<%BANNER%>

Grounding Explanations

Permanent Link: http://ufdc.ufl.edu/UFE0022645/00001

Material Information

Title: Grounding Explanations The Role of 'in Virtue of' in Philosophical Inquiry
Physical Description: 1 online resource (122 p.)
Language: english
Creator: Butchard, William
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Philosophy -- Dissertations, Academic -- UF
Genre: Philosophy thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In this study, I discuss a type of explanation that shows up in a variety of contexts, including philosophy and ordinary discourse. Consider the question, 'What makes a person tall?' Here are two possible answers one could offer: Having the right genes makes a person tall. Having a height that is significantly greater than average makes a person tall. The first explanation is causal. It says that a tall person's genetic structure is what causes her to grow up to be tall. But the second claim explains a person's being tall in terms of what it is to be tall in the first place. Clearly, the later would be a non-causal explanation. In this work, I discuss what bearing an understanding of such 'grounding' explanations has on our understanding of philosophical analysis. I argue that a clear understanding of grounding explanations points the way to a solution to the so-called paradox of analysis. According to discussions of the paradox, claims to the effect that a certain concept should be analyzed in terms of other concepts seem, on the one hand, to be informative but seem, on the other hand, not to be even potentially informative because they say that the concept should be analyzed in terms of itself. I argue that we should understand analysis claims in terms of grounding explanations and that when we do this, we have the beginnings of a solution to the paradox of analysis. I also discuss the theory of truth-making. Truth-maker theorists ask the question, 'What makes claims (beliefs, etc.) true?' T his is a request for a grounding explanation. Typically, a theory of truth-making says that, for every truth, there exists an entity that makes it true (e.g., a fact or states of affairs). There are, however, problematic cases. If truth is grounded in the existence of entities, then what sort of entity could ground the truth that there is no Santa Claus? In the study, I develop a theory of truth-making that grounds truth in something other than facts or states, which enables us to handle problematic cases of the sort just mentioned.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by William Butchard.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Witmer, D Eugene.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022645:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022645/00001

Material Information

Title: Grounding Explanations The Role of 'in Virtue of' in Philosophical Inquiry
Physical Description: 1 online resource (122 p.)
Language: english
Creator: Butchard, William
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Philosophy -- Dissertations, Academic -- UF
Genre: Philosophy thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In this study, I discuss a type of explanation that shows up in a variety of contexts, including philosophy and ordinary discourse. Consider the question, 'What makes a person tall?' Here are two possible answers one could offer: Having the right genes makes a person tall. Having a height that is significantly greater than average makes a person tall. The first explanation is causal. It says that a tall person's genetic structure is what causes her to grow up to be tall. But the second claim explains a person's being tall in terms of what it is to be tall in the first place. Clearly, the later would be a non-causal explanation. In this work, I discuss what bearing an understanding of such 'grounding' explanations has on our understanding of philosophical analysis. I argue that a clear understanding of grounding explanations points the way to a solution to the so-called paradox of analysis. According to discussions of the paradox, claims to the effect that a certain concept should be analyzed in terms of other concepts seem, on the one hand, to be informative but seem, on the other hand, not to be even potentially informative because they say that the concept should be analyzed in terms of itself. I argue that we should understand analysis claims in terms of grounding explanations and that when we do this, we have the beginnings of a solution to the paradox of analysis. I also discuss the theory of truth-making. Truth-maker theorists ask the question, 'What makes claims (beliefs, etc.) true?' T his is a request for a grounding explanation. Typically, a theory of truth-making says that, for every truth, there exists an entity that makes it true (e.g., a fact or states of affairs). There are, however, problematic cases. If truth is grounded in the existence of entities, then what sort of entity could ground the truth that there is no Santa Claus? In the study, I develop a theory of truth-making that grounds truth in something other than facts or states, which enables us to handle problematic cases of the sort just mentioned.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by William Butchard.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Witmer, D Eugene.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022645:00001


This item has the following downloads:


Full Text

PAGE 1

1 GROUNDING EXPLANATIONS: THE ROLE OF IN VIRTUE OF IN PHILOSOPHICAL INQUIRY By WILLIAM A. BUTCHARD A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 William A. Butchard

PAGE 3

3 ACKNOWLEDGMENTS I would like to thank John Biro, John Palm er, Greg Ray, and Robert Hatch for their helpful comments. I am particularly indebted to Gene Witmer for many fruitf ul discussions and for carefully reading and commenting on vari ous versions of this dissertation.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3ABSTRACT.....................................................................................................................................6 CHAP TER 1 INTRODUCTION....................................................................................................................82 THE CENTRALITY OF IN VIRTUE OF...........................................................................122.1Non-causal Explanations............................................................................................... 122.2Deflationary Attitudes toward the Notion in Question................................................. 142.3The Grounding Metaphor..............................................................................................172.4The in Virtue of Locution...........................................................................................182.5Substantive Theses Involving Grounding..................................................................... 223 ELIMINATING THE METAPHOR OF GROUNDING....................................................... 253.1Locating Determination in Conceptual Space............................................................... 253.1.1Lawful Determination....................................................................................... 263.1.2Causal Determination........................................................................................ 293.1.3The Possibility of a General Account of Determination................................... 303.2IVO Determination........................................................................................................323.2.1In Virtue of as an Adverbial Modifier............................................................ 343.2.2Adverbial Explanation...................................................................................... 393.3Worries about the Very Idea of Determination............................................................. 423.4In Virtue of and Necessitation.................................................................................... 474 TRUTH-MAKING, GROUNDING, AND ONT OLOGICAL COMMITMENT.................. 524.1The Truth-Maker Principle........................................................................................... 524.1.1Truth-Making and Ontological Commitment...................................................544.1.2Problematic Cases fo r the Truth-Maker............................................................594.2The Proposed Formulation............................................................................................ 634.3Truth-Making and Supervenience.................................................................................705 THE PARADOX OF ANALYSIS......................................................................................... 775.1An Epistemic Puzzle about Analysis............................................................................ 785.2Various Suggested Solutions to the Puzzle................................................................... 795.3A Second Puzzle...........................................................................................................855.4Equivalence and Substitution........................................................................................875.5The Proposed Solution..................................................................................................915.6Revisiting the Epistemic Puzzle.................................................................................... 97

PAGE 5

5 5.7Some Problematic Cases............................................................................................... 986 PROSPECTS FOR FUTURE APPLICATIONS.................................................................. 1026.1Essence........................................................................................................................1026.2Structural Universals................................................................................................... 1116.3Non-mereological Modes of Composition.................................................................. 117LIST OF REFERENCES.............................................................................................................119BIOGRAPHICAL SKETCH.......................................................................................................122

PAGE 6

6 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GROUNDING EXPLANATIONS: THE ROLE OF IN VIRTUE OF IN PHILOSOPHICAL INQUIRY By William A. Butchard August 2008 Chair: D. Gene Witmer Major: Philosophy In this study, I discuss a type of explanation that shows up in a variety of contexts, including philosophy and ordinary discourse. Consid er the question, What ma kes a person tall? Here are two possible an swers one could offer: Having the right genes makes a person tall. Having a height that is significantly gr eater than average makes a person tall. The first explanation is causal It says that a tall persons ge netic structure is what causes her to grow up to be tall. But the second claim explains a person s being tall in terms of what it is to be tall in the first place. Clearly, the later would be a non-causal explanation. In this work, I discuss what bearing an unde rstanding of such grounding explanations has on our understanding of philosophical analysis. I ar gue that a clear unde rstanding of grounding explanations points the way to a solution to the so-called para dox of analysis. According to discussions of the paradox, claims to the effect th at a certain concept should be analyzed in terms of other concepts seem, on the one hand, to be in formative but seem, on the other hand, not to be even potentially informative becau se they say that the concept s hould be analyzed in terms of itself. I argue that we should understand analysis claims in terms of grounding explanations and that when we do this, we have the beginnings of a solution to the pa radox of analysis.

PAGE 7

7 I also discuss the theory of truth-making. Truth-maker theorists ask the question, What makes claims, beliefs, etc. true? This is a request for a grounding explanation. Typically, a theory of truth-making says that, for every truth, there exists an entity that makes it true (e.g., a fact or states of affairs). Th ere are, however, problematic cases. If truth is grounded in the existence of entities, then what sort of entity could ground the truth that there is no Santa Claus? In the study, I develop a theory of truth-making that grounds truth in some thing other than facts or states, which enables us to handle problematic cases of the sort just mentioned.

PAGE 8

8 CHAPTER 1 INTRODUCTION In this study, I discuss a st yle of explanation that show s up both in philosophy and in ordinary discourse. I call such explanations grounding explanations. An exam ple of a grounding explanation someone might make in or dinary discourse would be the following: As an instructor, you have an ob ligation to apply a common set of standards when grading your students. A couple of examples from philosophy would be the following: Human beings are conscious in virtue of their physical features. Actions have their moral status in virtue of their consequences. Each of these three claims provides, in effect, an explanation of the fact that something has a certain feature. However, explanations come in a variety of sorts. Some explanations are causal, some are not. The claims in question, tho ugh explanatory, surely are not causal. If I say that as an instructor a person has certain obligat ions, I am not saying something to the effect that being an A causes being a B The term as indicates an explana tion to the effect that one thing is grounded in prior to or more basic than, another. In this case, I am saying something to the effect that being an instructor grounds the presence of certain obligations. While grounding explanations are a part of the conceptual machinery behind much of our day-to-day thought, they are the focus of philosophy. The very target of philosophical inquiry, one might say, is an understanding of what grounds what. I have three basic goals. The first is to demonstrate the importance of grounding explanations for philosophy. The second goal is to clarify grounding explanations by comparing them with causal and nomic explanations. The thir d goal is to show that a clear understanding of grounding explanations enables us to make progress with respect to substantive philosophical

PAGE 9

9 issues in cases where groundi ng plays a prominent role (e.g., truth-making and the paradox of analysis). The study has five chapters. In the second ch apter, I show that grounding explanations show up in a variety of contex ts, and I identify some of the expressions commonly used to advance such explanations. I also address certai n deflationary attitudes one might have towards grounding explanations, attitudes to the effect that as, in virtue of, and related expressions simply indicate that there is some sort of explanation available without speci fying a specific type of explanation. The worry behind such an attitu de would be that the expressions in question indicate that some sort of explan ation is present, but they do not indicate that there is any special sort of grounding explanation present. I argue that these sorts of doubts are misguided. In the third chapter, I make a suggesti on about how to eliminate the metaphor of grounding. I argue that getting cl ear about the notion of determination can give us a better understanding of natural law, causa tion, and explanations that app eal to these notions. I argue that it is possible to give a unifi ed account of the various phenomena we tend to think of as being accompanied by determination. I also address vari ous worries some might have about the very notion of determination. In the fourth chapter, I disc uss the theory of truth-making. Truth-maker theorists ask the question, What makes claims, beliefs, etc. true? This is a request for a grounding explanation. Typically, a theory of truth-making says that, for every truth, there exists an entity that makes it true. The candidate entities for tr uth-makers are, for example, facts and states of affairs. There are, however, problematic cases. If truth is grounde d in the existence of entities, then what sort of entity could ground the truth that there is no Santa Clau s? In this chapter, I use the account of grounding that emerges in the third chapter to develop a theory of tr uth-making that grounds

PAGE 10

10 truth in something other than facts or states of affairs while still satisfying the intuition that truth is, in some sense, grounded in how things are. This move away from explaining truth in terms of facts or states of affairs puts us in a positi on to handle problematic cas es of the sort just mentioned. In the fifth chapter, I discuss what bear ing an account of groundi ng explanations has on our understanding of philosophical analysis. I ar gue that the account of grounding contained in the third chapter points the wa y to a solution to the paradox of analysis. According to discussions of the paradox, any claim to the effect that a certain concept should be analyzed in terms of other concepts will turn out to be equivalent to the claim that the concept should be analyzed in terms of itself. As a result, analysis claims cannot be informative. I argue that we should understand philosophical analysis claims in terms of grounding ex planations and that when we do this, we can see how to ground a featur e (or concept) in terms of what we might call its constituents, and we have the beginnings of a solution to the paradox of analysis. In the final chapter, I discuss the prospects for further applicati ons of the account of grounding explanations I develop in the earlier chapters. I lay out some prominent theories of essence, focusing on theories that appeal to the notion of a things identity conditions. Theories that appeal to identity conditions are sometimes sp elled out in terms of the set of features that make it the thing that it is This is a non-causal usage of make, but, whereas it is easy to understand how some of a things features make it have other features it is not at all clear what it means to say that a things features make it the thing that it is I discuss the prospects for a plausible conception of essence that appeals to identity conditions without saying things to the effect that a things essence is what make it the thing that it is. I also make a suggestion about how to make sense of the notion of a structured universal. I spell ou t the beginnings of an

PAGE 11

11 account to the effect that, when it comes to univers als, we should avoid lite ral talk of parts and wholes and think in terms of generalizations abou t which universals are inst antiated in virtue of which.

