EAAN in the case of moral knowledge
April 12, 2011 — 14:48

Author: Alexander Pruss  Category: General Religion and Life  Tags: , ,   Comments: 22

I’ve never been strongly moved by Plantinga’s EAAN’s general sceptical conclusions allegedly following from naturalism and evolution.  It has seemed to me that on the best causal (sketches of) accounts of intentionality, it’s pretty much guaranteed that a significant portion of our empirical beliefs are true.  I have serious problems with these causal accounts, but given the accounts, EAAN does not appear that persuasive to me.  

However, I think one can use EAAN-type arguments for a more limited conclusion, namely that if naturalism and evolution are true, then certain important kinds of knowledge are seriously threatened, specifically moral (and maybe more generally normative) knowledge (I think certain kinds of modal and metaphysical knowledge are also threatened, and it may be that metaphysical naturalism falls within the class of threatened knowledge).

The standard naturalistic evolutionary story about how we get moral beliefs is something like this.  Certain kinds of beliefs about what one ought to do promote the fitness of communities and individuals.  Consequently, as a result of certain mimetic and/or genetic evolutionary processes, we have roughly the moral beliefs we do.  There might be causal intermediaries like propensities for making certain kinds of moral inference.  

But notice a crucial difference between this explanation and evolutionary explanations of our ordinary empirical beliefs.  In the ordinary empirical case, Plantinga’s critics can say we are selected for propensities to have tiger-presence beliefs in the presence of tigers, because there is an obvious fitness benefit from having such beliefs when the beliefs are true.  One might worry about details here, but the story has an initial plausibility.  However, in the case of moral beliefs, the benefit of having the beliefs does not come from the beliefs’ being true.  

In the moral case, assuming naturalism and evolution, at best we have a Gettier case instead of knowledge.  If we are lucky, there is a large overlap between those moral beliefs that promote fitness and those moral beliefs that are true.  Our moral beliefs, based as they are on natural propensities to believe, may be justified.  But they are not knowledge, because the connection is too coincidental on this story.

To see that the connection is coincidental, consider this story that is meant to be parallel to the story about moral beliefs. Outside of our community, there is a dark forest. People who go deep into the forest never come back. Eventually, we evolve (mimetically and/or genetically) a propensity to believe that the depths of the forest are full of tigers, and this propensity keeps us out of the forest. In fact, there are tigers deep in the forest, but they are nice tigers and never eat people. The reason people who went deep into the forest never come back is not because the tigers ate them, but because boa constrictors killed them. Maybe we have a justified and true belief that there are tigers in the forest, but it is at best a Gettier case.

Objection: It is a nomically necessary truth that there is a large overlap between possible fitness-promoting moral beliefs in communities like human ones and true moral claims.  This nomically necessary overlap might, for instance, be found in the case of a variety of metaphysically necessarily true claims about the wrongness of freeriding.
Response: Modify my tiger story by adding the supposition that there turns out to be a nomically necessary connection between the presence of tigers and the presence of boa constrictors (perhaps it’s nomically necessary that tigers only exist where there are particular kinds of trees and hat there are boa constrictors wherever there are such trees)  We still have a Gettier case, I think.  
I can imagine a reliabilist objecting here, though.  After all, now the process (individual, cultural or genetic) of forming beliefs about the presence of tigers on the basis of facts sufficiently correlated with the presence of boa constrictors, such as disappearances of people, may count as reliable.  I am inclined to think this judgment tells against those versions of reliabilism that make the judgment, and the only plausible varieties of reliabilism will be those that say something like this: in general the process of forming beliefs about the presence of one dangerous animal merely on the basis of the presence of a different kind of dangerous animal is relevantly unreliable.
Here’s a different twist.  Take the tiger / constrictor case, and suppose that nobody has actually ever met a tiger or constrictor and survived to converse with others.  The people in the village speak a language like English and say things like: “We must not go deep into the forest, as there are xingas there.”  They all believe propositions they express with sentences like: “There are xingas deep in the woods”, and their beliefs have an evolutionary explanation involving the disappearances of people eaten by boa constrictors.  What does “xinga” mean?  It may mean boa constrictor or, perhaps, scary thing in the woods.  But it does not mean tiger.  The tigers are plainly coincidental, whether or not their presence in the forest is nomically connected with the presence of the boa constrictors.  And of course the point remains if the word they happen choose sounds like “tiger” instead of sounding like “xinga”–their sound-alike to “tiger” will not mean tiger.
By the same token, if an English-like linguistic community’s beliefs expressed by phrases like “One xot to cooperate” arose solely because of a connection with fitness and had motivational force, it is dubious whether that “One xot to cooperate” would mean one is morally obligated to cooperate.  It might be more reasonably taken to mean that cooperating promotes fitness–that would seem the better choice of reference magnet.  And what goes for their “xot” goes for our “ought”.  But in English “One ought to cooperate” does not mean that cooperating promotes fitness.  Hence the evolutionary naturalist story is false.
  • You might like to look at a recent paper by Kahan:
    Kahane, Guy. 2011. Evolutionary Debunking Arguments. Noûs 45 (1):103–125.
    and two forthcoming papers by myself and my coauthor Paul Griffiths:
    Griffiths, Paul E, and John S. Wilkins. In Press. When do evolutionary explanations of belief debunk belief? In Darwin in the 21st Century: Nature, Humanity, and God, edited by P. R. Sloan. Notre Dame, IN: Notre Dame University Press.
    Wilkins, John S., and Paul E. Griffiths. In Press. Evolutionary debunking arguments in three domains: Fact, value, and religion. In A New Science of Religion, edited by J. Maclaurin and G. Dawes. Chicago: University of Chicago Press.
    both of which are available at PhilPapers.
    We argue that Plantinga’s EAAN fails for science, and Kahane that it succeeds for morality, but that moral antirealism is the better solution.

