The Factivity of Reasons and Evidence

One challenge for philosophers who want to maintain an equivalency thesis (e.g., Kearns and Star 2008, 2009) between normative reasons and evidence is that evidence is non-controversially regarded as always-factive whereas reasons are controversially regarded as always-factive. For Michael Smith and Jonathan Dancy a normative (justifying) reason for p does not require p to be true (Hornsby 2007: 294 n. 8). Jennifer Hornsby argues, along with Clayton Littlejohn (2010) in his Notre Dame book review of Fantl and McGrath (2009), that normative reasons are factive. By contrast, I’m not aware of anyone who argues that evidence is non-factive. Beliefs about evidence might be fallible, but evidence is almost always regarded as facts that are true.

    A move an equivalency theorist might make is to claim evidence justifies belief with regard to truth-directedness. A belief p is justified in relation to normative reasons for holding p. From a third-person perspective only reasons that are true will justify belief. Such reasons align with evidence because their justificatory power derives, in part, from their factivity. This, however, is not a move an equivalency theorist will be comfortable with. There are also subjective reasons, and these reasons are not always factive. Such a theorist posits a symmetry thesis:

  • Symmetry: Normative reasons and evidence are symmetrical with regard to subjectivity and objectivity.

According to Symmetry one can speak of subjective reasons and subjective evidence along with objective reasons and objective evidence. What is objective is always factive and what is subjective is not always factive (or, is non-factive). One problem with this neat division is that cases (like the Coop-Petrol case) apply pressure to this divide. It seems that normative reasons must always be factive. Littlejohn (2010) mentions that Fantl and McGrath (2009) are committed to saying:

(1)   What justified Coop in giving Audrey the toxic stuff was that there was gin and tonic in that glass.

And, (1) entails (3):

(3)   Coop was justified in giving Audrey the toxic stuff (in part) because there was gin in the glass.

A factive theorist can point out that (1) entailing (3) is only a problem if p is true and q is true. That is, Coop was only justified in his reason for acting as he did if there was gin, and not petrol, in the drink he gave Audrey. However, there was not gin in the drink, so it’s not true that ‘p because q’ is true. Thus, Coop was not justified in his action. Fantl and McGrath try to block (1) entailing (3) by resorting to subjectively-based (2):

(2)   What justified Coop in giving Audrey the toxic stuff was that there was gin and tonic in that glass, as he thought at the time.

I agree with Littlejohn that (2) doesn’t block the entailment. That Coop thought that the clear stuff was gin and tonic doesn’t prevent the implication in (1) that there was gin and tonic in the glass and this reason justified Coop in acting as he did. So, it does not block (1) from entailing (3). If one takes Littlejohn’s suggestion that normative reasons are factive, the next question becomes whether evidential beliefs must be true. Timothy Williamson thinks evidential beliefs must be true, whereas Jim Joyce (2004) argues against this idea.

    For Williamson, “If e is evidence for h, then e is true” (2000: 201). If something is inconsistent with the facts, then it’s not evidence because, “No true proposition is inconsistent with my evidence, although I may think that it is” (2000: 201). According to Joyce, Williamson holds his factivity view because he conflates subjective and objective senses of justification. Williamson is only focusing on the objective (third-person) form of justification, so it’s no wonder that beliefs justified by evidence must always be true. Evidence is always factive for Williamson, so that which it justifies must also be factive.

    Joyce, like Fantl and McGrath, ask us to consider the subjective sense of ‘justifies’ in terms of a subject’s reasons that she possesses for a belief. These justifying reasons need not always be factive, and, as Joyce points out, an agent will only mistakenly consider evidence justifying a belief if there’s a false belief present. For instance, this occurs when a juror holds the false belief that a witness is reliable and believes based on that false belief that the witness’ testimony is evidence for the innocence of the defendant even though the witness is lying.

    I’ll close by pointing out a couple of assumptions underlying the Joyce and Fantl-McGrath response. These assumptions might be undermined as a way of arguing for the factivity of normative reasons and evidence.

  • Assumption 1 (AS1): ‘Objective’ goes with ‘factive’, whereas ‘subjective’ goes with ‘non-factive’.

(AS1) implies that ‘objective’ is agent-neutral (applying to external facts), whereas ‘subjective’ is agent-relative. Another way of understanding ‘objective’ is in terms of ‘abstraction’. Even if something is abstracted away from an agent it may none-the-less be relative to an epistemic situation and by implication be relative to the mental state of a (hypothetical) agent. So, ‘objective’ can still be agent-relative in that sense. It might be that an actual agent can inhabit such an epistemic state. What matters is that an actual agent is not required to inhabit such a state for there to be evidence or reasons.

  • Assumption 2 (AS2): An agent only has (possesses) evidence or reasons subjectively. Objective evidence (or reasons) is evidence there is, where subjective evidence is evidence an agent has.

Contrary to (AS2) it’s possible to argue that subjective access to evidence, through armchair reflection, is not required for evidence possession. If evidence is factive, then an agent may possess such evidence even though an agent is not subjectively able to recall or grasp that evidence.

53 Comments

Add yours →

  1. Hi Chris,

    I have some more specific comments, but, first, I wonder if you can tell me where you get your sense that it’s “uncontroversial” that evidence must be true. Both the thesis and the historical claim strike me as untrue. For one thing, all those who hold that experience is evidence won’t think that. Also, about anyone who holds a recursive theory of evidence won’t hold it, for we can have justified false beliefs.

    Russell holds the view that evidence can be false explicitly, toward the end of “On Intuitive Knowledge,” I think false propositions can fit Chisholm’s definition of evidence, and I think Richard Jeffrey thought falsehoods could be evidence.

    Conee and Feldman certainly don’t hold the factivity thesis, it’s not clear to me that BonJour does either. And a proposition wouldn’t have to be true to play the three primary functional roles of evidence: that which rational people rely upon in belief formation, that which is a sign of truth, that which is an arbiter of disputes.

    So maybe I’m totally misunderstanding something here. It’s happened before.

    • Hi Trent…Thanks for the references. Perhaps some philosophers have endorsed false propositions counting as evidence. I’m open to the idea that there is controversy, but I’m understanding “controversial” in terms of what is “non-controversially desirable.” It may turn out that endorsing the non-factivity of evidence is a consequence of one’s theory, but I do not take it to be desirable to advocate false propositions counting as evidence in reference to the goal of securing true hypotheses, a goal which is pragmatically and epistemically desirable. Endorsing the opposite thesis seems undesirable because it leads to examples like: it is not the case that the wall is red is evidence that the wall is red. Even though being falsely appeared to can *seem* like evidence from one’s perspective, it is not evidence for the truth of the hypothesis if, in fact, the wall is green and there’s just a red light shining on the wall. We have all sorts of false beliefs and experiences that do not count as evidence for a range of hypotheses. I’m not sure what counting them as evidence gets us with regard to the goal of securing true hypotheses. I’ve heard an iterative argument before: mounting up falsehoods may eventually lead to an approximation of the truth of the hypotheses. But, I’m unmoved by such arguments.

      I’d also like to hear the “specific comments” you mentioned. Thanks.

  2. A couple of things.

    1. I don’t know what it means for something to be “count[ed] as evidence in reference to the goal of securing true hypotheses”. If E epistemically supports P then E is evidence for P even if your goal is to seduce unsuspecting junior faculty into publishing their dissertation with you (sorry, the subject of a recent scam).

