Disconcerting Denouements of Darwinian Dilemmas

The revenge of moral knowledge

This the last time I pass off essays written for other purposes as blog posts, I swear! (At least until the summer.) This time we’re defending moral realism against evolutionary debunking. This might also be considered a rebirth of the moral knowledge argument that i absolutely DESTROYED a couple of weeks ago (though I wrote what you’re about to read some time before that post). Enjoy!

1. Introduction

Very few still doubt that humans are the result of a process of evolution. It is likewise quite clear that this process is guided primarily by a concern for survival and reproduction (i.e., evolutionary fitness). But if our cognitive faculties are the result of such a process, how can we trust the beliefs formed by these? In her article “A Darwinian Dilemma for Realist Theories of Value,” Sharon Street (2006) argues that while our faculties are generally reliable, they are not reliable within the domain of value judgements, and that this should make us skeptical of moral realism. My thesis here is that Street’s Darwinian dilemma leads to unacceptable skeptical conclusions when applied to epistemology, and solving the epistemological problem will likely also solve the moral one.

The article will proceed as follows: §1 explains the metaethical positions relevant for our purposes; §2 summarizes Street’s argument and briefly attempts to diagnose the root of the skeptical worry; §3 parodies Street’s argument, for epistemology, and argues that the result here is unacceptable; and §4 sketches two possible solutions to the epistemic problem, and argues that we should expect solutions to the epistemic problem to solve the moral problem as well.

2. Moral Realism and Anti-Realism

Let us first get clear on the positions in question. Overall, we may classify metaethical positions into realist and anti-realist ones. Moral realism holds that there are stance-independent moral facts and/or properties—i.e., there are some moral propositions that are true, and the truth of these does not depend upon the attitudes of agents. Anti-realism is simply the negation of this.

Realism may be further divided into naturalism and non-naturalism. Naturalist realism holds that moral facts and/or properties are natural facts and/or properties. Non-naturalist realism is realism plus the negation of natural realism—i.e., moral facts and/or properties are non-natural. It is not perfectly clear what we should take “natural” to mean, though a plausible account is that natural facts are those that can be investigated through scientific methods (whatever we take that to include). This implies being able to be part of causal relations, though it is plausibly not a sufficient condition. Many moral realists will also claim that moral facts provide us with categorical reasons—as opposed to hypothetical reasons. While the force of hypothetical reasons depends on agents’ desires (e.g. “if you want to pass this exam, you ought to write a good article”), categorical reasons apply to all agents, regardless of desires (e.g. “you ought not murder”).

For completeness, I will also note that anti-realism is generally divided into non-cognitivism (the view that no moral utterances are truth-apt), error theory (the view that, while truth-apt, all moral propositions are false), and subjectivism (the view that some moral propositions are true, though their truth is stance-dependent). All the positions mentioned here can of course be further subdivided, but that level of detail will be unnecessary for our purposes: assessing Street’s Darwinian dilemma, as we will now do.

3. The Darwinian Dilemma

The starting-point of Street’s argument is an empirical one: that many of our moral judgements promote evolutionary fitness if followed. For example “the fact that something would promote one's survival is a reason in favor of it” and “the fact that someone is altruistic is a reason to admire, praise, and reward him or her” (Street, 2006, p. 115). Creatures with the disposition to form these sorts of preferences would generally have a better chance of reproducing than creatures with, say, the reverse preferences, or any number of arbitrary preferences. To be precise, Street does not propose that our actual moral judgements are directly influenced by evolutionary pressures. Rather she claims that what she calls our “basic evaluative tendencies” are directly influenced by evolutionary pressures, and that these in turn heavily influence our actual moral judgements. I will grant these evolutionary claims for the purposes of this article.

