Home » Library » Modern Library » Reply to Martin on Atheism and Morality

Reply to Martin on Atheism and Morality


Jeffery Jay Lowder

Michael Martin is the author of Atheism, Morality, and Meaning. I recently wrote a review of that book.[1] I concluded that although Martin’s book is a welcome addition to the literature on the relationship between atheism and morality, it does contain some significant shortcomings. In a recent essay, Martin responds to both my review and a review by Taner Edis.[2] According to Martin, not only are my criticisms “inaccurate,” but some portions of my review are even based on an “unsympathetic reading” of his book. In this piece, I will defend my objections against Martin’s replies.

Atheism and the Motivation for Being Moral

One of the well-known worries about atheistic ethics is that atheism provides no assurance that justice will triumph over evil, and thereby weakens atheists’ motivation to behave morally (hereafter the “weakened motivation objection”). It is important to be clear on precisely what this worry is about. The problem is not that it is impossible for atheists, by virtue of their atheism, to be motivated to behave morally. Rather, the alleged problem is that, all other things held equal, atheists have less motivation to be moral than their theistic counterparts.

In my review, I argued that Martin’s discussion of this worry is all too quick: he argues that the mere possibility of the triumph of justice is sufficient to answer this worry, while the issue is one of probability. In reply, Martin writes:

Lowder also is concerned that I do not meet the objection that atheists cannot have motivation for being moral.

Unfortunately, as my above remarks should make clear, Martin has misconstrued the issue. For the record, I believe that Martin does succeed in meeting the objection that atheists cannot have motivation for being moral (hereafter, the “impossible motivation objection”). Rather, my point is that Martin’s response to the impossible motivation objection is not successful as a response to the weakened motivation objection.

Moreover, by portraying the issue in his book as one of impossibility, Martin fails to accurately represent the worry actually defended by contemporary theistic philosophers. For example, Robert Adams, who is widely regarded as the most prominent defender of moral arguments for the existence of God as well as the divine command theory, argues that a naturalistic worldview weakens moral motivation. As Adams puts it:

Having to regard it as very likely that the history of the universe will not be good on the whole, no matter what one does, seems apt to induce a cynical sense of futility about the moral life, undermining one’s moral resolve and one’s interest in moral considerations.[3]

The impossible motivation objection is weak and is not pressed by contemporary theistic philosophers (though it probably is pressed by theistic laymen). In contrast, the weakened motivation objection, which is stronger than the impossible motivation objection, is defended by contemporary theistic philosophers. In that sense, Martin’s discussion is based on an extremely uncharitable interpretation of theistic worries about atheistic metaethics.

Let us return to Martin’s reply to my review. Martin next discusses the role of externalism in his defense of atheistic metaethics.

Interestingly enough, [Lowder] latches on to only one small strand of my defense that atheists can be motivated–and misunderstands it. For the record, I advocate externalism: the view that moral facts are not inherently prescriptive and that if a moral fact is motivational, this is contingent both on what the moral fact is and on the psychological state of the agent (p.37).

However, in my original review I clearly acknowledged that Martin is an externalist. Moreover, it is unclear why Martin thinks externalism helps to provide a successful answer to the theist’s worry. Indeed, even in his reply to my review, Martin continues to ignore the effect that theism or atheism can have on the psychological state of the agent. Theists (or at least traditional theists) believe that the triumph of justice is guaranteed by God. In contrast, atheists have no such belief. Surely the theist’s belief and the atheist’s lack of belief can and often do influence their moral motivation. According to one version of the weakened motivation objection, while atheists have various secular motives for being moral, the theist will always have (at least) one more motive to be moral: the theist can have all of the motives for being moral that the atheist has, plus additional theological motives not available to the atheist, including gratitude to God and a motivation to go to Heaven. So it still seems to me that there are interesting arguments about the effect of atheism on moral motivation, arguments that Martin does not address in his book.

Martin also protests that I did not mention his points defending the possibility of atheistic moral motivation.

I also refute two arguments against the possibility that atheists cannot be morally motivated (pp.27-34) and in the process indicate that atheists can be motivated by everything from high moral ideals to prestige (p.31). In addition, I show that atheists might be motivated to work for the triumph of justice even if this were unlikely so long as such triumph were considered a great value (p.33) and I also say that in some cases atheists might be motivated to be moral even if it were known for certain that evil would triumph (p.33). (italics mine)

I did not mention those points because they are not relevant to the weakened motivation objection. The fact that atheists can or might be morally motivated does not in any way answer the objection that atheism allegedly weakens atheists’ motivation to behave morally. For example, while an atheist might be motivated to work for the triumph of justice so long as such triumph were considered a great value, a theist could be even more motivated to work for the same outcome.

