly, the propositions one believes provides defeasible reasons for believing the propositions they knowingly entail. Engel characterizes his positist view as a form of nondoxastic coherentism that is compatible with other meta-epistemic views. In particular, Engel is keen to show that his view is compatible with the existence of basic beliefs grounded in sense experience and also compatible with a version of infinitism that holds that inference itself is justification enhancing. (Ibid., pg. 141)

So much for a general canvassing of Engel's aims. He intends to defend the legitimacy (though not justificatory exclusivity) of positism, accepting van Cleve's definition (pg. 146). Before delving into his arguments, I found there to be a lot of terminological nuance, enough to warrant a whole section of Engel's essay. I suppose this is to be expected in cases where not much has been said on a subject, so it's necessary to look at what Engel is specifically defending. I'll also break up my following sections to correspond to the main sections of Engel's article.

Terminology

Roughly, the principle distinction between the doxastic and propositional justification is that doxastic justification is concerned with the justificatory relationships among and statuses of one's actual beliefs, not merely potential propositions that could or should serve those purposes for an epistemic agent. The relationship between propositional and doxastic justification corresponds, respectively, to ex ante and ex post justification. This is the preferred terminology of Engel, who accepts Alvin Goldman's definitions in “What is Justified Belief?” (1979, pg. 21): “The ex post use occurs when there exists a belief, and we say of that belief that it is (or isn't) justified. The ex ante use occurs when no such belief exists... Here we say of the person, independent of his doxastic state vis-à-vis p, that p is (or isn't) suitable for him to believe.” Ex ante or propositional justification determines whether one may or should [have] acquire[d] a belief; ex post or doxastic justification determines whether a belief one already accepts should be kept or discarded. Accordingly, doxastic justification is sometimes said to be parasitic on propositional justification in that having the former entails having the latter but not vice versa. With this in mind, Engel proposes to evaluate positism in terms of ex ante justification (2014, pg. 148).

Engel further intends to evaluate positism with respect to “personal justification,” meaning he is concerned with the circumstances under which a person is worthy of epistemic praise for believing that p (cf. Engel, “Personal and Doxastic Justification in Epistemology,” 1992). Engel says that this sort of epistemic justification occurs when one believes “that p only if she has an undefeated reason for believing that p” (pg. 147). In this context, her reasons must be “internally accessible;” that is, her reason “to believe that p is a consideration, from [her] egocentric point of view, that suggests that p is true” (pg. 147). 

Engel then provides a sort of useful glossary of terms. Notably, a belief or experience is considered “basic” if a subject is “noninferential” and “immediately justified” in having it (pg. 148). Doxastic [foundations of coherence] theories assume “only beliefs can serve as reasons for other beliefs,” whereas nondoxastic theories reject this assumption (pg. 148). Most of what else Engel mentions may be here passed over, as their primary function just seems to be to show the number of contrasting theories of epistemic justification in terms of structure and nature. The specific theory that is worth mentioning is the one Engel defends. The following closes out this section:

...there is a third form of coherence theory that epistemologists have not recognized. Like foundations theories, modest nondoxastic coherence theories acknowledge that some reasoning is linear and admit that some beliefs are basic, for example, simple perceptual beliefs. Unlike foundations theories, however, modest linear noncircular nondoxastic coherence theories insist we can be justified in holding nonbasic beliefs that do not ultimately trace their justification back to basic beliefs. In section 3, I argue that a Positist version of this sort of Modest Coherentism (PMC) allows us to solve the regress problem for ex ante justification as it most frequently arises. (pg. 149)

Clearly, Engel's views indeed live up to his self-titled description as an epistemic “ecumenical beast” (pg. 157). He thinks epistemic justification can be had in a number of ways, but while some of these variations are interesting, I will restrict myself to commenting on why and how he thinks positism suffices.