    2. You say “Endorsing the opposite thesis seems undesirable because it leads to examples like: it is not the case that the wall is red is evidence that the wall is red.” Why think that? It doesn’t immediately follow from “Some false propositions can count as evidence.”

    3. You say “We have all sorts of false beliefs and experiences that do not count as evidence for a range of hypotheses. I’m not sure what counting them as evidence gets us with regard to the goal of securing true hypotheses.” I’m assuming that “experiences” is not in the scope of “false,” but I don’t understand the force of the claim that we have some experiences that don’t count as evidence for some hypotheses. That seems trivial. I’m an empiricist. A popular empiricist doctrine is that experiences are ultimate evidence (there’s a great article by Firth by the title “Ultimate Evidence” which discusses empiricist theories of evidence). The experience as of seeing a red wall counts as evidence that there’s a red wall whether the wall is red or not. I think that false beliefs count as evidence arises from two very plausible theses:

    1. Once evidence is what is evident to one.
    2. What is evident to one can be false.

    I don’t think “p is evident to S” entails “p”. I’m a phenomenal concervative, a credulist, a commensensist (“If it walks like a duck, quacks like a duck…”). So I find some reasoning relevantly like this just fine.

    It looks like a birthday present, and I have no reason to doubt this, so probably it’s a present. And if it’s a present, it must be my birthday (suppose 99% of the time there are presents in my presence it’s my birthday). So, probably, it’s my birthday! “Happy birthday to me!”

    Now suppose it’s not my birthday and you wonder why I have just declared it is, you say: “What’s your evidence that it’s your birthday?” I’m going to say: “I got a present.” Ex hypothesi “I got a present” bears an objective epistemic supports relation to “It’s my birthday.” I don’t want any more out of the term ‘evidence’ than the active relatum of the supports relation. And my experience objectively supported something that objectively supported my conclusion (transitivity of confirmation does not hold universally, but this has nothing to do with the kinds of cases that are exceptions). And from the subjective side “I got a present” is what I conditioned upon when updating my probability for “It’s my birthday.” I don’t want anything more from evidence on the subjective side than conditionalization.

    Another way to think of it is that my evidence for a proposition should include my *premises* for it. And I can certainly reason from false premises.

    I think the intuitive connections between evidence, epistemic support, conditionalization, and premises weave a tapestry which includes some false beliefs as evidence.

    Nevertheless, for the empiricist all ultimate evidence is experiential and this seems to bypass the whole debate at the fundamental level.

  3. You raise a number of important points. I think I’ll address them in separate replies so that I can explain my responses without trying to get to everything at once. You could also intersperse replies if you’re so moved.

    In point (1) you mention the important thing is whether “E epistemically supports P,” as this is what makes E evidence for P. A distinction might clarify things and better explain my argument. To borrow terms from Carnap (1962), I was using a ‘classificatory’ notion of confirmation (support) and you responded with a ‘quantitative’ notion of confirmation. On the classificatory reading, whether E “counts” as evidence for P is a binary relation. P is either confirmed by E or it’s not. On the quantitative reading, P is confirmed by E to degree R. I’m assuming you understand epistemic support probabilistically, and that if E is positively relevant to P, which is indicated by increasing its probability, then E is evidence for P. With this distinction in place let me provide my argument in quick and dirty form:

    1. Credences are estimates of truth-values (Joyce 2005).
    2. A credence is assigned to a hypothesis conditional on a body of evidence: Prob(P|E).
    3. Evidence E confirms the truth-value of P from a classificatory perspective if E makes P more likely than its negation (~P).
    4. Thus, E “counts” as evidence for P only if the strength of E (in relation to P) crosses a well-defined threshold (.5).
    5. If E consists of falsehoods, then E will have a low credence value in relation to P because credences are estimates of truth-values.
    6. This means E will never cross the well-defined threshold.
    7. Thus, E will never “count” as evidence for P.

    There are other ways of understanding the Carnapian concepts I used. Contemporary renderings are as follows: incremental vs. absolute confirmation (Maher 2005), categorical vs. gradational scales of fitting the facts (Joyce 2005).

  4. Regarding your point (2) there might be a better way of situating my claim. You may recall Feldman’s (2005: 111-113) use of Pollock’s distinction between undercutting and rebutting defeaters. Feldman uses the distinction to fend off skeptics who claim that perceptual evidence is not good evidence in red light cases. Feldman uses the red light case to identify a different type of undercutting defeater. It’s not that there’s no evidential connection between color appearance and the actual color; instead, as Feldman says, “It remains true…that a thing’s looking red is a good reason for believing that it is red” (112). The evidential connection remains intact, but the undercutting evidence (i.e., a red light pointing at the wall) indicates that the typically reliable connection between ‘appearance’ and ‘actual’ cannot be relied on in this case.

    My claim is that false perceptions or appearances are not and should not be “counted” as evidence. There are a couple of ways for me to go. I could claim something analogous to what Christensen (2010) says about higher-order evidence, namely, that the defeater destroys the connection between the evidence and the conclusion. So, in red light cases, there is no evidential connection between ‘appearance’ and ‘actual’. This is to embrace the skeptic’s position whole stock. Or, I could adopt an attenuated reply that claims the connection between evidence and conclusion is left intact, so false appearances are to some degree evidence, but the defeat in such cases casts doubt not on the evidence in such cases per se but casts doubt on the general reliability of inferring from ‘appearances’ to what is ‘actually’ the case. To assume otherwise is to beg the question by assuming empiricism in order to support an empirical understanding of evidence. In this way, the red light case indicates that the senses cannot be trusted. I like endorsing a version of Christensen’s claim: undercutting destroys the evidential connection.

    A motivation for this claim is that it avoids overly subjective, unnecessary qualifications. This is because false propositions about appearances result in qualifications if one wants to tell a coherent theoretical story or count them as evidence (i.e., from her perspective, given what she knew, based on the condition of her senses, etc.). Though I’m not a full-blown rationalist, and I’m open to being persuaded in an empiricist direction, I take the unreliability of the senses to indicate the need to not count false propositions as evidence. That is, the undercutting defeater destroys the evidential connection.

  5. Sorry, just got internet again.

    “5. If E consists of falsehoods, then E will have a low credence value in relation to P because credences are estimates of truth-values.”

    That strikes me as wholly misguided since *estimates* of the probability of falsehoods can be arbitrarily high.

    If I *estimate* that Pr(p)=high, then even if p is, unbeknownst to me, false, I’ll still conditionalize on it, and rightly so.

    • No worries. Good hearing from you.

      That seems right at a given time, in a given epistemic situation. An agent can only work with what she’s got, and it may result in arbitrarily high probability estimates and something counting as evidence which later turns out to be false. I might amend #5 to include conditionalization over time by which the credences will accurately settle to reflect the actual state of the world. As the agent gathers new empirical data the falsehoods will wash out as they fall below the threshold and, thus, no longer support the truth of the hypothesis. This is an attempt to stick within a subjective Bayesian approach. However, I prefer to endorse a more objective account of probability in which case the *estimates* are not tied to what an agent knows (or can cognitively access) at a given time.