Martin also objects that I do not mention the discussion in his book about the problems with theistic moral motivation. In his words, “even if atheists have motivational problems, so do Christians (pp.177-180).” Although I agree with Martin on this point, it does not refute the weakened motivation objection. There could be problems with theistic moral motivation and yet the weakened motivation objection could be correct: although there are problems with theistic moral motivation, it could still be the case that there are even more problems with atheistic moral motivation, such that atheism still weakens moral motivation. Additional argumentation is needed to show that the problems with theistic moral motivation either counterbalance or outweigh the problems with atheistic moral motivation. Along those lines, Martin does make one point in his book that is relevant to the present discussion:

As we have also seen, the homicide and crime rate is much higher in the United States, a religious country, than it is in Western Europe and Scandinavia, where religion is far less prominent. It would seem that the threat of divine punishments and rewards does not work very well. (p. 178)

While Martin is correct about this, there is more that needs to be said. In addition to the prospect of Heaven and the threat of Hell, there are other motivators available to the theist but not to the atheist. I shall say more about these other motivators later on. (Of course, there are also other demotivators available to the theist but not to the atheist.) Moreover, it is not clear that comparing the homicide and crime rates of a single religious country, the United States, against Scandinavia, is a representative sample of either religious or secular countries. (That Martin’s comparison involves a representative sample of both religious and secular countries needs to be shown, not assumed.) And while the threat of divine punishments and rewards may not be an effective motivator, this may be in part due to the irrationality of the unmotivated: it may be the case, just as Martin suggests in his book, that people are not motivated by the threat of divine punishments and rewards because those threats are so far off in the future.

Martin also takes issue with my statement, “If the desired outcome of an action is unlikely, then one may well be unmotivated to perform it.” In reply, he writes:

But this neglects my point that a small chance of success combined with the great value of the outcome can be motivational. If the high probability of an outcome were necessary for motivation, no one would be motivated to buy lottery tickets, and social reformers would not be motivated to work against overwhelming odds.

But notice I never claimed that the high probability of an outcome is “necessary” for motivation. All I claimed was the following:

If the desired outcome of an action is unlikely, then one may well be unmotivated to perform it. Indeed, in the event that the desired outcome is highly unlikely, one may even be highly motivated not to perform it!

I agree that some people in some situations are motivated to perform an action even if the desired outcome is unlikely. However, in addition to the probability and value of the desired outcome, the cost of an action is also a factor. Martin’s example of buying lottery tickets is telling in this regard. Many people are motivated to buy lottery tickets, but usually the cost of doing so is low. Relatively few people, apart from irrational gambling addicts, would be willing to spend, say, half of their entire annual salary on lottery tickets. This distinction is relevant since, if we assume that conflicts between morality and self-interest are possible, the cost of some moral actions can be high. This example shows another way in which theists worry that atheism weakens moral motivation: while atheists may be motivated to be moral when doing so involves little or no inconvenience, according to this objection, atheists have less motivation for performing acts of self-sacrifice than theists. As for Martin’s other example of social reformers, social reformers are surely motivated not only by the value of the ultimate desired outcome, but also by the prospect of intermediate progress.

Finally, suppose, for the sake of argument, that atheists know with absolute certainty that evil, not justice, will ultimately triumph. What would be the implications of that certainty on atheistic moral motivation? In my review, I criticized Martin’s discussion of that question on the grounds that it does not address the issue of motivation (and specifically, the weakened motivation objection). In his response, Martin says that we should interpret his answer to that question as follows: an atheist “could be motivated to adopt certain virtues in order to practically cope with evil” (my emphasis), even in the situation where the atheist knows with absolute certainty that evil will ultimately triumph. Unfortunately, even Martin’s preferred interpretation doesn’t address the weakened motivation objection. Rather, it seems to confirm the objection. If atheists know with absolute certainty that evil will ultimately triumph while theists believe that justice will ultimately triumph (and everything else is held equal), one would expect that atheists (or at least rational atheists) would have weaker moral motivation than theists (or at least rational theists).