PAGE 12

12 CHAPTER 2 THE CENTRALITY OF IN VIRTUE OF 2.1 Non-causal Explanations Consider the question, What m akes a person tall? This question would normally be asked in ordinary discourse and perhaps scien tific discourse. The speaker is asking for an explanation of the fact that some people are tall. Here is a possible answer one could offer: Having the right genes makes a person tall. Were someone to respond in this way, she would be offering a causal explanation. It says that someones genetic structure causes her to grow up to be ta ll. The causal facts supply the explanation. This is not th e only sort of response one could give. Someone could respond by saying this: Having a height that is significantly gr eater than average makes a person tall. This second response introduces a very different sort of explanation. It explains the fact that some people are tall in terms of what it is to be tall in th e first place. Clearly, to say this would be to offer a non-causal explanation. In the first answer, the term makes introduces a set of factsthe causal factsthat are supposed to explain why certa in people are tall. Are we saying something analogous when we offer the sec ond sort of answer? If so, we should be able to say what the relevant explan atory facts are. Surely they are not causal. Having a height significantly greater than average does not cause a person to be tall. In this case, one might say, it is the definitional facts (whatever those come to) that c onstitute the basis of the explanation. But it is not at all clear that definitional facts are always the basis for grounding explanations. Suppose I say, with respect to a ball that its being square explains the fact that it has a shape. This is certainly not a causal explanation. But bei ng square is not part of the definition of having a shape. Nor is having a shape part of the definition of being square. An appropriate explanation

PAGE 13

13 in this case would be to the eff ect that being round is a determin ate feature with respect to the determinable feature having a shape So, we have a case where the relevant facts are definitional and a case where they are not definitional. Is a unified account possible? There is a set of expressions that show up in a variety of contexts, including philosophy and ordinary discourse, which intr oduce non-casual explanations of the grounding sort. Some of the expressions turn up primarily, if not exclusiv ely, in philosophy, while othe rs turn regularly in both contexts. The expressions, though perhaps not synonymous, all seem to be used to advance the same sort of explanation. Consider the following series of claims: The balloons having a shape is nothing more than its being round. The balloons having a shape consists in its being round. The balloons being round counts as having a shape. These all sound perfectly appropriate, but the ma nner of speaking does not apply to the causal case. We would not move from saying things such as (1) and (2) to saying th ings like (3), (4), or (5): 1. The balloons touching the branch caused it to pop. 2. The balloons touching the branch brought it about that it popped. 3. The balloons popping is nothing more than its touching the branch. 4. The balloons popping consists in its touching the branch. 5. The balloons touching the branch counts as its popping. Clearly, there is a common notion behind the expr essions caused and brought it about. There is a genuine kindcausationwhich enables us to explain one thing in terms of another. In this sense, various explanatory expressions (expression which introduce explanations) can be given a unified treatment. What isnt so clear is whether we should beli eve that the non-causal explanations we have been considering can be give n a unified treatment. That is, it is not clear that there is a common explanatory notion behind such expressions as nothing more than, consists in, as, and counts as To state the problem in a s lightly different way, while it seems

PAGE 14

14 clear enough that the expression s in question introduce non-causal explanations, it is not clear whether non-causal explanations (or a large subset thereof) can be given a unified accountone which says more than that they are non-causal (and perhaps not law-like). Is there a common set of non-causal facts (i) that are present in explanations that appeal to definitions, determinables and determinates, and other non-causal phenomena; and (ii) make for a unified picture of these non-causal explanations, i.e., that ma ke for an account of these expl anations in terms of a unified kind analogous to causation? I believe that it is possible to supply this sort of account. The details of the account shed light on substantive philosophical problems such as the ones I mention above. 2.2 Deflationary Attitudes toward the Notion in Question There are so me deflationary attitudes one might have toward the notion the project purports to illuminate. Someone might think that while there are explan ations that do not say that something stands in the causal relation to the explanandum they form such a disparate group that no unified account is possible. While the causal cases are unified by the fact that such explanations are cases in which f acts bear a causal relation to the explanandum (causation being a genuine explanatory kind), in th e non-causal cases (in particular the cases I have been calling cases of grounding) there is no particular explanatory phenomenon being picked outthere is just a lack of causation. If a co llection of things are all non-Fs, that does not automatically give us a unified kind (a kind that ca n play the appropriate explanat ory role). It does not even guarantee the presence of a closely knit family of kinds. The thesis, then, is that in virtue of, count s as, etc. enable us to say that one thing explains another, but they allow for the presence of many different (and not interestingly related) explanation-making facts. The onl y restriction is that the facts are not ones that involve an

PAGE 15

15 object standing in a causal relation to the explanandum Thus when we offer an explanation by saying (1) we are simply claiming that there is an explanation present: 1. The balls being round counts as (or is nothing more than) having a shape. And in this case, the explanatory fact in question would be, for example, this: x has P x has Q and Q is a determinate with respect to the determinable P And when we say (2) we are simply claiming that there is an explanation present: 2. The balls having structural feature F counts as (or is nothing more than ) its being fragile. In this case, the explanatory fact might be something like x has P x has Q and Q realizes P Finally, when we say (3) we are, once more, saying that there is an explanation present: 3. x's being a bachelor counts as (or is nothing more than) being an unmarried male. In this case, the explanatory fact would be the following: x has P x has Q and Q defines P In this way, the presence of several apparently disparate examples of non-causal explanation-making facts suggests that there is no unified phenomenon in the non-casual cases. The relevant expressions introduce the notion of explan ation without introduc ing any particular type of explanation. We have no reason to think th at the facts that account for the presence of an explanation constitute a unified ki nd. The basic point, then, is that being non-causal is not itself a unified kind, and there is no hope for a unified account of grounding explanations in the absence of some other unifying notion. There is, therefore, a serious burden of proof on anyone who thinks a unified accoun t is possible. A related attitude is that the cont ent of claims to the effect that B counts as A consists in A etc. depends on the sort of case being discussed. According to this deflationary thesis, the

PAGE 16

16 expressions in question--nothing mo re than, consists in, co unts as, etc.do not introduce the notion of explanation itself. Sometimes the expr essions introduce definitions, sometimes they introduce the notions of determinables and determin ates. Because of this variation, no unified account is possible. I believe these deflationary pi ctures to be misguided. Let s begin with the first. The expressions consists in and c ounts as do indicate a style of explanation, not the notion of explanation itself. We can see this if we consider the following two claims: If someone is tall, that is because she has the right genes. If someone is tall, that is because she has a height that is signif icantly greater than average. These are both intelligible claims, but they are in a sense ambiguous as they stand. Each has a causal reading and a non-causal reading. We can see this in light of our inability to substitute counts as talk for because talk. Attempting to do so in this case gives us the following claims: Having a height that is significantly greater than average counts as being tall. Having the right genes counts as being tall. The second claim should sound off. This, I suggest, is due to a genuine ab sence of equivalence. This is plausible given the result we get if, inst ead, we attempt to replace b ecause talk with talk of causation. That give s us the following claims: Having a height that is significantly greater than average causes being tall. Having the right genes causes being tall. In this case, the problem is with the first cl aim. Here the claim is more than strange soundingit is obviously false. Clearly, the subs titution fails because the explanation we are looking for is non-causal. But substituting counts as in this case was successful. There is, therefore, a ready explanation for the apparent failu re to substitute counts as in the first case. It is simply incorrect to say that having the right genes counts as being tall. The reason is that the counts as locution (unlike becau se) does not simply introduce th e notion of explanation. If

PAGE 17

17 we want to recast the above claims in terms othe r than because, we should use counts as talk in one case and causal talk in the other. This gives us the following claims: Having a height that is significantly greater than average counts as being tall. Having the right genes causes being tall. Neither counts as nor causes, we can say, is interchangeable with because. Rather, they serve to disambiguate it, since the one expression introduces a causal notion, while the other introduces a non-causal notion. In this sense, c ounts as claims and cau sal claims introduce more fundamental explanatory notions than that introduced by because. This line of reasoning applies to the other locutions we have considered We cannot, for example, substitute is nothing more than for because. These considerations do not show that there is a unified type of explanation that the relevant expressions introduce. That is, even if the expressions do not simply introduce the notion of explanation in general, and even if they clearly do point to non-causal explanatory facts, it may be that the non-casua l explanatory facts in question do not form a unified kind. So, addressing the first deflationary attitude does not automatically supply an answer to the second. However, the metaphor of grounding appears to apply to the noncausal explanations we have considered. In each case, it seems appropriate to say that one set of facts grounds another. The appropriateness of the metaphor may indicate the pres ence of a unified kind. Let us look at this metaphor in more detail. 2.3 The Grounding Metaphor While the grounding m etaphor is appropriate for the non-causal cases I have been discussing, it is not appropriate for cases of causation. We cannot see causation as a relation that grounds one fact in another. We can see this if we note the appropria teness of a different metaphor, that of levels. Suppose one billiard ball hits another and causes it to move six inches.

PAGE 18

18 The explanans and explanandum are at the same levelthe objects mentioned are of the same sort, both billiard balls, and one causes the other to do move. Causation, of course, can figure in an inter-level explanation. The causal facts about the micro-constituents of a billiard ball explain its macro-features, but this is because th e micro-constituents of the ball are its constituents a non-causal notion. Causation, therefore, is not what makes the e xplanation inter-level. On the other hand, if we say something to the e ffect that what makes a person tall is having a height that is significantly greater than average, we are e xplaining someones being tall in terms of what it is to be tall in the first pla ceand in doing this we are relying on the notion of a height and the notion of an average. In some sense, these are lo wer-level phenomena. They are more basic than or prior to th e phenomena we are trying to ex plain. The reason that it is appropriate to apply the grounding metaphor, then, is that grounding suggests levels. This is not to say that all non-causal explan ations are due to the presence of grounding. A scientific explanation th at appeals only to non-causal laws is a non-causal expl anation, but it is not a grounding explanation. If A co-varies with B in a law-like fashion, it does not follow that A grounds B or vice versa. Moreover, generalizations that figure in grounding explanations have the force of logical or conceptual necessity, someth ing law-like claims lack. An account of this sort of explanation would, of course, require us to remove the grounding metaphor. But the metaphorin the restricted application just sp ecifieddoes seem to point to something common to the non-causal cases we have been focusing on. If I say that A is nothing more than B that A consists in B, or that B counts as A, in each case the explanations these facts underwrite areto invoke the second metaphoresse ntially inter-level. 2.4 The in Virtue of Locution So far, the exam ples we have considered ha ve been day-to-day claims. In day-to-day contexts, we ask things like, What counts as harassment? or What ma kes this cheating?

PAGE 19

19 These questions invoke very ordinary notions. At least, there is nothing th eoretical about them. The notion of grounding, however, is at the heart of what philosophe rs do. One could go so far as to say that to engage in philosophical enquiry is to search for grounding explanations. Coming to understand the nature of these expl anations, by eliminati ng the grounding metaphor, therefore, has meta-philosophical implications. In philosophy, i n virtue of and by virtue of are, as a matter of sheer frequency, central among the expressions used to introduce the notion of grounding. Indeed, the standard characterizations of key philo sophical notions contain these expressions. The analytic/synthe tic distinction, for example, is characterized by Quine in the following way: But Kants intent, evident more from the use he makes of the notion of analyticity then from his definition of it, can be restated thus: a statement is analytic when it is true by virtue of the meanings and independ ently of fact. (Quine 1951, p. 43) If a claim is true by virtue of meanings, then we have an explanation for the truth of the claim, and that explanation is surely not causal. It is a claim to the effect that meanings ground analyticity. Here is another example. Brian Mclaughlin a ppeals to in virtue of in characterizing causal laws: It would not follow that the nomic properties of the special sciences are epiphenomenal, even if the Causal Priority of the Physical and Physical Comprehensiveness hold. Even given the truth of the doctrines in question, the laws of the sp ecial sciences in question may be causal laws, laws in virtue of which events participate in causal relations. (Mclaughlin 1989, p. 131) If causal laws are laws in virtue of which events participate in causal relations, then the laws do not cause the events to participate in causal relations Here in virtue of is supposed to introduce an explanatory fact that explains why events stand in causal relations in the first place. There are also examples in epistemology. Pollock says the following about good reasoning:

PAGE 20

20 I have been arguing that many of the individual bits of data on which our epistemological theory is founded will, in a certa in sense, be self-evident (more accurately introspectible). By virtue of knowing how to reason, we know ho w to tell right reasoning when we see it, and that provides us with our data. (Pollock 1986, p. 218) The claim here is not to the effect that knowing how to reason causes one to be able to know good reasoning when one sees it; ra ther, the claim is a stronger one to the effect that being good at reasoning grounds being able to id entify good reasoning when we see it. In virtue of claims, then, do show up in a variety of contexts within philosophy itself. This, of course, does not show that they play an important role in philosophy, or even that the usages we have seen advance a common style of explanation. As far as these examples go, it may just be a handy locution. But it may be more than that. It is not obvious that the claims made in these quotations can be made without in virtue of and related ex pressions, and the fact that the explanations in the cases we have looke d at are not causal expl anations suggests that there might be some further feat ures these explanations share, features which may reveal a common style of explanation. On occasion, the locution (or the notion it may introduce) is discussed explicitly. Kim, for example, discusses the following example: When Socrates expired in the Athenian pris on, Xantippe became a widow. The onset of Xantippes widowhood was determined by the death of Socrates. As we might say, Xantippe became a widow in consequence of, as a result of, or in virtue of Socrates death. (Kim 1974, p. 22) Though Kim does not focus on the in virtue of locution, he does take it to indicate an important type of notion. He says: These noncausal dependency relations are pervas ively present in the web of events, and it is important to understand thei r nature, their interrelations, and their relation to the causal relation if we are to have a clear and complete picture of the ways in which events hang together in this world. (Kim 1974, p. 22)

PAGE 21

21 There are, then, plenty of examples of philosophical claims that are appropriately expressed in terms of the in virt ue of locution. Why does in virt ue of occur so frequently in philosophy? One reason, I want to suggest, is the ease with which it enables us to give voice to philosophical disputes. Consider an example. Suppose we have a particular action x, which is a right action. We might characteri ze a potential dispute between tw o philosophers in terms of the following positions: x is right in virtue of being entailed by a moral principle that would result in the best consequences x is right in virtue of being entailed by a moral principle everyone would accept from the original position Both philosophers can agree that x is right. They can also agree on what each takes to be the thing that grounds the rightness of an action. Both can agree, for example, that according to the first position rightness is grounded in the fact that it is entailed by a moral principle that would result in the best consequences. But there is a disagreement. It consists in the fact that each philosopher accepts her own gro unding claim and rejects the others. Here is a second example. One way to characterize neutral monism is in terms which grounding claims its advocates accept and which ones they reject. Consider the following grounding claims: Mental facts obtain in virtue of neutral facts. Physical facts obtain in vi rtue of neutral facts. Mental facts obtain in virtue of physical facts. Physical facts obtain in vi rtue of neutral facts. While neutral monism is the view that the firs t two grounding claims are true and the second two are false, physicalism is the view that only the third claim is true. The in virtue of locution, then, does make it easy to give vice to a variety of philosophical thesis. But the locution is more than just a handy tool. A close examination of the

PAGE 22

22 locution, as it appears in philos ophical discourse, shows that th ere are substantive philosophical issues that hinge on the correct formulation of grounding claims. 2.5 Substantive Theses involving Grounding Explanations in the form of grounding claims have, like other claims, what we might call a logical form. Sometimes it is necessary to make explicit appeal to e xplanatory notions when giving a philosophical account of something. Disc overing the logical form of a grounding claim can, as a result, help solve philosophical puzzles Consider the notion of truth. Intuitively, at least, every truth corresponds to a fact, for facts are what make propositions (and beliefs, sentences, etc.) true. Talk of making in this context is, of course, not talk of causation but of grounding. Truth, we can say, is grounded in fact (if the corr espondence theory of truth is basically correct). Some think that to explain truth we need to appeal to truth-makers. According to such an account, to explain truth, we have to ask what makes a proposition true A truth-maker is supposed to be an entity whose very existence grounds the truth of a proposition. This leaves us with a puzzle about how to handle negative truths Does a correspondence theory of truth commit us to negative stat es of affairs? After all, if A corresponds to B we have to quantify over A and B Therefore, if A is a negative truth, there must exist something that grounds itand that something must be a funny sort of absence. So, anyway, some have argued. In deciding what the candidates are for truth-ma kersor whether the thes is in question, in its proper form, would even mention truth-makers it is important to be clear about what a grounding explanation is supposed to accomplish and what form it needs to take in order to be successful. I believe that talk of correspondence can be expre ssed in terms of a grounding notion that frees us from the problematic commitments incurred by talk of co rrespondence in the form of truth-making.