      For example, on Salmon’s version of frequency evidence whether e is evidence for p depends on whether the “weights” (as understood by Salmon and Reichenbach) are such that w(p|e) > w(p), or w(p|e) > 1/2. This is an objective rendering of evidence that does not depend on a subject’s knowledge or beliefs.

      I’m not saying I endorse the frequency understanding of evidence. I only use it as an example to show that the subjective Bayesian understanding of probability and evidence is not the only game in town. And, endorsing an objective account of evidence avoids arbitrarily high estimates of probability based on empirical data an agent lacks given her epistemic situation.

  6. On your other comment, it sure sounds to me like you’re endorsing skepticism.

    It just doesn’t seem complicated to me. Red appearances defeasibly support that there’s a red thing. If we get defeat, the balance of evidence changes. It’s just so simple, and I see not a single problem for it.

    • “…it sure sounds to me like you’re endorsing skepticism.” I’m not endorsing global skepticism only local skepticism about the initial reliability of the senses and appearances. The skepticism is about what I’m willing to initially grant. To qualify as evidence, and for an agent to earn her beliefs, they must be defeasibly established over time. In the red light example we quickly gain evidence that ‘appearance’ does not match ‘actual’. Why say, “what once was evidence is no longer evidence because it was defeated by further evidence?” It is no credit to the agent as a good epistemic agent that things appeared to her a certain way initially. Evidence and beliefs must be earned. That is, they must possess positive epistemic status, and such status can only be established over time. The unreliability of the senses warrants such initial skepticism.

      What I just said denies “very plausible” thesis (1) in your comment #3 above. Evidence is not what is evident to one. There’s a distinction between what is ‘evident’ (appears to be the case) and what is ‘evidence’ (is actually the case). An analogous distinction is between what is ‘intuitive’ and what is an ‘intuition’. Just like many intuitions are not intuitive (e.g., complex propositions) many pieces of evidence are not evident (e.g., undiscovered relevant facts). So, false beliefs do not count as evidence because they belong to what is ‘evident’ but not what counts as genuine ‘evidence’.

  7. Neither objective nor subjective probabilities are relevant to epistemology. That’s why they call them *epistemic* probabilities. Objective probabilities are relevant only if we know them and calibrate.

    Epistemology is *normative* and normativity must be sensitive to agents situations. Or at least the only kind of normativity I’m interested in is the kind we can actually hold people responsible for, and that’s always going to depend on an agent’s total perspective.

    • I find your comment interesting given your commitment to empiricism. Given that evidence finding and justification is an empirical enterprise I’m not sure why epistemology would not stand to gain from an empirical understanding of probability and evidence. I think epistemologists can learn from philosophers of science in this regard. It might also make epistemology more relevant to science. There’s even a paper that supports this idea. It’s by Peter Achinstein and it’s titled, “Why Philosophical Theories of Evidence Are (And Ought to Be) Ignored by Scientists.” (Phil. of Science)

      I think epistemology will come-of-age when it moves out of the first-person domain and third-person epistemology becomes the norm.

      It also seems to be a false dilemma that only if epistemology is limited to epistemic situations can people be held responsible for what they know. Otherwise, they cannot. This implies that we are not rationally criticizable for things outside the scope of what we know, things for which we are unaware. This line of thinking lends itself to an ‘ignorance defense’. I was not aware that P based on what I knew at the time. I think ignorance is not a defense even when it comes to epistemology. People can be held accountable for things they do not know in a given situation. Taking this position, hopefully, motivates agents to be better epistemic agents through fact finding and belief revision.

  8. Unsurprisingly, I don’t buy the distinction between the intuitive and intuition either.

    “The unreliability of the senses warrant such initial skepticism.”

    What is that judgment based on? What *could* it be based on?

    If you don’t give seemings prima facie normativity, you’ll never bootstrap your way out of it.

  9. If you look at the early logical positivists, I think you’ll see the situation vis a vis science is quite the opposite. Remember that the Vienna Circle began as the “Earnst Mach Society” and Mach’s position is very similar (perhaps maximally similar) to the one I’m advocating. I strive for a view as close as possible to logical empiricism, and it was born, at least in part, of the desire to have an epistemology fit for science.

    We can never escape the first-person perspective, so any epistemology in the third person would be parody. And I am the anti-Achenstein. In my Oxford Bibliography on evidence, I think I list a few replies which I find decisive. I think one of them is by Sherrilyn Roush.

    No one can ever be held accountable for what they did not know. Ever. What they can be held accountable for is *not knowing it*. That does all the explanatory work.

  10. I find it interesting that you endorse logical empiricism, a theory that was discredited by the middle to late 1970s. Some reasons for the decline of logical empiricism include: a breakdown of the view of language that undergirded many empiricist ideas, holistic arguments, problems with developing inductive logic, the growth of history and psychology in the philosophy of science, and pressure from scientific realism.

    Regarding language, a difficult question for logical empiricists is whether scientific language only tracks the flow of experience or patterns in observables? Is a sentence describing a hidden structure of the world meaningful? Is a ‘quark’ sentence merely tracking patterns of observation? How can this be if quarks are largely unobservable? Were scientists talking nonsense prior to observation of the fundamental constituents of matter? Much of science seems to posit the hidden, yet real, structures of reality, which are only later observed as technology catches-up to science. Empirical observation is an antiquated model of scientific discovery and progress.

    We’ll have to agree to disagree regarding the first-person question. I think many problems in epistemology arise from insistence on a first-person perspective. I think a third-person perspective injects an ideal observer or a logically omniscient being (to borrow a phrase from Swinburne 2001: 64) in the right places. This allows for greater objectivity and apriority in the space of logical possibilities and inductive probabilities. The shift from the first-person to the third-person involves better approximations of the ideal.

    As you might have guessed, I do not find the Roush reply to Achinstein regarding ‘positive relevance’ decisive. You might read the Achinstein 2004 (Phil of Sci) reply to Roush. While I’m not a devotee of Achinstein, as I disagree with some of what he says, I do find his move to couple objective epistemic probability with an explanatory conception of evidence attractive. Even if I don’t agree with all the details I do think he’s heading in a positive direction.

  11. Hey Chris and Trent,
    I’m late to the party, but I wanted to say a few things in support of factivity.

    In “Evidence” C&F say that they want to remain neutral on the question as to whether evidence is propositional, so given their further insistence that “ultimate evidence” is connected to experience, they might defend one of two views:

    V1: Evidence consists of the propositions that are the contents of experience.
    V2: Evidence consists of propositions that describe a subject’s experiences.

    I don’t see that they argue against the factive conception of evidence. Indeed, they seem to implicitly assume it when they argue against McDowell. In arguing that he’s wrong to say that our evidence can include “worldly” propositions (e.g., there is a tree over there), they say that in the case of hallucination, someone could have that experience even if there’s no tree. The only reason I think _that_ observation does the work they think it does is if we assume that subjects have the same evidence in hallucination and perception _and_ evidence is factive. (If evidence were not factive, there’s no reason to say that in the case of hallucination where it looks as if p, p can’t be in your evidence because ~p.)

    Fwiw, I think V2 is implausible given C&F’s gloss on ultimate evidence as evidence you do not need evidence for. The scope of non-inferential knowledge includes propositions about the external world. As for V1, if that’s the view, I think C&F are in trouble. The linguistic evidence that propositional evidence is factive is pretty strong. I could be wrong, maybe the non-factive team could offer examples of evidence ascription where the evidence is propositional but the ascription is non-factive. I’ve never seen such an example, so I suspect that the non-factivity team thinks that evidence is non-propositional.