Of course, even if the weakened motivation objection is correct, this does not make God’s existence more probable than not. The weakened motivation objection might be true while at the same time there is no God or immortality. If the weakened motivation objection is right, then we have a public policy problem about whether promoting atheism will have deleterious social consequences; what we don’t have is an intellectual reason to reject atheism.

Atheism and the Justification for Being Moral

In my review, I criticized Martin’s book on the grounds that it does not discuss the implications of atheism on the justification for being moral (hereafter, “moral justification”). Martin agrees that he did not directly or explicitly discuss moral justification in his book, but is ambivalent about whether he should have discussed it. However, the issue of moral justification is probably the single most common concern among laymen about atheistic metaethics. Moreover, independently of atheism, the justification for being moral is of interest to many moral philosophers. Therefore, I think the issue of moral justification cries out for a discussion in a book-length treatment of atheistic metaethics.

In his reply to my review, Martin briefly discusses moral justification. Martin believes that there are cases where the demands of morality conflict with a person’s self-interest. Moreover, Martin believes that atheists in such situations are unjustified in choosing morality over self-interest. However, Martin does not discuss the implications of his moral justification position combined with his moral motivation position. In a nutshell, Martin’s position amounts to this: “Atheists may desire to be moral generally, but they are unjustified in acting on that desire in specific situations when being moral conflicts with self-interest.” (To be fair to Martin, although he qualifies his statement by saying, “if morality is justified in terms of self-interest,” he never indicates another way in which morality might be justified.) This doesn’t seem to me like much of a defense of the morality of atheists; indeed, I fear that Martin’s comments will only serve to confirm people’s worst fears about atheistic morality. Moreover, although Martin doesn’t “think there is much to say,” I think there is much more that could be said on the topic in order to deflate the issue.[4] It is unfortunate that Martin missed such an important opportunity to do so.

Martin also considers the implications of Christianity for the issue of moral justification. He writes:

However, even if God exists, insofar as Heaven is a reward for faith and not for moral behavior, it is also true for Christians: being moral would not be correlated to heavenly reward. My discussion in Chapter 17 is relevant to this point.

Even if Martin is correct that, for Christians, being moral is not correlated to heavenly reward, heavenly reward is not the only possible reward for Christians who behave morally. Christians (and theists generally) can find a “reward” in the satisfaction that comes from knowing they have behaved in a way pleasing to God. Moreover, even if Heaven is a reward for faith and not for moral behavior, it is still the case that sacrificing one’s own life is less costly for a Christian than it is for an atheist. And Martin’s point does not apply to theists who don’t require being moral as a condition of some supernatural benefit.

Atheism, Moral Realism, and Moral Objectivism

Ordinary usage of the terms “objective morality” and “subjective morality” assume that the concepts are incompatible with one another. In my review, I criticized Martin’s preferred definition of “objective morality” on the grounds that it fails to exclude subjectivism. Whereas Martin defines objective morality as the belief that there are moral facts, subjectivism allows that there is a fact of the matter about what is good or valuable. Interestingly, Martin never denies that his preferred definition of “objective morality” fails to exclude subjectivism. Instead, he provides multiple objections to Sayre-McCord’s definition: (1) “Objective morality” has several senses, but I assume there is only one meaning; (2) I do not explain why Sayre-McCord’s definition captures ordinary usage; (3) Sayre-McCord’s definition contradicts ordinary usage since religious believers assume that theistic morality is objective; and (4) Sayre-McCord’s definition contradicts ordinary usage since philosophers regard the Ideal Observer Theory as objective but Sayre-McCord’s definition denies this.

However, these objections are flawed. (1) I agree that “objective morality” has several senses, but Sayre-McCord’s definition (or something like it) is the most natural sense in the context of moral arguments for God’s existence. (2) I explained that Sayre-McCord’s definition captures ordinary usage because it excludes subjectivism. Moreover, it captures the metaphysical aspect of objectivity implicit in theistic arguments. Martin should know from his own replies to moral arguments for God’s existence that when theists talk about “objective moral values,” they are making an ontological claim (objectivity as correspondence with objects) and not an epistemological claim (objectivity as impartiality).[5] (3) Yes, Sayre-McCord’s definition excludes voluntaristic versions of theistic morality from being objective, since his definition of moral objectivism requires that the truth-conditions of moral claims make no reference to anyone’s subjective states, including God’s subjective states. Sayre-McCord’s definition could be easily modified to perfectly represent common usage if we replaced “anyone’s subjective states” with “any human’s subjective states.” Yet Sayre-McCord’s definition is superior to the modified definition, since it is ad hoc to make an exception just for God: just about the only people who argue for such an exception are defenders of the divine command theory or some form of theological voluntarism. Moreover, with or without the modification, Sayre-McCord’s definition much better reflects common usage than Martin’s definition, since the former focuses on the ontological sense of objectivity whereas the latter focuses on the epistemological sense. Finally, as for (4), whether the Ideal Observer Theory (IOT) is considered objective depends on the sense of objectivity adopted by the speaker. Taking objectivity to be correspondence with objects, it is possible to develop specific versions of the IOT according to which morality is objective, but it is also possible to develop specific versions of it according to which morality is not necessarily objective (i.e., Firthian). Thus the IOT as such is neither determinately objective nor determinatively subjective.

