Philosophical Essays against Open Theism – ch. 4: Alexander

By | April 11, 2024

Philosophical Essays Against Open Theism (Routledge Studies in the Philosophy of Religion)This is part four of eleven in a series responding to the essays in Ben Arbour’s edited volume, Philosophical Essays against Open Theism (Routledge, 2019).

In this post I tackle chapter 4 by David Alexander, “Open Theism and Origins Essentialism: A New Argument Against Open Theism” (pp. 71–79).

In this essay, Alexander argues for a thesis that he calls generalized origin essentialism (GOE) and then uses that thesis along with what he (following Jonathan Kvanvig) calls the asymmetry thesis (AT) to generate an objection against all “standard” versions of open theism.

I will begin by presenting Alexander’s argument and then examine his key premises, particularly GEO. I will argue that Alexander’s argument fails in at least three major ways.

The GOE argument against open theism

Alexander kindly formulates his argument in numbered statement form (p. 76). I’ve slightly modified it here for clarity’s sake, e.g., by specifying that it is token events and token identity that Alexander has in view and by making the inferential links and assumptions more explicit than he does. I have also inserted line 7 because it is a key inferential step that Alexander inadvertently omits. These modifications are intended to steelman the argument.

  1. There is some [token] future event F determined by the past and present, and God knows F. (from AT)
  2. For any [token] event E, the causal ancestry of E is essential to the [token] identity of E. (from GOE)
  3. If something X is essential to the [token] identity of something else Y, then Y implies X. (assumption)
  4. Hence, for any [token] event E, the [token] identity of E implies the causal ancestry of E. (from 2, 3)
  5. For any event E, if God knows E, then God knows whatever E implies. (assumption)
  6. Hence, God knows whatever F implies. (from 1, 5)
  7. Hence, God knows the causal ancestry of F. (from 4, 6)
  8. There are future contingent events in the causal ancestry of F. (assumption)
  9. Hence, God knows some future contingent events. (from 7, 8)

I now begin my analysis of the GOE argument. As should be evident from my reconstruction, the argument is valid (i.e., the logic is perfectly good) but there are five independent assumptions (lines 1, 2, 3, 5, and 8) that need to be examined.

Line (1) and the Asymmetry Thesis

What Alexander calls the “Asymmetry Thesis” (AT) derives from the work of Jonathan Kvanvig, whom Alexander quotes as follows (p. 71):

[T]he Asymmetry Thesis [is] the thesis that the part of the future that is determined by present and past events is secure in truth value and falls within the scope of omniscience whereas the parts of the future that remain undetermined by the present and past do not fall within the scope of omniscience and perhaps are not secure in truth value.

Kvanvig believes that open theists are committed to AT. Alexander concurs. I disagree. While there are some open theists who could accept AT as it stands, not all open theists would. concur as well, with the caveats that not every aspect of AT would be endorsed by every open theist. Many open theists, myself included, would emphatically reject the idea that the undetermined parts of the future fall outside the scope of God’s omniscience. We would say, instead, that God perfectly knows the future as settled to the extent that it is settled and as open to the extent that it is open. Setting aside the problematic aspects of AT as Kvanvig states it, the crucial idea for the sake of Alexander’s argument is that the future is alethically open in part because it is causally open in part and alethically settled in part because it is causally settled in part. As Alexander glosses it (p. 71),

AT = “Portions of the future are determined by the past and present, and portions of the future are not so determined.”

Stated in this way, I think all open theists could accept AT.

Now, Alexander claims that line (1) derives from AT. He also makes clear (on p. 72) that it is token future events that he has in mind. But does line (1) follow from AT? No! In the first place, AT says nothing about individual or token events. It talks instead about parts, portions, or aspects of the future. In the second place, the determined portions of the future that AT references cannot plausibly be understood by open theists as event-tokens as opposed to event-types. And, finally, in the third place, it can be argued that there are no future event tokens at all.

So, let’s talk about the token/type distinction. Tokens are concrete individuals. As such, they have to be fully specified. Types, in contrast, are abstract kinds. They need only be partly specified. For example, an event-type, like a total solar eclipse, is conceived of it as a certain kind of event, namely, one in which the sun casts a complete lunar shadow upon the Earth. So conceived, event-types abstract from lots of details, such as precisely where and when the event-type occurs (if it ever does) and which concrete individuals it involves. In contrast, an event-token, like the total solar eclipse that just occurred on April 8, 2024 is a fully concrete event that includes the actual sun, moon, and Earth throughout the span of the event. It also includes all of the specific places on Earth that were in the path of totality and all of the people, animals, and objects that participated in the eclipse in one way or another.