So it looks like we have a perfectly good explanation of why we make the moral judgements we do. This does not by itself present a challenge to moral realism—everything about us is presumably explained in some sense through evolution. The challenge lies in the fact that the explanation of our basic evaluative tendencies—and thus of our actual moral judgments—doesn’t track the moral facts. In most domains, evolution generally seems to track truth; a creature with unreliable senses will generally do worse than one with reliable ones. This means that evolution will generally be truth-tracking when it comes to observations of medium-sized objects on earth. But whether or not it is actually wrong to murder my children will not affect whether it is conducive to my fitness not to murder my children, and thus whether or not I actually believe that it is wrong to murder my children. Likewise for any moral fact, meaning our moral faculties are not truth-tracking.

Street now presents a dilemma for realists. The first horn of the dilemma is to deny that there is any relation between moral facts and evolutionary pressures. If this is right, then we should have no reason to trust our basic evaluative tendencies. There may still be a little room for the realist here. While basic evaluative tendencies have a strong influence on our particular moral judgments, there are still other factors—such as upbringing—that influence these. If one such factor somehow tracks moral truth, then the realist has a response to the argument. One candidate is that rational reflection plays this role. But as Street points out (2006, p. 124), moral reflection only consists in assessing certain evaluative tendencies in terms of other evaluative tendencies, and so if the starting point is not truth-tracking, there is no reason to expect the considered result to be.

The second horn of the dilemma is of course to affirm a relation between evolutionary pressures and moral facts. In doing so, the realist will be proposing a scientific theory to explain the facts about creatures and their attitudes, which makes reference to the moral facts. But this will make the realist’s theory open to scientific scrutiny, and here the anti-realist theory wipes the floor. Since our basic evaluative tendencies so nicely line up with what we would antecedently expect from looking at the non-moral facts, the realist theory does not provide anything in terms of explanatory power. Additionally, the realist theory postulates things that the anti-realist theory doesn’t, making it less parsimonious. All of this suggests that we have no reason to believe the realist theory, and should simply discard it. Thus neither horn of the dilemma gives the realist much hope of being justified in trusting their normative judgements—and since these horns exhaust the logical space, we are justified in rejecting moral realism.

While Street doesn’t do this, I will here attempt to diagnose what exactly leads to the skeptical worries. I think the problem is that our moral beliefs are not sufficiently sensitive. By a sensitive justification I mean one that satisfies the criterion: P is justification for Q only if P is more likely if Q is true than if Q is false. I think this principle will be highly intuitive for many. As an illustration, imagine that your only reason for thinking it is 8:00 is that your watch shows that time. You now find out that your watch is stopped. This would mean that it would be equally likely to show that time, regardless of whether or not it really is 8:00—and exactly for this reason you are no longer justified in your belief. I will then call a belief insensitive just in case none of the justifications for it are sensitive. It is beyond the scope of this article to defend sensitivity as the correct diagnosis of the Darwinian dilemma, though what I argue in this article can likely be fitted into alternative diagnoses.

If our best theory would make our moral beliefs insensitive, as Street suggests, then we cannot be justified in thinking any particular moral proposition is true. But if that is right, it is hard to see what justification we could have for at least one moral proposition being true—especially since justifications for moral realism usually proceed from particular moral judgements.

As a side note, this should make it clear that naturalist realists don’t escape the problem. The claim of a naturalist theory is that moral property P is identical with natural property N. Thus what you would need for a difference in moral facts would be for P not to be identified with N. But whether or not P is identical with N, or some other natural property M, does not have any impact on the causal history of the natural world; N-things do what they do, and M-things do what they do, regardless of which (if any) is identical to P. So naturalism is no more sensitive than non-naturalism.

It is now clear what a realist response to the Darwinian dilemma must consist in. It must 1) develop a theory that explains how the justification for our moral beliefs can be sensitive, and 2) show why this theory is preferable to insensitive alternatives. In the following I shall do this in reverse order.

3. Darwinian Self-Defeat

I shall now argue that the reasoning Street employs in arguing that our moral beliefs are insensitive, can equally show that certain epistemic beliefs are insensitive. But whereas we might coherently reject our moral beliefs, we cannot do the same for the epistemic ones. In §4 I shall argue that solving this problem will likely also solve the moral one.