Atheism and the (Firthian) Ideal Observer Theory

There are multiple versions of the IOT. Martin adopts the version advocated by Roderick Firth. In my review, I criticized Martin’s appeal to the Firthian version of the IOT. I began my discussion of the IOT by quoting Martin’s description of the Ideal Observer (IO). I then pointed out that “when Martin describes the ‘Ideal Observer,’ many people will think ‘God.'” This is not just a problem with laymen misunderstanding the theory; many commentators in the secondary literature on the IOT have remarked that the IO sounds like another name for God.[6] In his reply, Martin complains that someone who has read only my review might assume that the IO is simply a hypothetical God. But clearly I did not portray the Firthian IO as a hypothetical God, since I quoted Martin’s entire description of the IO, and that description lacks many of the divine attributes, including omnipotence, omnibenevolence, etc. Perhaps I should have made explicit all of the differences between the IO and God, but I didn’t consider it necessary. My point was that in spite of all of the differences pointed out by Martin, many people will think the IO sounds like God: although the IO is not identical to God, IO is attributed with so many super-human abilities (e.g., omniscience with respect to nonethical facts, omnipercipience) that it sounds more like a supernatural being than a regular human. I believe that this remark is entirely consistent with (and based upon) a highly sympathetic reading of Martin’s book.

Next, I offered three objections to Martin’s IOT. (1) In the absence of information about what would influence a Firthian IO’s reactions, it is unclear how the Firthian IOT justifies any normative claim over its denial. (2) If the Firthian IOT is to be interpreted as a form of ethical naturalism, then the Firthian IOT does not provide an ultimate ontological foundation for moral properties. (3) Using a variant of the Euthyphro dilemma to clarify the role of the IO, I argued the Firthian IOT is not a complete metaethical theory, since it does not offer an explanation for moral facts and properties. In retrospect, it now seems to me that (2) and (3) are not independent objections; rather they collapse into a single issue. Nevertheless, I shall consider Martin’s reply to each objection in turn.

In reply to (1), Martin asserts that normative claims are empirical claims that can be verified by approximating the properties of an IO. However, Martin misunderstands my objection. Approximating the properties of an IO will enable us to determine the truth of normative statements if and only if the IO’s reaction is either determined or made probable by the properties of an IO. That is precisely the point in dispute. (Martin speculates that I might be on a quest for certainty, but I am not. My point is that the properties of a Firthian IO do not even make it probable that the IO will contemplate things like honesty with a feeling of approval.) Again, consider the moral claim (MC), “Honesty is morally good.” There seems to be no logical contradiction or even improbability in saying that a human could approximate the properties of an IO and yet contemplate honesty with a feeling of indifference or even disapproval. In the absence of information about what would influence a Firthian IO’s reactions (i.e., whether the Firthian IO is honest and values honesty), it is still unclear how Firthian IOT makes it probable that MC (and not its denial) is justified. One person could approximate the properties of an IO and contemplate honesty with a feeling of approval, while another person might contemplate honesty with a feeling of indifference.

Turning to (2), Martin claims that I “misunderstand” error theory; IOT is incompatible with error theory. He writes,

According to John Mackie’s classic discussion of it, the objectivity is an ingrained belief but this ingrained belief is false. There are no moral facts although people firmly believe that there are. (John Mackie, Ethics, Penguin 1979, p. 49). By contrast, on the IO Theory ethical statements can be objectively true since there are moral facts.

