Ontic and Epistemic Doubt B: Halakhic Aspects (Column 323)
In the previous column we discussed the conceptual and scientific distinction between epistemic doubt and ontic doubt. Despite the confusion that quantum description arouses in us, it turns out that Talmudic commentators employ precisely this distinction as well. True, they do so in the legal sphere rather than the physical one; yet, as we shall now see, it is very easy to understand and clarify this confusing distinction in the legal domain. In this column I will illustrate it in two halakhic contexts—“bereira” (retroactive clarification) and kiddushin that are not amenable to consummation. In the next column I will discuss the link between these two sugyot and show an interesting implication for the phenomenon of quantum collapse, as treated in the previous column, within halakha.
Epistemic Doubt in Halakha
Halakha often deals with situations of doubt. For example, a case in which two pieces of meat are before me—one pork and one kosher—and I must decide whether to eat one of them, and if so, which. This is a state of doubt, and we resolve it according to the rules for resolving doubts. Even if I find a piece of meat lying on the sidewalk in the city center and I am unsure whether it is kosher or not, we may use indications to help resolve the doubt. If a survey shows that most butcher shops in the city sell kosher meat, then strictly speaking it is permitted to eat the meat I found. This is an example of the rule for deciding doubts (following the majority).
The Mishnah (Yevamot 118b) presents another case of doubt:
“[If one] betrothed one of five women and does not know which one he betrothed, and each says: ‘He betrothed me’—[according to Rabbi Tarfon] he gives a bill of divorce to one and leaves [the marriage settlement] between them and withdraws; Rabbi Akiva says: This is not a way to remove oneself from transgression—he must give a bill of divorce and a ketubah to each and every one.”
“[Similarly, if one] robbed one of five [people] and does not know from whom he robbed, and each says: ‘He robbed me’—[according to Rabbi Tarfon] he may place the stolen item between them and withdraw; Rabbi Akiva says: This is not a way to remove oneself from transgression—he must pay each and every one.”
A similar situation can arise when a man betroths a particular woman but forgets which of the five it was. Or when a person sends a messenger to betroth a wife for his son from a particular town (for example, as Abraham’s servant Eliezer was sent to find a wife for Isaac in Haran), and the messenger betroths someone and dies, and the woman’s father who received the kiddushin also dies; neither the groom nor anyone else knows who the betrothed is. In such a case all the women are considered, out of doubt, to be his wives; he is forbidden to marry the sister of any of them, and all of them require a get in order to marry someone else. This is a state of doubt to which the regular rules of doubt apply.
“Kiddushin Not Amenable to Consummation”: A Dispute of the Amoraim
By contrast, the sugya (Kiddushin 51a–b) discusses “kiddushin that are not amenable to consummation” (kiddushin not amenable to consummation). Consider a case where a father of two daughters gives money to a man to betroth one of them to him without specifying which. Note the difference from the previous cases: here we are dealing with sisters, not just any two women. The problem that arises is that if Rachel is the one betrothed to me, then Leah is my wife’s sister and it is forbidden for me to have relations with her; and if Leah is the one betrothed to me, then Rachel is my wife’s sister and relations with her are forbidden. In short, after performing this foolish act, I find myself unable to cohabit with either of them, since each is potentially my wife’s sister. This is the situation termed in halakha “kiddushin not amenable to consummation.”
What is the status of such kiddushin? Abaye and Rava dispute this:
“It was stated: Kiddushin that are not amenable to consummation—Abaye said: they are kiddushin; Rava said: they are not kiddushin. Rava said (as explained by Bar Ahina): ‘When a man takes a wife and comes to her’ (Deut. 24:1)—kiddushin that are amenable to cohabitation are kiddushin; kiddushin that are not amenable to cohabitation are not kiddushin.”
