Introduction
Arguments from evil purport to show that some fact about evil makes it at least likely that God does not exist. Skeptical theism is a popular response to arguments from evil, and—if true—is held to undermine a key inference often used in such arguments. The aim of this article is show that everyone should endorse skeptical theism—or, at least, the skeptical component of it. In Section 1, I explain how arguments from evil work, and illustrate how skeptical theism undermines them. And in Section 2, I put forth an argument for the truth of skeptical theism. The upshot is that everyone should endorse the skeptical component of skeptical theism, and hence everyone should reject arguments from evil.
1. Skeptical Theism and Arguments from Evil
Take some evil E. A God-justifying good for E is a good G that is such that God would (at least possibly) be justified in bringing about (or allowing) E for the sake of G. A good is a God-justifying good if and only if (i) it outweighs E and (ii) it could not have obtained without E (or something as bad as or worse than E) obtaining. This is a standard understanding of God-justifying goods (what Michael Bergmann [Reference Bergmann, Clark and Rea2012] calls God-justifying reasons and what others call morally justifying reasons). For example, William Rowe says that ‘an omniscient, wholly good being would prevent the occurrence of any intense suffering it could, unless it could not do so without thereby losing some greater good or permitting some evil equally bad or worse’ (Reference Rowe1979: 336) and Michael Bergmann says ‘a good state of affairs G—which might just be the prevention of some bad state of affairs E*—counts as a God-justifying reason for permitting an evil E if and only if (i) G's goodness outweighs E's badness and (ii) G couldn't be obtained without permitting E or something as bad or worse’ (Reference Bergmann, Clark and Rea2012: 11, footnote 5). (Many other philosophers operate with this understanding of a God-justifying good, see, e.g., William Alston [Reference Alston1991], Daniel Howard-Snyder [Reference Howard-Snyder and Howard-Snyder1996], Hud Hudson [Reference Hudson2014a], Stephen Maitzen [Reference Maitzen, Dougherty and McBrayer2014], and William Rowe [Reference Rowe1991, Reference Rowe and Howard-Snyder1996, and Reference Rowe2006].)
Some arguments from evil rely on something like the following inference:
(1) We know of no God-justifying good for an evil E.
(2) Therefore, probably, there is no such good. (The most prominent proponent of this inference is William Rowe [Reference Rowe1979].)
Since it is assumed that God's existence is incompatible with gratuitous evil—evil for which there is no God-justifying good—it follows that, probably, God does not exist. (While this assumption is common and will be taken for granted in this article, a growing number of philosophers reject it [e.g., Hasker Reference Hasker2008; Rubio Reference Rubio2018; Mooney, Reference Mooney2019; Sullivan Reference Sullivan, McBrayer and Howard-Snyder2013; and van Inwagen Reference van Inwagen2006].) Skeptical theists reject the inference from premise (1) to (2): they claim that our (supposed) lack of knowledge of a God-justifying good does not render it probable that there is no such good, and hence this version of the argument from evil fails. (Skeptical theists have argued that skeptical theism undermines other arguments from evil as well, see, for example, Bergmann [Reference Bergmann, Flint and Rea2009] and Howard-Snyder and Bergmann [Reference Howard-Snyder, Bergmann, Peterson and Arrogan2004].)
