For the most part, what follows can be read and understood independently. Nonetheless, in order to see the subject from a wider stance, I’d advise reading the first four parts. Let’s have a short summary of the series and see where we are now. We started with G. Frege’s distinction between sense and reference (part I), then continued with Bertrand Russell’s analysis of definite description (part II), P. F. Strawson’s pragmatic turn (part III) and finally, albeit slightly detached from the main bone of contention, which was the status of definite descriptions, S. Kripke’s account of proper names (part IV). This post will try to follow the “rise and fall” of another polemical subject which, from the moment of its first appearance in K. Donnellan’s article “Reference and Definite Description” (1966), quickly became a hotspot. To be sure, in the seventies and eighties, refuting Donnellan was many philosophers’ favourite pastime. It will be interesting to show how, eventually, the very movement that tried the most to turn away from his account, ended up saluting his trailblazing intuitions. I am hoping that from what follows the reader will extract what is needed to put together a feasible answer to The Whisky Puzzle. Every now and then I will refer to the two questions formulated there.
A word about the rather unusual citation format you’ll see. I will use the year of the first appearance at the beginning of each section. However, in most of the cases, I consulted a reprinted version of the articles edited in (Davis, 1991), where the articles were not ordered chronologically but alphabetically (no idea why). Therefore, the hyperlinked in-text citation leads to the original source, but the page numbers are from the aforementioned book.
This is a loooong post. I see no way I could have made it any smaller. Splitting things into separate posts didn’t feel like a better idea either. I suggest chewing it patiently with a big cup of tea on the side. If you have any questions, post a comment.
The startup: Donnellan’s distinction
There are in fact two functions of definite descriptions, two ways in which definite descriptions can be used, and both Russell and Strawson are guilty of not noticing this natural distinction. This was Donnellan’s main thesis (Donnellan, 1966). According to this thesis, when we say “The F is G”, the expression “the F”[1] can be used in two different ways: either attributively or referentially. When used attributively, a definite description states something (“G”) about whoever or whatever is the “F”. The word “attributive” is meant to mark that the speaker employs an attribute (an essence, Donnellan writes) to designate whoever successfully counts as having that property. If at the scene of a crime, we come across Smith’s battered body and remark “Smith’s murderer is insane”, we mean by “Smith’s murderer” not any particular individual, but whoever it was that murdered Smith. When used referentially, “the description enables his [the speaker’s] audience to pick out whom or what he is talking about and states something about that person or thing” (p. 54) To follow the classic example, suppose we are in a courtroom where Jones is on trial for the murder of Smith, and Jones is behaving strangely. We might say “Smith’s murderer is insane,” meaning to refer to that man over there behaving so strangely. So in this case, we are not any longer referring to whoever murdered Smith, but to a particular individual, a particular man we see in front of us. Notice right away the similarity with Mary’s utterance in the puzzle.
One might ask: aside from the intuitive specific/unspecific distinction, what is it so important about the two uses? The crucial difference, according to Donnellan, is that in the referential use, it doesn’t matter if the man we refer to by “The F” is in fact “F”. So the speaker can successfully refer to Jones in the courtroom case even if Jones didn’t kill Smith or even if, say, nobody fits “The F”, i.e. nobody murdered Smith (he actually committed suicide). In the crime scene case, however, this is not possible: since the speaker refers to “The F” only under that description, if nobody fits the “Smith’s murderer”, then the speaker cannot have said something true (if the description occurred in an assertion) or generally cannot have performed an act felicitously (e.g. a directive “Bring me the banana from the drawer”). As a general rule in Donnellan’s account, if two propositions differ according to the function of their descriptions (referential or attributive), then their truth-conditions will behave differently in case no entity fits the description employed[2].
Finally, Donnellan stresses (quite confusingly) that, although the two uses behave differently in the way explained above, it is not a semantic ambiguity. It was indeed startling that, after finishing a section where he explained a variation in truth-conditions (which is almost the definition of semantic ambiguity), we went on to say that it didn’t “seem at all attractive to suppose an ambiguity in the meaning of the words” (p. 59) What then is the source of the difference? To this question, Donnellan states (tentatively) that, since it is a difference in intentions, it must have something to do with the way language users make use of expressions, so he coined it “pragmatically ambiguous: the distinction between rules that the description plays is a function of the speaker’s intention” (p. 60). “Nevertheless,” he continues, “the burden of proof is surely on the other side”.