PAGE 23

23 Moreover, there are philosophical claims that are less obviously grounding claims but are, upon reflection, reasonable candidates for grounding cl aimsand, as we will see, when we state these claims explicitly in terms that indicat e grounding, we shed light on a second philosophical puzzle. Consider the notion of analysis. An anal ysis is a claim (or proposition) that states the conditions under which a concept applies. The analysis of being a brother for example, is typically cast in something like the following terms: x is a brother iff x is a male sibling The terms on the right side (male and sibli ng) are supposed to be in some sense more basic than, or prior to, the terms on the left side. Basicality and priority are not strictly modal terms. It is very plausible that they are grounding notions. If this is corre ct, then an analysis is a claim to the effect that anything th at is a brother is a brother in virtue of being a male sibling. As we will see, this cannot be correct, strict ly speaking, because to be a brother is to be a male sibling, and a feature cannot be had in virt ue of itself. Notice that the same basic puzzle applies the biconditional formulation we just saw. If it is the co rrect analysis, then it is equivalent to (i.e., the same in content as) the following: x is a brother iff x is a brother This is a statement of the paradox of analysis As we will see, thinking about the analogous puzzle when it comes to in virtue of claims whic h are not analyses can shed a good deal of light on how to dissolve the paradox of analysis. In pa rticular, it may be that putting grounding claims in their correct form suggest a way of freeing ourse lves from formulations of analyses that allow the substitution responsible for the puzzle in the first place. These are the two examples that I will focus on in the dissertation. As we will see in the final chapter, getting clear about the prope r from of grounding claims willpotentially, at

PAGE 24

24 leastshed light on a variety of other philosophical claims (e.g. claims about structured universals, non-mereological compositi on, and various essentialist theses).

PAGE 25

25 CHAPTER 3 ELIMINATING THE METAPHOR OF GROUNDING What does the term grounding introduce? There are three possibi lities. The first is that it introduces the notion introduced by in virtue of and related expressions. The second option is that it introduces determination. Th e third is that it introduces a pa rticular sort of determination. Let us consider these options. Grounding is not the notion I am a ttempting to explicate in this dissertation, for the latter notion, as I ague later in this chapter, is not a relational notion, while grounding obviously is. Nor does grounding introduce determination in general. If it did, then it would apply in cases where the presence of de termination is due to causation, which, as we have seen, it does not. Therefore, it must refer to a special sort of determinationthe sort that is due to certain non-causal phenomena. Because that phenomenonwhich is the one I am attempting to explicateis so frequently intro duced by the in virtue of locution, I will I will use IVO as a convenient label. Gro unding is determination due to IVO. 3.1 Locating Determination in Conceptual Space To say that a for m of determination is non-causal is a purely negativ e characterization. I want to attempt to supply a positive account of non-causal determination. But what is determination? I will not try to offer a se t of necessary and suffi cient conditions for the application of this notion. Nor will I take it as comparatively basic with respect to IVO. That the latter strategy would be misguided is sugge sted by the correctness of only one of the following claims: If there is determination presen t, that is in virtue of th e presence of laws, causation, or IVO. If there is a law, causation, or IVO present, th at is in virtue of th e presence of determination.

PAGE 26

26 Which of these claims is true? Were we to take determination as basic, we would be endorsing something like the second claim. I find it diffi cult to believe this claim. If there were already determination present, then noting the presence of laws, causation, or IVO would not contribute to the explanation. So, determination is not more basic than the other three notions. It is not something that should be introduced by a condition in an account of (in a set of necessary conditions for) any of the other notions. My strategy will be to locate determination in conceptual space by noting some of the features determination displays in various examples. There are three sources of determination I will consider: laws, causation, and IVO. I will discuss laws and causation first and propose a unified acc ount of determination in these cases. Then I will argue that the account can be adjusted in such a way as to cover determination in cases where it due to IVO. 3.1.1 Lawful Determination In what sense do laws determ ine how things ar e? The sense of determination mentioned in this question is not the one introduced by determi nistic in the claim that the world is governed by deterministic laws. To say that laws determine how things are is to invoke the notion introduced when we say that the world is governed by laws. Even a world with nothing but indeterministic laws would be governed by those laws. When we say that a world is governed by laws, what we mean is something like this: Which laws of nature there are determin es which property distribution there is. What we are saying here can be contrasted with a sort of Hu mean picture, which says the following: What property distribution there is determ ines which laws of nature there are. Clearly, whether the world in indeterministic or not, we are forced to choose between these two claims. When we say that a law that is indeterministic, all we mean is that it is statistical

PAGE 27

27 (and that it cannot be reduced to a non-statis tical law), and when we say that a law is deterministic, we mean that it is not statistical (o r that, if it is statistical it can be reduced to a law that is not statistical). However the world is with respect to these sorts of laws, the laws determine property distributions, not vice versa In what sense do laws determine property distributions? First, note that law-like determination does not require cau sation (or a causal law). A la w does not have to be a causal law in order for it to determine property distributions. Consider the following (admittedly abstract) example. Suppose there is a non-causal law, L to the effect that all A s are B s. Then we can explain the occurrence of a particular B in terms of L But notice a subtlety here. L does not explain the occurrence of B in terms of a particular A L tells us why it is necessary that B occur, given that A occurred, but that does not give us an explanation of B in terms of A L places constraints on possible distributions of A s and B s, so, of course, given L if an A occurs, it is necessary that a B occurs. This is clear if we co nsider a slightly different law, L* and understand L by analogy. L* says that all A s are B s and all B s are A s. Then, if a particular A occurs, it is necessary that a B occur. And L* says that it is necessary that B occurred, given that A occurred. But L* also says that it is necessary that A occurred, given that B occurred. There is an explanation for the fact that there are no A s that are non-B s, and for the fact that there are no B s that are non-A sbecause L* places constraints on possible distributions of A s and B s. But clearly in this case neither A nor B explains the other in light of L* They are, exp lanatorily speaking, on a par. If we say that A explains B then we cannot say that B explains A And if we say that B explains A we cannot say that A explains B The law, because it says that it is necessary that either occurred given that the othe r occurred, supplies an explanation of why there are no A s that are non-B s (and vice versa ). The explanation, again, is that there is a law that

PAGE 28

28 places constraints on possible distributions of A s and B s. And the only difference between L and L* is that L contains a one-way conditional and L* contains a bi-conditional. The important point is that this is not an explanatio n of why something occurred in terms of non-laws or, better yet, while the explanation mentions non-laws ( A and B ), it does not appeal to determination between non-laws. The determination it introduces holds between the law and the property distribution. So, in cases of determination due to nomic connections, there is, fo r certain events, an explanation of why they occurred. Each event oc curred because there is a law that says that it had to occur given that certain other things occurred. This does not introduce any determination relation between A and B In a world where there are only correlational laws, we have no reason to point to the initial state and explain other stat es in terms of a determination between the initial state and later states (together with the laws). We could just as easily point to later states and explain the initial state in terms of a determinatio n relation between the late r states and the earlier states. In such a world, there is no determin ation between states. Nevertheless, the laws determine the property distributions. This is an example of pure lawful determin ation, because something is determined by a law even though there is no causa tion between events. This is not to say that laws are sufficient for property distributions. You n eed individuals, for example, and laws are not sufficient for which individuals exist. Two worlds with the sa me laws can contain different individuals. Nor do laws tell us why properties are distributed over one set of indivi duals rather than another. Presumably, there is an explanation for any partic ular distribution, so the laws are only part of the explanation. Nevertheless, we want to say something to the effect that laws determine property distributions (and not vice versa ). The laws themselves are not sufficient for a given

PAGE 29

29 distribution, but they are a part of something that is sufficient for that distribution, and their being a (suitable) part of something that is sufficient for the distribution is what makes for determination. If the laws are probabilistic, then they are still part of something that is sufficient for the property distribution. It is just that chance is an element as well. 3.1.2 Causal Determination It is a part of our comm on sense conception of the world that there are events that are brought about by other events. Common sense, therefore, dictates that explanations of the sort just discussed do not exhaust the explanations of the sort that figure in day-to-to-day (and scientific) explanations. Th ere are cases in which it makes perfect sense to explain B s in terms of a determination relation they bear to A s and not just in terms of law-like constraints on possible distributions of A s and B s. Cases in which there is causation between events are the obvious ones. In such cases, we may appeal to a law (perhaps to the effect that all A s cause B s) in order to explain the occurrence of a B but we also appeal to determination between events. We can contrast this view (a full-blown concepti on of causation) with a deflationary conception. In very general terms, one could adopt one of two attitudes toward cau sation. One could adopt the position that if A causes B then there is a nomic correlation to the effect that whenever there is something of one sort, there is something of another sort. There is on this deflationary conception, no principled difference between causa l laws and correlational laws. The ones we call causal may be laws that make reference to successive times, but this does not imply that causes bring about or determine their effects. In other words, according to the purely nomic conception of causation, causation does not i nvolve determination be tween eventseven though there may be such a thing as lawful determ ination, which is genuinely explanatory. Since causation is a relation that is s upposed to hold between events (e vents being non-laws), any view

PAGE 30

30 that equates causation with mere nomic determin ation leaves out a part of the common sense conception of causation. I will assume th at the full-blown conception is correct. We saw that the claim that laws determine property distributions does not commit us to saying that laws are sufficient for property dist ributions. We can say something similar about causation. The claim that A caused B does not commit us to the claim that A is sufficient for B To be sure, if A caused B then there is a sense in which B must occur given that A occurred. But this is not because A itself is sufficient for B It is not as if truth that B occurred falls out of the truth that A occurred. Sometimes, for A to cause B it is necessary that other conditions obtain conditions that obtain only contingentlyagainst which A brings about B These background conditions are not ones that would figure in an anal ysis of causation. If A determines B there is a set of claimsnamely, the (causal) explanation of B s occurrencewhich entail the truth that B occurred, but the occurrence of A is insufficient for the occurrence of B (the claim that A occurred does not entail the claim that B occurred), even if A caused B .1 3.1.3 The Possibility of a Genera l Acco unt of Determination On the basis of these two examples, I want to suggest the following general (though rudimentary) account determination. Then I will ar gue that this picture applies to the cases due to IVO. In both of cases, there is a notion present (e ither that of a law or that of causation) that is explanatory because when it is present, there is determination. We can state the matter in general terms. Whenever determination is present there is some other explanatory notion present (e .g., that of causation) such that whenever this notion applies, the claim that it applies is a co njunct of a claim that entails the explanandum (a claim to the effect that a certain event occurred). Here we characterize the additional notion in terms of explanation. As a result, we have characterized determination in terms of explanation. This 1 For a defense of this position see Mackie 1965.

PAGE 31

31 suggests that we do not have an analysis of determination. For we want to say that whenever there is an explanation present that is b ecause there is determination present, not vice versa But were we to define determination in terms of explanation, we would be committed to saying that determination is present because there is an explanation present. Despite not supplying an analysis, a direct appeal to e xplanation in this way does help us situate determination in conceptual space. The picture shows that we cannot identify determination with either of the other explanatory notions, for it is present whenever either of the notions is present, and they are distinct notions. Nor can we identify determ ination with the necessi tation that is present whenever one of these other explanatory notions is presentfor that is necessitation in the form of entailment between propositions, and determinati on is a relation that holds between events in the causal case and between laws and property distributions in the nomic case. The general picture here is th at laws and causation should be understood, in part, in terms of the sufficiency that accompanies them, and that determination (a notion that smacks of sufficiency) should be understood in terms of laws and causation. When we say, with respect to law-like and causal cases, that A determines B then, we are committed to at least this much: 1. There is a claim C that entails the claim B occurred and has A occurred as a conjunct (i.e., A is a part of somethi ng that is sufficient for B ); and 2. C is a claim that introduces a further explanatory notion (e.g., that of a law). To put it slightly differently, whenever A determines B an explanatory notion (distinct from that of determination itself) applies to A and B a notion that applies only when determination is present. Moreover, when that further notions applies, A is a part of something that is sufficient for B

PAGE 32

32 We could eliminate the direct appeal to explanation and give a disjunctive account of determination. Rather than referring to the pr esence of a further explanatory notion, we could simply say that either causation or a law is at work. Were we to take this route, the account would equate talk of determination with talk of sufficiency accompanied by the appropriate causal or nomic link. Since there is no direct ap peal to explanation, this account reads more like an analysis, though its disjunctive character may make it seem that the key notions have nothing significant in common. One might argue, however, that they do have something significant in common, on the grounds that a dire ct appeal to explanation was possible in the first place. In other words, since causation and laws are both explanatory notions, we have grounds for thinking that the disjuncts in the account of de termination do have something significant in common: they are both explanatory. For our pur poses, we do not need to decide the matter. Whichever route we take, we will have the resour ces for developing an account that applies not only to causal and nomic case of determination but to IVO cases as well. 3.2 IVO Determination Is it po ssible to fit determination of the IVO so rt into this picture? With respect to the crucial elements of the picture, IVO determination fits rather well. It may seem, at first, that it does not fit. Someone might argue that, while when A causes B A does not suffice for B when A in virtue of B B does suffice for A For example, if some x is a brother ( A ) in virtue of being a male sibling ( B ), then the claim that x is a male sibli ng entails the claim that x is a brother. Again, if something has a color (A ) in virtue of being red (B ), then B suffices for A So, on the basis of these examples, it may seem that IVO de termination clearly does carry sufficiency in a way that causal determination and nomic determina tion do not. However, to argue in this way would be a mistake. Consider the following generalization: For any x, x is a brother in virtue of being a male sibling

PAGE 33

33 It seems true enough. But notice that th e following claims seem true as well: For any x, if x is a brother, x is a brother in virtue of being male For any x, if x is a brother, x is a brother in virtue of being a sibling Someone might object to these partial explanatory claims on the grounds that neither being a male nor being a sibling is sufficient for being a brother The objection would be that claims to the effect that something has P in virtue of having Q can be true only if having Q is sufficient for having P This would suggest that, if I say that so meone is a brother in virtue of being male, what I am saying is, strictly speaking, false. But I believe that there are true in virtue of claims to the effect that something has P in virtue of having Q where having Q does not suffice for having P In such cases, however, having Q is an essential ingredient of something wh ich does suffice for having P Thus, having the property of being male is insufficient for having the property of being a brother but there is a sense in which being a male is a part of something which is sufficient for being a brother. The suggestion is that even these claims are covere d by the general account I have offered. The in virtue of locution introduces an explanatory notion, and when it ap plies, there is determination. Moreover, there is something that is sufficient for the explanandum In this case, it is the explanandum itself. But that is not a problem. The general account does not say that the explanandum is determined by the thing that suffices for it. Rather, it is determined by a part of whatever suffices for it. The sufficiency is nece ssary for the explanation (and the determination), but that does not commit us to saying th at the thing that is sufficient for explanandum in any way brings it about. This is not to say that B never suffices for A in the IVO cases. If something is red, that is sufficient for being coloredfor it is true that it is colored solely in virtue of being red. A difference between causal and nomic determination, on the one hand, and IVO style