Now, when open theists say that the future is partly settled or partly determined, they mean (nearly always) that it is certain types of events that are settled or determined. Christian open theists believe, for example, that Christ will one day physically return. We think it is fully settled, because God is fully committed to ensuring that it happens, that Christ returns. But we don’t think all of the details of that event are settled, such as precisely when it occurs, who is alive at the time, etc. Indeed, it is precisely because open theists believe the future is significantly (albeit only partly) open, that they believe many of the details of future events remain open. So, for nearly any conceivable future event F that is determined by the past and present, open theists are going to think of these as event-types, not as event-tokens. What God knows in knowing that F will occur is that an F-type event will occur. For this reason alone, Alexander’s argument fails, and fails at the very first step. Even if open theists grant AT, they should reject the idea that line (1) follows from AT. Alexander illicitly makes line (1) seem more acceptable to open theists than it should by suppressing the “token” qualifier, which I inserted in square brackets.

Moreover, though somewhat more controversially, an open theist can follow Arthur Prior in his 1968 essay “Identifiable Individuals” and argue that there simply are no tokens of not-yet-existent individuals. For example, I do not as yet have any grandkids. Perhaps I never will. Who knows? In any case, it is not clear how I could possibly token (individually refer to) any of my future grandkids. They simply aren’t around for me to point to one and say “Look at that cute kid. His name is [name]. He was born on [date] in [place] and weighed [pounds].” The bracketed placeholders make clear that many of the details are not specified. Indeed, they aren’t even specifiable. If this line of thinking is right, then, since (given ontological presentism) future event-tokens are not-yet-existent individuals, it follows that there are no future event-tokens, period.

Line (2) and Generalized Origin Essentialism

What Alexander calls “Generalized Origin Essentialism” (GOE) is a thesis that he derived from the work of Robert Koons, whom he quotes as follows (p. 71):

Generalized Origin Essentialism (GOE): For a class of entities, if E is an entity from that class, and if C is causally upstream of E, then C’s existence or occurrence was logically necessary for the existence or occurrence of E.

Alexander takes GOE to mean that “token events logically imply whatever is in their token causal history” [emphasis mine—note the underlined words], or, more pithily, that “the token present necessitates the token past” (p. 72).

At the outset it should be noted that GOE is a generalization of origin essentialism (OE) over an entity’s entire causal history. OE is widely regarded by philosophers as a plausible thesis with respect to certain classes of entities, particularly organisms. For example, it seems plausible that you, the particular (token) organism that you are, has an identity that is tied to having the specific biological parents that you did. If you had had other biological parents, then perhaps you would not and could not have been numerically identical to the you that you in fact are. More generally, even if we reject the idea of personal numerical identity over time (as I think we should), we can still argue plausibly that you-now are what you are because of your recent causal history and so, in some sense, you carry that history with you. But those reflections point merely to OE. OE only says that entities of certain kinds carry (entail) part of their causal history. GOE goes much further by holding that entities of those kinds carry (entail) their entire causal history, stretching back, perhaps, all the way to the beginning of time.

Now, regarding GOE and its role in Alexander’s argument, the first question to ask is whether GOE implies line (2). Not obviously, because Koons’s statement of GOE doesn’t specify which classes of entities GOE applies to. Whether GOE applies to event-tokens is not stated explicitly by Koons. But let’s set this issue aside. Koons’s reference to an “occurrence” of E may be taken to imply that this is an event. Moreover, his reference to “entities” may be taken to imply that these are tokens.

The second question to ask is whether Alexander’s gloss on GOE (“the token present necessitates the token past”) is equivalent to GOE. Here I can confidently say, “No.” Alexander conflates generalized origin essentialism (GOE) with what we might call holistic origin essentialism (HOE). There are many reasons for affirming HOE. For example, those who hold (as most open theists do) to a presentist ontology, need a way to ground truths about the actual past. If (per presentism) only present reality exists, then whatever grounds truths about the past must be included within the totality of present reality. Thus, the token present necessitates the token past. That’s HOE, not GOE. It’s “holistic” because it says that the totality of present reality carries its actual past (its history or origin) along with it. I’ve argued in print that God’s memories are the most plausible repository for the memory of the past. The key point to notice, however, is that HOE does not say, like GOE, that every token event carries its entire causal past along with it. It only says that the all-encompassing present event carries its past.

Alexander presents (pp. 73–76) many reasons why we, and open theists in particular, should accept GOE. But many of the reasons he gives, such as the just-mentioned need to ground the past given presentism (pp. 73–74), the need to “explain the asymmetry between the past and the future” (p. 73), and discussions of whether God must create the best possible world (p. 74), are reasons for HOE, not GOE. Other reasons that Alexander offers for GOE, such as some reflections from William Hasker concerning what’s required for personal identity over time (pp. 74–75) and the time traveling “grandfather paradox” (p. 75) are merely reasons for OE, not GOE. By conflating these three principles (OE, HOE, and GOE), Alexander makes GOE seem a lot more plausible than it is.

Once OE, HOE, and GOE are distinguished, however, it should be obvious that GOE is much less plausible than either OE or HOE. It’s one thing to say (with HOE) that the totality must carry its past with it, it’s another thing to say (with GOE) that every token event must do so. And it’s one thing to say (with OE) that some entities carry part of their causal history than to say (with GOE) that they carry all of their causal history.