    One worry about the non-propositional view is that our evidence ascriptions do tend to be propositional in form (e.g., What’s his evidence for believing she’s guilty? That her prints were on the weapon, that she was seen on the night of the killing running from the scene of the crime, and that she admitted to the killing to a friend.) Of course, we also speak of things like fingerprints and bloody gloves as evidence, so we need a way to fit this altogether.

  12. Sorry, just a quick follow up. I thought V2 was implausibly limiting in terms of the evidence it recognized. I thought this wasn’t a feature of V1, but inconsistent with their mentalism given the factivity of evidence.

    There was something Trent said that I found surprising. I thought “evident” was factive.

    A: He told me that it was self-evident.
    B: It is to him and it was to me.
    A: But you now think it is false.
    B: Exactly.

    Trent, I think, thinks this is coherent. Maybe he thinks ‘p is evident’ or ‘p is self-evident’ conversationally implies p but doesn’t entail it. I have doubts. Consider:
    A: He said that it was evident.
    B: He was right. Not only is it evident, it is true.

    In this second exchange, B’s remark looks a lot like:
    (i) He knows it. Not only that, he believes it.
    (ii) He knows it. Indeed, it is true.

    Redundant conjunction is evidence of entailment. You can reinforce what is conversational implications:
    (iii) She has been sober for days. Indeed, she has been sober her whole life.

    The point about the functional role of evidence is interesting. It might be that non-truths or falsehoods can play those three roles perfectly fine (the sign thing worries me), but what about the role of evidence in explanatory inference? Suppose you know p is part of your evidence. Do you know that, assuming that p admits of explanation, there’s some q such that ‘p because q’ is true? If so, we have an argument for factivity. ‘p because q’ is true only if p and q are true. ‘Because’ statements are factive in the explanans and explanandum positions. It seems strange to think that even when you know p is part of your evidence, you’re not in a position to infer that if p is not brute, p can be explained by something else.

    Another role evidence seems to play is in explanation of normative standing. So, for example, ‘You shouldn’t have jumped to the conclusion that the killer forced his way in because the window was broken from the inside’. Here a bit of evidence (that the window was broken from the inside) is what explains why a belief had a certain normative property. Owing to the factivity of ‘because’ point, the bit of evidence explains only if the relevant proposition is true.

  13. Rumors of LE’s demise are greatly exaggerated: Another example of the difference between philosophers getting bored with a view vs. actually refuting it.

    That a historical time slice of LE’ers were ALSO AND UNNECESSARILY committed to other theses which failed is irrelevant to the view itself.

    The subjectivity dog won’t hunt: it’s an objective fact that I have certain sensations and there’s no escaping it. You can’t *get* to a third-person perspective. There *isn’t* one. It doesn’t *exist*.

    The idea of “objectivity” as a “view from nowhere” is chimeral. We have nothing other than our experiences to go on. They are our only guides in belief.

  14. “The linguistic evidence that propositional evidence is factive is pretty strong.”

    ***facepalm*** I’m about done with that game. NONE of the linguistic arguments strike my ear as any good at all. I have grown to despise linguistic arguments for anything. Our ears are shattared. Let the X-phi people sort out what’s supposed to sound good or bad.

    Being evident is most certainly not factive.

    Earl has said he’s got no problem with a version of V2. They’re fluid on it, so it’ll be hard to catch them in inconsistency–I’ve tried–because they’ll point out (factive) that when they say X it’s because in the context they’re assuming Vn, but of course if you assume Vm instead then X goes.

    In my chapter in _Evidentialism and its Discontents_ (to be posted hopefully the end of next month), I argue for a propositional theory of evidence (and at the FEW and MEW will be arguing that this makes evidentialism false, we should be first and foremost empiricists or we should define evidentialism as a view about the *source* of all evidence: experience).

    At the MEW especially I’ll be focusing on the relative merits of V1 and V2. Earl doesn’t like V1 because he’s inclined to be a non-conceptualist about experience.

    Lots of details and many ways to model it. I’m for the Canberra plan.

  15. Hi Trent…I did not provide rumors of the demise of LE. I provided reasons for the demise of LE (e.g., an incompatible alternative theory like scientific realism) and the inability of LE to account for the semantic value of a scientific sentence (i.e., a sentence involving a real yet unobservable entity). It remains a live question how improved-LE handles the semantics of such sentences? Perhaps your answer involves the theses improved-LE is no longer “unnecessarily” committed to. References? You’ve only mentioned Max Plank and Roderick Firth. Perhaps I’m unaware of someone who has reinvigorated LE or made it credible again.

    I’m not linking “objectivity” to a “view from nowhere” that fails to do justice to what it means to be a thinker who has experiences from a particular perspective. I’m simply being an epistemic externalist: justifiers are not restricted to accessible elements of first-person experiences. Evidential propositions can be true or false independently of what an agent believes about the propositions or facts. There is room for first-person experience and the type of evidence this endorses. However, this is not the *only* (or most important) type of evidence. The most important type of evidence is that which is veridical in that it is true irrespective of what agents believe about the evidence. This is where factivity and an explanatory connection do some of the theoretical work.

    I mentioned the notion of logical omniscience. This notion does not introduce metaphysical spookiness into the theory. It’s not a gap that cannot be bridged from a first-person to a third-person perspective. Rather, it is a view C&F argue against in “Evidence” 2008: 94-95. The idea is that evidence stands in objective probabilistic relations to the propositions. C&F dismiss this view because, “it suggests a kind of logical omniscience that justified believers need not have” (94). I think C&F mistakenly criticize this view. They claim the view requires agents to be logically omniscient or understand all the entailment relations in their body of evidence; otherwise, they do not understand or know that which is entailed by their evidence. C&F assume cognitive accessibility is required in order for entailing evidence to be justifying evidence. However, the view is not committed to the idea that the agent herself needs to escape her first-person prison, hover above her experience, and break free from her cognitive limitations. This begs the question against the view by superimposing an epistemic internalist norm on the view, a norm which the view does not endorse, and then claims that the view fails because it violates the norm.

    This is also not to say that the view is incompatible with internalism. In fact, Swinburne (2001) posits logical omniscience as a theoretical construct and goes on to argue for a type of epistemic internalism.

  16. Trent…Your reply to Clayton’s comments got me wondering about some things. If you dismiss linguistic data as probative because you have, “grown to despise linguistic arguments for anything,” then how do you argue for the truth of empiricism? What counts as evidence for your view? I’m hoping it’s not an argument for the “existential thrownness of our existence” or that experience is all we’ve got. How do you avoid circularity or avoid baldly asserting, “we should be first and foremost empiricists?”

    Further, doesn’t your assertion that “our ears are shattered” work against your empirical/perceptual model of evidence? If we cannot trust our first-person experience about what sounds right to our ears, then why not turn everything over to the X-Phi people? Is this not an admission that perceptual experience (and the cognition that makes sense of it) is suspect? It seems truth needs to be outsourced to a survey-based third party.

  17. Dude, you’re packing *all kinds* of things into LE that were just contingent baggage of those who happened to promote it at one point in history. Seriously!

  18. Circularity???

    The evidence for evidentialism is of the same kind as all evidence: it seems true. It’s the simplest view that vindicates our intuitive judgments of what justified and what’s not.