The Epistemic Regress Problem for Ex Ante Justification

Engel begins section 2 of his article with an important historical point: 

...the epistemic regress problem was put forth as a skeptical challenge designed to undermine the very possibility of rational belief. In its contemporary guise, the regress problem has been formulated as an argument from elimination in favor of foundationalism - one designed to show that basic beliefs must exist if we are to be epistemically justified in believing anything at all. (pg. 149)

One comment that should be made at this point also relates back to Engel's epistemic ecumenicism, and that is with regards to the weak standard Engel has set for the goal of epistemic justification to be met. I don't have a problem with much of Engel's terminology outlined in the previous section, but if all that it takes for one to be justified in coming to believe p is that he have an undefeated (but not undefeatable) reason for it, and if what suffices as a reason is that it merely suggests to his mind that p is true, then one could easily imagine scenarios in which two people could be justified in coming to believe contradictory propositions. Does Engel think this qualifies as rational belief? According to the original context of the skeptical challenge to which Engel alludes, I don't see how he could. But then I fail to see how the skeptical challenge can be addressed by Engel in the first place. It would be one thing if Engel wished to propose positism in a different context, but in the beginning of this section he has made it clear he knows the original point of the regress argument and believes PMC provides a solution. But unless he were willing to argue against the much more stringent (indeed, infallibilistic) skeptical standard of epistemic justification - which he does not do here, in any case - that would require he deal with the skeptical challenge on its own terms, not his. The force of this point will, I hope, become clearer in the following discussion.

Next, Engel lays out premises of the regress argument for ex ante justification:

A1. S is justified in coming to believe that p iff either (1) S is immediately justified in coming to believe that p or (2) S is mediately justified in coming to believe that p. 

A2. S is mediately justified in coming to believe that p iff (1) S has a doxastic reason q for p (where q might be a conjunction), and (2) S is ex post justified (either mediately or immediately) in believing q. (pgs. 149-150)

He then reasons how a foundationalist would use these premises to argue that foundationalism must be true. Clearly, he doesn't believe this succeeds. He makes three points, none of which I believe are persuasive.

Firstly: “Even if sound, all RA shows is that there must be basic beliefs, if we are to have any justified beliefs at all, and it remains a theoretical possibility that justification skepticism is correct” (pg. 151). One may claim “no one is justified in believing anything,” but those words only have intelligible, definite meaning in the context of the claimant being epistemically justified in assigning said meaning. Otherwise, it could mean anything, including “everyone is justified in believing anything,” In other words, justification skepticism cannot be consistently maintained, and any argument to the contrary presupposes this is true, Engel's own argument to this effect included.

Secondly:  “...even if there are basic beliefs, as RA allegedly shows, in order for foundationalism to be correct, there must be enough basic beliefs to support the structure of our justified nonbasic beliefs, and RA does nothing to show the latter” (pg. 151). He goes on to say he does believe perceptual beliefs are properly basic (I'm not sure if Engel distinguishes between basic beliefs and properly basic beliefs) if they are grounded in perceptual experiences, but he argues many nonbasic beliefs we think we are justified in believing, like moral and philosophical beliefs, cannot be traced back to basic beliefs. This is backwards, and Engel repeats this mistake a few times. A foundationalist reasons from foundations, not to them; he only is in a position to know what is a justified nonbasic belief because he was first in a position to know what is a justified basic belief. Assuming certain nonbasic beliefs are justified and then using that to rule out the possibility of there being a sufficient number of justified basic beliefs to account for these is to put the cart before the horse. If this is the external critique it appears to be, a foundationalist will find it unpersuasive simply because he does not operate from the same bases as Engel, so to speak.