Consider, for example, the famous Ship of Theseus puzzle. The ship has its individual planks replaced one-by-one until the whole ship is replaced. The replaced planks are them assembled into a new ship alongside the now fully-refurbished Ship of Theseus. The question is then posed: Which ship, if any, is identical to the original Ship of Theseus? The one that preserves the original materials in the same arrangement, or the one that preserves greater continuity with the original? My inclination regarding this puzzle is to say “neither” because complex composites like ships just aren’t the right kind of thing to have strict token identity conditions over time. If we ask about the identity conditions of the ship-at-a-particular-instant, then it seems right to say that it has identity conditions, namely, having the exact parts that it has at that instant. But it is simply false, I maintain, that the ship maintains token identity over a span of time during which the constitution of the ship has been intrinsically changing. Likewise, I say, non-totality events, which can span a long period of time, may lack token identity conditions, especially while the event is still ongoing. During WWII, for example, was there a fully specified fact of the matter as to how it was going to play out? That’s very doubtful, unless some sort of global determinism is true. Indeed, it’s only in retrospect that we can even define (by stipulation) a start and end point for that event. In other words, until the event is over-and-done with, it seems to have the character of an event-type, something open to further specification, rather than an already fully specified event-token. Furthermore, given the above-noted concerns about not-yet-existent individuals, it is arguable that future events aren’t in principle tokenable at all. Even if we can somehow token future events, however, why think that every event must carry its entire causal history along with it? That seems incredibly implausible. My left ankle bears scars from a recent fracture and surgery, but it surely doesn’t carry a complete memory of everything that’s ever happened to it (every footstep I’ve ever taken!). If there’s a complete memory of that event, it lies in my mind, or much better (because my memory is incomplete and imperfect) in God’s mind, not in the ankle itself or the physical stuff involved in the fracture event.

In sum, while I believe there are plausible arguments for OE, HOE, I see no compelling reason to accept GOE, particularly when applied to future event tokens.

Lines (3) and (5)

Moving on, I have no objections to lines (3) and (5). These are both highly general and abstract claims that look like truisms. Line (3) seems to follow from the notions of essentiality and identity. Line (5) seems to follow from the nature of divine omniscience.

Line (8), the Causal Ancestry of Undetermined Future Events

The final assumption of the GOE argument, line (8), says that there are future contingent events in the causal ancestry of F. Recall, from (1), that F is a token future event that is already causally determined by the past and present. So, to evaluate (8), we need to ask ourselves how an event can be both determined and still have future contingent events in its ancestry. I see only one way: It is determined that F comes to pass, but it is not yet determined in what way F comes to pass. We might suppose, for example, that it is determined that I travel to work tomorrow, but that it is not yet determined precisely how I get there. I have my usual route that I take to work, but let’s suppose it’s a future contingent whether a major car accident occurs along that route such that, should it occur, I would not be able to get to work in that way. For simplicity, let’s suppose that there are no other relevant considerations concerning how I get to work. In that case, it’s a future contingent whether I get to work by my usual route or whether I get there by some other route, but it’s also determined that I do get to work one way or another. The question is whether it is coherent with this scenario to suppose (a) that my traveling to work tomorrow is a token future event and (b) that GOE applies to that event. I will now argue that neither (a) nor (b) is coherent. If an event is determined to occur, but how it comes about is not yet determined, then we’re dealing with an event-type, not an event-token. Furthermore, even if we were dealing with an event-token, GOE could not apply to it. In short, line (8) of Alexander’s argument is inconsistent with lines (1) and (2).

In the first place, then, if my going to work tomorrow is determined but how it comes to pass is not yet determined, then the nature of the event remains somewhat generic. We can think of three events here: that I go to work tomorrow (generic), that I go to work tomorrow by route A (one specification), and that I go to work tomorrow by route B, where B≠A (another specification). What’s determined is the generic description, but not either of the more specific descriptions. Since what’s determined is a generic description that is open to further specification, we’re dealing with an event-type, not an event-token. Hence, if event F is as line (8) describes, then line (1) is false. F is not a token future event.

In the second place, if my going to work tomorrow is determined but how it comes to pass is not yet determined, then, given line (2), the fact that I go to work tomorrow must include a specification of how I get to work, since that is part of its causal ancestry. But it obviously doesn’t include any such specification. Because it remains an open question how I get to work, the identity of that event (my going to work tomorrow) cannot include a specification of how I get there. Otherwise, its being determined that the event occurs would entail its also being determined how it occurs. So, if event F is at line (8) describes, then line (2) is false. F’s identity does not include its entire causal history.


Alexander’s GOE argument against open theism utterly fails. It has three independent assumptions—lines (1), (2), and (8)—that are either easily deniable by an open theist (lines 1 and 2), or that are incompatible with other assumptions in the argument (line 8). As noted, Alexander (inadvertently, I presume) masks the implausibility of his formal argument on p. 76 by suppressing the “token” specification in lines (1)–(4) and by conflating GOE with OE and HOE.

Leave a Reply

Your email address will not be published. Required fields are marked *