    April 12, 2011 — 17:48
  • Thanks a lot for the references!
    In regard to Kahane, moral realism is surely much more plausible than the conjunction of evolution with naturalism. And it’s crucial for the debunking move that one have not only evolution but also naturalism. For instance, given evolution plus theism, the causal premise becomes doubtful–it seems quite likely that evolutionary development would be truth-tracking in regard to moral claims if God set up the evolutionary development. A similar point applies given evolution plus optimalism, and perhaps even given evolution plus Aristotelian teleology.
    In other words, the following seems a much more reasonable position than Kahane’s:
    1. If evolution and naturalism hold and moral objectivism is true, we have no moral knowledge.
    2. Moral objectivism is true.
    3. We have moral knowledge.
    4. Therefore, either evolution or naturalism (or both) is false.
    5. Evolution isn’t false.
    6. So, naturalism is false.
    I do have some questions about your “When do…?” paper.
    i. Is the existence of a Milvian bridge sufficient for knowledge, or only for justification? I was targeting knowledge in my post.
    ii. Do you think there is a Milvian bridge in the case of large-scale metaphysical beliefs, such as that naturalism is true?
    iii. Do you think there is a Milvian bridge in the case of normative epistemological claims, such as that one ought not believe unjustified claims or that one ought not both believe the denial of a tautology?
    And a final remark. One place my argument is weak is in regard to something like Objection 1. Specifically, I am worried about David Lewis’s remarks about knowledge of necessary truths in On the Plurality of Worlds. Lewis argues that there is a sense in which you can’t accidentally get a necessary truth right. I am wrong in no world in which I believe that 2+2=4. Now, fundamental moral truths are, very plausibly, necessary truths. So in every world where I believe that humans ought not freeride (with whatever “ceteris paribus” type qualifications this needs to make it a necessary truth), I thereby believe a truth. So in some sense it’s not an accident that my belief is right–this belief couldn’t be wrong. I am inclined to think Lewis is wrong here–looking for worlds where this belief is false is the wrong way of examining accidentality. What do you think?

    April 12, 2011 — 18:32
  • There are two papers along this same line of thinking, one I’m particularly fond of, Mark D. Linville’s “The Moral Poverty of Evolutionary Naturalism” and Tom Crisp’s “An Evolutionary Objection to the Argument from Evil”