Thirdly: “Take the current debate between foundationalism, coherentism, infinitism, and positism. No matter which of these positions you believe is correct, you won't be able to trace this philosophical belief back to properly basic beliefs” (pg. 151). Engel cites Plantinga in his footnote to this comment, further adding that this would make foundationalism self-refuting. But why Engel thinks, for instance, that the belief “divine revelation is self-authenticating” cannot be basic, I don't know. I think that belief could put one in a position to make the regress argument for foundationalism. If Engel's flat denial is due to people disagreeing about things regarding God, I don't see how that is relevant. People can disagree about everything. Does that imply there can be no incorrigible beliefs? No. Here again the importance of discussing infallibilism is demonstrated.

Justification Ex Nihilo

Having spelled out the regress problem and addressed why he thinks foundationalism fails, Engel turns to his own view, which is primarily targeted at explaining how reasoning itself can be justification generating rather than justification transmitting.

In the first subsection, Engel makes the argument that one is “justified in believing q” if he also believes “that p and that p entails q,” even if he is unjustified in believing either that p or that p entails q (pg. 152). I don't think this is true. The internal logic of the individual should lead him to believe q, but this is an entirely separate question from whether or not he is justified in believing q. This is true even on Engel's already watered down definition of epistemic justification as one's having an undefeated reason for q, not to mention more robust perspectives. If one encounters a defeater for p and still believes that p and that p entails q, would Engel say that he is still justified in coming to believe q? If so, he must revise his definition of epistemic justification. If not, he must admit that it is not “his commitment simpliciter to p and p entails q that rationally commits him to q” (pg. 153). It appears the second is the route Engel opts for, as in the next subsection he prescribes N2, a norm stating “If you believe that p and that p entails q, and you care about whether q, then if you have no defeaters for p as a reason for q, believe q” (pg. 155). So I missed the point of this subsection.

In his second subsection, Engel discusses justificational opacity and norm followability. He first discusses what epistemic norms are, writing that “S is justified in coming to believe that p if and only if S would not violate any epistemic norms by coming to believe that p” (pg. 153). If epistemic norms are usually merely permissive rather than mandatory, as Engel says they are, then I would say that rather than calling S justified in coming to believe p, it makes more sense to say he is just permitted to believe p. Epistemic justification connotes a favorable epistemic status, whereas epistemic permission norms seem more neutral. Regardless, Engel next argues that 

On the plausible assumptions that (i) we ought to be able to reason the way we should reason and (ii) ought implies can, the correct epistemic norms, whatever they are, must be such that we are actually capable of following them and guiding our beliefs in conformity with them. Accordingly, an adequacy constraint on any regulative epistemic norm is that it be followable. A person cannot be criticized for failing to follow a norm that is impossible to follow, for no one can be expected to follow an unfollowable norm. (pg. 153)

To begin with, notice that Engel is willing to accept plausibility assumptions only when they suit his purposes. Justification skepticism isn't plausible, but that was nevertheless cast as a mark against foundationalism. Plausibility should work both ways in these discussions. But more importantly, freedom doesn't ground or follow from norms. This would transition into a discussion of moral ontology, which I have written about elsewhere (link). Suffice it to say as a person who thinks determinism is an epistemic necessity and that persons do have moral obligations, I disagree with this assumption.

In any case, Engel mentions these two assumptions to argue that premise A2 of the regress argument for ex ante justification can be formulated as unfollowable norm N1: “For any belief B1 that you hold, employ B1 as a reason for some new belief B2 only if you are justified in believing B1” (pg. 153). There several problems with this. Firstly, the aforementioned distinction between permission and justification becomes evident, as one may be permitted to believe B2 even if B2 cannot be deduced from justified belief B1. This is permitted as long as ~B2 also cannot be deduced from a justified belief, so N1 only holds if it is further stipulated that one wishes to be justified in coming to believe B2. That is very different from just wishing to be permitted to hold B2. But further, even if it is “psychologically unrealistic” (pg. 154) to separate our justified and unjustified beliefs, it is not unfollowable, it's just hard. We typically don't want to do that except in philosophical or apologetic contexts. But the point is that even on Engel's “plausibility assumptions,” A2 can hold.