PAGE 34

34 determination, on the other, then, is that in the IVO case, sometimes A is identical with, rather than being a part of, the thing that suffices for B Again, this is not a problem because the sufficiency claim about A and B is not supposed to indicate determination from A to B The sufficiency is just part of the explanation, sinc e it is a condition on the presence of determination. There is another respect in which IVO determination is different from law-like determination and causal determination. There are c onsiderations that sugges t that the terms that introduce IVO determination are different from causal terms and law-like terms in that IVO terms do not introduce a relation that holds between laws, facts, stat es of affairs, or what have you. When grounding explanations are properly fo rmulated, the key terms act as adverbial modifiers rather than re lational terms. For this reason, if there is a general account that covers law-like determination, causal explanation, and I VO determination, it will not be formulated in terms of the determination of events. 3.2.1 In Virtue of as an Adverbial Modifier I want to suggest that the in virtue of locution is f undam ental among the expressions which introduce IVO explanations. It is fundame ntal in the sense that it, unlike counts as does not play the grammatical role of a relational term and this re veals an important conceptual distinction between types of determination. In virtue of may, to some, appear to be straightforwardly a relati onal term. Consider the following claims once more: The balloon has a shape in virtue of being round. The balloons touching the branch caused it to pop. The second of these is a causal claim, and, for that reason, it seems to commit us to using sentence nominalizations. The expression it to p op is, of course, not a sentence nominalization, but we tend to think of causation as something th at relates events, and the underlying structure of events, as the relata of causation, is made salient by the use of sentence nominalizations, since

PAGE 35

35 these expressions make the causally relevant prop erties apparent. Notice th at we are indifferent to a substitution of the balloons popping for it to pop. The event of the balloons touching the branch we say, caused the event of the balloons popping Someone might be tempted to say that, by anal ogy, there is in the first claim an implicit its immediately following in virt ue of, giving us the following: The ball has a color in virtue of its being red. I see no reason to think this is the correct reading. I want to suggest a reading that is, for a principled reason, in line with the reading we would naturally gi ve to certain action sentences, namely, sentences which say that an agent did one thing by doing something else. Consider the following claim: I turned on the light by flipping the switch. This claim, I suggest, says something at least in the same ballpark as I turned on the light in virt ue of flipping the switch. It does not say My flipping of the switch caused my turning on the light. At first glance, it may not be obvious that this causal reading is incorrect. To the extent that it sounds like the correct reading, ho wever, it is, I suggest, being c onfused with claims such as the following: My flipping of the switch cau sed the light to turn on. My flipping of the switch caused the turning on of the light. These claims, of course, say something different from the action claim in question, for they do not say that I am doing one thing by doing something else. Rather, they say only that my doing something caused something else. For a discussion in favor of a non-causal reading of the sorts of action claims I discuss above, see Davidson 1971, pp. 56-61. Note also that Austin

PAGE 36

36 (1955, p. 120) associates the by locution with causal claims. For example, he defines perlocutionary acts in terms of the achieving of certain effects by saying something. This usage of by, however, is consistent with the non-causal reading. By saying something, I may cause something to happen, but, in virtue of this causal connection, I may also achieve something. That is, the connection between saying something and achieving something can be non-causal even if the achievement was due in part to a causal connection between my saying something and some later event. Just as we can say, for ex ample, that my flipping of the switch caused the light to turn on and that in virt ue of this causal connection, I turned on the light, we can say that my telling a joke cause the event of someones laughing and that I changed the persons mood in virtue of the causal connection between my telling a joke and the persons laughing. Concerning nominalizations, notice that we would not say the following: I turned on the light by my flipping of the switch. This claim does not sound natural at all. It is much more natu ral to say that I turned on the light by flipping the switch And, again, I want to suggest that the nominalized version sounds off because it is conceptually confused. In discussing a related matter, Jonathan Bennett has the following to say: We can say: He silenced the rattle by ti ghtening the screws, but the genetive-object version, He silenced the rattle by his tighteni ng of the screws, is poor English. (Bennett 1988, p. 214) Bennett claims that we can sa y, He silenced the rattle by means of his tightening of the screws. But he thinks that saying this is analogous to saying, He silenced the rattle by means of a push and a tug, which he suggest s should not be understood as ak in to claims which indicate that an agent is doing one thing by doing somethi ng else. It seems that there is a principled distinction to be drawn here. In by means of a push and a tu g, what follows by means of (namely, a push and a tug) is a noun phrase. Similarly, tightening in by means of his

PAGE 37

37 tightening of the screws is a noun (due to th e possessive). Since th ese are noun phrases, the claims in question are, strictly speaking, not claims to the effect that an agent did one thing by doing something else. So, by and by means of are not interchangeable. There is, however, a use of by that captures claims of the sort we have in mind. On this use, by should be followed by a verb. If I say, I am tightening the screw by turning the screwdriver, we can see turning as a verb (in which case there is no implicit nominal ization following by). On this use of by, just as it doesnt make sense to say, He silenced the ra ttle by a push and a tug, it doesnt make sense to say, He silenced the rattle by his tightening of the screws. Similarly, since flipping acts as a noun in I tu rned on the light by my flipping of the switch, it does not, according this reading of by, indicate that the agent did one thing by doing somethi ng else, as it does in I turned on the light by flipping the switch. I suggest, then, that the by claims without the nominalizat ion sound more natural because they capture the claim in question better than the nominalized version. If this is correct, we should take these sentences at face value and dr op the nominalization following by. I want to make the further suggestion that, si nce it is plausible that by cl aims and in virtue of claims introduce notions belongi ng to the same family, we should not say this: I turned on the light in virtue of my flipping of the switch. Rather, we should say the following: I turned on the light in virt ue of flipping the switch. The suggestion, once again, is that dropping the no minalization leaves us with a more naturalsounding sentence because doing so provides a better way of expr essing the claim in question. No doubt there are differences between by an d in virtue of. The expressions are not interchangeable. For example, the follow ing non-action claim surely sounds off:

PAGE 38

38 The ball has a color by being red. However, the claim is at least in the right ballpark. To see this, we can recast it in terms of instantiation, retain the by locution, and generate a slightly better formulation: The ball instantiates the property of having a color by instantiating the property of being red. This sounds better, though by vi rtue of or in virtue of w ould sound better still. Using by in isolation is, strictly speaking, better saved for cases wher e there is an agent involved. Nevertheless, these observations suggest that, desp ite their differences, b y and in virtue of express notions belonging to the same family. A nd if we have reason to believe that by does not introduce a relation, then we have some reason to think that in virtue of does not introduce a relation either. The upshot of this discussion, then, can be i llustrated with the follo wing contrast. Suppose I move my arm and a cup falls off the table as a re sult. Here are two ways one might try to give an account of what happened: He caused the cup to fall off the table. His moving his arm caused the cup to fall off the table. Each is a claim with the form A caused B, but in the first account A is a person, while in the second account A is an event (his moving his arm ). We can ask whether causation always relates events or whether it sometimes relates other sorts of entities, such as persons. We can ask similar questions when it comes to grounding explan ations. Here are two ways we might try to state our initial grounding explanation: A person is tall in virtue of having a height that is signi ficantly greater than average A person is tall in virtue of her having a height that is si gnificantly greater than average The second treats in virtue of as an expression that introduces a relation what we might call the grounding relation. The expres sion her having a height that is significantly greater than

PAGE 39

39 average refers to an object (a property instantiation or state of affairs). But the first claim does not treat in virtue of as introducing a relation at a ll. It treats is as an adverbial construction. The rough idea is that to have a feature in virtue of is to have the feature in a certain way To say that a feature is had in a way is to say how it is had. If this is th e proper form of a grounding explanation, then we can see the claim that a person is tall in virtue of having a height that is significantly greater than average as being analogous to claims such as The deputy made the arrest on behalf of the sheriff. If a deputy can make an arrest unnecessarily or the deputy makes the arrest on someone elses behalf then the deputy is making the arrest in a certain way Likewise, to say that somethi ng has a feature in virtue of is to say how the feature is instantiated, and this is to say that it is had in this or that way. If this is correct, then grounding explanations are unlike causal ex planations in that, while caus al explanations introduce the causal relation, grounding explanations not introduce a determination relation at all. 3.2.2 Adverbial Explanation Why is adverbial m odification e xplanatory, exactly? Let us consider some examples. If I say that someone walked quickly, we have adverbial modification that tells us how something was done (or the way in which it was done), but it is not explanatory. When it came to actions, we saw that claims to the effect that someone turned on the light by flipping the switch the adverbial modifier also tells us how something was done, but it doe s so in the sense that it indicates the method or means by which someth ing was done. Saying how something was done in this sense is, of course, is explanatory. This explains why it is inappr opriate for a magician to respond to the question, How did you do that? by saying, Quite well, thank you. To respond in this way would be to dodge th e question. It would be to dodge a request for an explanation. To supply the explanation is to reveal the method. Of course, another magician might do the same trick using a different method. The magici ans do the very same trick, but they do it in

PAGE 40

40 different ways because they use different methods. This is not to say that talk of doing one thing by doing another is synonymous wit h, or analyzable in terms of, talk of methods. I am just pointing to a usage of how that is different from other usages in that it when it applies it is also appropriate to talk of methods and means. We can understand claims about inanimate obj ects by analogy. We saw that, although a claim about an inanimate object, such as The ball has a color by being red sounds off, it is plausibly in the same vicinity as actions claims of th e sort we have been considering, and as a result can be understood by analogy. It does not sound unnatural at all to say that the ball instantiates the property of having a color by instantiating the property of being red We can capture both cases with ordinary talk of ways. In saying that a person turned on the light by flipping the switch, we are describing the way in whic h she turned on the light. If someone did not know how to turn on the light, you could instruct them to do it in this or that way. In saying that a ball has a color in virtue of being red, we are saying that it is colored in one way as opposed to other ways. This is intuitive enough, but what does talk of ways come to? In ordinary speech, we say that things are certain ways. In defending a persons behavior, for example, I might say, Well, thats just the way he is. We al so say that an object is different ways at different times. I might say, The city used to be that way, but it is not that way anymore. Th ese sentences include the expressions the way an d that way. In the second case, we could have made the same claim by substituting for that way an expr ession that introduces a specific feature. We could have said, The city used to be exciting, but it is not exciting anymore. In the first case, however, it seems

PAGE 41

41 that need a referring expression. What does the way refer to he re? The obvious answer is that, if it refers, it refers to a property. So, it seems th at to say of an object th at it is a certain way is just to say that it has a certain feature. And to say that it has a feature in this or that way is, in some cases, to say that it has the feature in virtue of having some other feature. So, we have determination in the absence of a determination relation because there are adverbial constructions which are used to say how something was doneand in cases where something like a method is revealed, we have determination. And it seems that this way of speaking is naturally cast in te rms of in virtue of talk. E xplaining cases of IVO that do not involve an agent may call for a term other than h ow a property is instanti ated, but talking about the way in which the feature is had gets at the some sort of explanation, since we can understand these cases in terms of the action cases. Returning to the grounding metaphor, it is a ppropriate, in part, because there is determination. But, of course, there is causa l determination, so if grounding is supposed to capture something non-causal, we need to sa y why the grounding metaphor is particularly appropriate for non-causal explana tion. Here are two suggestions. When we think of causes and effects, we typically (though perhaps not always) think of them as arranged sequentially. In the normal cases at least, causes precede effects. This is not the case when it comes to non-causal explanations of the sort we have been consider ing. Somethings having a color, for example, is not preceded by its being red. The grounding suggest s the metaphor of levels Levels are not the sorts of things that are arranged sequentially. They are stacked, and some levels are closer to the ground floor than other. The grounding metaphor, th en, captures the non-se quential character of the explanandum and explanans in cases of IVO determination.

PAGE 42

42 Here is the second suggestion. The metaphor also suggests support. If one thing grounds another, it supports it. There is a sense in which something s being red supports its having a color. Having a color is not ontol ogically independent of having this or that determinate color. With respect to this kind of determination, the explanandum is not ontologically distinct from the explanans, unlike cases where determination is due to cause and effect. In brief, then, what the grounding metaphor captures about these cases of IVO determination is its non-sequential character and the ontological dependency that holds between explanandum and explanans. I should stress here that the explanandum and explanans are not, on this account, related by the IVO notion itself. It is not a relational no tion. They are related by IVO determination. 3.3 Worries about the Very Idea of Determination Why m ight someone be hesitant to accept the idea that there is such a thing as determination? Here are four reasons someone could offer: First worry Determination commits us to the view that there are necessities in nature. That is, it commits us to the view that there is a form of necessity that holds between entities that are not propositions. If, for example, event A determines event B then A necessitates B Nothing in the proposed general picture of dete rmination suggests that there is a worrisome sense in which there are necessary connec tions between non-propositions. The account acknowledges only a very modest sense in which B is necessitated when A determines B Necessity figures in explanations that introduce determination. If for example, there is a causal explanation of B in terms of A then there is a set of claims about A (and about the laws of nature, surrounding conditions, etc.) su ch that the claim that B occurred (or will occur) can be derived from this set of claims. This is necessity in form of entailment. None of this requires necessity between non-propositions. Laws of nature, causation, and IVO ar e all determination phenomena, and this is reflected in the commonalities the account notes. As noted above, the notion of

PAGE 43

43 determination, whatever it comes to, involves the notion of necessita tion, and the proposed account spells this out in term s of entailment relations betwee n propositions whenever laws, causation, or IVO are at work. The account does not identify determination with any sort of necessitation. Second worry A similar worry is as follows. The idea of determination is problematic for the same reason that the idea of an essence is problematic. The notion of essence is, in part, modal, for essential properties are necessary to the things that have th em. Necessity should be treated not as an entailment relation between pro positions but as an operator within the scope of which is a conditional. The worry about essences would be that it is impossible to capture the necessity that is supposed to accompany essences by treating that necessity as an operatorand that, for the same reason, it is impossible to think of the necessity that accompanies determination as properly expressed with an operato r. If x has property P essentially, then there is something modal about x which we capture wi th an adverb, necessarilyas in x has P necessarilyand of course we shou ld be skeptical about existen ce of such features. And we cannot simply say Necessarily, x has P in place of x has P necessarily. Although we have a necessary truth, and we ha ve removed the adverb that appeared to introduce a feature that must be had in a wa y that can be characterized only in modal terms, this account of essence forces us to admit too many prope rties as essential. We would have to admit that it essential to x that it be accompanied by the number 2. Without introducing some further

PAGE 44

44 condition which does not make a direct appeal to the notion of essenceso the objection would go--such an account will be too weak. This line of argument does not apply to the ca ses we have been considering, for we have already introduced a further condition that makes fo r determination rather than simple necessity. The further condition is that there be an explanatory notion at work that is distinct from that of determination. In brief, we have already done, w ith respect to determination due to causation, laws, and IVO, what this objection presupposes cannot be done with respect to essence. (In the last chapter, I discuss various ways in which one might try to make sense of the notion of essence by fitting it into this general picture.) Third worry Talk of determination is hopelessly metaphorical. When we try to say what we mean when we say that one thing determines another, we inevitably begin saying things to the effect that A makes B occur or that A brings about B But, of course, while a person (qua agent) can make another person do something, inanimate objects cannot. To talk as if an inanimate object (or even a non-agent) can force anothers hand invokes anthropomorphic imagery, and if we cannot eliminate the metaphor when discussing determination, we should be skeptical about the presence of determination in nature. Is talk of determination hopelessly metaphorical ? Let us begin with a particular form of determination. Suppose I say that A caused B When asked what I mean by caused, I begin to invoke metaphors such as A made B occur. But I could be doing more than one thing here. One the one hand, I could be attempting to subs titute expressions that are more-or-less synonymous with caused; on the other hand, I coul d be substituting expressions that I know are not synonymous with caused in an effort to di rect my listeners attention on something that is present when the expressions caused and makes apply. The objection either presupposes that