Consider the principle, P, that “out of two theories that make the same predictions, you should prefer the simpler one.”¹ Street herself seems to employ P in arguing against the second horn of the dilemma: the problem with the theory that evolution tracks moral facts is that it makes the same predictions as the anti-realist theory, but does so in a more complex way. This principle is also crucial to many arguments in philosophy and science in general. A theory can almost always be gerrymandered into making the same predictions as another.² The only way to adjudicate between theories in this case is with something like P. But P looks just as susceptible to the Darwinian dilemma.

Street’s argument is supposed to target epistemic normativity as well (2006, Note 2), meaning this is no surprise for her. But I do not think she escapes the problem so easily. As Jonas Olson (2011), who defends an epistemic error theory, explains, skeptics of epistemic normativity still have a reason for giving arguments. Instead of them giving arguments to the conclusion that you ought to believe X, they simply argue to the conclusion that X is true. Thus they do not take arguments to have categorical normative force, but only hypothetical normative force. That is, when Street is arguing against moral realism, she is not arguing that “you ought to believe in anti-realism,” but rather that “if you want to have true metaethical beliefs, then you ought to believe in anti-realism.” While this does not rely on the prescriptive principle P, it does rely on a related descriptive claim, D, that “out of two theories that make the same predictions, the simpler one is more likely to be true.” Just like with P, it is not at all clear how the truth or falsity of D could be sensitive.

On the face of it, it makes sense that evolution would favor a belief in simpler theories. After all, it takes more processing power to figure out the consequences of a more complicated theory, meaning creatures who believed more complicated theories would either have to spend more time processing information, or would have to invest energy into developing more powerful cognitive capacities in order to accomplish the same task, both of which would decrease fitness. Thus we have an evolutionary account of our belief in D. At the same time, these evolutionary pressures would not have been tracking the truth or falsity of D. For them to have done this would require that the probability of evolution occurring one way over the other had been affected by the probability that one of two predictively equivalent theories is true. But for two theories to be predictively equivalent implies that the probability of evolution occurring one way over the other is the same, regardless of which is true. So the relative probabilities of these two theories cannot by itself make any difference to evolutionary pressures.

We thus have a Darwinian dilemma for the believer in D: They either deny a relation between evolutionary pressures and epistemic principles like D, in which case our belief in D probably is not sensitive. Otherwise they affirm a relation. But since evolutionary theory already predicts our belief in D without reference to its truth, any theory which does reference its truth will make the same predictions, but in a more complicated manner, and should thus be rejected. We can not accept this argument for two reasons: 1) D and similar principles are foundational to our epistemic practices, and denying them would lead us to a radical sort of skepticism, and 2) the argument itself presupposes D to argue against it, meaning we cannot accept the argument if we accept it. The claim here isn’t that this shows that D is actually true, but only that D cannot be rationally rejected.

One objection to the argument may be that a principle of simplicity like D simply follows from the basic axioms of probability, which we surely can know: The probability of the conjunction A & B must always be lower than the probability of A by itself.³ This means that adding more content to a theory trivially makes it less probable. But we must be careful here with what our principle states. What Street needs is that the probability of A & B is greater than the probability of A & not-B,⁴ which is not equally trivial. After all, this style of argument is not simply intended to show that we should withhold judgement with regards to whether some extra entity exists, but that we are justified in rejecting its existence.

Another suggestion may be that D is simply a special case of a more general principle, the truth of which is sensitive. This may be something like the principle, E, that “simpler theories are more likely to be true, all else being equal,” which entails D. If simpler theories were not more likely to be true, then believing E would make creatures less fit, since they would have a higher chance of believing false theories, some of which might lead them to get killed through making false predictions.