    I didn’t say there was no such thing as a good linguistic argument. I’m just implying that they’re of no use in Philosophy because we’ve rigged our intuitions with theories. Part of one’s evidence for the current unreliability of their senses could consist in sensory evidence.

    You have to remember I’ve presented on this stuff all around the world for years now and I’m just sick to death of “he said” “she said” linguistic arguments. “That sounds terrible!” “No! That sounds fine!” I’ve had enough.

  19. Trent,
    I guess I’m not as skeptical of the linguistic arguments or the arguments from the role evidence plays in explanation, but then I guess I’d want to know two things about the non-factive view. What’s the argument for it? How do we talk about evidence on that view?

    The remarks you gave about the functional role of evidence don’t support factivity, but they don’t support non-factivity either. They seem neutral. Myself, I found the linguistic arguments for factivity convincing only after realizing that in explicit talk about evidence makes it difficult to talk of evidence as if it is non-factive. My early view was that a subject’s evidence just consisted of the subject’s non-factive mental states, but the view that evidence was propositional and evidence ascriptions were factive was something I accepted later rather unwillingly.

    Two more functional role arguments to consider. There’s the one Williamson offers. Your evidence determines evidential probabilities. On a non-factive conception of evidence, nothing rules out having inconsistent evidence, but then you can’t define evidential probabilities on inconsistent bodies of evidence. The factive conception rules that out. The other has to do with epistemic possibility. If we understand epistemic possibility as relative to a subject’s evidence and epistemic necessity likewise, we can say that p is epistemically possible for S if ~p is consistent with S’s evidence and p is epistemically necessary for S if ~p is inconsistent with S’s evidence. There are many variants on this, but if ‘must’ is strong and expresses epistemic necessity, the account needs a factive conception of evidence.

    One last thing about linguistic intuitions. I can’t dismiss the work that linguists do, they’re experts and I trust the experts. I understand that the data might be messy, in which case we don’t have a good linguistic argument. But, I wonder if the data is that messy. I’m willing to wager that most will come to the view that ‘evident’, like ‘obvious’ is factive. If the intuitions of most pointed in that direction, would that sway you or do you really dislike linguistic arguments?

  20. “Contrary to (AS2) it’s possible to argue that subjective access to evidence, through armchair reflection, is not required for evidence possession. If evidence is factive, then an agent may possess such evidence even though an agent is not subjectively able to recall or grasp that evidence.”

    Agreed, sort of. I have a discussion of armchair access in a piece coming out in Synthese. I don’t think Armchair Access is particularly well motivated and I certainly don’t think it should worry Williamson as the line between what’s accessible given all your resources vs. what’s accessible given only some looks to be sort of arbitrary. I should note that the armchair access team doesn’t always deny factivity of evidence. Indeed, they often assume it in running arguments from error. So, someone who thinks evidence has to be armchair accessible might say that your evidence for believing that there’s a cat in the corner is that it looks as if there is a cat in the corner. They will say that this can only be true if it actually looks as if there is a cat in the corner. They are factives about evidence who think we have a more limited stock of facts that get into our evidence than someone like Williamson does.

  21. Trent…Thanks for the clarification. I was worried about circularity with regard to empiricism: empiricism is true; therefore, all (ultimate) evidence is empirical. Your preferred philosophical methodology sounds something like IBE + reflective equilibrium. Is that correct? If so, I’m wondering what you think about the moral psychology literature on the systematic unreliability of our “intuitive judgments of what’s justified and what’s not.” For example, Haidt (2001) claims that automatic moral intuitions generate our moral judgments. The intuitive judgments are downstream from the automatic intuitions. Because our automatic intuitions tend toward bias the post hoc judgments that characterize them also tend to be distorted. Haidt even argues that, “Our moral reasoning works like a lawyer seeking evidence, not like a judge seeking truth…We make up justifications post hoc, which we present as though they were the causal reasons that led to our initial judgment” (Haidt 2002: 54).

    I mention the moral psy literature because you highly regard the X-Phi literature. They employ similar methodologies, and the moral psy literature seems to tell against your trust of intuitive judgments about what’s justified and what’s not.

  22. Clayton…Thanks for your comments. Going back to your original comment, I’m not sure C&F would pick between V1 and V2, even though you suggest they need to. Their view that “all ultimate evidence is experiential” leaves open the question why perceptual and memorial experiences are justifying. They recognize the need to tell a more illuminating story. They do so in terms of best explanations. That is, for C&F, perceptual experiences justify propositions about the world, “when the propositions are part of the best explanation of those experiences that is available to the person” (2008: 97). Coherence must hold between, “propositions that assert the existence of the non-doxastic states that constitute one’s ultimate evidence and propositions that offer an optimal available explanation of the existence of that evidence” (98).

    I take this to mean that coherence between V1 and V2 is necessary in order for their view of evidential support to do its work. That is, propositions that are the contents of experience (which assert the existence of one’s ultimate evidence) (V1) must cohere with propositions that describe a subject’s experiences (by offering a good explanation of the subject’s experiences) (V2). This makes perceptual evidence support different propositions to different degrees based on how those propositions cohere with one’s ultimate evidence. So, to answer your question, I think C&F are committed to evidence (though not “ultimate” evidence) consisting of propositions, perhaps contrary to what they say otherwise. This is because in the set of evidence will be propositions that are part of the best explanation of the agent’s experiences.

    I think there are a number of problems with this view. To mention one, it seems C&F are committed to a foundationalist picture of justification (i.e., ultimate evidence is that which stops the regress or can justify without needing to be justified). Yet, ultimate evidence can only do its evidential work in relation to propositions that best explain the subject’s experiences. This seems to paint a coherence-based picture of justification. Are C&F proposing some type of founherentism view of justification? If not, then how do they resolve discrepancies between a picture that combines foundationalism and coherentism? Perhaps I’m just unaware of their discussion of such a worry.

  23. On its face, the complicated view you sketch at the end makes it seem as if they are stuck with the view that knowledge of the external world is inferential, but Feldman is pretty explicit that if knowledge of the external world is possible, non-inferential knowledge of the external world is possible. (He has a paper on epistemic possibilities and something or other in a recentish Wiley publication.) As far as I can tell, they have at some point or other defended the following claims in print:

    1. Non-factive mental duplicates always share the same evidence. (2004 book)
    2. Liberal foundationalism-we can have non-inferential knowledge of the external world. (Foundational Possibilities paper)
    3. Factivity-evidence consists of true propositions. (Implicit in their criticism of McDowell in the paper in the Q. Smith collection)
    4. IKSE-If you know p non-inferentially, p is part of your ultimate evidence (UE is, after all, evidence you do not need evidence for and on most views).
    5. Experientialism-our ultimate evidence is experience. (Evidence paper in Q. Smith collection)

    I’m sure some of these views are true. I’m skeptical that all 5 could be. The best thing to do, IMO, is deny 1 and modify 5 so that both propositions about experience and the propositional contents of experience get into the evidence when the situation is right.

  24. Thanks for laying out C&F’s views. What I described doesn’t seem to commit C&F to the view that “knowledge of the external world is inferential”. They officially endorse claim #2. First, I’ll try to resist the claim I made from committing C&F to inferential knowledge of the external world.

