Ontic and Epistemic Doubt VI: A Methodological Loop (Column 327)

In this column, which concludes the series, we will discuss a methodological note that surfaces in several of the topics addressed in recent columns and offer a few examples. It concerns a loop-type argument, and by the end we will see that in some of its instances there is the difficulty I called in the previous column “the ontological fallacy.”

First Example: R. Shimon Shkop’s view about a doubtful pesik reisha from the case of date-palm brooms

R. Shimon Shkop in his Sha’arei Yosher, Gate 3, ch. 26, explains R. Akiva Eiger’s distinction as follows:

The reason, as it appears to me, is that R. Shimon’s concession in the case of a pesik reisha is not because wherever it is a pesik reisha it is deemed as intended on account of the fact that he certainly knows it will occur and therefore it is as if he intended it. For if so, where it is doubtful whether it is a pesik reisha, it ought to be permitted, since in such a case the matter is doubtful to the actor; yet in a doubtful pesik reisha the doubt concerns a Torah prohibition.

At the beginning of his remarks he presents two explanations for why pesik reisha obligates even when one lacks intent (aino mitkaven): (a) the pesik reisha renders him as one who intends; (b) the pesik reisha creates a linkage between the act and the actor even though he does not intend it. At the start of Column 325 I noted that among later authorities it is common to understand that R. Akiva Eiger’s view—that a doubtful pesik reisha is prohibited in an epistemic doubt (i.e., that the exemption of aino mitkaven applies only to an epistemic doubt)—depends on these two understandings. If the pesik reisha makes him “intending,” there seems no reason to distinguish between the two types of doubts; this is apparently the view of the Taz. R. Akiva Eiger assumes that a doubtful pesik reisha is prohibited because, in his opinion, pesik reisha obligates due to the second understanding: the linkage between acts is determined on the ontic plane rather than the epistemic. That is precisely what R. Shimon Shkop assumes.[1]

Immediately thereafter he brings a proof for the second position (R. Akiva Eiger’s approach) from the sugya of the date-palm brooms (Shabbat 95a):

See Shabbat 95a and Tosafot, where R. Eliezer and the Sages disagree regarding a broom—whether it is a pesik reisha. On the face of it, if their dispute concerns the condition [of inevitability], how can one say that the person performing the act knows with certainty that it will be so? And how could there be a dispute about something that depends upon the doer’s knowledge? Rather, certainly the matter of pesik reisha does not depend at all on human knowledge; rather, if in reality it is a pesik reisha, it is forbidden even if the actor does not know what the outcome of the action will be. For the main reason in pesik reisha is that in such a case the act is ascribed to that [outcome] even without the doer’s intention, and only where it is not a pesik reisha does R. Shimon hold that the act is called by that name only through intention: the act of plowing is called such only if it is done with the purpose of plowing; but when one drags a bench with the purpose of moving the bench, the act is named “relocating a bench from place to place.” But when one cuts off an animal’s head—even without the intention to kill—the act itself is called by its name.

In that sugya the Tannaim (R. Eliezer and the Sages) disagree about a case in which a person sweeps the ground with palm fronds and, in the process, leaves fall off. The question is whether he is exempt because he does not intend that (he intends to sweep, not to detach leaves), or whether it is a pesik reisha and therefore liable.

R. Shimon argues: if liability in a pesik reisha stems from the doer’s certain knowledge (and not from the necessity of the physical–ontic linkage), how can one Tanna forbid it on the grounds that detaching leaves is inevitable while another Tanna holds it will not occur? How is the sweeper supposed to know with certainty that leaves will detach if there is a Tanna who, in his assessment, says it will not happen? Of course, a person can assess that in such a situation leaves will detach, but we cannot treat that assessment as certain knowledge when there is a Tanna who assesses otherwise. There is “certain knowledge” only where it is impossible that any reasonable person would think otherwise. But here there is a Tanna—surely a reasonable person—who assesses that the leaves may not detach. From here R. Shimon Shkop proves that the distinction between pesik reisha and aino mitkaven lies on the ontic plane and not the epistemic (the person’s cognition). The dispute is whether, in reality, it is clear that leaves will detach; therefore, even if there is a dissenting view, this does not change the fact that, according to the other Tanna, in reality leaves will certainly detach (the other Tanna is, presumably, mistaken). Put differently: the claim that “every reasonable person knows” that leaves will necessarily detach is refuted by the mere existence of a dissenting (reasonable) view that holds they will not necessarily detach. Thus, R. Shimon Shkop proves that pesik reisha cannot mean epistemic certainty; rather, it concerns ontic certainty.