PAGE 45

45 I am attempting to appeal to synonymy or that I am drawing an analogy between causation and the agent-making features in cases of compulsion. But it is not clear that I am appealing to synonymy in this way. It is, I th ink, plausible that we can un derstand the substitution in the following way. I am carrying out the substitution in order to high light something that is present in cases of mere causation and cases of agency. Presumably, causation is present in both cases, and that is not what I am trying to highlight when I carry ou t the substitution. Nor is the thing I am trying to highlight something that makes for agency, which is what makes the makes locution appropriate when it is. What I am trying to highlight, according to this suggestion, is the presence of determination. The shift to metaphorical language helps me highlight the determination because it is obviously present in those cases. Cases of mere causation and cases of agency both involve determination, so we can highlight the presence of determination by drawing an analogy between mere causation and cas es of causation that i nvolve agency. If I make someone do something, then the person wa s in some sense compelled, and however we understand compulsion in this case, it should be understood in terms of determination. So, there is no presupposition here to the effect that m akes, brings about a nd the like are synonymous with causal talk. And when it comes to talk of determination itself, when we use makes and brings about we are using terms that apply to certain types of determination, inviting the listener to understand determination by example. Someone might argue that the term determina tion is like makes in that, strictly speaking, it applies only in cases where there is agen cy and that its application is metaphorical in other cases. We say things to the effect that the tennis player was determined to win. In this case, we mean something to the effect that sh e was compelled (from with in) to do everything in

PAGE 46

46 her power to win. If she does win, her being determ ined supplies at least a partial explanation of why she won. And, someone might say, when we sa y that one event determined another, we are invoking a metaphor that appeals to cases of psychological determination of this sort. Even if this is righteven if all literal talk of determination is tied to psychological compulsionthis does not show that there is not a non-psychological phenomenon common to all of the cases we have been considering. Talk of compulsion and so forth could introduce the notion indirectly. That is, compulsion, de spite not standing for this common notion, nevertheless manages to introduce it (pragmatically, perhaps) beca use of a conceptual connection it bears to compulsion. If this is right, then it is pl ausible that there are non-psychological terms which introduce the notion in th e same way. Talk of causati on, for example, may introduce it even if it does not stand for it. So, even if determination ha s a literal application only in psychological cases, there may still be vocabulary av ailable which does not directly stand for the notion but manages to introduce it because of the conceptual connections it bears to certain psychological notions. This would explain why we substitute expressions that smack of agency when we try to say what we mean by cause. There is somethi ng in cases we call mere causation that calls for the comparison with compulsion. That commonality is what I am calling determination. So, even if much of the language we use when we tr y to explicate causal talk in day-to-day terms is psychological, that does not exclude the pres ence of a non-psychological phenomenon in both psychological and non-psychological cases. That we can, plausibly, fit laws, causation, and IVO into a common framework suggests that there is such a notion.

PAGE 47

47 Fourth worry Determination does not carry any e xplanatory weight. There are other notionsfor example, that of a law of naturetha t explain why things are the way they are, and these notions can be characterized w ithout any mention of determination. Is determination explanatorily superfluous? In this chapter, I have noted common features to various styles of explanation. Whether A causes B there is a law, A that explains B or B in virtue of A A is a part of something that is sufficient for B (and in some case itself suffices for B ). I have been suggesting that these commonalities consist of phenomena that give us a better understanding of laws, causation, and IVO. In each case, we were able to identify these phenomena (e.g., the presence of sufficiency) wit hout making a direct appeal to explanation. This suggests that the commonalities are phenomena that make explanation possible in the first place. They are the phenomena that tell us how laws, causation, and IVO can be explanatory. The unified account helps us make sense of laws, etc. by situating them in a common framework. This framework, therefore, gives us a better unders tanding of other explanat ory notions (like that of a cause). To that extent, the framework I have supplied is not expl anatorily superfluous. Moreover, to the extent that th e framework lines up with our pretheoretic intuiti ons about what determines what, it makes sense of talk of determination. 3.4 In Virtue of and Necessitation There is on e more deflationary attitude one might adopt. I have sketched an account of the relation between such notions as that of determ ination, sufficiency, and IVO. According to this account, if A in virtue of B then B is, in some sense, a part of something that is sufficient for A Someone might suggest a differe nt account of the relation between IVO and necessity. In particular, someone might suggest that we spell out IVO in strictly modal terms. According to this account, to say, for example, that being red counts as having a color is simply to say that being red necessitates ( is suffici ent for) having a color. That is one might argue that what we

PAGE 48

48 have in mind when we say that being re d counts as having a color is the following generalization: Necessarily, for any x, if x is red, x has a color Such a simple necessitation account cannot wo rk, for there are obvious counterexamples. Consider the following claim: Necessarily, for any x, if x is self-ide ntical, x is accompanied by the number 2 This conditional is true, but it is not in line with our intuitions about in virtue of. That it is not is easily seen when we note that the following conditional is true as well: Necessarily, for any x, if x is accompanied by the number 2, x is self-identical In general, it seems that if A explains B then B does not explain A But entailment allows for cases where A entails B and B entails A So, entailment is insufficient for explanation. One might be tempted to refine the account by introducing asymmetric necessitation For example, one might say that something has F in virtue of having G just in case the following modal claim is true: Necessarily, for any x, if x is F then x is G ; and it is not the case that, necessarily, if x is G x is F This account is inadequate, as shown by the following true conditional: Necessarily, for any x, if x has a color and is six feet tall, x has a color Surely, the claim that x is colored and six feet tall entails the claim that x is colored, and the claim that x has a color does not entail the claim that x has a color and is six feet tall, so we have an instance of asymmetric necessitation. But having a co lor and being six feet tall does not explain having a color, so we do not want to say that x has a color in virtue of having a color and being six feet tall. The presence of this sort of case shows that we ca nnot understand in virtue of in terms of asymmetric necessitation.

PAGE 49

49 To address this sort of case, we might try to spel l out a notion of minimal sufficiency We might, for example, spell out minimal sufficiency in the following way, which is designed to rule out the conjunctive case we just considered. Suppose F minimally suffices for G just in case the following modal claim is true: Necessarily, for any x, if x is F then x is G ; and there is no pair < F* G* > such that x is F iff x is F* & G* ; and either (i) F = F* ; or (ii) F = G*. This formulation handles the counterexample to the previous formula tion, since it rules out the pair < having a color being six feet tall >, which contains a feature that is identical to the feature we are trying to explain. The account, however, is nevert heless inadequate. We can see this if we change the example slig htly. Suppose that x is red. We now have an example that is in line with the notion of minimal sufficiency just introduced. The claim that x is red and six feet tall entails the claim that x is colored, and neither being red nor being six feet tall is identical with having a color But the formulation is at odds with our intuitions about when to apply in virtue of, for we do not want to say that x is colored in virtue of being red and six feet tall. We might try replacing the identity claims F = F* and F = G with modal claims, giving us the following: Necessarily, for any x, if x is F then x is G ; and there is no pair < F* G* > such that x is F iff x is F* & G* ; and either (i) necessarily, if x is F* x is G ; or (ii) necessarily, if x is G* x is G. This form ulation handles the case we just considered. The pair < being red, being selfidentical> contains a feature that is sufficient for having a color but is not identical with it. But

PAGE 50

50 the account is too strong. Suppose that x is se lf-identical and that x is accompanied by the number 2. We have introduces two features, being self-identical and being accompanied by the number 2 There is also the further (conjunctive) feature, being self-identical and accompanied by the number 2. Nothing could lack any of these thr ee features, so having any one of them entails having the other two. If, for example, x is self-identical, x is self-identical and accompanied by the number 2. However, we also want to say that x is self-identical and accompanied by the number 2 in virtue of being self-identical (a nd in virtue of being accompanied by the number 2). So, we have a case where we want to say that one feature is had in virtue of another, despite the existence of a pair of the sort this formulation excludes. So, once again, we have a counter-example. Suppose we combine the notions of asymmetr ic necessitation and minimal sufficiency, giving us the following modal account of in virtue of: Necessarily, for any x, if x is F then x is G ; it is not the case that, necessarily, for any x, if x is G x is F ; and there is no pair < F* G* > such that x is F iff x is F* & G* ; and either (i) necessarily, if x is F* x is G ; or (ii) necessarily, if x is G* x is G. The example we just considered involves features that nothing could lack. This last account of IVO rules out that case because such features do not entail one anot her asymmetrically. However, the suggested account is too strong. If x is red, then x has a colora case of asymmetric necessitationand we want to say that x has a color in virtue of being red. But the current account prohibits us from saying this, for there is a pair of feat ures of the sort the formulation rules out. For ex ample, there is the pair < being accompanied by the number 2 being accompanied by a true proposition with content x is red>. This pair is always had by x

PAGE 51

51 whenever x is red, and the second member entails that x has a color. Despite the presence of this pair, we want to say that x has a color in virtue of being red. It may be possible to continue adding restri ctions in this way until we find a notion of minimal sufficiency that has the appropriate ex tension. But proceeding on a case-by-case basis in this way is beginning to look blatantly ad hoc As a result, it is di fficult to believe that minimal sufficiency is guiding our intuitions. We might try to look to the notion of releva nce entailment. The idea would be that to explain a claim, it is not enough to show that it is entailed by anot her claim. The claims must be relevant to each other. Setti ng aside trivial cases of relevance, su ch as x has a color and entails x has a color, we should say that there is an explanation just in case the one claim relevantly entails the other. So, x is red relevantly entails x ha s a color, and this su pplies an explanation. But notice that there are ready counter-examples to this approach. Suppose that something has a color. And suppose that God knows that it has a co lor. Then the claim that God knows that it has a color entails (relevantl y) the claim that it has a color. But the claim that God knows that something has a color does not explain why the object has a color. We would not say that something has a color because God knows it has a color. An account of explanati on in terms of idea of analytic entailment faces the same problem. It is analytic that, if God knows that x has a color, then x ha s a color. But, again, Gods knowing that something has a color does not explain the fact that it has a color. In general, then, it seems that modal notions are not rich enough to provide a unified treatment of grounding explanations.

PAGE 52

52 CHAPTER 4 TRUTH-MAKING, GROUNDING, AND ONT OLOGICAL COMMITMENT In Chapter 3, I argued that in virtue of and related expressions introduce a type of determination that is best underst ood in terms adverbial modification rather than relational terms. That is, instead of saying that the ball has a color in virtue of its being red, which treats in virtue of as a relational expression, we should (and in fact normally do) say that the ball is colored in virtue of being red (dropping the nominalization). Assuming that the use of the nominalizations in question carries ontological weight, not being forced to use them has philosophical implications. In this chapter, I argue that ou r ability to avoid the nominalizations in question puts us in a position to formulate what I will call the truth-maker principlea thesis which seems true but has proved difficult to endorse without incurring problematic commitments. In particular, the principle seems to commit us to nega tive states of affairs. I will begin this chapter with a brief discussion of the truth-maker prin ciple and the trouble it causes for truth-maker theory. Then I will argue that the considera tions in the last chapter point the way to a satisfactory formulation of the principle. Then I illustrate ho w this theory of truth-making explains negative truths and gene ral truths without introducing probl ematic state of affairs. In the final section, I discuss the relation betw een truth-making and modality, including some plausible supervenience claims. 4.1 The Truth-Maker Principle Consider th e proposition . W hy is it true? It is true, no doubt, because some dogs bite. Trivial as this response may s ound, it supplies a genuine explanation. To say that A makes B true is, of course, not to say that A causes B to be true. Rather, it is to say that B is true in virtue of A where in virtue of introduces a non-causal explanation (see Armstrong 2004, p. 5). The explanation is non-causal, since it appeals to a logically sufficient condition for

PAGE 53

53 the truth of the proposition, namely, the condition that some dogs bite. The logical sufficiency alone, however, does not supply the explana tion. The conjunctive condition that some dogs bite and some dogs are spotted also logically suffices for the truth of , but this condition does not explain the truth of the proposition. Some conditions, we might say, merely suffice for the truth of the proposition, while other conditions make the proposition true .1 Truth-maker theorists try to capture the notion of a truth-making condition with a generalization to the effect th at, if a proposition is true, there must be something that makes it true Call this the truth-maker principle. There are at least two motivations for thinking that a generalization to this effect is correct. One is th at it does some justice to the intuition that drives the correspondence theory of truth. If the truth-maker principle is correct as stated, then to any true proposition there corresponds an entity, namely, its truth-ma ker. The other motivation is that it does justice to the (related) intuition that truth is grounded, i.e., th e intuition that facts about truth are in some sense determined by other facts. As we will see, truth-maker theorists who accep t the principle in its unrestricted form encounter some difficulty in trying to explain every truth in terms of the existence of an entity. Explaining negative truths (e.g., th e truth that Mars is not perf ectly round) in this way may require introducing negative st ates of affairsentities that, one might think, have a peculiar structure. I argue that the presence of the difficult cases in question is not due to the generality of the truth-maker principle but to a misundersta nding of the notion of making something true. I argue that the in virtue of locution enables us to formulate the truthmaker principle in a way 1 Part of what makes a given pr oposition true is its being the very proposition it is. That is to say, assuming propositions have their content esse ntially, a propositions own nature is relevant to its being true. I will suppr ess this point throughout the chapter on the assumption that it is obvious.