    A helpful distinction is between ‘justifying’ and ‘being justified’. C&F were talking about how that which is foundationally ‘justified’ can go about ‘justifying’ propositions about the world. For non-inferentially justified perceptual experiences to further justify propositions about the world those propositions need to “assert the existence of the non-doxastic states that constitute one’s ultimate evidence” (98) *and* cohere with propositions that best explain one’s ultimate evidence (i.e., best describe the subject’s experiences).

    C&F view this as an upshot of ultimate evidence: lifting up the hood on ultimate evidence shows it to have the ability to justify propositions with a coherence relation (i.e., by paying heed to ultimate evidence and providing the best explanation of the evidence). Some propositions about the external world are justified inferentially, but they derive their ultimate justification from that which is non-inferentially justified. So, all knowledge is ultimately non-inferentially justified, even though some propositions will be derivatively justified.

    The weirdness I was trying to point out is that non-inferentially justified evidence can only inferentially justify propositions if those propositions exhibit a coherence relation with regard to ultimate evidence. This makes the entire justificatory prowess of the evidence set depend on coherence to some degree; otherwise, you have foundations that don’t do any work in justifying. And, it gets even dicier.

    Two claims C&F make in the same section in the “Evidence” paper are confusing:

    (1) The truth of the contents of a memory experience may be part of the best explanation of the experience itself (97-98).

    (2) What is justified for the person includes propositions that are part of the best explanation of those experiences available to the person (98).

    (1) makes it sound like the factive contents of experience can be self-justifying. That is, the contents can be part of the best explanation of the experience, and, as such, be justified by the experience (because the contents can be justified by the experience and at the same time do the justifying of the explanation). Is the factivity of experiential contents self-justifying? That seems a weird consequence of claim (1).

    Claim (2) indicates that what is ‘justified’ for the person includes propositions. If these propositions are ‘justified’ what stops them from further ‘justifying’? So, how is propositional justification and inferential knowledge of the external world not a natural consequence of the justificatory structure of their view? So, contrary to what I said at the start of this comment, I’m not sure how C&F resist the idea that “knowledge of the external world is inferential”. They can’t retreat to ultimate evidence because it is part of the justifying structure of ultimate evidence that lets in inferential coherentism and propositional knowledge of the external world.

  25. “IBE + reflective equilibrium”

    Spot on.

    “I mention the moral psy literature because you highly regard the X-Phi literature.”

    Uh, no. Not at all. I think it’s got a looooonnnnnggggg way to go to much respectability. And I’m not sure I’d care even if they did it “right.” The answers to such questions are a loooooonnnngggggg way from reflective equilibrium.

  26. “Yet, ultimate evidence can only do its evidential work in relation to propositions that best explain the subject’s experiences. This seems to paint a coherence-based picture of justification.”

    A. Not necessarily Rich seems to like to think of the IBE part as a part of the epistemic support relation.

    B. So what? Earl considers himself a kind of coherentist. In “The Basic Nature of Epistemic Justification” (2004, ch. 2, originally, I think, 1988) he make this quite explicit. The last line is worth the price of admission.

    Kvanvig and Riggs have showed how taking experiences as ultimate evidence and coherentism go together. Vogel’s got relevant stuff on this as well (and Pryor’s stuff in a way).

  27. “1) The truth of the contents of a memory experience may be part of the best explanation of the experience itself (97-98).”

    Uhhh, can you read that to me, ’cause I don’t see that there.

  28. “On a non-factive conception of evidence, nothing rules out having inconsistent evidence, but then you can’t define evidential probabilities on inconsistent bodies of evidence”

    Why think either of those things? Foley has discussed the latter.

    And you know what I think about epistemic possibility.

  29. Clayton, I should have added, “You *must* be joking!” :-P

  30. 1. Evidence is that which indicates to one the truth of a proposition.
    2. Indications can be misleading.
    3. So Evidence can be misleading.

    Experiences have “assertive contents” think of the phrase “the testimony of the senses.” That’s our ultimate evidence–or propositions to that effect if you prefer. Testimony can, alas, be false. Hence, evidence can be false.

    • (1) assumes evidence as “scientific evidence” not “justifying evidence”. Scientific evidence doesn’t equal justifying evidence for C&F because you can have scientific evidence without having a reason to believe that which the evidence supports. C&F endorse justifying evidence, not scientific evidence. Justifying evidence requires the conjunction of the evidence with info about the connection between the evidence and the proposition. Such justifying evidence cannot be misleading because it, of necessity, is a reason to believe the conclusion in any possible situation, even when one has other evidence defeating the conclusion (see C&F, “Evidence”, p. 90).

  31. Earl prefers to say that all ultimate evidence consists in “non-falsehoods”. You can’t have a falsehood as an item of ultimate evidence. That should be meeting you half-way. You should be satisfied with that. :-)

  32. 1) The truth of the contents of a memory experience may be part of the best explanation of the experience itself (97-98).”

    Re: Ref, my bad, I was looking in “Evidentialism” not “Evidence”

    “(1) makes it sound like the factive contents of experience can be self-justifying. That is, the contents can be part of the best explanation of the experience, and, as such, be justified by the experience (because the contents can be justified by the experience and at the same time do the justifying of the explanation). Is the factivity of experiential contents self-justifying? That seems a weird consequence of claim (1).”

    No, not at all.

    The idea is much, much simpler than that. Suppose you have a memory experience that p. In normal circumstances, the best explanation of your having that experience is that p is true and you’re remembering accurately. That’s it.

  33. “Such justifying evidence cannot be misleading because it, of necessity, is a reason to believe the conclusion in any possible situations, even when has other evidence defeating the conclusion ”

    Non sequitur.

  34. Hey Trent,

    “Clayton, I should have added, “You *must* be joking!” :-P”

    Well played.

    I don’t actually know what you’d say about “must”, whether you think must is factive. von Fintel and Gillies think “It must be that p” is true only if p is true, and given your view on epistemic nec. and poss., it’s hard to see how to explain this if e’s aren’t T’s, as it were.

    I was reading another attack on Williamson last night and thought that this argument against this objection seemed promising. It’s premised on assumptions that Trent might not like and C&F might reject, but I think it is interesting. It gets to the functional role of evidence point.

    In his Phil Studies piece, Rizzieri argues that cases of justified, false belief cause trouble for W’s view. You justifiably but falsely believe that you locked the door to the office (LD) and believe on this basis that no one can get in there (O). He writes, “If Williamson’s proposal that (E = K) is correct then (LD) cannot serve as an evidential ground for (O). This generates problems for (E = K). The first difficulty is that it is very plausible that (LD) does partially constitute my evidence for (O). After all, I am justified in believing (LD), (LD) supports (O), and an explicit inference from (LD) is my most immediate basis or ground for (O).”

    If this is a counterexample to W’s view, we have to assume:
    (1) That I have just locked the door is evidence that nobody can enter my office.

    But, this entails:
    (2) Because my door is locked, it is more likely that nobody can enter my office.

    But, owing to the factivity of ‘because’, we get:
    (3) My door is locked.

    But, (3) is false. If (big ‘if’) you think evidence is propositional and any bit of evidence is evidence-for something because it raises its probability, it looks like we’ll be able to argue in this fashion:
    (i) p is evidence for q.
    (ii) Because p, q is more likely.
    (iii) p.

