Reconsidering a Source Theory of Reference-Fixing

Originally written for PHIL 1860 Sense and Reference with Prof. Richard Kimberly Heck.

In Naming and Necessity, Saul Kripke presents the following picture of the use of names: first, there is an “initial baptism” (or dubbing) in which an object receives a name, and subsequently the name is “passed from link to link” in a “chain of communication” wherein future speakers acquire a name from its present users (Kripke 96). On this picture, then, when I discuss the sinking of the Titanic, the name “Titanic” refers to the ship by virtue of a causal chain extending from the day champagne was broken over its bow, through various speakers, writers, and whatever history book I learned about the disaster from as a child, to me. Kripke acknowledges that there are exceptions to this picture, such as the case in which someone has no intention to use a name in the same way that he learned it, such as the case of someone naming their pet aardvark Napoleon (Kripke 96). Moreover, he stresses that this is just a picture—not a full-fledged theory—of reference, that there may well be many places where tightening up is required to account for every case, and that he is “sort of too lazy at the moment” to do so (Kripke 93).

A significant such problematic case is that of unintentional reference shifting, famously encapsulated by Gareth Evans with the example of Madagascar, whose reference shifted from the traditional African meaning of the area around Mogadishu on the mainland to the modern meaning of the island off the southwest coast. Despite Kripke’s insistence to the contrary, Evans dubs the above picture the “Causal Theory” and takes the Madagascar example and reference shifting as a phenomenon to be “decisive” against it (“Causal” 195). In his analysis of the same phenomenon, John P. Burgess disagrees. Here, I will attempt to get as clear as possible on the moral of such examples, mostly agreeing with Evans. I will also offer some discussion and criticism of Evans’ alternative picture.

I think it is important to start by sketching that view, so that it is clear what turns on the explanation of the examples. Evans agrees with the “Causal Theory” that the fundamental nature of names is causal. However, he disagrees that the original source of the causation is the dubbing and that the proximate source is the acquisition of the name from another speaker. Rather, on his view, the object itself is the cause, by virtue of being dominantly responsible for the speaker’s body of information (“Causal” 199). In other words, when a speaker claims that “Napoleon lost at Waterloo” he refers to Napoleon because his internal concept of the person of Napoleon (his mental image of a conquering general, perhaps picturing David’s painting of him crossing the Alps) was in fact created because, originally, of Napoleon’s deeds. In this way, Evans returns from Kripke’s view in which the relata are dubbing and use, to the descriptivist view in which the relata are description and object. The crucial difference between the old descriptivism and Evans’ view is that the relation is causal, following Kripke, rather than one of fit. That is, the speaker refers to Napoleon not by virtue of Napoleon’s fitting his mental description, but rather by virtue of Napoleon being the cause of that description, or rather, his being the “source” of that description, a status that is causal but will require some further elucidation. For this reason, I will subsequently refer to Evans’ view as the “Source Theory.” To be clear, then, this example is not meant to decide between a causal and non-causal framework, and it is certainly not meant “to support the description theory” despite the claim in (Kripke 163).

Now, let us turn to the case of reference shifting. The historical details of the Madagascar case are given with more care and precision by Burgess. On his analysis of the history, Marco Polo used the name “Madagascar” to refer to the same region as the indigenous people, i.e. the environs of Mogadishu. However, he mistakenly believed this region to be an island. A few hundred years later, European cartographers who were aware of the fact that there is only one island off the west coast of Africa believed Polo to be referring to that island by “Madagascar” and the reference was thereby shifted. Thus, this example constitutes a case in which each link in the chain intended to use the name in the same way as their forebears, but the European cartographers failed to do so (which means this case is not explicable in the same manner as the case of Napoleon the aardvark).