    April 12, 2011 — 19:46
  • hiero5ant

    Well, to the extent that evolution can be said to have a ‘beneficiary’, it is neither individuals nor communities, but replicators.
    There are pretty obvious parallels here with Sharon Street’s ‘Darwinian Dilemma for Realist Theories of Value’, and I think overall the argument is a success. But the target of that success needs to be more clearly delimited: it only establishes skepticism about 1) realist theories with 2) correspondence models of truth in which 3) moral “beliefs” are perceptual, and hence are made true or false in virtue of the world.
    A good expressivist has a truly elegant response in denying the descriptive character of normative judgments while accepting a deflationary view of truth. She can let evolutionary psychologists and social scientists tell their story of how we come to have the desires we have without being troubled that their theories never seem to reference any mind-independent facts “causing” them to be true, or invoking Yahweh as a “Trooth Fairy” who sneaks in at some unspecified point in metazoan evolution to add a moral volt-meter to our sensory apparatus. (Indeed, even if it turns out there is intelligent inter vention in our cognitive evolution, supernatural or otherwise, to give us certain moral propensities, she can still deny that moral truths have any metaphysical justification outside of themselves.)

    April 12, 2011 — 21:57
  • Michael Bergmann points me by email to Sharon Street’s “A Darwinian Dilemma for Realist Theories of Value”.

    April 13, 2011 — 7:49
  • So it looks like there are a number of people who have published on this, and all the papers that I’ve so far looked at basically agree that the following are not all true:
    1. Evolution occurred pretty much as science describes it.
    2. Naturalism is true.
    3. Moral realism is true.
    4. We have moral knowledge.
    This is very interesting in connection with the traditional atheist contention that the moral argument fails because atheists can account for morality just as well as theists. It’s looking as if they can do it only at the cost of a non-realist theory, and that is a high cost indeed (both because it violates intuitions and because of the Geach problem).

    April 13, 2011 — 7:56
  • Dianelos Georgoudis

    You write: “It has seemed to me that on the best causal (sketches of) accounts of intentionality, it’s pretty much guaranteed that a significant portion of our empirical beliefs are true. I have serious problems with these causal accounts, but *given* the accounts, EAAN does not appear that persuasive to me.”
    Given that naturalistic accounts of intentionality do not really work, I don’t see how they defeat or come close to defeating Plantinga’s reasoning in EAAN.
    In any case, a perhaps overlooked property of the EAAN argument is that its basic premise about the unreliability of cognitive faculties evolving on N&E (i.e. on the naturalistic interpretation of the evolution of intelligence) is scientifically verifiable: We already have software that simulates the Darwinian evolution of very primitive cognitive faculties. I think that in the near future it should be possible to make a computer simulation of mechanisms whose behavior is driven both by beliefs (i.e. models of their environment) and wishes (i.e. goal orientated causal properties), which simulation will then enable us to actually observe whether their beliefs accurately track their environment or not. My guess is that when we do that we’ll experimentally confirm, at least as a first approximation, that Plantinga’s reasoning is sound.
    “I think certain kinds of modal and metaphysical knowledge are also threatened, and it may be that metaphysical naturalism falls within the class of threatened knowledge”
    I agree. In fact I think that if the purpose is to show that naturalism is self-defeating then using the EAAN is like shooting a rabbit with a canon. A more targeted (and thus perhaps stronger) version of the EAAN would roughly go like this:
    1. On N&E the evolution of intelligence is an unguided process driven exclusively by the fitness of behavior.
    2. On N&E evolution takes place in the physical world in which metaphysical truths play no role.
    3. Therefore, on N&E holding true metaphysical beliefs will neither increase nor decrease the fitness of behavior.
    4. Therefore, on N&E evolution will not produce reliable cognitive faculties about metaphysics.
    5. Therefore, on N&E one should not believe in metaphysical beliefs produced by one’s cognitive faculties.
    6. Naturalism is a metaphysical belief.
    7. Therefore, on N&E one should not believe in naturalism.
    8. Therefore, given the truth of natural evolution, naturalism is self-defeating (even if it happens to be true)
    Sometime back I wrote to Plantinga about this idea. I just looked it up, and I find that the way I put it there is quite concise: “After all, exactly the same evolutionary history of humanity according to E could have taken place in any number of very different objective realities (some of them naturalistic some of them non-naturalistic worlds), and hence the [on naturalism] by definition unguided process of E would not select in us the cognitive faculty to discover in which world we actually exist.”
    I asked Plantinga what he thought, to which he responded in a very brief but mildly positive manner.
    One way or the other, the EAAN is a great argument and it is a good bet that it will be discussed for a long time.