    So, evidence is factive if it is propositional and it makes that which it is evidence for more likely.

    Yes, some will get off at propositionality, but insofar as objections to W often grant propositionality, it looks like they cannot if those objections assume both propositionality and non-factivity unless they also deny that having evidence explains higher probability of the propositions supported by the evidence.

  35. I certainly don’t think epistemic-must is factive. In fact that seems crazy to me. I think epistemic modalities all work off evidence and so “epistemic must p” =df “Pr(p/my total E) = 1 – e.

    “If this is a counterexample to W’s view, we have to assume:
    (1) That I have just locked the door is evidence that nobody can enter my office.

    But, this entails:
    (2) Because my door is locked, it is more likely that nobody can enter my office.”

    I certainly don’t accept the next move–“factivitiy of because” but I also can’t see my way clear to endorse the inference from (1) to (2) here .

    As I read (2) the “because” is just what Feldman calls a “premise indicator term” and so what’s fingered here is simply a validitiy claim. But, of course, validity doesn’t entail premise-truth.

  36. “‘p because q’ is true only if p and q are true.”

    Clayton…Is it possible to resist the above argument for the factivity of ‘because’? I can imagine ‘p because q’ being true despite it not being the case that both p and q are true. Consider the following: the bird is on the birdfeeder because it looks as if the bird is on the birdfeeder. It might be true that it looks as if the bird is on the birdfeeder, and this might provide a good explanation of why it’s the case that the bird is on the birdfeeder, but it might be false that the bird is on the birdfeeder owing to a perceptual hallucination. Perhaps falsehoods can still play the role of evidence in explanatory inference. Would you take a Williamson route in response and claim you only know p (as part of your evidence) if p is factive?

  37. I agree with Trent [surprise :-)]. I need to see something stronger for the “factivity of because”. I also jump off at the second part of the “big if”. P increasing the probability of Q is not a sufficient condition for P to be evidence-for Q. It’s not sufficient because of all the slight-increase-in-probability counterexamples (see Achinstein 2001). It’s also possible to argue increase-in-probability is not necessary for something to be evidence-for. This is argued for using counterexamples where the probability doesn’t change yet it makes sense to call something evidence-for something else. In addition, there are cases where the probability decreases yet it makes sense to call something evidence-for something else. We know that Williamson endorses positive relevance as evidence-for. But, it seems possible to resist evidence-for on other grounds and resist the generalization of the argument to make sense of how evidence is factive.

  38. Chris, that must be a typo! You *must* have left out the “not” in that sentence!

    I’m officially making this my last comment before we disagree again!

    “I know she’s coming to the party because she said she was” when I misunderstood her.

  39. “It might be true that it looks as if the bird is on the birdfeeder, and this might provide a good explanation of why it’s the case that the bird is on the birdfeeder, but it might be false that the bird is on the birdfeeder owing to a perceptual hallucination”

    Hmmmmm…. That seems wrong to me.

    A: Explain why the bird is on the bird feeder.
    B: But it’s not, it flew away.
    A: Well, I don’t believe you. But even if the bird isn’t on the feeder, why is the bird on the bird feeder?

    I don’t know how to deal with people like A, do you?

    The oddity of exchanges like that seems to support the factivity of ‘because’. Another reason is that if you append ~p, you get something that seems manifestly contradictory. The other is if you append p or q, you get a redundant conjunction which is taken to be confirming evidence of entailment (See Stanley’s assertion and certainty piece). So, for example,

    (1) The peasants revolted because the taxes were too high.
    I say that this entails:
    (2) The peasants revolted and the taxes were too high.
    Consider:
    (3) The peasants revolted because the taxes were too high. Not only that, the peasants revolted.
    That seems a lot like:
    (4) He knows that she’s out of town. Not only that. she is out of town.
    That seems unlike:
    (5) He hasn’t smoked in days. Not only that, he hasn’t smoked in years.

    I’d also say that these look roughly equivalent:
    (6) The taxes were too high. That’s why the peasants revolted.
    (7) The peasants revolted because the taxes were too high.

    I’m pretty sure that (6) entails that the taxes were too high. Again, the usual tests need to be introduced to see if (6) and (7) are really equivalent. It seems like a contradiction to assert (6) & (~7) or (7) & (~6). It seems like a redundant conjunction to say, “Not only (6), but (7) as well”. If someone said, “I know (6), but I have my doubts about (7)”, we’d have our doubts about him.

    As for the probability raising idea, two quick points. The first is that the sufficiency of probability raising isn’t what matters, what matters is necessity. So, if you thought that p is evidence for q only if p raises the probability of q, then you get the inference from p is evidence for q to q is more likely because p. You’re right that this might be too crude a picture for evidential support. Maybe it’s just often the case that evidence for something raises its probability. That’s alright, once you get the basic idea of the argument, we can kick away the ladder. Suppose for example you think p is evidence for q only if p is some sort of reason for believing q. Then you get ‘p is evidence for q’ only if ‘Because p, there is some reason to think q isn’t false’. Or, put in whatever necessary condition you think there is for evidential support and the factivity of ‘because’ kicks in. [I had assumed that probability raising was necessary for something to be evidence for something else because W assumed it in KAIL and it was assumed in a paper criticizing him.]

    So, the recipe. Suppose C is a necessary condition q must satisfy for p to be evidence for q. It looks like we’ll be able to reason as follows:
    (i) p is evidence for q.
    (ii) because p, q meets C.
    (iii) p.

  40. A quick one.

    At noon:

    (1) I got the job because of my superior talents.

    You at two:

    CC: Um, did I hear you say, “I got the job because of my superior talents”?
    CL: Yes.
    CC: And you subscribe to the knowledge-rule of assertion?
    CL: Of course!
    CC: So, you shouldn’t have said what you said at noon.
    CL: Nonsense, I knew what I said was true.
    CC: But you didn’t get the job! And, to be blunt, you don’t have superior talents!!!
    CL: Irrelevant. If what I said at noon doesn’t entail that I got the job, you can’t argue from the fact that I didn’t get the job to the conclusion that I didn’t know what I said was true at noon.
    CC: [Beating CL mercilessly with a copy of KAIL]

    And at this point, I think CC is right. CL is talking nonsense. So, another data point. You can’t say later that I knew earlier p b/c q once it is clear that either ~p or ~q. It doesn’t matter if earlier you justifiably believed with great confidence that p and q were true. It doesn’t matter that in nearby possible worlds p and q stand in the right relations to be explanans and explanandum.

  41. Hi all,

    The Philosophy Carnival pointed me here, so I apologize for breaking in late. Clayton, we’ve discussed the relevant inference elsewhere, I think — the one from

    (1) What justified Coop in giving Audrey the toxic stuff was that there was gin and tonic in that glass.

    to

    (3) Coop was justified in giving Audrey the toxic stuff (in part) because there was gin in the glass.

    When we say “what justified S in phi-ing is p”, we’re talking of course about justifying reasons. So, what this means is this: S’s justifying reason for phi-ing is p. Construing (1) in this way gives us (1*):

    (1*) Coop’s justifying reason for giving Audrey the toxic stuff was that there was gin and tonic in that glass.