Burgess thinks the structure of such a case is as follows: some speaker uses a name to refer to an object, but another speaker mistakes this use and thinks it is meant to refer to some other object (with which the second speaker is already acquainted). Then, the second speaker acquires the name as a name for the second object (Burgess 198). The reason why this occurrence is not more common, says Burgess, is that there is usually a corrective step in which the newly acquired name is challenged by more interaction with the dominant use, and so the speaker learns of her error. The specific feature of the Madagascar case that allows for the successful shift in reference (and makes the example “recherché”) is that there was no such intermingling between traditional African and new European uses of the name, and thus no opportunity for correction of the mistake (Burgess 199).

This seems like a successful analysis of the phenomenon, but Burgess’ conclusion that because of it the example is not fatal for Kripke’s picture is somewhat more mysterious. In particular, it is somewhat unclear how exactly Burgess (or Kripke) would incorporate this analysis of the phenomenon into the Causal Theory. Although Burgess does not say so explicitly, it seems that in such an integration, the step in which the name is acquired as a name for a new object must be a new, separate dubbing that itself is the start of the subsequent causal chain. It cannot instead be a different kind of link in the original causal chain, because reference transfer is simply unsuccessful, and so my use today of “Mogadishu” to refer to the island cannot causally stretch back any further than the cartographers. At the very least, the conditions Burgess provides for reference shifting, and the account of how the reference transfer step might fail and instead become the site of the shift, give no reason to think that such a shift can be incorporated into the old communicative chain rather than starting one anew via another dubbing. Although the second dubbing explanation does seem to account for the case, it muddies the waters considerably when it comes to understanding what a dubbing actually is. It demonstrates that a dubbing might be unintentional, and furthermore that it might be the case that no one on Earth is aware that it took place.

This analysis, wherein the mistake constitutes a second dubbing, seems very strange when applied to a case without some of the particular structural features of the “Madagascar” case, such as the “switched-at-birth” case that Evans presents at the same time (“Causal” 196). Although Burgess admits that such other cases might require “different explanations” (Burgess 200) it seems difficult to get around the fact that a second dubbing will have to occur at some point in the explanation of any reference shifting case (or at least, Burgess’ analysis similarly gives us no reason to believe the contrary.) Let us consider, then, this second case of Evans in which two babies are born, named by their respective mothers, and then switched in the nursery accidentally. Evans concludes then that “the man universally known as ‘Jack’ is so called because a woman dubbed some other baby with the name” (“Causal” 196). Here, there are not even two speakers as in the Madagascar case, and there is not a clear point of reference transfer failure that can be pointed out as the second dubbing. The problem is clear: when was the second baby dubbed “Jack”?

The intensity of this problem increases when we consider an extension of the case: suppose the mistake is subsequently discovered and the babies are returned to their original mothers. At what point does “Jack’ remain a name for the second baby after the switch back? Most would agree that if the mistake were discovered after many years, the names would not switch back with the babies, but they would if the mistake were discovered before the families were discharged before the families left the hospital (perhaps after a few days of calling the second baby “Jack”). The exact line is not clear, nor does it really matter (it probably also depends heavily on the specifics of the situation) but the ambiguity itself represents a serious problem for the Causal Theory. If the name “Jack” became a name for the second baby through a second dubbing (even one that was inadvertent and unbeknownst to all) then how can there be such a vagueness as to when the name sticks and when it doesn’t? On the Causal Theory, the names switch back if the switch happens before the second dubbing, and they don’t if it happens after. Thus, ambiguity as to when the name sticks or not is exactly confusion as to whether a second dubbing took place. This is problematic for two reasons. First, the reply that it is unclear, or up for debate, or ambiguous whether the names stick within a particular fully fleshed out version of the case is not available to the Causal Theory. Either a dubbing took place, or it didn’t. It is simply incoherent, on this Kripkean interpretation, to say that the names are in any sort of middle ground state. Secondly, the act of dubbing is moving from mysterious to dubious. Elaborating a more specific theory of dubbings on the original version of Kripke’s account was acceptably elided because of various in-the-weeds details about the success of such a speech act, and so forth, but we more or less “know it when we see it.” Here, however, we find ourselves with no such theory in hand and a case in which there theoretically must have been a dubbing, but there are no salient candidates to be found.