Methodological issues: Kripke’s intervention
In the paper “Speaker Reference and Semantic Reference” (Kripke, 1977), the aftermath of the Russell-Strawson-Donnellan dispute is carefully analyzed from a methodological viewpoint: “My concern,” Kripke notes at the start, “is not primarily with the question: is Donnellan right, or is Russell (or Strawson)? Rather, it is with the question: do the considerations in Donnellan’s paper refute Russell’s theory (or Strawson’s)?” (p. 77)
So, according to Russell, if there is no referent satisfying F in “The F”, then the sentence “The F is G” is false[3]. According to Strawson, if the same situation, the speaker has said (NB: the speaker, so we switch to truth-conditions for propositions instead of sentences) something which is neither true nor false. According to Donnellan, if “The F” is used referentially, then the utterance “The F is G” not only refers successfully to an individual x (which may or may not in fact be F), but the speaker can successfully predicate G of x. This is quite intuitive, Kripke seems to uphold: the classical analyses were assigning truth-value (and truth-value gaps) solely on the basis of the semantic meaning of “The F”, i.e. “on the basis of the [property of] someone else whom no one was talking about” (p. 78). Let me turn to our puzzle to exemplify this. For now, let us change “Smith” into “the man drinking whisky” so that the example is in line with what the theories are talking about. I will return to the original form eventually. Let us then have Mary say (1’) “The man drinking whisky is a journalist.” If the man whom Mary is referring to is actually drinking ice-tea, than little does it matter that he is a journalist: the empty description turns the sentence false (or renders a value-gap). Moreover, if there is a man drinking whisky (say, behind Mary, and Mary cannot see him) but he is not a journalist, then Russell would consider (1’) to be false regardless of the fact that Mary didn’t speak or try to speak about the man behind her (she couldn’t have, she did not know he was even there).
While Kripke readily acknowledges the distinctions above, he is anxious to point out a breach in Donnellan’s argument. Suppose we agree that the man behind Mary does not make (1’) false. In other words, that agreeing with Donnellan, we say that what matters indeed for the truth-conditions of referentially used definite descriptions cannot be some person whom nobody is speaking about. But is the opposite applicable? If the ice-tea-drinking man, to which Mary is referring – albeit using a misplaced definite description, is indeed a journalist, does this make (1’) true? Kripke notices that nowhere in Donnellan’s paper is a clear-cut affirmation of this, and thus, methodologically, we must ask if Russell’s theory is indeed refuted by Donnellan: if indeed there are different Russellian conditions for the referential uses.
In order to clarify some properties of the concepts involved, Kripke elaborates on two points. First he introduces the distinction between “semantic referent” and “speaker referent”, then he recasts everything so as to fit into the general tableau of Gricean pragmatics (see this figure). The semantic/speaker referent distinction – though not the last word in the matter, as we shall see – is an important one for the puzzle. The semantic referent is given by what the words, i.e. the designator, mean in a certain idiolect or language plus certain contextual determinations[4]. The speaker referent on the other hand, is determined by a special sort of intention plus some Gricean pragmatic principles that are general to communication. So let’s see how the distinction works (keep in mind the puzzle):
Suppose a speaker takes it that a certain object a fulfills the conditions for being the semantic referent of a designator, “d.” Then, wishing to say something about a he uses “d” to speak about a; say, he says “Ï•(d).” Then, he said, of a, on that occasion, that it Ï•’d in the appropriate Gricean sense (explicated above), he meant that a Ï•’d. This is true even if a is not really the semantic referent of “d.” If it is not then that a that a ’s is included in what he meant (on that occasion), but not in the meaning of his words (on that occasion). (p. 84)
For Kripke, the referential/attributive is simply a matter of the way these general intentions function with respect to a definite description. In the simple case, the speaker wishes to refer to the semantic referent (i.e. he uses “Smith” to refer to Smith – NB: not as Mary does!) In the complex case, “he has a specific intention, which is distinct from his general intention, but which he believes, as a matter of fact, to determine the same object as the one determined by his general intention.” (p. 85) (i.e. he wishes to refer to the man “over there” but believes that he is Smith – NB: as Mary does) Obviously, the “complex” case is Donnellan’s referential use, and the “simple” – Donnellan’s attributive. In the complex case, the speaker referent and the semantic referent may coincide, but they need not – as when we use “Smith’s murderer” to refer to that man over there.