Skeptical theism comes in many varieties (for different examples, see Cullison Reference Cullison, Dougherty and McBrayer2014, DePoe Reference DePoe2017, Hudson Reference Hudson, McBrayer and Howard-Snyder2013 and Reference Hudson2014a, and Wykstra Reference Wykstra1984 and Reference Wykstra and Howard-Snyder1996). However, in this article, I will be exclusively concerned with one particular type of skeptical theism. For the purposes of this article, a skeptical theist is a monotheist who affirms the following thesis:
skepticism: We have no good reason to think that the goods and evils that we know are connected to some instance of evil are representative, in respect to [value], of the actual goods and evils that are connected to said instance of evil. (Perry Hendricks Reference Hendricks2019: 116)
There is some controversy about how to understand the ‘connected’ clause in skepticism, as defined above. For example, Kirk Durston (Reference Durston2000) seems to think understanding the connection causally is sufficient, whereas Paul Draper (Reference Draper2013) and others think that for skepticism to work, the connection must be understood as a necessary one. Since my argument will work for either view, I will not enter this dispute here. Goods and evils are taken to be states of affairs, and different states of affairs have different values. Take some disvaluable (bad) state of affairs X. If a state of affairs Y outweighs X and Y could not have obtained without X, then Y (at least possibly) justifies one in allowing X. More exactly, one is justified in allowing X only if the total value of the set of states of affairs that are produced by X is greater than the total value of the set of state of affairs that would have obtained had X not occurred. Thus, it is a necessary condition for God to be justified in allowing X that the set of states of affairs produced by X is more valuable than the set of states of affairs produced by ~X. (It is implied here that the set of states of affairs produced by X could not have obtained via ~X.) In the terminology introduced above, if the value of the set of states of affairs produced by X is greater than the value of the set produced by ~X, it is a God-justifying good. The inference from (1) to (2) claims that this necessary condition is probably not satisfied, and hence there is probably gratuitous evil, and hence God probably does not exist.
In determining whether there is a God-justifying good for X, we need to know the total value of the set of states of affairs connected to X and the total of the set of states of affairs connected to ~X. As such, skepticism, as defined above, amounts to the claim that we have no good reason for thinking that the states of affairs that we know are connected to some instance of evil are representative, in respect to value, of the actual states of affairs connected to the prior mentioned instance of evil. Crucially, this (skepticism) entails that we have no good reason to think that the total value of the set of states of affairs actually connected to some instance of evil E falls within a particular range of value: if we have no good reason for thinking that the value of the states of affairs actually connected to E fall within the same value range (are representative) as the states of affairs connected to E that we know of, then we have no good reason to think that the total value of the set of states of affairs actually connected to E falls within a particular value range. And if we have no good reason to think that the set of states of affairs actually connected to E fall within a particular value range, then we cannot infer that there probably is no God-justifying good for E because to say that is to say that the total value of the set of states of affairs connected to E is probably lower than the total value of the set of states of affairs that would have come about if ~E obtained. (I illustrate this further at the end of section 2.2.)
More abstractly, the reason many hold that skepticism undermines the inference from (1) to (2) is due to the idea is that an inductive inference is only justified if one has good reason to think that the sample one is inferring from is representative. If there is no good reason to think the sample is representative, then the inductive inference it supports is unjustified. Some examples will help illuminate why this is so.
Suppose that Mary, a scientist, conducted a study that showed that drug D cured cancer in the patients who participated in the study. And suppose that Mary inferred from the fact that the drug cured the patients involved in her study that D would cure all cancer patients. However, suppose that Mary did not give any indication of the steps she took to ensure that her sample of cancer patients used in her study is representative of all cancer patients. And suppose further that when pressed in person, Mary did not produce any good reason for thinking that her sample of cancer patients is representative. In such a case, one should not accept the conclusion of her inference: while her conclusion may be correct, one is unjustified in accepting it on the basis of her inductive inference.
Consider another example (borrowed from Hudson Reference Hudson2011 and Reference Hudson and Kvanvig2014b): I suspect there is a rabbit in my garden, so I hire the A1 Rabbit Extermination Team to investigate. The team arrives and begins to look around for rabbits. After twenty minutes, team members come and tell me that they have checked my garden, and there are no rabbits in it. ‘How do you know this?’ I ask. They reply that they checked part of the garden and found no rabbits. They inferred from their finding no rabbits in that part of the garden that there are probably no rabbits in the whole garden. If they have good reason to think that the part of the garden they checked is representative of the whole garden in respect to rabbit population, then this inference is justified, and I should be happy with the job they have done. I want to put my mind at ease, so before paying them for their work, I ask ‘Why should I think the section of the garden that you checked is representative? What steps did you take to ensure that it was?’ In response, they give me no good reason to think that it is representative and are not able to tell me any steps they took to ensure that the section they checked is representative. I am not impressed by their response. The right move here is to hire a new pest control service. The A1 Rabbit Extermination Team has not given me good reason to think that the team has done the job well: the team members have given me no good reason to think that there are no rabbits in my garden, and I am not justified in accepting their inference. Indeed, since they have no good reason to think that the section they checked is representative, they are not justified in accepting their own inference either.