According to Abaye, the kiddushin take effect, and both women require a get; the groom is prohibited to both because with respect to each there is a doubt of karet-level prohibition as the sister of his (possibly) married wife. Nevertheless, Abaye holds this does not prevent the kiddushin from taking effect. The status, for Abaye, resembles the earlier case of one who betroths one of five women and forgets which—there is a doubt, and we act stringently. Rava, however, rules that such betrothal does not take effect at all, because such kiddushin are not amenable to consummation. Both agree on the factual point that the kiddushin are not amenable to consummation; they disagree only whether this prevents the kiddushin from taking effect. Notably, this is one of the six Ya’al Kegam cases in which the halakha follows Abaye (the letter kuf represents this sugya).
What Are “Kiddushin Not Amenable to Consummation”?
Before continuing, let us clarify the concept itself. At first glance, it means kiddushin that cannot be realized through physical relations. This, in fact, is how Rashi seems to understand it, learning from the verse “When a man takes a wife and comes to her” that kiddushin must be fit for cohabitation. If so, it is not clear why Kiddushin do not take effect with those liable to karet (e.g., one who attempts to betroth his sister or mother): according to Abaye, couldn’t kiddushin still take effect even if cohabitation is forbidden? Indeed the very sugya indicates that if one betroths Rachel and then Leah, Leah is not betrothed because she is his wife’s sister—showing that kiddushin do not take effect where there is karet. The dispute concerns only the case where he betroths “one of the two” unspecified.
Tosafot (ad loc., s.v. “kiddushin she’einan mesurin lebi’ah”) present the opposite question on Rashi. With those liable to a mere negative prohibition (lav)—e.g., a priest and a divorcée—the Tannaim dispute, and the halakha rules that kiddushin do take effect. If “amenable to cohabitation” were a physical possibility requirement, why would kiddushin take effect there, since cohabitation is forbidden? Tosafot therefore explain that the problem is not the physical impossibility of relations but the legal result generated by the act of kiddushin: in the sisters case, the kiddushin themselves produce an interdiction that blocks cohabitation for both, turning an act whose function is to permit a woman to her husband into one that prohibits her to him. Where the prohibition pre-existed independent of the act (e.g., lavim), the kiddushin can still take effect. The Rashba (Kiddushin 51a) similarly writes that “kiddushin not amenable to consummation are not kiddushin—specifically where the kiddushin complete (generate) the prohibition; but for those subject to lavin and positive prohibitions, although not amenable to consummation, they are kiddushin.”
In other words, the issue is not that the kiddushin cannot in practice be realized, but that the act of kiddushin yielded an outcome opposite to its telos: instead of opening a path to marital relations, it closes it. Where the blockage is independent of the act, the kiddushin can take effect.
Is This Doubt Epistemic or Ontic?
At first glance, this appears to be an ordinary case of doubt: one of the two sisters is betrothed to him, we simply don’t know which, so we apply the rules of doubt. But this is incorrect. First, in the case of three sisters, why not apply majority to permit him at least two of them? Moreover, the situation can lead to contradictory permissiveness (like the “two paths” paradox): he could “choose” one and live with her, yet both were simultaneously under the same prohibition. This indicates that we are dealing not with epistemic doubt but with ontic doubt—an indeterminacy in reality itself—where the regular rules of doubt do not apply. We will elaborate shortly, but first let us challenge the assumption of ordinary (epistemic) doubt from another direction.
As is well known, the Rishonim disagree whether the rule “a biblical doubt is treated stringently” is itself biblical. The Rambam holds that biblically, a doubt is lenient and the stringency is rabbinic; the Rashba views the stringency as biblical. If so, according to the Rambam, why should our case be “kiddushin not amenable to consummation”? Each woman is but a doubtful sister of his wife, and biblically he could cohabit with each; only rabbinically would he be restrained. Yet even Abaye agrees that one may not cohabit with either, i.e., that the kiddushin are indeed “not amenable to consummation.” This suggests that even the Rambam would agree here that biblically he may not cohabit with either. Why? Because the “doubt” here is of a different nature; the usual rules of doubt do not apply.
R. Shimon Shkop: Sha’arei Yosher (Gate 3, ch. 22)
R. Shimon Shkop articulates this sharply:
In the case of one who betroths one of five women… there is a deep inquiry: in what manner does the status take effect upon one of them? If the betrother did not specify, how can we say that in true reality one particular woman was designated for kiddushin? By what reason would Heaven designate one woman more than the others?