Thus, Kripke concludes, a pragmatic analysis can be given – in terms of intentions – for what the speaker brings to the truth-conditions of his propositions using (differently) definite description. It is not, in any case, a different function that the description have, but a different usage they may fulfill. In this sense, Kripke showed that Donnellan’s account does not run counter to Russell’s. But is the separation of simple intentions from “complex” intentions accountable?
First pragmatic attempt: Searle’s aspects
Of all the things John Searle has done for the history of language understanding, his account of referential/attributive distinction might be the least successful (Searle, 1979a). To be clear: in spirit, the ideas and the directions are applaudable. It is, as the subtitle says, the first attempt to give a coherent account for the R/A distinction from a pragmatic standpoint – namely that of speech act theory. However, the actual account Searle arrives at is vague and almost pre-theoretic.
Searle starts by squaring the past discussion with the vocabulary of speech-act theory. He then undertakes to show that: “both alleged uses are referential in the sense that they are cases of referring to objects, the only difference is in the degree to which the speaker makes his intentions fully explicit in his utterance” (p. 121) Whatever the background, every theorist agrees that there is at least one property of every object we are capable of referring to. Obviously, it doesn’t have to be strictly speaking a property of that object but rather a linguistic representation the speaker has of that object. Searle tries not to step over any cognitive theory so he dubs this the “aspect” under which we refer to an object.
So Donnellan’s referential use of definite description is to be explained as a speaker “picking out one aspect” which he or she believes will enable the hearer to pick out the same object. This is not to say that if the object does not (“truthfully”) possess that aspect, the speaker would have said something false. That is because the “used aspect” – Searle calls it the secondary aspect – is only part of what is meant, and not part of what is said. “Just as in the indirect speech act cases […] in the referential use of definite descriptions one performs the act of referring to an object as satisfying the primary aspect by way of performing an act of reference expressing a secondary aspect” (p. 126). The primary aspect is part of the truth-conditions – and if nothing has that aspect the speaker didn’t refer successfully – while the secondary aspect is not (just as in the classic case where Jones need not be the murderer of Smith in order for the speaker to successfully refer to him in the courtroom).
Though very skeptical about Kripke’s own analysis (by that time, Searle was very much interested in the subject of intentionality – see Searle, 1983), Searle arrives at more or less the same methodological conclusions. Referring to the semantic vs. speaker reference explained as general vs. specific intention, Searle notes quite friendly that: “Consider the definite description ‘The man eating a ham sandwich on the top of the Empire State Building at 10 a.m., June 17, 1953’. Kripke tells us that in my idiolect the semantic referent of this designator is given by my general intention to refer to a certain object whenever the designator is used. I can only say that I never formed and do not have any such general intention, and I venture to guess that you haven’t either.” (p. 130)
As a consequence, although argued for from a different perspective, Searle arrives at the same conclusion which Kripke arrived at two years earlier: Donnellan’s account does not touch upon the Russell vs. Strawson analysis. The fact that one can mean more than one can say, that is, the fact that there are cases where the speaker expresses a secondary aspect and not the primary one is irrelevant because the phenomenon cannot be accounted for in terms of semantic analysis (be it Russellian or Strawsonian).
The reader will be right to point out that no such conclusion was necessary. The aspect-analysis, even if closer to our intuition than both Donnellan’s and Kripke’s version, is ad hoc and thus unable to properly explain the phenomenon. What Searle basically says is that, in the referential case, the “what is said” is different from “what is meant”. This did not need any further emphasis. In fact, it is any language user’s first intuition. About Mary’s utterance “Smith is drinking whisky tonight” we were hesitant to say candidly “True!” or “False!” particularly because we felt that she, in some way or another, meant something else (more, less, doesn’t matter) than she said.
Rescuing the Gricean account
In “Referential/Attributive” (Bach, 1981), Kent Bach offers an improvement of Searle’s ideas and I believe it’s safe to say that, within the Gricean conception of communication, this is more or less the final saying. As we shall see, this doesn’t mean that this is The Answer. We’ll see in the last section why Bach’s follow-up, along with the previous accounts from Searle and Kripke, should be understood as Gricean.