Burgess’ analysis does correctly clarify that posing the challenge of reference shifting by pairing a case like Madagascar with a case like Kripke’s original Gödel-Schmidt case (Kripke 84) and challenging the Causal Theory to explain why reference shifting occurs in one but not the other is confused. Burgess cites multiple examples of philosophers framing the issue this way (Burgess 196), including both works contemporary to his own paper, and Evans’ original introduction of the Madagascar case (“Causal” 196). With Burgess’ clarification of the failed reference transfer step in mind, it is hard to see a reason why one would expect a reference shift in the Gödel-Schmidt case, except in a revised version like the one Burgess himself gives (Burgess 199). The problem, instead, is that there is little way to account for the necessary second dubbings that reference shift will in general require. In fact, my view on the problem of reference shift for Kripke is not that it is a “demonstration by case” in the sense that it is a case for which a Kripkean theory straightforwardly predicts an answer that conflicts with common usage and intuitions, but rather that it applies immense theoretical strain. Whereas before analyzing such examples it was reasonable to think that a follower of Kripke might be able to produce an adequate theory of dubbing if pressed, afterwards it is clear that such a task is impossible. Nothing will fill the role that the Causal Theory requires of dubbings, if it is to be able to explain reference shift.

Let us now consider, by contrast, how Evans might account for reference shift. In the case of Madagascar, the explanation for why modern use of the name refers to the island while that of Africans 700 years ago referred to the area around Mogadishu is straightforwardly that the island is the dominant source of our information about “Madagascar” whereas for them it was the mainland. Continuing to bracket for now specifications of the theoretical terms “dominant” and “source”, this explanation makes it clear that this example presents no mystery for the Source Theory. An account of how such a difference in dominant sources, i.e. why the reference shift happened, should probably look exactly like that of Burgess. As far as I can tell, nothing in Burgess’ description of how Mogadishu came to be acquired as a name for the island, either historical or theoretical, is incompatible with Source Theory, and I think Evans would do well to agree with Burgess’ analysis of the conditions necessary for reference shift to occur, and his explanation of its rarity.

The switched-at-birth case provides a deeper elucidation of the structure of the Source Theory. Unlike the Causal Theory, Evans does have a story to tell about why there might be ambiguity about whether the given names stick in the case two babies switched in the nursery and then switched back a month later. To decide whether “Jack” refers to the biological child now returned or to his counterpart, the theory asks which is the dominant source of the family’s information. Although they have been with the latter for a month and the former only a few hours, it seems as though “this is, in fact, my child” is such a crucial part of the mental picture that it is reasonable to provide either answer to such a question. Thus, it is reasonable to expect some disagreement and ambiguity about which boy is “Jack.” Likewise, if the child has grown to be 15 years old, or if the switch back happened in the hospital, the answer to the dominance question is obvious and agrees with the obvious common intuition.

Although this alleviates the problems that the Causal Theory had with this example, it raises an obvious question of its own for the Source Theory: what are we even talking about when we ask which one is “dominant”? The notions of “source” and “dominance” lend the theory some flexibility, but are also somewhat of a weakness. On the dominance count, Evans notes that being the dominant source of information is “not simply a function of amount of information” and that “detail in a particular area can be outweighed by spread” and that “reasons for being interested in the particular item at all will weigh” (“Causal” 201). This weighing approach rather than straightforward criteria is present in the above discussion of the switched-at-birth case, and indeed the cases that Evans gives to illustrate the idea of dominance are very similar in structure. For example, he notes that if Napoleon had been replaced by an impostor, it would matter considerably at what point during Napoleon’s career the replacement had taken place. Determining to which person “Napoleon” refers would involve a weighing of the general’s pre-replacement and post-replacement careers and their effects on the popular and individual conceptions of “Napoleon” to determine dominance (“Causal” 202).

To be fair to Evans, he acknowledges similarly to Kripke that his goal is not a comprehensive theory but rather a general picture (“Causal” 198). We have seen, however, that a good way to press on such a picture is to exhibit the strain that its unexplained theoretical terms are under. The central notion of “source” to Evans’ picture seems to be under some such strain.