Engel then attempts to bolster his case by providing memorial beliefs as an example of how A2 can't be followed: “Given the justificational opacity of memorial beliefs, we can't follow the epistemic norm (i.e., N1) implicit in A2. Consequently, A2 is false, where the regulative conception of justification is concerned” (pg. 154). Setting aside what points I've already made in regards to unfollowability, Engel's short discussion of memorial beliefs really seemed out of place in the context of his article. For starters, doesn't any reasoning session engage in memorial beliefs? As temporal creatures, it appears to me we take time to think any proposition. Doesn't that mean each belief we have is in some sense memorial, as we have to remember that to which each concept in a proposition refers? Or does Engel believe propositions and even arguments can be comprehended instantaneously? The latter appears dubious to me - Engel doesn't even define what he thinks a memorial belief is, let alone address any of these relevant concerns - but the former calls into question Engel's assertion that “we can't tell introspectively which of our memorial beliefs we are justified in holding and which we aren't” (pg. 154), as he must have assumed the opposite in believing himself to be justified in arguing this position. Either way, I think it was a mistake to squeeze one page worth of material about a heavy metaphysical topic like memorial beliefs into a discussion about the structure of justification and think it decisively or forcibly shows anything. I certainly did not follow the logic of the following statement from subsection 3: “Since we can't tell introspectively which of our memorial beliefs we are justified in holding and which we aren't, we rightly regard all memorial beliefs as having a default (albeit defeasible) permissible-to-reason-from status” (pg. 155). 

He ends this second subsection by providing N2 (defined above) as an alternative method of reasoning to N1, promising that the next section will explain why people do and are right to reason in thusly. Now, I don't necessarily have a problem with N2. I would say it's a fine permission norm. Engel even makes the useful point (and again on pg. 157) that Peter Klein, a notable infinitist about propositional justification, seemingly accepts N2 as a permission norm about doxastic justification and has really committed himself to positism. But to recap above criticisms, as a solution to the regress problem in the context of the skeptical challenge, I think it doesn't suffice. 

Unfortunately, this next subsection mostly explains how people reason, not why it is right. Engel notes we often reason from premises we don't question until we encounter a defeater, at which point we see if there are any defeaters of said defeater. How we proceed from there depends on what we find. He compares our private reasoning to our public reasoning as well as computer simulations. This is descriptive language. The closing paragraph is the closest thing I found to justification for N2:

Reasoning in this way is extremely efficient. If we had to stop and reevaluate our beliefs each time before we reasoned from them, we would draw very few conclusions. The point is not merely that we do regularly engage in default reasoning but that, given both the efficiency and the self-correcting nature of such reasoning, it is entirely rational for us to do so. Objection: But won't engaging in default reasoning make us prone to countless irremediable errors? Response: No, for when we employ default reasoning, we're constantly engaged in what John Pollock (1986, 56-57) describes as “primed research” - the subconscious monitoring of our reasoning, constantly being on the lookout for reasoning errors and potential defeaters for our reasons. In this way faulty reasoning and faulty beliefs are constantly getting corrected, as new information become available. (pg. 156)

From this account, it seems the purpose of positism is utilitarian. If the point of reasoning this way is to arrive at more conclusions rather than truly justified conclusions, it does make more sense to allow for more premises than basic beliefs. And given the existence of permitted beliefs in addition to mandatory ones, this is fine. But then positism becomes less about solving the regress problem and more about its indirect value in being unwittingly used in other, less strict contexts than epistemology. 