Although recast in a slightly different theoretical framework, Bach’s explanations are quite straightforward. Following Grice[5], Bach distinguishes between “making a statement” – which is reminiscent of Grice’s communicating – and “saying something” – which is exactly Grice’s saying. In speech act terms, the making of a statement is the illocutionary act – in which Bach stresses the presence of an attitude (belief, in case of a statement, desire in case of a directive etc.) – while saying something is the locutionary act. “A speaker might utter ‘Ostriches can’t fly,’ thereby saying that ostriches can’t fly. If he is expressing any attitude literally, it is the belief that ostriches can’t fly. He could be making a literal statement to that effect, but he might not be. Instead he might be making a statement nonliterally, perhaps expressing the belief that cowards make bad pilots” (p. 24, my emphasis)
The Searlean picture is now a bit clearer. When uttering “The F is G”, what is said is always that the F is G. If it is used attributively, what is said = what is stated. In the sense above, the belief, i.e. what is stated, is that what is said is the case, which in turn is, if Russell is correct, something like “there is an x which is uniquely F and x is G”. If “The F is G” is used referentially, what is said is still the Russellian statement, but what is stated (a “further statement”, in Bach’s terms, p. 25) is different. “If the speaker is using ‘the F’ referentially, he must be thinking of the individual he is talking about in some other way than as the F. Let us represent this other way as the designator ‘d’, which may but need not be a definite description” In other words, when used referentially, the speaker states some belief or other about d – in our case, the belief that it is G, although he is not saying that “d is G”. As we’ve seen above, what is said is always “The F is G”.
This, according to speech act theorists, was the source of so many confusions: it is intuitively odd that we can state some belief not by means of expressing it. In the Whisky Puzzle, that Mary can state something about the man in front of her, not by expressing the proposition/saying that “The man in front of me is such-and-such”, but by saying “Smith is such-and-such” This brings us to the last question.
What about “near misses”? What if, as it is in our puzzle, what is said is False (or fortuitously true of some other d’ fitting the description “The F”), but what is stated is true. Well, the answer should be clear: “The whole point of using ‘the F’ referentially is to get the hearer to think of the object one is talking about no matter how he does it. The hearer is supposed to exploit the speaker’s use of ‘the F’ to figure out which object is being talked about, and that object need not be the F. Referential success requires only that there be such an object d.” (p. 27) In other words, since the speaker is not expressing the belief that F is G but merely the belief of (or about) d that it is G – the fact that “The F is G” is true or false of someone else is irrelevant since it is not, to put it again in speech-act terms, the primary illocutionary act, not the stated belief (in Bach’s terms). This, as I’ve pointed out at the beginning, is more or less la chute du rideau of theorists working in the Gricean framework.
Post-Griceans: Carston’s explicature and Recanati’s contextual theory
In a nutshell, post-Gricean starts by seeing as problematic something which Gricean pragmatists – speech act theorists (Searle, Bach) or not (Kripke) – readily assumed: namely that “any pragmatically determined aspect of utterance interpretation, apart from disambiguation and reference assignment, is necessarily an implicature” (Carston, 1988, p. 39). The word “implicature” here is meant as to cover everything which is apart from “what is said” (Searle’s primary aspect, Bach’s stated belief, Kripke’s specific intention etc.) Let’s make some sense of why this is problematic by reviewing a few ideas from Robyn Carston’s “Implicature, Explicature and Truth-Theoretic Semantics” (Carston, 1988) and Francois Recanati’s “The Pragmatics of What is Said” (Recanati, 1989a). After that, we’ll check how Francois Recanati deals specifically with the referential/attributive distinction from a not-that-Gricean point of view in “Referential/Attributive: A Contextualist proposal” (Recanati, 1989b).