    There is no clear inference from (1*) to (3). If you don’t like the use of “p justifies S in phi-ing” as a shorthand for “S’s justifying reason for phi-ing” then just go with the latter. Personally, I think it’s fine because “p justifies S” is ambiguous (it seems to me) between a causal reading according to which some fact causes S to be justified and the justifying-reason reading. On the former reading, it’s only true if p is true. On the latter reading it’s not. It’s the latter reading that matters. It’s justifying reasons that we’re saying can be false.

    I see no reason why p being a justifying reason you have for phi-ing entails that p, no more than p being a belief you have entails that p.

    For example, when there is no gin and tonic in the glass, there is something weird about

    1**) That there is gin and tonic in the glass is his justifying reason for giving Audrey the toxic stuff.

    But whatever is weird about 1** is also weird about 1***

    1***) That there is gin and tonic in the glass is the thing he believes.

    But there is clearly a sense in which 1*** can be true even when there is no gin and tonic in the glass. I don’t see a reason yet why 1** can’t be true in that same sense.

    Of course, there’s Christopher’s evidence point. I’m not in love with the idea of allowing falsehoods as evidence, either. My intuition that Coop’s justifying reason for giving Audrey the toxic stuff is that there is gin and tonic in the glass. I don’t have the intuition that that there is gin and tonic in the glass is evidence Coop has. I guess my intuitions run afoul of the Equivalence thesis.

  42. Hey Jeremy,

    I’m glad that we might be on the same page when it comes to the factivity of evidence. Quick question about justifying reasons and “because” claims. Consider your:

    (1*) Coop’s justifying reason for giving Audrey the toxic stuff was that there was gin and tonic in that glass.

    Abstracting a bit, I’m interested in claims of this form:

    (1a) C’s justifying reason for V-ing was that p.

    To my ear/eye, that does seem to entail:

    (2a) Because p, C had a justifying reason for V-ing.

    So, instances of it:

    (2*) Because there was gin in the glass, Coop had a justifying reason for giving Audrey the stuff.
    (2**) Because there was gin in the glass, something justified Coop in giving Audrey the stuff.
    (2***) Because there was gin in the glass, there was some reason for Coop to give her the stuff.

    My own intuitions, fwiw, is that (1*) entails (2*)-(2***).

    Your belief case is interesting. You said that this could be right even if there was no gin:

    (1***) That there is gin and tonic in the glass is the thing he believes.

    This sounds weird to me, but it might just be because it is sort of stilted English and not because it is semantically defective. Here’s a potential difference between “belief” and “justifying reason”.

    (3) Coop’s justifying reason for giving Audrey the toxic stuff was that there was gin and tonic in that glass but he doesn’t know that there was gin and tonic in the glass.
    (3′) Coop gave Audrey the glass for the reason that there was gin in it but he didn’t know that there was gin in it.
    (3”) Coop gave Audrey the glass for the reason that there was gin in it but there was no gin in it.
    (3”’) Coop’s justifying reason for giving Audrey the stuff was that there was gin and tonic in it, but I don’t know if there was any gin in it.
    (4) Coop’s belief was that there was gin and tonic in the glass, but Coop doesn’t know that there’s gin in the glass.
    (4′) Coop believes that there was gin and tonic in the glass, but he doesn’t know that there’s gin in the glass.
    (4”) Coop believes that there was gin and tonic in the glass, but I don’t know if there was or not.
    (4”’) Coop believes that there was gin and tonic in the glass, but there wasn’t.

    According to Unger and Hyman (maybe), if p is your reason for V-ing, you know p. I think that p can’t be your justifying reason unless it is your reason. Myself, I think this is a bit trickier than Unger thought when he wrote _Ignorance_, but it is interesting that there could be an asymmetry between belief and reasons ascription that comes out when combined with knowledge ascriptions.

  43. I don’t find myself all that tempted to say that 1a entails 2a. I do find it more plausible to say that 1a entails 2a*:

    (2a*) Because p is justified for C, C has a justifying reason for V-ing.

    So, an instance:

    (2*#) Because it was justified for Coop that there was gin in the glass, Coop had a justifying reason for giving Audrey the stuff.

    But (I think, though I guess you do not) that it was justified for Coop that there was gin in the glass. So it fits just fine that Coop had a justifying reason for giving Audrey the stuff.

    I’ve always liked that you agree that “p can’t be your justifying reason unless it is your reason”. I think so too. I’ve always thought that this should make my case really easy with you. Because all I’d need to add, it seems to me, are these premises:

    A) When Coop gives the stuff to Audrey, he does so for some reason (HIS reason).

    B) If Coop gives the stuff to Audrey for a reason it will be the reason expressed in the best recapitulation of his psychology: “R, so I’ll give the stuff to Audrey.”

    C) Coop’s psychology is not best recapitulated by replacing R with “I think that there’s gin in the glass” nor with “It’s justified for me that there’s gin in the glass” nor with “It looks like there’s gin in the glass.”

    D) The only plausible alternative to the options in C is: “There is gin in the glass”

    So, Coop’s reason for giving the stuff to Audrey is that there is gin in the glass. And, given what we agree about, he can’t have any justifying reason that isn’t his reason. So, if he has a justifying reason, it’s that there is gin in the glass.

    But,

    E) He is justified in giving the stuff the Audrey and it is a reason that justifies him.

    So, that there is gin in the glass is his justifying reason.

    I’m guessing you will reject…. the second conjunt of E? No, earlier, because you’re not going to like that there is gin in the glass being Coop’s reason. So B? Or the inference from A-D to the first sentence of the subsequent paragraph?

  44. Hey Jeremy,

    I probably would reject A and say instead:

    A’) When Coop gives the stuff to Audrey, he takes himself to have a reason to do so.

    The fallibility he has concerning the facts extends to the fallibility he has in describing his reasons/the reasons he has for acting. No gin means no reason. That’s what B-D suggest (when you add in the extra stuff that I think supports the factivity of justifying reasons).

    You deny that there is any entailment from 1a to 2a, so it should be true that:
    (#) It’s not because there was gin in the glass that Coop had a justifying reason to give Audrey the stuff but Coop’s justifying reason for giving her the stuff was that there was gin in the glass.

    Is that true on your view? I like B-D and don’t like (#), so I’d reject A and E. That’s maybe not the best way to describe cases like this, but bracketing that, I’m just curious about (#).

  45. Gotcha. I should have known it was A. So, Coop doesn’t have a reason for giving Audrey the stuff?

    Now,

    (#) It’s not because there was gin in the glass that Coop had a justifying reason to give Audrey the stuff but Coop’s justifying reason for giving her the stuff was that there was gin in the glass.

    Yeah, it sounds bad. But I think it’s true in this case. Put “the proposition” between “was” and “that” and it sounds a whole lot better to me.

  46. (#) It’s not because there was gin in the glass that Coop had a justifying reason to give Audrey the stuff but Coop’s justifying reason for giving her the stuff was the proposition that there was gin in the glass.

    Sounds strange to me, still. But, that’s not going to surprise you.

    Coop doesn’t have a reason for giving Audrey the stuff? Given B-D, he couldn’t have any reason but the fact that there’s gin in the glass. So, we agree on everything but one small point.

  47. There is no fact that there is gin in the glass, so that can’t be his reason. But that there’s gin in the glass could be his reason. (One nice feature of this — it would allow good-case-Coop and bad-case-Coop to have the same operant reasons. Maybe you don’t think that’s such a nice feature.)

    But, hell! I’ll take our almost-everything agreement and be happy.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s