In respect to arguments from evil, and in particular the inference from premise (1) to (2), the question is whether our sample of goods and evils (i.e., states of affairs) we know of is representative. And the relevant property in respect to representativeness of our sample of goods and evils is the value of the goods and evils (state of affairs) we know of (see Hendricks Reference Hendricks2019). Skepticism, however, says that we have no good reason to think that our sample of goods is representative in respect to value. Hence, given that an inductive inference is only justified if we have good reason to think its sample is representative, skepticism undermines arguments from evil that rely on an inductive inference like the one above. While there is more to say about this normative premise, the purpose of this article is to show that skepticism is true, not defend the normative premise. (For more on this normative premise, see Alston Reference Alston1991; Bergmann Reference Bergmann2001; and Hudson Reference Hudson and Zimmerman2006, Reference Hudson and Kvanvig2014b, and Reference Hudson2017.)
Obviously, only theists can be skeptical theists. However, skeptical theists think that everyone, including atheists, should accept the skeptical component of skeptical theism. That is, skeptical theists think that everyone should endorse skepticism, and hence everyone ought to reject arguments from evil that rely (tacitly or explicitly) on an inference like that of (1) to (2) (see above). Unfortunately, reasons for endorsing skepticism run slim. To motivate skepticism, sometimes we are given an analogy about parents and children (e.g., Wykstra Reference Wykstra1984) or an analogy about rabbits and gardens (Hudson Reference Hudson and Kvanvig2014b). Other times, we are reminded that human beings are cognitively limited creatures (e.g., Alston Reference Alston1991 and Bergmann Reference Bergmann2001). However, these analogies and reminders will (likely) not pressure the antiskeptical theist to adopt skepticism: those who do not already endorse skepticism are unlikely to be persuaded by these analogies and reminders. Perhaps due to the tendency of skeptical theists to stay at the intuitive level, direct challenges to skepticism are difficult to come by, with Benton, Hawthorne, and Isaacs (Reference Benton, Hawthorne, Isaacs and Kvanvig2016) being an exception (see Hendricks [Reference Hendricks2019] for a response). Below, I aim to remedy this situation: I will put forth an explicit argument for the truth of skepticism that shows that everyone ought to endorse it.
1.1 Clarifying Scope: What I Will and Will Not Argue
The topic of this article is concentrated exclusively on why we should endorse skepticism. There are many other issues that pertain to skeptical theism and the problem of evil that I will simply gloss over. This is in part because these issues have been addressed elsewhere and in part because addressing those issues would take this article far astray from its central thesis. In this section, I will briefly make explicit the scope of this article.
Since I have argued elsewhere (Hendricks Reference Hendricks2019) that skepticism is a plausible way to understand skeptical theism as it is advocated by Michael Bergmann (e.g., Bergmann Reference Bergmann2001), I will not rehearse those arguments here. Instead, I will take for granted that skepticism approximates to the skeptical theism Bergmann advocates. Additionally, it has been claimed that skepticism, as defined above, entails some sort of excessive skepticism: many argue that skeptical theism entails skepticism about morality, divine revelation, and even the external world (e.g. Richard Gale [Reference Gale and Howard-Snyder1996], Stephen Maitzen [Reference Maitzen, Dougherty and McBrayer2014], Erik Wielenberg [Reference Wielenberg2010], and Hud Hudson [Reference Hudson and Kvanvig2014b and Reference Hudson2017]), and many have responded to these skeptical worries (e.g. Bergmann [Reference Bergmann, Clark and Rea2012], Daniel Howard-Snyder [Reference Howard-Snyder and Kvanvig2009], Michael Rea [Reference Rea, McBrayer and Howard-Snyder2013], and Hendricks [Reference Hendricks2018, Reference Hendricks2020, and Reference Hendricksforthcoming]). I will not enter this dispute here, for it would take far too many words to do the problem justice. Instead, I will simply note that I think worries that skeptical theism entails excessive skepticism are overblown and that there have been plausible responses given to these worries (see references above). What I will do in this article is this: I will put forth an argument for the truth of skepticism: I will try to show that everyone, theist or not, should endorse skepticism.