There is much to be said about the principles of neo-Gricean pragmatics, but I think for our present purposes we should stress this: the gap between the linguistic meaning (what the sentence determines by virtue of its syntax and conventional meaning of the words) and what is said is wider than it was previously believed (Recanati, 1989a, p. 98). In the past, Gricean pragmatics was ready to “let in just whatever is necessary” to bring the linguistic meaning to a propositional form – to a truth-evaluable proposition. So, for example, the sentence “I am in the city now” is not truth evaluable aside from contextual determinations such as “Who is speaking?”, “When did he say that?” “Which city does he refer to?” etc. So what is said – explicatures, what is literally expressed – need contextual information too[6]. In this case, “there is simply not a neat correlation between a semantics/pragmatics distinction and an explicating/implicating distinction” (Carston, p. 37)
Fig. 1 – Post-Gricean Pragmatics
You can imagine that the theoretical apparatus used to support this is a bit more complicated, but the picture itself is just that: when underdetermined (even if the semantic interpretation renders a complete propositional form), what is said is contextually influenced. We could say that Gricean pragmatists didn’t take to seriously Grice’s hedge when he said, in “Logic and Conversation” that “In the sense in which I am using the word say, I intend what someone has said to be closely related to the conventional meaning of the words (the sentence) he has uttered” (Grice, 1975, p. 44, my emphasis). OK, now let’s see how this relates to the referential/attributive distinction and finish this humongous post.
Differences aside, we have two theories (Recanati, 1989b). First, the Naive Theory (Donnellan) tells us that there are different truth-conditions so the difference between referential and attributive is semantic. In the attributive use, the proposition is true if and only if there is a unique F and it is G (the Russellian statement), in the referential use, the proposition is true if and only if there is an object which is G (the object being or not being F is irrelevant. Second, the Implicature Theory (Kripke, Bach, Searle etc.) says that the proposition is the same in each case – for e.g. Bach, what is expreesed is always that “The F is G” –, but the difference arises at the level of implicatures, that is, there is a difference in what is communicated. What is communicated is the same with what is said (see picture above) in the attributive use, but different in the referential use. So according to the Implicature Theory, the difference has to do not with what speakers say, but with what they mean: in referential case, that an object d is G, in attributive case that ‘there is an x which is uniquely F and F is G’.
As we have seen above, this alone resolves the puzzle (at least after a fashion). The ‘constituent’ of the truth-conditions is ‘d’ in one case ‘The F’ in the other. So, if the speaker mistakenly believs that ‘d’ is F and uses F to refer to d, then – to use Kripke’s terminology – even though the semantic referent is misplaced, the speaker’s referent will be ‘d’ and ‘d’ will be part of the truth-conditions.
This is only applicable in Gricean picture with two levels of meaning: what is said and what is implicated. Methodologists (as Recanati calls them) think that when a sentence like the “The F is G” is used to mean two different things there must be either: (a) a difference in what is said – so a semantic ambiguity – or (b) a difference in what is implicated. Grice’s Modified Occam’s Razor (“Do not multiply senses beyond necessity”) made them opt, in some way or the other, for the second explanation. In working with Grice’s principle, however, they implicitly stretched the Naive Theory so as to make it into (a) an Ambiguity Theory. Do you remember that Donnellan said quite precisely that he is not thinking of the difference as a semantic ambiguity? Could it be, then, that Donnellan’s account is in fact consistent with Grice’s Modified Occam’s Razor? Recanati believes so.
As I stressed in the beginning of this section, the Naive Theory holds that (a) truth-conditions are different and (b) what is said is different. It is only the Ambiguity theory that holds both (a) and (b) and (c) the sentence is semantically ambiguous. So the question Recanati sets about answering is this: can one hold (a) and (b) but not (c)? The Contextualist Theory tries to do just that.
So, going back to the post-Gricean picture in Fig. 1, we see three levels of meaning: Sentence meaning, What is said, and What is communicated. In order to hold a Contextualist theory, then, one must show that the difference is at the second level (what is said) and not at the first level (which would be an Ambiguity Theory) nor at the third level (which would be an Implicature Theory). To show that linguistically there’s nothing ambiguous about “The F is G”, we must simply go back to the Russellian analysis and see that the truth-conditions for the sentence (NB: not the truth-conditions of the proposition expressed by uttering the sentence in a specific context) are not ambiguous. “The F is G” can be taken to mean “There is an x which is uniquely F and F is G” or ∃x![F(x)→G(x)].
This out of the way, the Contextualist hypothesis says that “A sentence ‘The F is G’ is context-sensitive and expresses either a singular proposition [about d] or a general proposition [about “The F”] depending on the context of the utterance” (Recanati, 1989b, p. 231). The sentence, thus, is not ambiguous, but it can express different propositions with respect to different context. Going back at Fig. 1, the Contextualist theorist will simply say that the “pragmatic ingredients” of what is said differ.