In this example we see that the very existence of view A refutes view B that opposes it. Note that this is not about the arguments of view A refuting view B, but that the mere existence of view A does so. Hence this is a methodological loop. Below we will see additional examples of such loop-type logic.

Second Example: A legal analogy

A similar logic may be seen in disagreements between judges. As is known, conviction in criminal law requires proof beyond a reasonable doubt. At the same time, the law itself contemplates the possibility of disagreement among the judges and rules according to the majority. Now consider the following: three judges sit on a murder case. Two hold that the defendant is the murderer; the third says he is not. In such a case, as noted, the law convicts according to the majority. Yet there is here a judge—unquestionably a reasonable person and trained jurist—who says the defendant did not murder. Can we say that, in the eyes of the other two judges, there is no trace of doubt? The absence of reasonable doubt means that no reasonable person would suspect that perhaps the defendant did not commit the murder. But before our eyes stands a trained, reasonable jurist who says the defendant did not murder. In such a situation we should acquit on account of doubt, even though the majority holds he is the murderer. For example, in the long Zadorov saga, this was indeed the situation and he was convicted by a majority. How can this be? Is it not correct to say that if there is a dissenting judge, then the very existence of such a view renders the situation a reasonable doubt regarding conviction?

It seems reasonable to distinguish between two different cases: (a) two judges say the defendant committed the murder and the third says he did not; (b) two judges say he committed the murder and the third says there is a reasonable doubt about his guilt. In the first case, conviction is difficult, for there is a judge who claims the defendant did not murder—surely a reasonable doubt. But in the second case, the dissenting judge does not claim the defendant did not murder; he only claims there remains a reasonable doubt (say, he is only 80% convinced). The dispute among the judges concerns the quality of the doubt (is it “reasonable” or not), not the factual reality itself. In such a case it is logical and proper to follow the majority and convict, for according to the majority the doubt here is minimal (not a “reasonable doubt”). This contrasts with case (a), where the dispute concerns reality itself rather than the quality of the doubt.

Indeed, Justice Danziger’s minority opinion in Zadorov was not that Zadorov did not murder, but that there was a reasonable doubt regarding conviction. In that situation one follows the majority. It is worth seeing descriptions of the debate between Justices Segal and Danziger in the Zadorov case here and here (though in my view the formulations there are imprecise, as they do not distinguish between the two decision types I set out; their essential dispute is as I described).

Rejecting R. Shimon’s proof

In this thread, which dealt with the concern of eating worms, I suggested a way to dismiss R. Shimon Shkop’s proof from the dispute about the date-palm brooms. I proposed that the Tanna who holds the case is not a pesik reisha may understand that not every sweep stroke necessarily detaches a leaf. The certainty that a leaf will detach exists only if we consider sweeping the entire house (at some point during the process a leaf will detach). In his view such a case is not a pesik reisha. By contrast, the other Tanna holds that even if, in the overall sweeping, it is clear a leaf will detach, that suffices to consider the case a pesik reisha.[2]

Accordingly, R. Shimon Shkop’s proof falls away, for on this proposal the Tannaitic dispute is halakhic (whether such a case is deemed a pesik reisha) rather than factual (whether the leaf must detach). On the factual plane both agree that for a single sweep there is no certainty of detachment, whereas for sweeping the whole house there is such certainty. The debate is a halakhic one: whether such a situation should be called a pesik reisha and obligate. This is not merely a hypothetical way to read the dispute; it seems, on the face of it, to be a more reasonable account of that Tannaitic disagreement. I do not belong to those who deny the existence of factual disputes; they are certainly possible and, in my opinion, do occur. But specifically here it is hard to accept that the Tannaim are simply disputing a matter of fact, for the factual question—whether sweeping detaches a leaf from the palm frond—depends on circumstances such as the nature of the ground, dryness of the leaves, the force of sweeping, and more. What sense is there in fixing universal, general rules here?! It is therefore more plausible that the dispute is halakhic rather than factual, and then it is indeed a principled and general disagreement.