    April 13, 2011 — 7:59
  • Here’s another interesting thing. You don’t need anything spooky in human evolution to give a theistic solution to the problem. In particular, you don’t need any divine addition of a “moral volt-meter”, to use the terminology of one of the commenters.
    Here is a sketch. Consider the empirical hypothesis that the following are true:
    1. Given initial conditions like those on earth and laws like those in our world, it is likely that intelligent beings would evolve.
    2. The evolution of non-social intelligent beings is unlikely (e.g., because of links between communication and intelligence).
    3. Intelligent social beings are likely to evolve fundamental moral beliefs centered on the claims that (a) reciprocity is morally called for; (b) we should care for our kin; and (c) we should be willing to make sacrifices for our community.
    Add the non-empirical hypothesis that:
    4. (a), (b) and (c) are true.
    Now, if God exists, he knows (1)-(4). In particular, he knows that if he creates a universe which contains planets with initial conditions like those of the earth and laws like those of our universe, there will likely evolve intelligent beings on those planets, and they will likely evolve fundamental moral beliefs centered on (a), (b) and (c). And so he creates such a universe with that end in mind.
    If the empirical claims (1)-(3) are true, the above theistic story has no empirical consequences contrary to our best evolutionary stories. But given the above theistic story we no longer have a Gettier case–it is no longer a coincidence, in the knowledge-relevant sense, that the moral beliefs that promote fitness are true. The truth of our moral beliefs is now a part of the explanation of why we have them, because God created a universe wherein true moral beliefs would likely evolve. (We might ask: What would God have done had evolution not gone the way he wanted it to? One answer is: He would then have miraculously intervened. But since it went the way he wanted it to, he didn’t need to.)
    I don’t endorse this story (I think it has theological–but not philosophical–problems in that I think we have good theological reason to think that God wanted to specifically create humans, and not just unspecified social intelligent beings), but it seems to be much superior to jettisoning moral realism.

    April 13, 2011 — 8:37
  • Mike Bergmann pointed me to this paper by Wielenberg arguing against this line of thought.
    Wielenberg’s position seems to be this. You can get truth-tracking because you have a pair of relations: (a) our having the cognitive faculties we do logically entails that we have certain rights; (b) our having the cognitive faculties we do causes us to believe that we have certain rights. Since the relationship in (a) is logical, it’s sufficiently close for truth-tracking.
    I think he’s wrong. Suppose Goldbach’s Conjecture (GC) is necessarily true. Imagine people (b) whose cognitive faculties cause them to believe that they have cognitive faculties and GC is true. Suppose there is some weird off-track selection pressure in favor of belief in GC (maybe some alien kills off everyone who doesn’t believe in GC, because the alien is convinced that GC is false and wants humans to have false mathematical beliefs). Moreover, (a) that they have cognitive faculties logically entails that they have cognitive faculties and GC is true. There is no truth-tracking here, but we do have both the (a) and (b) elements Wielenberg talks about.
    Nor will it help, I think, to replace entailment with relevant entailment. There can be non-obvious relevant entailments, and all we need to do is to choose one of them instead of the entailment in the above example.

    April 13, 2011 — 15:52
  • //”Given initial conditions like those on earth and laws like those in our world, it is likely that intelligent beings would evolve.”//
    I’m probably stepping out of bounds on account of what little formal education I have, nevertheless, I would reject this premise on a couple of points: (1) It is a “just so” argument that offers no satisfactory explanation as to how such conditions could occur given naturalism. (2) Considering natural evolutionists have yet to offer a tenable theory as to what these laws are, it seems question begging to suggest that intelligent beings would naturally evolve. There are no such known laws that have been discovered that could produce intelligent beings, it is only speculation at best.

    April 14, 2011 — 13:59
  • Alvin Plantinga’s argument seems very modest. He simply observes that we take for granted that we can and do know things. Given naturalism (no God, gods or anything similar) and evolution (survival weeding out random genetic mutations), we have a logical contradiction.
    He is not claiming to disprove naturalism or evolution, nor is he claiming to prove Christian theism. His argument is that it is an incoherent position for the naturalist, while completely consistent for the Christian theist to rely upon our cognitive faculties.