Such a reply would have been appropriate if I had claimed that the IOT as such is a form of error theory, but I made no such claim. My own view is that the IOT as such is incomplete, since it does not specify that it is either a success theory or an error theory. In my review, I noted that the hypothetical nature of the IO might provoke someone to worry that IOT is an error theory, and I then suggested a way of construing IOT according to which it is not an error theory. Although I did, as Martin notes, ask, “How, precisely, is IOT supposed to be different from error theory?”, I went on to answer the question! I stated that an IOT could be construed as a form of ethical naturalism, in which case that form of the IOT would be a success theory.[7] Since success theory is the opposite of error theory, it follows that IOT as such does not entail error theory.

However, the IOT as such is compatible with error theory, since error theory is the view that (a) moral claims are cognitive (i.e., moral claims are either true or false), and (b) no moral claims have the truth-value true.[8] Both the IOT as such and the Firthian IOT in particular are compatible with error theory, since neither version entails that at least one moral claim is actually true. Perhaps other versions of the IOT are incompatible with error theory, but if so that is because they entail that at least one moral claim is actually true.

Finally, as for (3), Martin argues that I have created a false dilemma. On Martin’s view, moral goodness is neither invented by the IO nor independent of the IO; rather, moral goodness is relational to the IO’s reaction to a given situation. However, I’m not sure that Martin’s reply is correct. Epistemologically, moral goodness is a relational property, since we identify moral goodness by trying to approximate the IO’s reaction. Ontologically, however, moral goodness is not relational, since it is identical to some complex natural property, the categorical ground of the hypothetical specified by the IOT (naturalistically construed).

Concluding Remarks

Martin is to be commended for writing a book-length defense of atheistic metaethics, that bridges the gap between the philosophy of religion and metaethics from an atheistic perspective. Although I believe that my original criticisms of Martin’s book stand, I still maintain that his book makes a much-needed contribution to the literature and deserves to be read by anyone interested in the relationship between atheism and morality. While Martin may not agree with my criticisms, I hope he at least now believes that I am not unsympathetic to his book.[9]


[1] “Review: Michael Martin, Atheism, Morality, and Meaning” 2003 (URL:https://infidels.org/library/modern/jeff_lowder/martin_review.shtml), spotted June 28, 2003.

[2] “On Two Reviews of Atheism, Morality, and Meaning,” 2003 (URL:https://infidels.org/library/modern/michael_martin/on_two_reviews.shtml), spotted June 28, 2003.

[3] Robert Merrihew Adams, “Moral Arguments for Theistic Belief,” in Rationality and Religious Belief, (ed. C. F. Delaney, Notre Dame, Ind.: University of Notre dame Press, 1979), p. 127.

[4] Cf. Peter Byrne, The Moral Interpretation of Religion (Grand Rapids: Wm. B. Eerdmans, 1998), pp. 34-47.

[5] I have in mind Martin’s critiques of William Lane Craig and Paul Copan, both of whom are concerned with the “ontological foundation” for objective moral values. See his “Atheism, Christian Theism, and Rape” 1997 (URL:https://infidels.org/library/modern/michael_martin/rape.html), spotted June 28, 2003; “Copan’s Critique of Atheistic Objective Morality” Philosophia Christi Series 2, 2/1 (2000): 75-89, republished electronically at (URL:https://infidels.org/library/modern/michael_martin/copan.html), spotted June 28, 2003; “The Naturalistic Fallacy and Other Mistaken Arguments of Paul Copan” (URL:https://infidels.org/library/modern/michael_martin/nat_fallacy.html), spotted June 28, 2003.

[6] E.g., Roderick Firth, “Ethical Absolutism and the Ideal Observer” Philosophy and Phenomenological Research 12 (1952): 317-345; Charles Taliaferro, “The Divine Command Theory of Ethics and the Ideal Observer,” Sophia 22 (1983), pp. 3-9; Paul Copan, “Can Michael Martin Be a Moral Realist? Sic et Non,” Philosophia Christi, Series 2, 1/2 (1999), p. 51, n. 25; Peter Williams, “Is There a Personal God?” Philosophers’ Magazine 8 (Autumn 1999): 22-23.

[7] However, even that was not quite right. I should have stated that if IOT were construed as a form of ethical naturalism and IOT were defined in such a way as entailing that at least one moral claim is actually true, then that form of IOT would constitute a form of success theory.

[8] Geoffrey Sayre-McCord, “Introduction: The Many Moral Realisms” in Essays on Moral Realism (ed. Geoffrey Sayre-McCord, Ithaca: Cornell University Press, 1988), p. 10.

[9] I am grateful to Glenn Branch, Keith Augustine, and Jim Lippard for helpful suggestions on a previous draft of this essay.

all rights reserved