I note that the thread there also raised the question of which actions can be considered parts of a single act for this purpose. For example, regarding eating a whole apple with a concern for worms inside, when one eats it in several bites: if in each bite it is not clear there will be a worm, but in the apple as a whole there are worms, the question is whether eating the apple counts as one act—in which case one could say there is a pesik reisha—or not.

Let me preface that at the end of this column we will see that R. Shimon’s proof is problematic on a principled level (it suffers from what I called in the previous column “the ontological fallacy”). Before that, however, I will bring additional examples of such a methodological loop (where the very existence of one view refutes the other). The next example returns us to the question of quantum collapse in halakha discussed in Column 324.

Third Example: Ritva vs. Tosafot Rid

In Column 324 I cited a dispute among the Rishonim about “quantum collapse” in halakha. The Ritva assumes that if a person imposes an indeterminate status (as in cases of bererah), he may, when he wishes, choose one of the two possibilities intermingled within it, and the law will follow that option. For example, a person who betroths one of two sisters, or divorces one of two wives, can at some point decide that it is this one rather than that one. Tosafot Rid disagrees and holds that one can choose between two possibilities only where one stipulated in advance that the law would follow the person’s later choice (according to the view that “yesh bererah,” when one betroths “that sister whom I will choose” at some later time).

There I already suggested that perhaps the Ritva is not essentially disagreeing with Tosafot Rid; rather, in his view, when a person imposes an indeterminate status he implicitly stipulates that he will have the option to choose one of the two options.[3] For example, when one betroths one of two sisters, his meaning is effectively: “I am betrothing that one whom I will later choose.” According to this suggestion, the Ritva too does not accept quantum collapse in halakha; rather, he holds that in all situations of indeterminate statuses there is an implicit stipulation. Tosafot Rid disputes him only on this technical point (in his view there is no implicit stipulation unless it is articulated). From this suggestion it follows that, for both, collapse is possible only when stipulated in advance (explicitly or implicitly).

In that column and in the fourth book of my series on Talmudic logic, I raised various reasons why this is not a plausible reading of the Ritva. First, it does not sound that way from his words. Beyond that, if that were indeed his intent, it is hard to see why he challenges the sugyot of bererah. He challenges, on sevara, why there is no bererah, i.e., why the person cannot choose. But if the challenge is based on an implicit stipulation, why not assume the Gemara holds there is no implicit stipulation and thus one cannot choose? The Ritva would then be challenging the Gemara on the basis of a rather speculative assumption. It is more reasonable that he views quantum collapse as a necessary consequence of indeterminate statuses and thus challenges the Gemara. Indeed, I cited the Avnei Miluim, who certainly did not understand the Ritva in that way (to the contrary, according to him there is clear proof that no stipulation is required to enable choice).

Parenthetical note: A dispute among Tosafists

It appears the Tosafists themselves disagreed on this point. In Yoma 55b a case of bererah without prior stipulation is presented (the Avnei Miluim I cited notes this), concerning two women who set aside two pairs of birds in partnership. Another case appears in Bekhorot 57a and Temurah 30a regarding partners who divided: one took ten lambs and the other nine and a dog; it turns out that one of the first partner’s lambs is the price of a dog and is forbidden for sacrifice.

In Tosafot s.v. “levror,” Bekhorot 56b, they write about the “price-of-a-dog” case:

“To choose one of them opposite the dog …” One may ask: what does this help, since it is unknown which is the one opposite the dog? And if you say because they are all equal in value, but if they are all equal then how can he select the one opposite the dog to permit the others? One must say that this is the Talmud’s logic: since there is bererah, such a matter can depend upon his will, for presumably this was his intention from the outset—that whichever he would later choose would be the one opposite the dog.

Here the rationale of the person’s presumed intent is explicit. That is, the fact that there is “collapse” in a case of an indeterminate status is a function of the presumptive intent of the one who imposed the status. When he imposed a status on ten lambs, it is clear his intention was to reserve for himself the option to select one of them. The assumption is: there is no bererah without prior stipulation; therefore, in cases where we observe such a phenomenon, there must have been an implicit stipulation.