Evans defines the source relation as a specific sort of causal relation, with some specification along the lines of epistemological reliabilism. That is, something is the source of some information if it causes a person to have that information in a way that is generally reliable for gathering information. This causal relationship between the object of thought and the person can generate some immediate worry. For example, what is the source of our information about the number 8? If it is the number 8 itself, then that seems to include a relatively serious metaphysical claim about numbers, imbuing them with causal powers and so forth. Alternatively, the Source Theory could appeal to a treatment like the one Evans gives for something like “the tallest man in the world.” If I deduce that the tallest man in the world is over 6 feet tall, then no one is the source of that information, and such cases can be dealt with via a pre-Kripkean Description Theory (“Causal” 200.) It also seems unpalatable, though, to claim too much in common between the description “the tallest man in the world” and the numeral 8.

Other cases build similar tension. What about fictional characters, is the source of the information the character itself? Or even more borderline, what about figures like Uncle Sam or the case already in the literature of Santa Claus? Not only does it seem somewhat problematic to say that fictional characters can have causal powers over humans (of course, it is obvious that works of fiction have such powers, but that is not the required claim here) but a dominance analysis of the meatpacker Samuel Wilson versus the legendary Uncle Sam seems to be on even shakier footing that the typical analysis.

Some further complication also occurs in the discussion of “deferential use.” For example, Evans discusses the case of a village in which a man (A) nicknamed “Turnip” had left long ago, and a new stranger (B) enters the village and is misidentified by the older generation of villagers (the only ones who knew A) as Turnip. Then, Evans thinks that depending on what information is passed down, there is significant potential for “Turnip” to become a name of B as the older villagers die, since for the younger villagers B will be the dominant source of information about Turnip (Evans 207). However, he makes an exception for “deferential use” in which the younger villagers are explicitly following the older villagers’ usage. It is not clear, however, how the details of such a case might be filled in. Suppose that the village has very strict norms about respect for the elderly, and disagreement with them is very unusual and frowned upon. In such a case, there might be some hesitation after the revelation of the separateness of A and B, out of fear of disrespecting the dead. Would this be enough for deference, or would the younger villagers actually have to think of themselves explicitly as deferring in use? Furthermore, given the squishiness of dominance, it seems plausible that this difficulty and the theoretical ugliness of adding exceptions for deference could be solved by restructuring dominance such that A is the dominant source for the younger generation when they are in fact engaging in a deferential practice regardless of the nature of their interactions with B, but perhaps such an approach would strain the notion of dominance too much.

These concerns are by no means fatal for the Source Theory, but they make more pressing the need for further elaboration. Evans presents a possibly alternative, possibly complementary theory of proper names in “Varieties of Reference” involving a core group of “producers” of a name and a periphery of “consumers” of the name, akin somewhat to the older and younger generations of villagers in the “Turnip” case (“Varieties” 377). That account may be just such an attempt to elaborate on (and revise) the Source Theory to alleviate these concerns, but I cannot analyze this later theory here.

In sum, I think I have shown that the problem of reference shifting is at least nearly as serious for a Kripkean Causal Theory as Evans thought. However, accounting for Burgess’ analysis shows that the pressure is more theoretical than empirical. Evans’ Source Theory handles such examples much better, but there are also other theoretical pressure points that leave it unsatisfactory as the end of the story for reference-fixing.

Works Cited

Bonevac, Daniel (2017). “Gareth Evans on the Causal Theory.” https://www.youtube.com/watch?v=fOK-pPxW7rg

Burgess, J. P. (2014). “Madagascar Revisited.” Analysis 74(2): 195–201.

Evans, Gareth (1973). “The Causal Theory of Names.” Aristotelian Society Supplementary Volume 47(1): 187–208.

Evans, Gareth (1982). The Varieties of Reference. Oxford: Oxford University Press.

Kripke, Saul A. (1980). Naming and Necessity: Lectures Given to the Princeton University Philosophy Colloquium. Cambridge, MA: Harvard University Press.