    April 14, 2011 — 14:12
  • Andrew Moon

    Hi Alex,
    Thanks for posting on this. I wanted to get the structure of your argument clear. Is it like this? (I use ‘warrant’ in Plantinga’s sense, as denoting what turns true belief into knowledge.)
    1) Naturalism and Evolution (assume for conditional proof)
    2) Therefore, either our moral beliefs are unlikely to be true or our moral beliefs are gettiered true beliefs.
    3) If our moral beliefs are unlikely to be true, then they are unwarranted.
    4) If our moral beliefs are gettiered true beliefs, then they are unwarranted.
    5) Our moral beliefs are unwarranted (2,3,4)
    6) If Naturalism and Evolution are true, then our moral beliefs are unwarranted (1-6, conditional proof)

    April 14, 2011 — 20:16
  • I was granting for the sake of argument that our moral beliefs are likely to be true, so I guess you could reconstruct it like that, though “warrant” isn’t a term I use.
    And then, as part of a bigger picture, I’d add:
    7. Many of our moral beliefs are warranted.
    8. Evolution is true.
    9. So, Naturalism is false.

    April 14, 2011 — 23:57
  • Andrew Moon

    I like ‘warrant’ because it’s easier to say, “that belief is warranted” than “that belief has that epistemic property which turns true belief into knowledge”. I think it made my reconstruction of your argument easier on the tongue. (No big point I’m making; perhaps it’s just a preference.)
    The reason I made the point about the moral beliefs being unlikely to be true is because I think that most people who use evolutionary debunking arguments have that conclusion in mind. Your appeal to Gettier cases makes your argument unique.
    Also, it seems really important. If you take out that disjunct and just assume that moral beliefs are likely to be true (granting that to the naturalist), then your argument’s not as interesting, I think. If I gave an argument to a naturalist with the conclusion, “So, most of the moral beliefs of people in the world are instances of Gettiered, true belief”, the naturalist’s response should be, “Okay, well, at least they’re true! I don’t really care whether or not they’re Gettiered.”
    You might press, “But it turns out, then, that most people don’t have moral knowledge!” He might respond, “What’s your argument that we have moral knowledge? I’m a moral realist. I think we have true moral beliefs. How much, theoretically, do I lose by denying moral knowledge of most individuals, so long as they have true moral beliefs?”
    So, I guess I might ask what’s so bad about a naturalist who embraces moral realism, Evolution, holds that we have mostly true moral beliefs, but denies that those beliefs are warranted. Is that naturalist embracing something implausible?

    April 15, 2011 — 1:32
  • anon

    Another paper to add to the list compiled so far:
    Justin Clarke-Doane ‘Morality and Mathematics: The Evolutionary Challenge’
    found here

    April 15, 2011 — 17:03
  • Michael Swanson

    Tom Crisp has a new paper that intersects with this topic:
    “An Evolutionary Objection to the Argument from Evil,” in Evidence and Religious Belief, eds. Kelly James Clark and Raymond Van Arragon (Oxford, forthcoming).
    There’s a hyperlink to a draft here: http://people.biola.edu/thomasc/thomasmcrisp/Vita.html.

    April 16, 2011 — 1:02
  • I think Craig is probably alluding to this argument in many of his popular level defences of divine command theory. Consider for example this claim “I just don’t see any reason to think that in the absence of God the morality evolved by Homo sapiens is objective.” Here he seems to have in mind a conditional. If God does not exists, then it’s unlikely that the morality evolved by homo-sapiens accords with an objectively true morality It’s of course quite undeveloped but I think something like it is alluded to.
    I wonder, also if Robert Adam’s hints at this and one could develop the argument in a way that it supports his position. In his writings on DCT Adams argues that “wrongness is the property which best accounts for the role assigned it by the concept” he then lays out several features the role of obligation plays. “the relevant property should be (a) a property of actions (b) objective (c) able to account for the wrongness of the major portions of actions we consider to be wrong. (d) play a causal role in our coming to know what is wrong. (e) be seen as supremely weight reason against doing the action” and so on. (Adams adds other criteria in Finite and Infinite Goods )
    Adams goes on to argue that a divine command theory is more plausible and attractive given theism, than any theory is without theistic assumptions. One reason for this Adam’s gives is that it provides an account of (d), because commands are communicated, and God could create our cognitive faculties and providentially guide history so that we became aware of what our moral obligations are. The line of argument you provide would be a way of strengthening this idea, and also suggesting reasons why a naturalistic account would struggle to make sense of this feature of moral obligation. On naturalism our cognitive faculties are not created to know truth and God did not providentially guide history and this undercuts us knowing that X is right because X is right.