So according to the Contextualist theory, the only element of generality in “The F is G” is that it can be used attributively while “[Proper name] is G” cannot. He dubs this “semantic potential”. The fact that “The F is G” can be used either way (referentially and attributively) does not mean that it cannot but be used in both ways. So our Mary, when she used “[Proper name] is G”, i.e. “Smith is drinking whisky”, happened to utter a singular proposition about the man in front of her, but she would have made the same mistake had she used a “The F is G” form in a referential way – i.e. “The man drinking whisky is Smith”. Let’s take Recanati’s example, maybe it’s clearer how “The F is G”, when used referentially, is similar to “[Proper name] is G”. If Giscard d’Estaing says “The present speaker is French” and “I am French” he utters the same singular sentence (namely, that “[Proper name] is G”, i.e. that Giscard d’Estaing is French) regardless of the fact that he used a “The F is G” in the first case and “[Proper name] is French”[7] in the second. The only difference between the two sentences is that “The F is G” can be used to express a general Russellian proposition. When it does not, we must attribute the “what is said” to the contextual ingredients
Short remark: notice that Recanati’s account is the only one which is indeed close to our (and your) initial intuitions. It is the only whan that recognizes that “there’s something fishy” about what is said, what is actually said, not that there are double meanings or implicatures. In oter words, Recanati’s account is the only one which allows the scenario in which if we go to our Mary and say: “Hey, look, you said something true of the man behind you, whom you didn’t even see”, the natural response would be “No, I have said something false about the man in front of me”!
[1] Note that “the F” is just a paraphrase. The actual expression occurring in natural language could be something like “his masterpiece” (“the masterpiece which is his”) or “John’s brother” (“the brother of John”). For now, there’s nothing wrong with assuming this as unproblematic. Refinements will be made when necessary.[2] “When a speaker says ‘The Ï• is ψ ,’ where ‘the Ï•’ is used attributively, if there is no Ï• we cannot correctly report the speaker as having said of this or that person or thing that it is ψ. But if the definite description is used referentially we can report the speaker as having attributed ψ to something” (p. 61) The contrast with Strawson’s approach is evident since Strawson posited that in case of denotation failure, the speaker would have said something “neither true nor false”.
[3] A more special situation occurs for “The F is not G”, but it is resolved in terms of scope ambiguity (see part II).
[4] If the designator contains indexicals, for instance, we will speak of semantic referent on that occasion or in that context. Among (or along with) this type of contextual determination, Kripke always mentions a “general (or “simple”) intention” to refer to what the designator refers (i.e. not to another special object by making use of that designator – case in which we are dealing with speaker referent).
[5] To be very specific, we would say that Bach is following the Searlean adjustments (in Searle, 1969) of Grice’s conception of meaning (Grice, 1957). Nonetheless the twofold division of “what is meant” vs. “what is said” is kept.
[6] Some might say: yes but this is just the semantic principle of assigning reference to variables (such as “I” or “now”) and can be given a semantic explanation. If this is meant as “the sentence determines how the variables should be semantically disambiguated”, it is perfectly true. Nonetheless, as both Carston and Recanati point out, the contextual determination is present even when the explicature is a propositional form (a fully-fledged, truth-evaluable proposition). If I say “I didn’t have breakfast”, I say “I didn’t have breakfast (this morning)” and implicate “I’m hungry” or “Give me food”. So the step from the linguistic meaning to what is said is not determined by the lack of propositional form – as it is in the case of indexicals – because “I didn’t have breakfast” is a complete proposition.
[7] Indexicals behave exactly like proper names in this respect and the fact that they have other linguistic properties in other cases is irrelevant here. Consider that he could have used “This guy here” (a demonstrative) or “G” (another proper name) to express a singular proposition with him as the logical subject.
What will follow might strike some as unchallenging. Those who are slightly acquainted with the beginnings of ordinary language philosophy might even notice the Austinian aroma of the whole perusal and treat everything as long since resolved. Yes, J. L. Austin too, in his William James lectures (Austin, 1962), and eveb some others before him (e.g. Gardiner, 1932; not to mention Arnauld & Nicole, 1683) tried to give an adequate answer to more or less the question (or type of question). Of course Austin came up with a terminology, made distinctions, proposed a theory, then another theory, then other proposed other theories and so on. 