This objection conflates evolution tracking X with X being correlated with what evolution tracks. A realist might similarly object that “evolution generally tracks truth, and moral truths are truths, so evolution tracks moral truths.” The problem here is that while truth is strongly correlated with what evolution tracks, it is not truth, but fitness, that evolution tracks and moral truths are irrelevant for fitness. Similarly, what evolution is really tracking when it tracks E is predictive accuracy. Thus whether or not simplicity generally makes theories more likely, would not affect whether simplicity were tracked for predictively equivalent theories, since these will have the same predictive accuracy. But it is exactly in the domain of predictively accurate theories that evolution would have to track whether simplicity increases probability, in order for P to be sensitive.

The argument that has been given here may be given for other principles as well. One obvious example is induction: creatures who did not believe that an event occurring many times in the past would make it more likely to occur in the future, would not have survived long, since induction has in fact held historically. At the same time, our evolutionary history cannot in principle have been tracking whether induction is actually reliable, since it happened in the past, and so cannot have been affected by whether something is likely to continue happening in the future. Seeing as we cannot rationally accept that our belief in D and similar principles isn’t sensitive, we are justified in looking for some theory that would make it sensitive.

4. Candidate Solutions

If the above argumentation is sound, then that in itself should make us skeptical of the Darwinian dilemma. Even so, it is not sufficient to show that the Darwinian dilemma is still not successful in the case of morality. Doing this will require doing the bread and butter work of assessing what theory does the best job at making epistemic beliefs like D sensitive, and then seeing whether that theory also makes moral beliefs sensitive. I will here gesture at two possible solutions to the problem, before considering what we can expect from solutions in general.

4.1 Axiarchism

One hypothesis that may allow D to be sensitive is axiarchism. This is the view that the world is largely or entirely determined by what is good (Bunnin & Yu, 2004, p. 65). With axiarchism, we have a neat explanation of why we know D: because it is good to know D. Were D not true, it would not have been as good to believe, and so we would have been less likely to believe it.

An important step of this is obviously that knowledge of D actually is good. If this is not right, then axiarchism will not get us any closer to being justified in believing D. This must either be because knowledge is intrinsically good, or because knowledge of D is somehow instrumentally good. But whatever explanation is given here, it will almost certainly also apply for moral knowledge. If knowledge is intrinsically good, then this includes moral knowledge; and if knowledge of the epistemic principles in question is instrumentally good, then moral knowledge will almost certainly be too—after all, moral knowledge would allow creatures to deliberately act morally.

Another question is by which mechanism reality is determined by what is good, given axiarchism. Different specifications of the theory will give different answers, from some nondescript force guiding the universe, to an omni-God being the creator of it. Such a specification will also determine which horn(s) of the dilemma the theory takes. But it is beyond the scope of this article to develop any full account.⁵

4.2 Rational Insight Theory

The other sort of response consists in postulating some special faculty which allows us to “grasp,” “get at,” or “just see” abstract epistemic truths like D. That is, we have some faculty that somehow allows us to access truths that could not otherwise have played a role in the history leading up to our deliberation. So while D could not in principle have had an effect on selective pressures, D is still true, and through our faculty of rational insight, we are able to just see that D is true. Thus our belief in D is sensitive, since we would not have grasped the truth of D, were D not true. This appears most compatible with taking the first horn of Street’s dilemma.

As with axiarchism, it is far from clear how exactly this is supposed to work, and this short sketch will sadly not be of much help. But however it does work, it would appear quite ad hoc to think that this faculty of rational insight would not also allow us to grasp moral facts, if such exist. What feature of epistemic principles makes these graspable, which doesn’t also allow moral truths to be grasped? The anti-realist will have to come up with some explanation for some difference, and argue why accepting this explanation is to be preferred.⁶

These very brief sketches will hardly be convincing to anyone who does not already accept the theories in question, but the idea is simply to give an idea of how a response can look. While there can be no in-principle argument for why a theory that solves the epistemic problem also must solve the moral one,⁷ it appears that principles like D have enough in common with moral facts, to expect that a solution to one would likely solve the other. Specifically, both sorts of facts are such that their truth or falsity cannot itself have made any difference to us; they need some help. A theory that explained our knowledge of D would need to posit some mechanism by which a fact that cannot have influenced our history, could still have been sensitive. But the charge against moral facts is exactly that they cannot have influenced our history, and so it would be strange if that same mechanism couldn’t also make moral facts sensitive. I can do no better here than make these sorts of vague gestures—and everything will come down to what the best explanation for our knowledge of the epistemic facts is. But I suspect that it will turn out that solving the epistemic problem without solving the moral problem will generally require extra ad hoc assumptions.