Rather, it seems more correct to say that in such a case—even in true reality—no one woman was designated for kiddushin. In such a case we cannot say “if Elijah comes [he will tell us],” because there is no single, determinate ‘truth’ to reveal. The prohibition here is not like other doubts in the world, where the Torah of doubt is ‘perhaps this is the [prohibited] one in reality’; here the matter is that the status arises due to a cause that generates it.
If a man betroths one specific woman out of the five and then forgets which, that is ordinary (epistemic) doubt: there is a true fact which is unknown to us (God knows which). But when he never specifies, there is no “in heaven” determined wife. This is not a case where there is a single correct answer that we simply do not know; rather, there is no such single answer. The situation is one of indeterminacy in reality—ontic ambiguity.
R. Shkop continues:
The cause of the halakhic status is the act of kiddushin—the giving of money and the declaration. Since he gave money and said: “Let one of these five women be betrothed to me,” such that it is fitting that one of them be in the status of a married woman, therefore each one is prohibited due to the act, not because we are in doubt that she might be the betrothed one more than the others…
That is, the prohibition on each is not born of an evidentiary doubt but from the very act that created this indeterminate status. There is no specific woman to whom the status actually attached; nevertheless, each is prohibited to marry others (and if they are sisters, each is the man’s “attenuated wife,” making the others “sisters of his wife”). In yeshiva parlance this is termed “definite doubt” (definite doubt): we have certainty about the indeterminacy of reality, as opposed to ordinary doubt (doubt about something definite), where reality is determinate but unknown.
Hence, as R. Shkop notes, there is no concept of bitul berov (nullification by majority) here: nullification applies when a distinct minority element is mixed into a majority. Here all elements are identically situated; there is no minority to nullify. Consequently, the usual majority rules do not apply.
The Connection to Quantum Theory
As we noted in the previous column, quantum cases feature ambiguity (ontic uncertainty), not epistemic doubt. It is striking that R. Shkop, who wrote before quantum mechanics, nevertheless arrived at a similar distinction in the legal realm—perhaps a zeitgeist of sorts. In fact, the legal analogy helps strip some of the mystique from quantum theory.
The case of kiddushin not amenable to consummation is identical to the two-slit experiment described previously. There we said the particle passes “through both slits,” by which we meant that its state is a superposition of the two classical alternatives. We represented this schematically as:
|ψ⟩ = α·|A⟩ + β·|B⟩
For our purposes, |A⟩ is the state in which Rachel is my wife and Leah is my wife’s sister; |B⟩ is the state in which Leah is my wife and Rachel is my wife’s sister. The post-kiddushin situation is the superposition of these two classical marital-status states. It is not that I am married to both sisters (even “attenuatedly”); that is impossible. Rather, the overall state is a sum of two classical states, each of which is perfectly determinate by itself. It is therefore more accurate to say that, with probability weights given by α and β, I am in one state or the other—not that both women are simultaneously my wives.
As in quantum mechanics, the coefficients α and β do double duty: they describe how the overall state is composed, and they determine the probabilities of measurement outcomes (collapse). If we could “measure” which woman is actually the wife, those coefficients would fix the probability of each outcome. In our legal case they are equal, producing an even (symmetric) ontic ambiguity.
The Sugya of Bereira (“Retroactive Clarification”)
There is a family of Talmudic sugyot that pivot on bereira. A classic example (Gittin 25a; Hullin 14a) concerns separating teruma from wine purchased from the Cutheans. A person buys a barrel of 100 log of wine from which no tithes have been separated, wishes to drink now (e.g., on Shabbat, when separating is forbidden), and declares: “The two log that I will set aside are teruma; ten are ma’aser rishon; nine are ma’aser sheni”—and then he drinks immediately. The Tannaim dispute whether this is permitted: according to the view that “there is bereira,” the later designation clarifies the earlier status; according to the view that “there is no bereira,” this does not work. The halakha, for biblical law, follows that there is no bereira.