Additionally, it seems positism only works within the context of a different theory of structural justification. Notice that Engel hastens to assure the reader that this view is self-correcting (retroactive) and includes primed research (proactive). While this never seems to get us to the point of infallible justification in believing anything, upon further consideration, it might already presuppose it. For in accordance with what standard[s] are we correcting or protecting our beliefs from faultiness? How is it we can sense when something we believe or encounter is erroneous? If whatever standard Engel would have in mind can itself be revised, this deflates his response to the posed objection: we might be correcting our beliefs according to a standard which itself needs or will be corrected. Who knows what is erroneous in such a case? If, on the other hand, the standard is a necessary one, then it seems we have a belief which has been justified by some other structural theory than positism, as positism is fallibilistic. Given Engel's epistemic ecumenism and prior defense of coherentism, this makes sense. But then once again, positism seems to be less about philosophical concerns and more about its potential usefulness as an add-on to some extant, self-sufficient theory of epistemic justification.

In the last subsection, Engel attempts to outline a sufficient condition for positist justification (pg. 156): 

(PJ) S is justified in coming to believe that q on the basis of her belief p, which she's ex post unjustified in believing, IF: 

(i) S believes that p,

(ii) S believes that p entails q (or that p makes q sufficiently probable), 

(iii) S appreciate the fact that (p & p entails q) is a reason to believe q,

(iv) S does not realize that she is unjustified in believing p,

(v) S has no reason to believe ~q, i.e., S is not aware of any rebutting defeaters for p as a reason for q,  

(vi) S has no reason to deny that p would not be true unless q were true, i.e., S is not aware of any undercutting defeaters for p as a reason for q, and 

(vii) S has no reason to believe ~p, i.e. S it not aware of any negating defeaters for p. 

I don't have much more to add to what I've already said. I will say that this seven-step process by which one can allegedly be justified in coming to believe q doesn't strike me as plausible or psychologically realistic per Engel's earlier remarks. I am particularly curious how a self-conscious positist - such as Engel appears to be - could exercise PJ(iv). Once you've done all that work in defending or coming to understand how you are justified on positist grounds, I would think it would be hard to fail to realize what one's unjustified premises are. That may just be because I tend to think about epistemology a lot, but then again, you'd probably have to think about epistemology a lot to become a positist in the first place. 

He finishes this section by noting p can itself come to be justified either by providing a reason for it, whether directly due to some new belief or because “the best explanation for why p has given rise to so many other beliefs that cohere with the rest of her belief system is that p is true” (pg. 157). It isn't clear whether Engel now believes, in light of this paper, which, if any, structural theory of epistemic justification can stand alone.

The Many-Solutions Solution to the Regress Problem and Conclusion

Engel wraps up his article with these two short sections, summarizing his position in them. The first section consists of what Engel perceives to be the positive and negative qualities of the various theories of epistemic justification, a discussion which can for these purposes be put aside, interesting though it may be, because it doesn't really concern positism.

In his conclusion, Engel clarified that his rejection of the transmission-only thesis is mainly grounded on the unfollowability of A2/N1, which is why I paid so much attention to the second subsection above. He finishes with a distinct point that wasn't mentioned in the body of the paper but is worthy of mention:

As for the arbitrariness thesis, PMC's combination of default reasoning and primed search for errors mitigates against the charge of arbitrariness. Although S might have acquired some belief B inappropriately/arbitrarily, the fact B has persisted, that is, B hasn't been purged due to countervailing considerations, despite S's having been on the lookout for such considerations, makes B more than just an arbitrary belief. The longer B survives this primed-search-self-monitoring process, the less arbitrary B becomes, True, even if B persists indefinitely, S won't be justified in believing B, unless she acquires an undefeated reason for B, but S needn't be justified in believing B in order for B to cease to be arbitrary. (pg. 158)

I would have been interested to see Engel address this more fully. For instance, at what point can B entirely cease to be arbitrary? What does it mean for B to be arbitrary or non-arbitrary, given that it can in both cases be unjustified? I didn't even realize this was a concern until the last paragraph of the paper.

To summarize my findings, I do not believe positism is a solution to the regress problem, nor do I think it can be synthesized with any other theory to nevertheless epistemically justify individuals in certain cases. It may have some function in the context of permitted beliefs.