2. The Preclusion Argument
In this section, I will introduce and defend three theses and show that they entail skepticism.
2.1 Connection
Consider the following thesis:
connection: if (i) S knows that some event E occurred, (ii) S has no good reason to think that there is no state of affairs x connected to E such that S knows nothing significant about x, then (iii) S has no good reason to think that there is no x connected to E such that the value of x is inscrutable to S.
The inference from (i) and (ii) to (iii) seems rather uncontroversial: if I know nothing significant about x, then the value of x will no doubt be inscrutable to me. Indeed, this seems to follow by definition. Let us say that a state of affairs is inscrutable just in case we know nothing significant about it. Rephrased, then, connection states that the value of an inscrutable state of affairs is inscrutable.
2.2 Preclusion
Now consider:
preclusion: if for some event E that S knows of, she has no good reason to think that there is no inscrutable state of affairs connected to E, then S has no good reason to think that the value of the states of affairs connected to E that she knows of are representative of the value of the states of affairs actually connected to E.
Preclusion, like connection, seems obviously true: if I have no good reason to think that there is no state of affairs x such that (a) x is inscrutable and (b) x is connected to E, then I have no good reason for thinking that all states of affairs connected to E fall within the same range of value. To illustrate this, suppose I am considering a set of states of affairs {w, x, y} connected to E, as detailed in figure 1. And suppose further that I have no good reason to think that there is no state of affairs, call it v*, connected to E such that the value of v* is inscrutable.
‘?’ = the value is inscrutable, and ‘*’ = I have no good reason to think it is not connected to E.
The fact that v* is an inscrutable state of affairs means that I have no good reason to think that its value falls in the range of {w, x, y}, and if I have no good reason for thinking its value falls in that range, then—because I have no good reason to think v* is not connected to E—I have no good reason to think {w, x, y} are representative, in respect to value, of the states of affairs actually connected to E. Indeed, if I did have good reason to think that the value of v* falls into the range of {w, x, y}, it would follow that it is not an inscrutable state of affairs: if I have a good reason to think that the value of v* falls within the range {w, x, y}, then I have at least some idea of its value (namely, that it is probably within the range of {w, x, y}), and therefore its value is not inscrutable. And hence the inscrutability of v* conjoined with the fact that I have no good reason to think that it is not connected to E entails that I have no good reason to think that the value range of {w, x, y} is representative of the states of affairs actually connected to E. Crucially, this means that I have no good reason for thinking that the total value of the set of states of affairs connected to E falls within any particular range, and this means that the total value of the set of states of affairs connected to E is inscrutable: because I have no good reason for thinking that v* is not connected to E and it is inscrutable—its value could be incredibly high, middling, or incredibly low—I have no good reason for thinking that the total value of the set of states of affairs connected to E falls within any particular range. For example, suppose v* actually obtains and is actually connected to E. Suppose further that the value of v* is incredibly high, say 1010,000. If that is the case, then, assuming the only other states of affairs connected to E are w, x, and y, the total value of the set of states of affairs produced by E (i.e., {v*, w, x, y}) is incredibly high and is not reflected by the value of {w, x, y}. This, again, is why skepticism is held to undermine arguments from evil (see section 1).
2.3 Event
Finally, consider the following thesis:
event: For any event (good, evil, ordinary) E, we have no good reason to think that there are no inscrutable states of affairs connected to E.
Event, I take it, is fairly obvious: while some event we know of may appear to have no connection to inscrutable states of affairs, this does not justify the conclusion that there are no such states of affairs connected to it: we have no good reason for thinking that some event we know of is not connected to an inscrutable state of affairs. This is because many connections are separated by long periods of time, making it more difficult to perceive them. For example, there is a long period of time between smoking (consistently) and getting cancer, which makes it more difficult to see the connection between these states of affairs.