    April 17, 2011 — 3:06
  • Matthew:
    Yes, but one must be careful with the claims that it’s unlikely that evolution would get things right. The following could both be true:
    (a) moral truths are necessary truths; and
    (b) there happens to be a correlation between the truth of a moral claim p and the contribution to fitness from individuals believing p, so that the kinds of evolutionary processes that produce rational animals that have moral beliefs are likely to produce rational animals that have central true moral beliefs.
    If this were true, then there is an objective probability sense in which naturalistically evolved moral beliefs are likely to be right. But the beliefs’ evolution still isn’t truth-tracking and we still have a Gettier case. (Imagine that, by a law of nature or even of metaphysics, most of the time where there is a wolf in sheep’s clothing standing in a field, there is a sheep behind it in a hole. The sheep-beliefs are still Gettiered.)

    April 17, 2011 — 8:37
  • hiero5ant

    I’d have to object in the strongest possible terms to the equivalence of antirealism with “debunking” or “lack of warrant”, which is why I just don’t see abandoning realism as a “steep price”. If anything, I would say it is a moral necessity.
    Yes, I would have to say that atheists’ claims to be able to ground morality are nonstarters, but only because I think the entire notion of “grounding” being a nice person in some metaphysical verbiage is fundamentally misguided. I look at those exchanges the way a nonbeliever looks at debates between catholics and protestants over sola scriptura vs. papal pronouncements. It’s an internal doctrinal struggle over sources of authority, both of which we’d all be better off without.
    If one can’t figure out that helping little old ladies across the street is the right thing to do without appeal to some grand cosmological architecture that says more about one’s own character than it says about whether “naturalism” is true.
    [Not to derail, but does anyone have any citations pointing to a serious or fundamental flaw in Blackburn’s solution to embedding problems? I’ve seen marginal technical criticisms and misunderstandings of it as fictionalist, but nothing so compelling that I would want to revise the whole of biology to accommodate an “intuition” that we’re beholden to nonhuman authority]

    April 19, 2011 — 13:45
  • I am not an expert on expressivist views. My necessary condition for a successful solution to the Frege-Geach problem are to solve the following varieties of it:
    1. Truth-functional operators. (“If trimming fingernails is permissible, so is a haircut.”)
    2. Quantification. (“There are no innocents whom it is permissible to kill.”)
    3. Truth-type operators. (“Everything she said about morality is true–she’s the expert on this!”)
    4. Modal operators. (“There is a possible world where it is permissible to feed arsenic to innocents.” (Say, a world where arsenic isn’t poisonous.))
    5. Probability and conditional probability. (“P(abortion is wrong | contraception is wrong) is higher than P(contraception is wrong | abortion is wrong).”)
    6. Knowledge-type operators. (“You’re culpable for doing something only if you knew or should have known it is wrong.”)
    7. Explanatory operators. (“Stealing from the French is wrong because stealing is wrong.”)
    Does Blackburn have a plausible story about all of these?
    In any case, I don’t see any reason why any revision to biology is required by what I said.
    Claims (a) and (b) in my response to Matthew cohere with everything we know in biology. Claim (a) coheres because it is entirely outside the realm of biology. And as far as I know, claim (b) coheres with our best theories of the evolution of cooperative behavior. If claims (a) and (b) are both true, then to get moral knowledge all we need is that God initiates the evolutionary process with a view to producing rational social beings that have true central moral beliefs.

    April 19, 2011 — 14:24
  • Mark Murphy

    A very helpful discussion of the prospects for noncognitivism generally and expressivism in particular is Mark Schroeder’s book Noncognitivism in Ethics. It makes very clear how serious the problems are for views like Blackburn’s and how expressivists have barely scratched the surface of providing a solution to the Frege-Geach problem.

    April 19, 2011 — 15:47
  • God, evolutionary psychology and moral realism

    Define moral realism as the claim there are objective moral truths and that we know some of them. Consider the argument: If no religious beliefs are true, moral realism is false. Moral realism is true. So, some religious beliefs are true. I won’t argue…

    August 5, 2011 — 16:08