By contrast, in Tosafot s.v. “ve’idach,” Temurah 30a (also cited by the Avnei Miluim), they ask:

It is astonishing: here it appears that if a prohibition is mixed with permitted items, one may take the prohibited piece and rely on bererah to say “this is the prohibited piece that has been removed;” therefore, in the case of “one who buys wine from the Samaritans” (Demai 7:4), “the two log I will later separate shall be tithe,” he relies on bererah and says “this is the ḥullin he is drinking.” Likewise, regarding the three women who bought bird-offerings jointly and one died—the Gemara simply asks in Yoma 55b: “let us take four zuz and throw them [i.e., remove the problematic portion], and the rest be permitted” on account of bererah. If so, in every case of prohibition mixed with permitted pieces, let us rely on bererah: remove the measure of the prohibition and the rest be permitted. Likewise, with sacrificial animals that became mixed with an ox to be stoned (Zevahim 80a) the ruling is ‘they must die’—why? Though living beings are significant and not nullified, still let us take one of them and say the rest are permitted.

Tosafot ask: in every stew where a forbidden piece falls among permitted pieces and is not nullified by the majority, let us remove one piece from the pot and it will retroactively be clarified that this was the forbidden one. The question presumes, from the Gemara, that there is bererah even when not expressly stipulated. These Tosafot plainly assume that clarification is a feature of indeterminate status as such, independent of the presumed will of the one who created it. Note that here we speak of a mixture produced by a forbidden item falling into permitted ones—this is not a person’s act that could be pre-stipulated to enable later selection—yet Tosafot still ask that bererah should help. That is, they do not view this as an implicit-stipulation mechanism but as a principle: an indeterminate status, by its nature, ought to be susceptible to later human clarification.

The conclusion is that even if the Ritva clearly appears—as the Avnei Miluim writes—to adopt quantum collapse in halakha (and not implicit stipulation), among Tosafot we find an approach that explicitly ties the matter to stipulation. I will now try to show that the approach that hangs it on implicit stipulation suffers from the same difficulty we saw above in the palm-broom sugya.

A logical difficulty

The position that depends the person’s later choice on an implicit stipulation assumes that when a person imposes an indeterminate status he necessarily stipulates to retain the option to choose. We noted (in a footnote) that without this one could not bypass the laws of stipulations (that for a stipulation to be valid it must be spoken, in doubled form, affirmative before negative, stipulation before act, etc.). On that basis it is very hard to accept an approach that sees such a case as an implicit stipulation. For those Rishonim who disagree certainly hold that even when a person imposes an indeterminate status he does not stipulate to retain a later choice. If so, how can one say that every reasonable person so stipulates? Those Rishonim would not stipulate, and they do not think others stipulate—and those Rishonim are certainly reasonable people. It follows clearly that it is not true that “every reasonable person” stipulates.

Again we see that the very existence of the dissenting opinion refutes the other view. If there are Rishonim who hold there is no implicit stipulation in such cases, the other Rishonim cannot claim that every reasonable person intends to stipulate implicitly. The very existence of Tosafot Rid and Tosafot in Temurah proves that Tosafot in Bekhorot are mistaken.

Seemingly, this difficulty compels the conclusion that this is not the Ritva’s intent. He does not mean that the matter rests on an implicit stipulation (and thus he would be disagreeing with Tosafot Rid only over an assessment of human intent); rather, in his view there truly is quantum collapse in halakha. That is, whenever one imposes an indeterminate status, by virtue of its very indeterminacy there is an option to choose one possibility even without an implicit stipulation.

Yet, as we saw, Tosafot in Bekhorot certainly did not hold so. It may be that they maintain that indeed everyone intends to stipulate, and those Rishonim who disagree and hold there is no implicit stipulation do not deny that this is what people intend; rather, such intent is ineffectual to create a valid stipulation or enable bererah. On that reading, again, the dispute is halakhic rather than substantive—which is of course possible—just as we saw in the previous example.