PAGE 54

54 that frees us from having to introduce states of affairs when explaining truth (accept, of course, in cases where the proposition is about states of affairs). I take it to be a requirement on the success of this approach that we be able to form ulate the truth-maker prin ciple in a way that does not commit us to the use of sentence nominaliz ations in key placesnamely, in those places where the nominalizations purport to re fer to states of affairs. I ar gue that, since in virtue of is an adverbial modifier rather than a relational term, framing the tr uth-maker principle in terms of in virtue of frees us from havi ng to explain negative and general truths in terms of problematic entities. If this is correct, th ere is nothing in the truth-maker pr inciple itself that requires us to appeal to such entities in order to give an account of truth-making that satisfies the intuitions that drive truth-maker theory. Indeed, according to the proposed account, it is possible to give a theory of truth-making for negative and general truths without introducing truth-makers at all. Whether the account covers other difficult cases (e.g., modal truths) is not a question I will try to answer. 4.1.1 Truth-Making and Ontological Commitment Truth-m aker theorists consider various types of truths and candidate entities for the ontological grounding of those truths. Consider the relatively simple truth that Earth is round. What are the candidates for the truth-maker in this case? It seems that there are three potential candidates: the state of affairs of Earths being round the event of Earths being round and the trope Earths roundness and the fact that Earth is round. Suppose the truth-maker of the truth that Earth is round is a state of affairs, namely, Earths being round The ingredients in the state of affair s are the concrete pa rticular, Earth, the universal being round, and perhaps the instantiation relati on. The truth that Earth is round is made true by the existence of Earth s being round. That is, the truth that Earth is round is true in virtue of the existence of the state of affairs of Earth s being round

PAGE 55

55 According to this view, the existence of all th e concrete particulars a nd universals does not fix all of the truths. Which states of affairs exist is not fixed by the existen ce of all the particulars and universals. For a given state of affairs to exist, particulars and uni versals have to come together in a certain non-mereological form of composition. The st ate of affairs of Earths being round is not the mereological sum of Earth and roundness. They have not only to exist but to be related in the right way.2 We can contrast this view with the view that the truth-maker for Earth is round is a trope or Husserlian moment. The truth-maker, according to this view, is the entity that is, basically, the state of affairs of Earths being round minus its concrete particular Earth. Call this entity This roundness. The view says that there is a concrete particular, Ea rth, and there is the property this roundness, where this roundness is a particul ar, not a universal. This roundness could not exist without being a property of Earth On this view, the existence of this roundness guarantees the truth of Earth is round, since, if this roundness exists, it is the roundness of Earth. And, in general, fixing the existence of all of the tropes fixes the truth generally. Another possible view is that this roundness, even if it exists, is not necessarily a property of Earth, though it is necessarily a property of so mething or other in other words, it cannot exist on its own, but Earth is not the only thing that can ground its existence. According to this view, the existence of Earth, this roundness, and the instantiation rela tion does not secure this truth, because these things still have to come together in the right way in order for the proposition to be true. The most obvious candidates for truth-makers are facts. It is not clear that facts can be distinguished from some other candi dates (e.g., states of affairs), but I will not try to decide that 2 See Armstrong 1997, p. 115.

PAGE 56

56 issue here. I want to focus on Davidsons Sling shot Argument against the very existence of facts.3 He formulates his argument in terms of correspondence, and I argue that if correspondence is understood in terms of truth-ma king, there is an erro r his reasoning. In particular, I argue that if co rrespondence is understood in terms of truth-making, then we have reason to be suspicious of one of the pr inciples of substitution Davidson relies on. Here are the two principles of substitution on which Davidson relies: The principles are these: if an expression corresponds to a fact described by an expression of the form the fact that P, then it corresponds to the fact described by the fact that q provided either (1) the sentences that replace p and q are logically equivalent, or (2) p differs from q only in that a singular term has been replaced by a coextensive singular term. (Davidson 1969, p. 42) These principles play a role in his slingshot ar gument against the existence of facts (or against the view that there are facts in addition to the one Great Fact). Take the expression there was a short circu it. Suppose that there was a short circuit and that this expression refers to the fact that there was a short circuit. Then everything is such that there was a short circuit. So the re ferent of {x: x is a natural numb er and is such that there was a short circuit} is identical with the referent of { x: x is a natural number}, namely, the set of all natural numbers. So, we can derive a true identity claim, namely, (1) {x: x is a natural number and there was a s hort circuit} = {x: x is a natural number} And this identity claim is logically equivalent to there was a short circuit. So, by our fist principle of substitution we get the following: (2) The fact that there was a short circuit is identical with the fact that {x: x is a natural number and there was a short circuit} = {x: x is a natural number} Now take any true sentence, for example, Grass is green and substitute it for there was a short circuit in {x: x is a natural number and there was a short circuit. This gives us {x: x is a 3 See Davidson 1969.

PAGE 57

57 natural number and grass is green}. This expression refers to the set of all natura l numbers. So {x: x is a natural number and grass is green} and {x: x is a natural number and there was a short circuit} corefer. But if they corefer, we can substitute one for another in (2), giving us the following: (3) The fact that there was a short circuit is identical with the fact that {x: x is a natural number and grass is green} = {x: x is a natural number} But {x: x is a natural number and grass is gr een} = {x: x is a natural number} is logically equivalent to grass is green, so th e fact that grass is green is iden tical with the fact that {x: x is a natural number and grass is green } = {x: x is a na tural number}. But (3) says that this last fact is identical with the fact that th ere was a short circuit, so the fact that there was a short circuit is identical with the fact that grass is green. The fact that there was a short circuit and th e fact that grass is green were arbitrarily chosen facts. So, in general, for any two expressions that purport to refer to a fact, if they each refer to a fact, they refer to the very same fact. Therefore, if there are facts, there is only one fact. Davidson states the lessons to be learned from this argument: Since aside from matters of corresponden ce no way of distinguis hing facts has been proposed, and this test fails to uncover a single difference, we may read the result of our argument as showing that there is exactly one fa ct. Descriptions like the fact that there are stupas in Nepal, if they describe at all, describe the same thing: The Great Fact. No point remains in distinguishing among various names of the great fact when written after corresponds to; we may as well settle fo r the single phrase corr esponds to The Great Fact. (Davidson 1969, p. 42) Davidson takes the Slingshot Argument to s how that, without some criterion other than correspondence by which to individuat e facts, we should conclude that there are no facts at all. But if we think of corresponden ce in terms of truth-making, it seems that we have grounds for being suspicious of one of Davidsons principles of substitution.

PAGE 58

58 If expressions correspond to facts by being made true by them, then it is false to say that true sentences all correspond to the same fact. The first principle of substitution is suspect because it seems to overlook the difference between truth conditions and truth-making conditions. At least, it may be that the principle seems true only of we ignore the distinction between truth conditions a nd truth-making conditions. The slingshot argument overlooks the intuitively correct claim that logically equivalent expressions do not necessarily correspond to the same fact (again, if we understand correspondence in terms of truth-making). Davidson s argument is not sensitiv e to the distinction between truth conditions and trut h-making conditions, and this generates a problem with respect to explanatory relevance. The expressions there was a short ci rcuit and {x: x is a natural number and there was a short circ uit} = {x: x is a natural number } are logically equivalent, but there is no way that we can say that the latter ex pression refers to the tr uth-maker of There was a short circuit. That would be like saying that the fact that there was a short circuit and 1+1=2 makes true the claim that there was a short circ uit. To say this would be to confuse truth conditions with trut h-making conditions. Here is an analogy. Consider the following claims: (4) The capes being red cau sed the bulls charging. (5) The capes being red and the capes bei ng red and such that 1+1=2 are logically equivalent. (6) The capes being red and such that 1+1=2 caused the bulls charging. Clearly the move from (4) and (5) to (6) is illegitimate. The reason has to do with causal relevance. That one expression is logically equi valent to another does not justify substituting in all explanatory contexts So, the event of the capes being red has all the same implications as the

PAGE 59

59 capes being colored and such that 1+1=2 but the property of being such that 1+1=2 is not causally relevant to the bulls char ging. We can put this in terms of in virtue of. Let us say that the capes being red is my favorite fact. It follows th at my favorite fact caused the bulls charging. We can even say that the bull charged because of my favorite fact. But we cannot say that the bull was caused to charge by my favorite fact in virtue of its being my favorite fact. Rather, we say that the bull was caused to charge by my favorite fact in virtue of its being the capes being red. If truth-maker claims are in virtue of claims then we can put the issue as follows. We would not say that there was a shor t circuit is true in virtue of the fact that there was a short circuit and 1+1=2. Nor would we say that {x: x is a natural number and there was a short circuit = {x: x is a natural number is true (solely) in virtue of the fact that there was a short circuit. Again, the first is true in virtue of the fact that there was a short circuit, and the second is true on virtue of the fact that the set of everything that is a natural nu mber and such that there was a short circuit is identical with the set of everyt hing that is a natural number. The Slingshot Argument seems to overlook the truth of these in virtue of claims. There is, however, another reason for avoiding talk of states of affair s in a general account of truth-making. While there may be propositions whose truth is properly explained in terms of states of affairs (namely, the ones th at are about states of affairs), if we appeal to st ates of affairs as a rule, we are confronted with problematic cases. 4.1.2 Problematic Cases for the Truth-Maker We have noted an intuitive for mulation of the truth-maker principle to the effect that, for every truth, there is something that makes it tr ue. The principle can be spelled out more precisely as follows: For any proposition

, if

is true, there exists some entity that makes

true

PAGE 60

60 Consider the proposition . Does there exist an entity that makes this proposition true? If so, the entity in question is not Mars itself, since there is a possible world in which Mars is not round. The proposition is ma de true, one might say, not by the concrete particular, Mars, but a state of affairs namely, Mars being round I argued earlier that grounding claims do not, in every case, have to carry sufficiency in order to be true. When it comes to truth-making, however, sufficiency is requi red. A truth-maker is an entity such that a certain proposition is true solely in virtue of the existence of that entity. For this reason, a truthmaker is an entity whose existence is sufficient for the truth of a proposition. One has to admit that this account has some prima facie plausibility. But there are some truths that make it difficult for the truth-maker theo rist to endorse a genera l principle on the basis of this kind of example. Consider the negative truth . If there exists a state of affairs that makes the proposition true, it seems that it does not have an ordinary structure such as xs being F Therefore, someone who wants to ackno wledge such negative states of affairs owes us an account of their structure. Armstrong attempts to give such an account of the structure of problematic states of affairs.4 His strategy, in brief, is to give an accoun t of the truth-makers for general truths and argue that negative truths can be explained in terms of general tr uths. Consider the general truth . What is its truth-maker? Armstrong points out that even a complete list of mammals that ar e in the room and human will not guarantee that every mammal in the room is a human being. Suppose that a and b are the only mammals in the room and that both are human. The state of affairs of as being a mammal, in the room, and human and bs being a mammal, in the room, and human does not rule out the presence of more 4 See Armstrong 2002, pp. 34-5.

PAGE 61

61 mammals in the room (some of which are not human), so it does not guarantee that all the mammals in the room are human. Armstrong, therefor e, sees the need to introduce an additional state of affairs to secure the gene ral truth. He characterizes this st ate of affairs in terms of what he calls the Tot relation. This is a contingent rela tion that holds between aggregates and properties. The aggregate of mammals in the room totals the property of being a mammal and in the room This state of affairs is supposed to secu re the general truth that every mammal in the room is human. Once we have secured this truth, we have secured the negative truth that there are no mammals in the room that are not human. A potential worry for a picture of this sort is that, to the extent that it is not ad hoc it is circular. We can put the worry in the following way. Do we have a handle on the Tot relation independently of the role it plays in Armstrong s account (that of pointing to a truth-maker for negative and general truths)? If we do not, then to introduce the relation for the reasons Armstrong does is ad hoc On the other hand, if we do have an independent grasp of the Tot relation, it is very plausible that we are mistaken in thinking that this notion is anything other than the notion of a general state of affairs its elf. Armstrongs account of negative and general truths, then, is open to the charge that it is either ad hoc or circular, depending on the nature of our grasp of the Tot relation.5 5 A second strategy for handling the pr oblematic cases is to argue that we dont need to appeal to totality state of affairs in order to explain negative truths. One could argue that negative truths are explainable in terms of positive states of affairs that are not general. See Cheyne and Pigden (2006) for an explanation of negative truths al ong these lines. Suppose Theaetetus is sitting down on the surface of the earth. Sit ting down on the surface of the earth excludes (or is inconsistent with) flying, so at any worl d where the positive state of affairs of Theaetetus sitting down on the earth exists, the proposition is true (see especially Cheyne and Pigden 2006, pp. 258-9). See Molnar (2000, pp. 73-5) for a crit ical discussion of accounts of negative truths that rely on the notion of exclusion.

PAGE 62

62 One might argue that we can sidestep these i ssues by endorsing a restricted version of the truth-maker principle, one which claims that only certain truths have truth-makers. Bigelow, for example, claims that while every positive truth has a truth-maker, negative ones do not.6 One worry here is that it violates the intuitions that truth I grounde d. For it seems that the truth of negative propositions, such as , are every bit as determined by how things are as any positive truth. Without a more detailed account of the truth of negative propositions, then, such a restriction on the truth-maker principle appears ad hoc These critical remarks may or may not be fata l to the accounts of truth-making discussed so far. I will not try to determin e that here. I want to focus on another approach to dealing with problematic states of affairs. The adverbial char acter of in virtue enables us to formulate the truth-maker principle in a form th at is both unrestricted and able to handle the problematic cases above. We began by considering an intuitive principle to the effect that if a proposition is true, there must be something that makes it true The truth-maker theorist typically spells this claim out by saying that the proposition is true in virtue of the existence of a state of affairs. This is supposed to capture both the correspondence intu ition and the intuition that truth is grounded. The adverbial reading of in virtue of claims enables us to appeal onl y to the existence of commonplace objects (such as properties and concrete particulars). Their existence alone is not, of course, supposed to be sufficient for the tr uth of the propositions in question. But the complete explanation for the truth of a given pro position will say that it is true in virtue of where is a claim whose truth suffices for the trut h of the proposition. Rather than being a claim to the effect that a state of affairs exists, says that certain objects exist and that they have 6 See Bigelow 1988, pp. 131-2

PAGE 63

63 certain features. As a result, there is nothing in the proposed formulation of the truth-maker principle to suggest that we need to introduce states of affairs to explain truth. 4.2 The Proposed Formulation Consider the truth . Why is it tr ue? Intuitively speaki ng, it is true because Mars exists and Mars is round. Suppose, in accordance with this intuition, one were to avoid nominalizations by endorsing th e following formulation: For any proposition

, if

is true, there is some entity such that

is true because that entity exists According to this thesis, instead of explaining the truth of by saying that the existence of Marss being round makes it true that Mars is r ound, or that is true in virtue of the existence of Mars being round, we should say that is true because Mars is round. Whereas the fi rst two claims contain sentence nominalizations, the third claim does not. As a result, states of affairs are not introduced.7 The claim says that the proposition is true because a there is an object wi th a certain feature. As a strategy for capturing the truth-maker principle, however, appealing to because will not work. We have already seen that it is ambiguous and does not introduce just the right notion.8 I suggest that we stay with the in virtue of locution and try to take advantage of its adverbial character. Were we to give a typica l truth-maker explanation along these lines, we could say that the proposition is true in virtue of the state of affairs of Mars being round. We 7 Hornsby (2005, pp. 34-9) discusses the commitm ents incurred by the use of because as opposed to other locutions. 8 This line of criticism also applies to accounts which say th at for something to make a proposition true is for there to be some explanation or other for the truth of the proposition. See Liggins (2005, pp. 111-5) for such an account. No te, however, that the current remarks are not aimed at those theories which enable us to av oid nominalizations by treat ing makes true as a sentential connective, as in a is red makes true the se ntence There is a colour a and b share. For an example of such a theo ry, see Melia (2005, pp. 82-3).