How is the status defined according to the view that there is no bereira? One might think the entire barrel remains untithed (tevel) until motza’ei Shabbat. Yet Rashi explains the prohibiting view differently: “He forbids [drinking], lest he have drunk teruma.” This implies that some halakhic “name” of teruma already rests, in a certain way, upon the wine now; we simply do not know which two log will eventually be designated. According to Rashi and those who follow him, the prohibition is not “pure tevel” but an ontic doubt—there is a thin layer of teruma-status spread over every potential two-log pairing in the mixture. This also explains why there is no nullification by majority: just as in the sisters case, there is no distinct minority element to be nullified; every pair is equally “eligible,” so usual majority logic does not apply.
Formally, the overall state can be described (for those who like symbols) as a superposition of many classical states:
|Ψ⟩ = Σi=1n αi · |i⟩
where each |i⟩ represents a specific pair of log designated as teruma and all the rest as chullin. In our case the coefficients are (conceptually) symmetric. The halakhic upshot is that, at least according to Rashi, the designation already “works” now in the sense that the name of teruma hovers over all potential pairs, though none is singled out. This is ontic doubt—an ambiguity in reality itself.
A practical consequence emerges in Gittin 25a: someone with two wives instructs a scribe to write a bill of divorce “for whichever wife exits the doorway first” tomorrow. Since a get must be written for the sake of the woman being divorced, the Gemara ties the validity of such a get to the dispute about bereira. Here, too, the parallel to the “one of two sisters” case is evident, and in the next column we will return to the topic of quantum collapse and its appearance in halakha.
Discussion
B. If that were correct, then even in a case where one betrothed one of two sisters it should not take effect for the same reason.
C. Indeed, that is the proof I brought from terumah. I did not understand your comment on my note 6.
So why, according to Rashi, is there bererah in a rabbinic doubt? After all, this is not a classic case of doubt where one can say: in a rabbinic doubt we rule leniently.
The connection between accepting bererah in rabbinic matters and the rule of a rabbinic doubt is itself disputed among the Rishonim. For example, some explained that the dispute whether there is or is not bererah was never resolved, and therefore in rabbinic matters we go leniently.
Many thanks to the Rabbi 🙂
I’m interested in asking: seemingly, in the case of “half slave and half free man” as well—are we also dealing here with an “ontic” doubt?
It seems to me that we are, because apparently he has both aspects together: both a full slave and a full free man.
Or perhaps this is an epistemic doubt? That in truth it is revealed before Heaven, only we are uncertain, etc.
This is certainly not an epistemic doubt, since we have no doubt whatsoever about him. He is half this and half that. This is no different from any jointly owned object that belongs to two owners. Simply, only half of him is owned by me. There is no problem here—neither epistemic nor ontic. There is no ambiguity here at all. He is both this and that. Here, for example, collapse is entirely irrelevant. No one can turn his half-slave status into a full slave or a fully free man (unless they emancipate him).
But from where does the Gemara know at all that there is such a thing as a law of doubt here? Why not say that this is simply not a form of kiddushin that was legislated, and so there is nothing here? I also asked this here: https://mikyab.net/%D7%A9%D7%95%D7%AA/%D7%A9%D7%90%D7%9C%D7%94-%D7%91%D7%A2%D7%A0%D7%99%D7%99%D7%9F-%D7%A7%D7%A0%D7%A1%D7%95%D7%AA. By the way, in the case of two firstborns that emerged together there is no collapse—so what would the Ritva answer from the next Tur?
And what I wrote there, “the prohibition,” should read “the punishment”—obviously.
I did not understand the question. A man performs kiddushin but directs it to one of two women. The natural result is that the kiddushin takes effect on both of them in a quantum way. You don’t need any source for that. On the contrary, Rava’s claim that it does not take effect is based on the reasoning that this is kiddushin not given over to intercourse, not on the fact that there is no clear addressee for the kiddushin. Therefore, if he were to effect kiddushin on one woman out of two who are not sisters, then even according to Rava there is kiddushin here despite there being no clear addressee for the kiddushin.