Moreover, many connections have not yet occurred (such as the connections between current events and future states of affairs), and many current events are connected to states of affairs that have obtained in the past, making it all the more difficult to discern them. For example, World War II ended the way it did largely due to Winston Churchill's actions. That particular state of affairs—the end of World War II—was connected to an event that is easy to gloss over: the position that Lady Randolph Churchill slept in on the night Winston Churchill was conceived. Had Lady Randolph Churchill slept in any other position, the route of the spermatozoa would very likely have been altered, which would make it very likely that her son would have had different chromosomes, making it such that her son would have been very different from the Winston Churchill we came to know, and this would make it very likely that World War II would have ended differently. When the end of World War II was current, it would have no doubt been difficult to see how that state of affairs was connected to the earlier, seemingly insignificant event of Lady Randolph Churchill sleeping in such and such a position. (I borrow this example from Durston [Reference Durston2000: 66].) Clearly, then, it can be difficult to perceive connections of present events with future (or otherwise distant) states of affairs.
Furthermore, some connections are complex and therefore difficult for human beings to perceive. For example, consider again smoking: many years ago, it was not obvious to anyone that smoking was connected to getting cancer: the connection between smoking and getting cancer was complex and took some time to discern. There are no doubt other connections between states of affairs that are complex and difficult (or perhaps impossible) for human beings to discern. Because of the complexity of connections, the fact that we cannot perceive (many) future connections and the fact that it is difficult to discern connections between a current event and a past state of affairs connected to it, we have no good reason to think that there are no inscrutable states of affairs connected to any event. (Indeed, because we know that for any event we observe there are very likely distant states of affairs connected to it that we know nothing about, we have good reason to think that there are inscrutable states of affairs connected to it.)
The above illustrates that event, as defined above, is true if we are thinking about connections causally. But there is also good reason to think event is true if we are thinking about connections in terms of necessity. In brief, this is because human beings are just not the types of beings who, if there is a necessary connection between states of affairs and events, will know about it and recognize it as such. For example, suppose—contrary to fact—that the mind is identical with, or supervenes on, the brain. If this is the case, then there is a necessary connection between brain events and states of affairs involving mental activity. But, of course, many do not see this necessary connection. Indeed, many claim to see that there is no connection. And this means that our lack of knowledge or recognition of a necessary connection between states of affairs is not good reason to think that there is none. Since it is unclear what other (good) reason one could give for thinking there is no necessary connection between states of affairs, this strongly suggests that there is no good reason for thinking that there is no such connection. While there is no doubt more to say here, I think this point is clear enough—we have no good reason to think that there are no inscrutable states of affairs connected by necessity to any event E. And hence, event, on either the causal or necessary understanding of connection, is secured.
2.4 From Connection, Preclusion, and Event to Skepticism
Event and connection entail that for any event E we know of, we have no good reason for thinking that there are not inscrutable states of affairs connected to E. However, this means that preclusion holds for all E we know of, that is, for any E we have no good reason to think that the value of the states of affairs that we know are connected to E is representative. But, of course, this is just a generalized way of stating skepticism: ‘instances of evil’ referred to in skepticism are just particular events, and event and connection, therefore, entail that for any instance of evil (event), we have no good reason to think that there are not inscrutable states of affairs connected to it. And from this, preclusion entails that we have no good reason to think that the states of affairs that we know are connected to some instance of evil are representative, in respect to value, of the states of affairs that are actually connected to said instance of evil. And this means that we have no good reason to think that the value of the states of affairs we know are connected to E is representative of the value of the states of affairs actually connected to E. And hence event, connection, and preclusion entail skepticism, which, as we saw above, undermines arguments from evil. More formally, we may put this argument as follows:
The Preclusion Argument
(3) For any E we know of, we have no good reason for thinking that there are not inscrutable states of affairs connected to E. (From event and connection.)
(4) Therefore, for any E we know, we have no good reason to think that that the states of affairs we know are connected to E are representative in respect to value of the actual states of affairs connected to E. (From (3) and preclusion.)
(5) Therefore, skepticism. (From (4).)
To deny the preclusion argument, one must deny event, connection, or preclusion. But this is an unenvious position to be in because these theses seem clearly true.
3. Conclusion
In this article, I have used the preclusion argument to show that skepticism is true. Critics of skeptical theism, therefore, owe us an explanation for why they think the preclusion argument fails.