PAGE 64

64 want to avoid states of affairs, so we will say instead that is true in virtue of the existence of something else. We saw that it is not true in virtue of the existence of Mars, since the existence of Mars does not secure the truth of the proposition. Nor does the existence of Mars together with the ex istence of the property of being round, since there are possible worlds where Mars and the property of being round exist but Mars does not instantiate that property. What we need to say here is that is true in virt ue of the existence of something that is both Mars and round. Of c ourse, the existence of is a no minalization we want to avoid. But since in virtue of is adverbial, we can avoid this nominalization and say that is true in virtue of co-existing with something that is bot h Mars and round. Equivalently, is true in virtue of being accompanied by something that is both Mars and round. Whereas because is followed by a sentence, in virtue of in this case is followed by a predicate that contains a referr ing expression, namely, the indefin ite description something that is both Mars and round. If the pr oposition is accompanied at a world by an entity that meets this description, the proposition is true at that world.9 We can explain negative truths in basically the same way. Consider th e truth . This propos ition is true in virtue of being unaccompanied by something that is both Mars and blue. Let us abstract away from these particular examples and look at a version of the truth-maker principle. We can st ate the truth-maker prin ciple in various ways, 9 For entity x to be accompanied by entity y, it may be necessary for x and y to be wholly distinct We see this restriction in Langton a nd Lewis (1983, p. 333). They stipulate that something is accompanied iff it co-exists with an object from which it is wholly distinct. Understanding TM in terms of this notion of accompaniment may place certain restrictions on what kinds of entities propositions might be. If for example, the proposition is a complex entity with Mars as a constituent, then the proposition cannot be accompanied by Mars, since the proposition is not wh olly distinct from it. If this is a problem, however, we can understand TM in terms of accompaniment as s imple co-existence, a relation which doesnt require its terms to be wholly distinct.

PAGE 65

65 depending on our commitments. Though it is not e ssential to the principle, I will appeal to possible worlds to facilitate the discussion. I suggest that we formulat e a simple truth-maker principle as follows (TM) For any possible world w and proposition

, if

is true at w then there is a property P such that i) there exists at w an entity x such that x has P and

is true at w in virtue of being accompanied by an x such that x has P ; or ii) there is no x at w such that x has P and

is true at w in virtue of being unaccompanied by an x such that x has P. TM commits us to propositions, properties, and (as formulated) possible worlds. But these are not things it would make sense to call trut h-makers, since their mere existence does not guarantee the truth of the right propositions to play the truth-maker role. In fact, TM does not require the existence of truth-make rs at all. Nevertheless, it capt ures the basic idea behind talk of truth-making. It connects the truth of propositio ns with the existence of entities, and the connection is an explanatory one due to the occurrence of the in virtue of locution. In formulating this general statement of TM, I qua ntify over properties. Th is meant to introduce nary relations as well. But it is important to see th at particular examples can conform to it without themselves quantifying over propert ies. Earlier, we explained the truth of some particular propositions by using the expressions is Mars, is round, and the like. These were not introduced as names for properties. Rather, they were used as predicates. As far as the examples go, there is nothing that requires us to introduce properties. But when we abstract away from these examples and try, by introducing TM, to sa y something about the truth of propositions in general, we end up quantifying over properties. Someone who wants to avoid commitment to properties in general might still accept the general spirit of TM w ithout endorsing TM as stated.

PAGE 66

66 Such a theorist can say that, while we can explain the truth of propositions on a case-by-case basis with the proposed theory, we cannot say wh at makes propositions true in general, since doing so requires us to introduce problematic entiti es. Were one to take the latter route, the proposed truth-maker principle woul d best be seen as a useful formula for explaining truth on a case-by-case basis. I will now illustrate how the principle works by applying it to a variety of cases. Let us begin with the comparatively unproblematic trut h we have already looke d at. Suppose is true. TM explains its truth in the actual world. In the actual world, Mars is round, so condition i) is satisfied. That is, there is a (conjunctive) property ( being Mars and round) such that, in the actual world, there is an x with that property, and the proposition is true in virtue of co-existing with an entity which has that prope rty. Given TM, there is no need to introduce a state of affairs to explain the truth of the proposition. There is Ma rs and the property of being Mars and round. And since, according to the descripti on introduced, Mars has the property in question, we have an explanation of the truth of the proposition. Now consider a negative truth. Suppose is true at world w Then TMs condition ii) is satisfi ed. There is a conjunctive property P being Mars and blue such that there is, at w no x such that x has that propert y, and is true, at w in virtue of being unaccompanied by an x such that x has that property. Given TM, there is no need to introduce a negative state of affair s to explain the truth of the proposition. It captures an intuitive explanation, namely, that the proposition is true because, whether or not Mars exists, there is nothing that is both Mars and blue. I should note that Lewis (2003) relies on the no tion of accompaniment to spell out a theory of truth-making in terms of his counterpart theor y. The view he advances is supposed to explain

PAGE 67

67 the truth of propositions without ap pealing to states of affairs. The theory he develops, however, is different from the one I develop here. The crucial difference is that his theory does not introduce in virtue of style explanations in key places. In brief, his theory is as follows. Consider a cat named Long. We can explain th e truth of in terms of the existence of a certain entity, namely, Long qua black The expression Lo ng qua black does not introduce a state of affairs. A ll it introduces is Long, Longs black counterparts, and property of being black. What counts as a counterpart for a give n object, of course, depends on the aspect under which we think of the object. The expres sion Long qua black refers to Long, but when we think of Long under the aspect of being black the set of counterpart s is not all of Longs counterparts but only the bl ack ones. Therefore, in every world containing Long qua black the proposition is true (see Lewis 2003, pp. 30-2) Lewis and Rosen (2003, p. 4) argue that we can explain negative truths in basi cally the same way. What makes the restricted negative existential true is the room (including all of its contents) qua unaccompanied by unicorns Similarly, what makes the unrestricted negative existential true is the world qua unaccompanied by unicorns The qua locution does not play the role in Lewis account that in virtue of plays in the account I develop. According to Lewis, Long qua black is a truth-maker for , but so is Long qua as he is Long qua as he is is Long qua black and hungry and loud and etc The existence of Long qua as he is certainly suffices for the trut h of the proposition, but to say that Long qua as he is explains why is tr ue is analogous to saying that the conjunctive condition that some dogs bite and some dogs are spotted explains the truth of . As we noted earli er, this conjunctive condition suffices for the truth of the proposition, but it does not explain it. By the same reasoning, we should not say that Long qua

PAGE 68

68 as he is explains the truth of , so the qua locution is not being used by Lewis in the way that I am using in virtue of. We should note, however, that Lewis (2003, pp. 28-9) claims that the notion of truth-ma king is explanatorily very thin, so he perhaps would not see this as a problem. Returning to the problematic cases, general trut hs are a bit more complicated. Consider a world w at which everything that is red is small. The proposition is, of course, true at that world. We can explain the truth of this proposition in terms of TMs condition ii). If at w everything that is red is small, then we can make the following claim: is true at w in virtue of being unaccompanied by an x such that x is red and x is not small Given the truth of this claim, TMs condition ii) is satisfied. There is a conjunctive property being red and not small such that there is nothing at w which has that property, and the proposition is true at w in virtue of being unaccompanied by something that has that property. Of course, one might be unhappy about accepting pr operties of this sort. If one adopts a rather sparse conception of properties, one might balk at such propert ies as the property of being not small or conjunctive properties involving them. As noted earlier, however, one might endorse the spirit of TM and avoi d commitment to properties in gene ral, so TM could still be of use to someone with such scruples. Moreover, there are other ways of accepting TM without accepting those particular properties even if one accep ts properties in general. For instance, one could complicate the truth-maker principle in a wa y that eliminates reference to the problematic properties in particular. To e liminate reference to negative prope rties, for example, one could add a third condition to the effect that there is a property pairP, P*such there is no x at w that has P and does not have P* and

is true at w in virtue of being unaccompanied by an x such that x has P and x does not have P* For example, the truth of
PAGE 69

69 small> can perhaps be explained in terms of the property pair being red being small If there is nothing at world w which has the first of the pair ( being red ) and does not have the second ( being small ), then the proposition is true in virtue of being unaccompanied by something which has the property of being red and does not have the property of being small In this way, instead of saying that there is nothing w ith the negative property of being red and not small we say that there is nothing that has the property of being red and does not have the property of being small The case Armstrong discusses is simply a (ver bally) more complicated example of this sort, and it can be handled in exactly the same way. Suppose is true. Then condition ii) is satisfied. There is a property being a mammal in this room and not human such that there is nothing which has that property, and the proposition is true in virtue of being unaccompanied by something which has that property. I am assuming that general propositions should be read as conditionals. For example, should be read as the conditional . This does not commit us to the existence of any red things. If we do not understand the proposition in this way, and see it as having existential import, th en we should see it as the complex proposition . TM will then apply to each of the propositions constitu ents as stated in the text. TM accounts for existentially quantified tr uths in the following way. Suppose the proposition is true. Th ere is a property P being a planet such that there is an x which has P and is true in virtue of being accompanied by an x which has P The truth is explained, in ot her words, in terms of the satisfaction of TMs condition i). Now consider the (unrestricted) negative existential . Suppose it is true at world w Then TMs condition ii) is satisfied. There is a

PAGE 70

70 property P being a unicorn such that there is, at w no x such that x has P and is true, at w in virtue of being unaccompanied by an x such that x has P Once again, given TM, there is no need to in troduce a negative state of affair s to explain the truth of the proposition. Now consider the restricted negative existential . Suppose it is true. Its truth is explained in terms of the satisfaction of TMs condition ii). There is a property P being a unicorn and in this room such that there is no x which has P and is true in virtue of being unaccompanied by an x which has P 4.3 Truth-Making and Supervenience I will now d iscuss the connection between truth-making and supervenience. Someone might try to avoid commitment to the problematic entities by appealing to s upervenience. I will discuss two such proposals. If truth is grounded, then th ere is something upon which truth supervenes. This is something any plausible theory of truth-maki ng will capture. Some theories of truth-making make central use of the notion of supervenien ce, either by defining truth-making in terms of supervenience or by weakening the truth-maker pr inciple by appealing to supervenience instead of the idea of a truth-making entity. We will look briefly at two theories that make central use of supervenience. Then we will see where the notion of supervenience fits into the proposed account. The proposed account does not make central use of the notion of supervenience, but it nevertheless explains the truth of certain plausible supervenience claims. Parsons (2006) spells out what he thinks is a workable th eory of truth-making within a nominalist framework. His account of truth-making introduces neit her properties nor states of affairs. The basic picture is this. Suppose the sent ence The rose is red is true. We can see the rose as the sentences truth-maker if we appeal to two notions: that of supervenience and that of

PAGE 71

71 a things nature Here is a version of the truth-maker principle he supplies when spelling out the notion of truth-making: For every true sentence, there is some thi ng such that the sentence cannot become false without a qualitative change, a non-Cambridge change, in that thing. (Parsons 2006, p. 327) The supervenience thesis is that there can be no change with respect to truth without some qualitative change in some thi ng. A qualitative change is a cha nge in an objects nature (the latter being the conjunction of its in trinsic properties at a given time).10 A things nature, according to this account, is something it has contingently (see Parsons 2006, p. 326). As a result, something can be the truth-maker for a sent ence at one time and not at another, even if the object still exists (see Parsons 2006, p. 328). The rose, for example, could cease to be a truthmaker for The rose is red by turning brown. A ppealing to supervenience in this way, Parsons thinks, makes it possible to give a theory of trut h-making that does not require states of affairs. John Bigelow (1988) does not defi ne truth-making itself in terms of supervenience, but he formulates the truth-maker principle in terms of supervenience in order to avoid problematic states of affairs. According to his account, th e correct version of the truth-maker principle does not say that for every truth there is a truth-maker. Rather, it sa ys only that truth supervenes on existence. Bigelow formulates the truth-maker principle as follows: If something is true, then it would be impossi ble for it to be false unless either certain things were to exist which dont, or else ce rtain things had not existed which do. (Bigelow 1988, p. 133) Bigelows thesis, I take it, is equivalent to the following: (S1) For any possible world w and proposition

, if

is true at w then, for any world w* such that

is false at w* i) there exists at w* an entity x such that x does not exist at w ; or 10 Parsons does not attempt to avoid property ta lk in spelling out his nominalist account.

PAGE 72

72 ii) there exists at w an entity x such that x does not exist at w* What the principle says, in brief, is that there can be no differen ce with respect to truth without a difference with respect to what exists. This holds for both positive and negative truths, so this version of the truth-maker principle may supply a way of discussing the difficult cases without introducing troublesome states of affairs. Moreover, S1 does some justice to the correspondence intuition. Although it does not say that there is, for every true proposition, an entity such that the proposition is true in virtue of th e existence of that entity, it does say that any difference in the distribution of truth values requires some other difference in the world. The problem with theories of the truth-making in which supervenience plays a central role is that, even if they free truth-maker theory fr om some troublesome commitments, they do so at a very high cost. Truth-maker theory is supposed to capture the intuition that the facts about truth are explained by various other facts, so a theory should tell us something to the effect that the facts about what exists determine or ground the facts about truth. The standard versions of the truth-maker principle capture th is intuition by saying, for exampl e, that every proposition is made true by a state of affairs or that every proposition is true in virtue of the existence of a state of affairs. But there are good reasons for thi nking that supervenience does not have the explanatory force of in virtue of and making. Indeed, one could go so far as to say that the supervenience of truth upon somethi ng else is interesting only in so far as it is evidence for an explanatory relation. But it is not itself explanatory. Rather, it is a co-variation relation between sets of properties (and sets of concepts, events, et c.) which is itself in need of an explanation.11 Nevertheless, S1or something close to itshould s eem correct to anyone who thinks there is a plausible version of the truth-maker principle. In general, the truth of the grounding 11 For a discussion of supervenience and explanation, see Kim (1974, pp. 147-8).