As for two firstborns, that is like terumah. Clearly there is no collapse, because he did not stipulate in advance, and therefore the collapse is only from that point onward. But with a firstborn, what matters is its status at the moment it is born. Therefore collapse is irrelevant there. Exactly like terumah, where you cannot drink the wine, or in kiddushin, where at the moment he betrothed them these were kiddushin not given over to intercourse.
But in today’s law, surely it is not like that (I assume). That is, there is an explicit use here of a metaphysical conception of legal effectuation, as also appears in that article on legal effectuation. But even if there are haluyot, ostensibly the ways they take effect could certainly be identical to a situation where there are no haluyot. Maybe the legal effectuation is classical?
On second thought, I accept it. This issue bothered me for a long time—wow, thank you.
By the way, I am not sure that in today’s law it is really like that. If I approach someone and buy from him one parcel of land out of his parcels (as in the case discussed in the responsum of the Rashba that I mentioned in the note), is it not correct that one of them is his by acquisition? And even if the court tells him to choose one now (a common-sense solution), one can still ask, from a legal standpoint, who was the owner from the moment of purchase until the moment of choice. For example, who pays the tax on each of the parcels until that moment? In short, I am not at all sure that you are legally correct. In my opinion there will be situations in which the law too will say the same thing (although of course where there is an obvious solution, like the Ritva’s collapse, clearly the law will adopt it, because there there is no metaphysics).
But if we are legislating whatever we want, I would say that until we reach a valid state, we remain in the previous valid state. Why complicate things? (And about this Scripture said: “It will raise the head of this judgment and the head of this judgment,” etc.) I don’t see any purpose this serves.
B, C understood.
I was mistaken regarding note 6.
We legislate the matter of acquisitions and kiddushin, but there are situations we did not think of. That is the result, and the judge has to decide what to do about it. The judge is not a legislator.
This is reminiscent of Rabbi Soloveitchik’s conception regarding the law of twilight: that it is not in the category of an ordinary doubt between day and night, but rather a separate state, like a superposition of day and night; he called this “an inherent doubt.”
Regarding note 3.
Seemingly, Rashash’s claim is that since we have identical entities here, it makes no sense to speak of a majority. But perhaps once the statistical coefficients differ, that itself defines them as different entities, and the option of nullification by majority would arise again, no?
That is not likely. The coefficients only represent the probability of obtaining each of the possibilities. They are not part of the definition of the possibility itself.
Beyond that, the basis for why there is no nullification here is that there is no mixture here. As I explained, this is not a mixture of women but a combination of states (each of which is a state of all the women). To define nullification between states rather than between objects sounds to me too speculative (though not impossible).
Clearly the coefficients represent only the probabilities. But usually (at least in a legal world) that would reflect the fact that there is also some difference between the entities or between the legal statuses that apply to them (otherwise why should there be a statistical difference). But since in any case this is a rather theoretical and not very distinct option, perhaps the discussion is not all that relevant.
Wonderfully elegant.
A. When the prohibition is unrelated to the kiddushin. According to this, perhaps one could discuss a case where he said to her: “So-and-so, acquire this courtyard from now on on condition that you never drink wine,” and her husband placed the get in that courtyard. Would that count as keritut or not? Perhaps according to Tosafot it would be keritut, and according to Rashi it would not. Though one could reject this, since the get is deficient insofar as its very validity is conditional, and not only its consequences.
B. As for the question why, according to Rashi and Abaye, kiddushin does not take effect in cases liable to karet: why can’t we answer along the same lines as the answer given for Rava, and say that kiddushin is defined within the framework of the law of interpersonal legal relations, whereas forbidden relations are outside the social domain and therefore it does not take effect—not because it is not given over to intercourse?
C. As for what you wrote in note 3, that this is so even when the weights are unequal: seemingly, isn’t the proof later on from terumah, which is not nullified in hullin, an explicit proof? And it seems to me that in these terms one could present the puzzle in note 6 as saying that if the practical weight is 0, then one can delete that term from the equation—and then one would have to say that even with weight 0, one does not delete it.
[The link has a typo. Instead of "from six" it should read "of six".]