Fourth Example: Renewing semikhah

By way of background, recall that only sages who themselves hold semikhah can confer semikhah on others (see Rambam, Sanhedrin ch. 4). This means semikhah occurs only “from above,” i.e., from God to Moses, from him to Joshua, and onward from the ordained to the next. As is known, at some point semikhah ceased (there is wide historical debate on the exact date), and since then we have had no judges empowered to rule on many areas of halakha. The Rambam, however, innovated—on sevara—that it is possible to renew semikhah “from below.” Thus he writes in his Commentary to the Mishnah, Sanhedrin 1:3 (and likewise, more hesitantly, in Hilkhot Sanhedrin ch. 4):

It appears to me that if there were agreement among all the sages and disciples to appoint one person as head over them, provided that this be in Eretz Yisrael as we said, then that person’s “academy” would be established for him and he would be ordained, and he could thereafter ordain whomever he wishes. For if you do not say so, it would be impossible ever to find a Great Court, since each of its members would in any case need to be ordained; and the Holy One, blessed be He, promised that “I will restore your judges as at first, and your counselors as at the beginning” (Isaiah 1:26). This will undoubtedly occur when the Blessed Creator directs the hearts of people and their merit and desire for God and for Torah increase, and their wisdom grows, before the coming of the Messiah, as is clarified in many verses.

He argues that the agreement of all sages in Eretz Yisrael can renew semikhah, and his proof is from the restoration of semikhah in messianic times: if semikhah cannot be renewed “from below,” how will “I restore your judges as at first” be fulfilled?!

On this basis, in the 16th century there arose a major dispute about renewing semikhah in Eretz Yisrael. The debate is recorded chiefly at the end of Responsa of Maharlbach, where his “Kuntres ha-Semikhah” appears. The sages of Safed, led by R. Yaakov Beirav, supported renewal and indeed convened and ordained R. Beirav, who then ordained five of his disciples. This chain continued for a few generations and then waned.[4] Opposite them stood the sages of Jerusalem, led by Maharlbach (R. Levi ibn Ḥabib), who opposed the renewal.

Maharlbach advanced many arguments against the sages of Safed, but one of them is pertinent here. He contended, among other things, that the mere fact that he and his colleagues did not agree to the renewal nullified the validity of the semikhah performed in Safed. Not because of the arguments he raised (for which he opposed the act), but because he and his colleagues did not agree. The Rambam’s condition was the agreement of all the sages of Eretz Yisrael; once there are dissenters, there is no general agreement, and the act lacks validity. This is again a similar methodological loop: the very existence of view A refutes the validity of view B.[5]

What did the sages of Safed answer? I do not recall having seen a direct reply (the documentation from both sides is in Maharlbach’s “Kuntres ha-Semikhah”). But perhaps they advanced the following claim: Maharlbach did not raise objections to the fitness of R. Beirav himself; he merely rejected the halakhic position that semikhah can be renewed. If so, this is simply a halakhic dispute, and in that we follow the majority. In a halakhic dispute there is no need for the agreement of all the sages of Eretz Yisrael; agreement is required only regarding the person being ordained, and on that there was no dispute. Therefore, once it is decided halakhically that renewal is possible (for the sages of Safed were the majority and also saw themselves as greater), there is no obstacle to ordaining R. Beirav. Note that here too the structure of dismissing the loop-type argument is similar: if the root of the dispute is not factual but halakhic, the methodological loop disappears.

A note from the ontological fallacy

We have seen several examples of what I called a “methodological loop,” i.e., cases where the mere existence of one view undermines the validity of the other. All these arguments are based on a sharp distinction between epistemology and ontology. They negate the possibility that we are dealing with epistemology (something pertaining to human cognition—because cognitive certainty is inconsistent with the existence of a dispute) and deduce from this that we must be dealing with ontology (reality itself).

Yet in the previous column we defined a third category of doubt: pseudo-ontological doubt. These are doubts that are, in truth, epistemic (cognitive), but are perceived by people as ambiguous states (ontic), and therefore halakha (and in some cases science) treats them as ontic doubts. I therefore called them pseudo-ontic doubts. I noted there what I called “the ontological fallacy”: taking this methodological assumption and treating it as a claim about reality. People who are unaware that viewing the doubt as ontic ambiguity is only a methodological convention, and who see it as a statement about reality itself (as if reality itself were ambiguous), commit the ontological fallacy.