Before concluding, I think it is worth making the dialectical structure and significance of what I have argued clear. What I have said in this article lends little to no positive support to moral realism. What it does do is give an indirect way of meeting Street’s challenge. If what I have argued in §3 is correct, then we have a strong reason to accept a theory like the ones described here. And if what I said above is true, such a theory will likely defuse the Darwinian dilemma for moral realism. Likewise, even if what I have argued in §3 is false, one who already accepts a theory like these, will not have to worry about the Darwinian dilemma for moral realism.

Some authors (e.g. (Crummett & Swenson, 2019)) have made arguments from moral knowledge to these sorts of theories, in the face of skeptical challenges like Street’s. I think this is a mistake—and it shows why the detour into epistemological facts is necessary. The only reason we can infer from the epistemic problem to a theory that solves it, is that it is self-defeating to give up principles like D on the basis of arguments that employ them. But there is nothing self-defeating about giving up moral knowledge. Seeing that our best theory would undercut moral knowledge, and inferring from this that there must be some other explanation for our moral knowledge, would be like finding out that my watch is most likely stopped, and then inferring from my belief in it being 8:00 that there must be some other explanation for how I know it is 8:00—my justification for my moral/temporal belief is exactly what has come into skeptical doubt.

5. Conclusion

I have been arguing that Street’s Darwinian dilemma leads to unacceptable skeptical conclusions when applied to epistemology, and solving the epistemological problem will likely also solve the moral one. This I have done by running a parody of Street’s argument, against a principle of simplicity, which much of our thinking—including the original Darwinian dilemma—relies on. Crucially, this principle is descriptive, rather than prescriptive, meaning it has “dialectical tooth,” even against global normative skeptics, such as Street. The conclusion of this parody is unacceptable, hence why we need some way out of it. But as I have then argued, a way out of this argument will probably also provide a way out of the original dilemma for moral realism, leaving moral realists with a way of escaping Street’s dilemma.

6. References

Bogardus, T. (2016). Only All Naturalists Should Worry About Only One Evolutionary Debunking Argument. Ethics, 126(3), 636–661.

Bunnin, N., & Yu, J. (2004). The Blackwell Dictionary of Western Philosophy (1st ed.). John Wiley & Sons, Ltd. https://doi.org/10.1002/9780470996379

Crummett, D., & Swenson, P. (2019). God and Moral Knowledge. In K. Vallier & J. Rasmussen (Eds.), A New Theist Response to the New Atheists (pp. 33–46). Routledge. https://philarchive.org/rec/CRUGAM

Cutter, B. (2023). From Moral Realism to Axiarchism. Midwest Studies in Philosophy, 47, 73–101. https://doi.org/10.5840/msp2023102741

Olson, J. (2011). Error Theory and Reasons for Belief. In A. Reisner & A. Steglich-Petersen (Eds.), Reasons for Belief. Cambridge University Press.

Street, S. (2006). A Darwinian Dilemma for Realist Theories of Value. Philosophical Studies, 127(1), 109–166. https://doi.org/10.1007/s11098-005-1726-6

1. Where parsimony increases simplicity.

2. E.g. by adding the auxiliary hypothesis of an evil demon making the other theory appear true.

3. Unless the probability of B given A is 1, in which case A&B has the same probability as A.

4. When B is a complicating claim, such as a positive existential.

5. (Cutter, 2023) defends axiarchism as a response to moral evolutionary debunking at length.

6. (Bogardus, 2016) defends this type of view as an option to escape moral evolutionary debunking.

7. A possible solution is a God who cares about epistemic—but not moral—beliefs.