Let us return now to aino mitkaven. Regarding our bench that someone drags and thereby makes a furrow in the ground, we saw that halakha treats this situation as ontic ambiguity rather than an epistemic doubt, but in truth this is pseudo-ontology. The real doubt is epistemic, not ontic (for each type of ground deterministically dictates the outcome). Now you can see that R. Shimon Shkop’s argument from the existence of a dissenting view collapses at once. The question is not whether on such ground a furrow will be made, but what a reasonable person thinks about the nature of the ground (for in pseudo-ontic doubts, halakha treats the reasonable person’s judgment as if it were reality itself). It is true that the reasonable person perceives this as ontic ambiguity; yet in essence it is epistemology.

The same holds for sweeping with date-palm fronds. There, too, the matter is pseudo-ontic ambiguity, not real ontology. If so, the existence of a Tanna who holds that leaves will necessarily fall off refutes even the pseudo-ontic option, not only the epistemic one. For even if one frames the doubt as an ontic dispute (as R. Shimon explained R. Akiva Eiger), in essence it is an epistemic dispute. Therefore, the claim that “on such ground a furrow will certainly be made” really means “a reasonable person thinks it will be made” (not necessarily that in fact it will be made). But if so, then the dissenting Tanna also undermines R. Shimon Shkop’s reading of R. Akiva Eiger (that this is an ontic doubt), not only the Taz. In this sense there is no advantage of pseudo-ontology over epistemology.

Of course, the examples that are genuinely ontic (not pseudo-ontic)—such as the dispute between Ritva and Tosafot Rid regarding “one of two sisters betrothed” or “one of ten lambs”—remain as they are. There the methodological loop truly proves that we are dealing with an ontic doubt rather than an epistemic one (for there it is genuinely ontology and not merely pseudo-ontology), and the arguments built on that loop remain valid.

In short: defining certain doubts as pseudo-ontic ambiguity is indeed useful for discussion in some contexts, but it is still important to remember that this is a methodological definition, not a claim about reality itself. Methodologically, it is very helpful to distinguish it from epistemic doubt, but factually it is wholly epistemic. Bottom line: R. Akiva Eiger distinguishes between two kinds of epistemology—epistemic doubt and pseudo-ontic doubt—not between epistemic doubt and ontic doubt. This distinction is made only for practical purposes; therefore, one cannot deploy, in such cases, the logic of loop-type arguments presented here. Applying the methodological loop to the broom case (to prove like R. Akiva Eiger against the Taz) is an example of what I called in the previous column “the ontological fallacy.”

[1] In my view this linkage is not necessary, as I explained to the questioner here. One must add that the matter also depends on whether the “intention” in question is will or knowledge. Plainly, the intention that obligates is will for the result; if so, I do not see why a pesik reisha indicates will. Some therefore interpreted that the intention in question is knowledge rather than will, and that in a pesik reisha there is knowledge that the prohibited result will occur. I noted at the start of Column 325 that several later authorities proved that both principles exist (according to R. Ḥayyim, for example, one is intention in Hilkhot Shabbat and one is intention in the rest of Torah law). If the “intention” is will, it is clear that liability in pesik reisha must be explained as the outcome being linked to the person, not that he is deemed intending.

[2] Simply put, the dispute concerns only a case where one swept the entire area of the house and a leaf detached. The question is whether, since in each small sweep there is no certainty of detachment, even sweeping the whole area counts as a pesik reisha or not. But if one swept only a small area and a leaf detached, then obviously this is not a pesik reisha.

[3] As is known, in the laws of stipulations several Rishonim write that in cases where the performer’s intention is completely clear (“matters in his heart and in the heart of every person”), there is no need to articulate the stipulation (see e.g. Tosafot s.v. “Devarim shebalev,” Kiddushin 49b). In such cases we presume an implicit stipulation even if it was not spoken. This is what jurists call an “implicit condition,” and that is the suggestion I raised above regarding the Ritva.

[4] It has already been noted that R. Yosef Karo, author of the Shulḥan Arukh, who himself was ordained by his teacher R. Beirav, does not mention in his works that there are ordained judges who can adjudicate.

[5] Here, it does not truly refute view B itself, but rather the validity of the act performed on its basis.