PAGE 73

73 claims to the effect that A in virtue B secures modal claims to the effect that something is sufficient for A If, for example, the car is red in vi rtue of having a red body, then the cars having a red body is sufficient for th e cars being red. And if proposit ion

is true in virtue of being accompanied by an x such that x has P then

s being true is sufficient for its being so accompanied. Is S1 an appropriate supervenience thesis? It is if we accept positive states of affairs, but not if we do not. Part of the significan ce of TM is that it does not commit us to negative states of affairs and general states of affairs; but it leaves th e question open as to whether there are positive states of affairs. If we assume that there are not even positive states of affairs, then we have to say that TM allows for a proposition to be true at one world and false at another even if the two worlds contain the same existents. Consider two worlds, w and w* Mars exists at both worlds. So does the property of being round. But at w is true, while at w* it is false. Despite the fact that the worlds contain the same existents, then, the proposition is true only at one of the worlds. So, if we assume that there are no positive states of affairs, we have to say that TM does not se cure the truth of S1. This is not a problem for TM, however, because if there are no positive states of affairs, S1, as formulated, is not plausible in the first place. If, on the other hand, we assume that there are positive states of affairs, then we should say that TM secures S1. If there are positive states of affairs, then, at w where is true, there exists the state of affairs of Mars being round. And at any world where is false, either there is no Mars or there is but it is not round. In either case, there is no state of affairs of Mars being round so the falsity of at a world requires a difference between what exists at that world and what exists at some world where the proposition is true. But notice that admitting positive states of affairs would not require us to reject the

PAGE 74

74 proposed explanation of the truth of . We can still say that there is a property P being roundand an xMarssuch that x has P and that is true in virtue of being accompanied by an x with P Introducing positive states of affairs secures S1, but that does not imply that we have to appeal to states of affairs in order to explain the truth of the proposition. As far as the proposed account goes, we can e xplain truth even if we suppose that there are not even positive states of affairs. Then we have to say that S1 is not the appropriate supervenience thesis. But in that case we can say that TM secures a supervenience claim in the vicinity of S1a claim we should, in fact, accept whether or not we think S1 is true. We can spell out this supervenience thesis as follows: (S2) For any possible world w and proposition

, if

is true at w then, for any world w* such that

is false at w* i) there exists at w an entity x such that x does not exist at w* ; or ii) there is some property P such that x has P at w but lacks P at w* ; or iii) there exists at w* an entity x such that x does not exist at w ; or iv) there is some property P such that x has P at w* but lacks P at w According to S1, truth supervenes on being, i.e., the facts about which propositions are true supervene on the facts about which things ex ist. According to S2, truth supervenes on which things exist together with what features they have So, even if we say that TM does not secure the supervenience of truth upon be ing, we can still say that it s ecures a plausible supervenience thesis. And, again, we want to accept S2 whet her or not we are committed to S1for, if truth supervenes (non-trivially) on anyt hing, then of course it supervenes on which things exist together with the features they have. And the supervenience of truth upon facts about which states of affairs exist does not rule out the supervenience of truth upon some more fundamental

PAGE 75

75 set of facts (in this case, the facts about the existence of the constituents of states of affairs and what properties they have). We can demonstrate that the trut h of TM secures the truth of S2. Consider any proposition

that is true at a world w Given TM, there are two possibilitie s. Suppose TMs first clause is true. Then proposition

is true in virtue of being accompanied by an x such that x has P Since truth-making is grounding that requires sufficien cy, it follows from this in virtue of claim that

s being so accompanied is sufficient for

s being true. In other words, we get the following bridge principle connecting the truth of TM with the truth of S2: (BP1) For any proposition

, if

is true at some world w in virtue of being accompanied by an x such that x has P then, for any world w* such that

is accompanied by an x such that x has P

is true at w* It follows from BP1and, therefore, from TMthat, at any world at which

is false, it is not the case that

is accompanied by an x such that x has P At any world at which

is false, then, either x does not exist at that world or x lacks P at that world. That is if TMs i) is true, then either S2s i) or S2s ii) is true, which implies that S1 is true if TM is true. As an example, consider the proposition . Suppose that it is true at some world in virtue of being accompanied by an x such that x is Mars and is round. Then, given BP1, at any world where is accompanied by an x such that x is Mars and is round, is true. Therefore, at a ny world at which is false, either Mars does not exist at that world or Mars exists but is not round. Now consider the second possibility. That is suppose TMs second clause is true. Then

is true in virtue of being unaccompanied by an x with P It follows that

s being so unaccompanied is sufficient for the truth of

. It follows, in other words, that BP2 is true: (BP2) For any proposition

, if

is true at some world w in virtue of being unaccompanied by an x such that x has P then, for any world w* such that

is unaccompanied by an x such that x has P

is true at w*

PAGE 76

76 It follows from BP2and, therefore, from TMthat, at any world at which

is false,

is accompanied by an x such that x has P At any world at which

is false, then, either

is accompanied by an entity by which it is not accomp anied at the world where

is true, or

is accompanied by a certain entity which exists at both worlds and that entity has some property P at the world where

is false which it lacks at the world where

is true. That is, if TMs ii) is true, then either S2s iii) is true or S2s iv) is true, which implies that S2 is true if TM is true. We can illustrate this type of case as follows. S uppose is true at a world in virtue of being unaccompanied by an x such that x is Mars and is round. At some worlds, the proposition will be so unaccomp anied in virtue of being unaccompanied by an x such that x is Mars. At other worlds, it will be so unaccompanied in virtue of being unaccompanied by an x such that x is both Mars and round (even though it is accompanied by Mars). Thus, depending on how things are at the wo rld where is true, at a world where the proposition is false, either there exists at that world an entity (Mars) that does not exist at the world where the proposition is true or there is an entity (Mars) which exists at both worl ds but has a property ( being round) at the world where the proposition is false which it lacks at the world where the proposition is true.

PAGE 77

77 CHAPTER 5 THE PARADOX OF ANALYSIS There is som ething that philosophers do when they attempt to answer the questions that interest them. Traditionally this has been calle d analysis. Sometimes it is called conceptual analysis. Whether philosophers do conceptual analysis is not something we need to decide here. Explicit philosophical analysis sometimes pr esents itself in the form of claims to the following effect: (A) x has P iff ____ (____ is supposed to be filled in with a set ne cessary and sufficient conditions for any xs having P.) For example, someone might offer the following analysis of knowledge: For any x, x knows that P iff x has a suitably justified true belief that P. An analysis is a type of grounding claim. It is a claim to the effect that some concepts (or properties) are in some sense prior to, or more basic than, other concepts (or properties). The notion of priority commits us to saying that analys is is asymmetric and irreflexive. That is, if A is the analysis of B then B is not the analysis of A ; and A is not the analysis of A Sometimes analysis claims are expresse d as identities, as in the following: The concept of knowing that P is identical with the concept of having a suitably justified true belief that P Like the bi-conditional formulation, this formulation does not capture the asymmetric character of analysis. One way of making the asymmetric character of analyses explicit is to makes a direct appeal to talk of analysis, as in the following claim: Having a suitably justified true belief that P is the analysis of knowing that P As we will see, all three of these formulati ons incur commitments which generate various puzzles about analysis.

PAGE 78

78 Before spelling out the paradox, I want to settle some terminological issues. When I speak of an analysis claim I mean a claim that invokes the notion of analysis. Analysis claims can be true, and they can be false. They can also be incomplete. If I say that being unmarried is part of the analysis of being a bachelor I am making an analysis cl aim which, though true, is not a complete analysis. If I say that being good looking is part of the analysis of being a bachelor I am making an analysis claim that is fals e. On the other hand, when I speak of the analysis I mean the one and only analysis claim which is both true and complete. And when I speak of an analysis I mean a claim which is the analysis of some notion. 5.1 An Epistemic Puzzle about Analysis Som etimes the expression paradox of analysis used to refer to a puzzle that an instance of, or at least analogous to, Freges puzzle. For our purposes, we can understand this puzzle in the following way. The following principle of substitution is prima facie plausible: For any expressions E1 and E2, if E1 and E2 express, respectively, the analysandum and the analysans of an analysis, then E2 can (in any non-quotational context) be substituted for E1 salva veritate. In light of this principle, cons ider the following proposed analysis: The concept of a brother is identical with the concept of a male sibling. The expression the concept of a male sibling expresses the analysans of the concept expressed by the concept of a brother, so, by substituti on, we can see the proposed analysis as being equivalent to the following: The concept of a brother is identical with the concept of a brother. Given the ability to substitute, the above two claims are equivalent with re spect to their content, so, given that the second claim is not potentially informative, the first is not potentially

PAGE 79

79 informative either. So, it looks as if it is impossibl e for an analysis to be informative. This is the way Moore, for example, characterizes the paradox.1 5.2 Various Suggested Solutions to the Puzzle Frege was not concerned with the paradox of anal ysis, of course, but his reflections on the nature of identity claim s bear on the paradox if it is understood as an instance of Freges puzzle. In Sense and Reference Freges solution is to apply the notion of a sense, or a mode of presentation.2 The idea is that we associate with expr essions like the morning star and the evening star not just a referent but also a way of presenting the referent. Thus we associate with these expressions descriptions such as the brig htest object visible for a short time near the horizon right after dawn and the brightest object visible for a short time near the horizon in the evening. These descriptions express th e sense of the morning star and the evening star respectively. What we have disc overed when we come to learn th e truth of The morning star is the evening star, then, is expressed in this claim: The brightest object visible for a short time near the horizon right after dawn is the brightest object visible for a short time near the horizon in the evening. The morning star is the eveni ng star has different cognitive signi ficance than The morning star is the morning star, then, because the former involves two different modes of presentation of Venus while the latter involves onl y one mode of presentation. What we learn when we discover that the morn ing star is the evening star is that two modes of presentation present the same object. Thus The morning star is the evening star expresses an astronomical discovery in that the statement is actually about Venus (and not the 1 See Moore 1942, p. 665. 2 See Frege 1982, p. 57.

PAGE 80

80 expressions the morning star and the eveni ng star). The fact discovered about Venus is mediated by the sense of the expressions we use to refer to Venus. Moore argues that analyses must be in some sense about words.3 He does not think that they are merely about words, since it is possible to know that two expressi ons stand in a relation (such as synonymy) without learning anything about the concepts they introduce, and Moore thinks analyses clearly tell us about concepts. On the othe r hand, an analysis cannot be a conjunction of two claims, one about words and one a bout concepts, since there is no reasonable candidate for the claim about concepts. Presumab ly, the point he is making is that the claim about concepts cannot be the original analysis clai m, The concept of a br other is the concept of a male sibling, and that there other reasonable ca ndidates. Nevertheless, he thinks, there must be some sense in which analyses are about both words and concepts. Cobb discusses, though rejects, a related solution.4 This solution distinguishes between primary and secondary information. The idea is that The concept of a brother is the concept of a male sibling and The concept of a brother is the concept of a brother convey the same primary information, which is an identity. So, they ar e equivalent in that respect. But they are neve rtheless informative because th ey impart different secondary information. The first imparts the in formation contained in this claim: The expression being a brother means the sa me in English as the expression being a male sibling. 3 See Moore 1942, pp. 665-6. 4 See Cobb 2001, pp. 421-2.

PAGE 81

81 The second imparts information c ontained in the following claim: The expression being a brother means the sa me in English as the expression being a brother. In this way, analyses are about both words and concepts and are informative despite our ability to substitute the re levant expressions. Fumerton discusses the paradox in the fo rm of Moores open question argument.5 He characterizes that argument as follows: 1. If x is good just means x is F then the question Is what is F good? has the same meaning as Is what q is F, F? 2. But these two questions obviously do not have the same meaning. Proof? One is clearly significant (requires deliberation to answer), the other, trivial (the answer to it is immediately obvious). Fumerton thinks this is a terr ible argument. The proof in step 2, he thinks, involves a nonsequator. That two statements differ in significance does not require them to differ in meaning, so the fact that a proposed analys is is equivalent to a tautology is not a reason for rejecting it. The basic idea is that if an analys is is correct, then of course it is equivalent to (the same in meaning as) a tautology, but this does not imply th at it does not convey information that is not contained in the meaning. Something these approaches to the paradox have in common is that they all grant that analyses are equivalent to some thing trivial and argue that they can nevertheless be informative because the information conveyed by the analysis cl aim is not part of its content. Since these responses grant that analyses are equivalent to tautologies, they do not treat them as grounding claims. If A is the analysis of B is equivalent to A is the analysis of A, then it cannot be a grounding claim, since something cannot ground itself. There are, however, suggested solutions 5 See Fumerton 1983, pp. 477-9.

PAGE 82

82 to the paradox that do not admit that analyses are equivalent to tautologies. This is the strategy I will pursue. I argue below that there is a distinct puzzle about grounding claims. The solutions we have looked at do not address this puzzle. I argue that if we solve this second puzzlea puzzle about how analyses could claims about groundingwe automatically have a solution to the first, epistemic puzzle. Before discussing that puzzle, I will say a bit about a second sort of response to the initial puzzle. Max Black challenges the view that analysis claims are identity claims, and in so doing challenges the view that analyses ar e equivalent to something trivial.6 As we will see, there are versions of the paradox that do not treat analyses as identity claims. Blacks suggestion is, nevertheless, instructive. He argues as follows. If analyses are claims to the effect that two nominally distinct concepts are the very same concept, then The concept of a brother is identical with the concept of a male sibling, is equivalent to, The concept of a brother is identical with the concept of a brother. And if this is right, the proposed analysis of being a brother fails. Black challenges the assumption that analyses are identity clai ms with an analogy. Consider the following mathematical claim: 21 = 3 x 7 If this were an identity claim, then we could subs titute for x 7 and preserve not just truth but the claim itself, giving us the following: 21 = 21 6 See Black 1944, pp. 263-267.

PAGE 83

83 This claim is true, but it is not the same claim as = 3x7. The reason is that the expression x 7 is not simply another name for the numbe r twenty-one. The claim introduces a relation which holds between three numbers, as he puts it, whenever the fi rst is the produc t of the other two. Thus, the logical form of the mathematical claim is not a=b, but, A (x, y, z), where A introduces a thr ee-term relation, not the two-term relation of identity. We should, he says, understand analysis clai ms by analogy. To say that being a male sibling is the analysis of being a brother is not to say that two no minally distinct concepts are the same concept but that there is a three-term relation between the concep t of a brother, the con cept of a male, and the concept of a sibling. In effect the expression being a male sib ling has been removed from the statement of the analysis. We say that the concep t of a brother is analyzed not in terms of being a male sibling but in terms of being a male and being a sibling Black looks for an intuitive characterization of the relation introduced by analysis claims. He suggests that talk of conjuncti on may be help (though he is not sa tisfied with this suggestion). He says that the relation A holds whenever concept x is the co njunction of concepts y and z. In a critical response to Black, Mo rton White (1945, p. 360) points out that is here indicates identity, which implies that A holds whenever concept x is identical with something (the conjunction of y and z). So, if we try to spell out A in terms of conjunction, we seem to be characterizing analysis in terms of identity. As we will see, the inadvertent appeal to identity is not central to Blacks basic suggestion.

PAGE 84

84 Some argue that we should spell out such a view in terms of structured propositions. Jeffrey King (1998), for example, thinks of pr oposition as complex entities whose structure should be spelled out in terms of properties, relations, and particulars, on the one hand, and lexical items and the (semantic) relations they sta nd in to their semantic values, on the other. An analysis, he thinks, is a proposition consisti ng of a complex entity, namely, a property (say, being a bachelor) and a second complex entity that represents that property and has among its constituents the constituents of the property in question ( being unmarried being an adult, being male ). The basic account is as follows. Prope rties and relations are complex, and they have other properties and relations as c onstituents. The rules of semantic interpretation take sentences as their syntactic inputs and map them to stru ctured propositions by ma pping the lexical items that are the constituents of sentences to their se mantic values, the latter being the constituents of propositions. The structure of a proposition is a re lation between its parts, and that relation is to be spelled out in terms that quantify over lexica l items. A proposition gets all of its structure from the relation that holds between the lexical items that make up a sentence (so, the semantic relations between the lexical items and their seman tic values do not add to the structure). Take the following sentence: For all x, x is a bachelor iff x is an unmarried adult male It expresses an analysis. The entities introduced by the left side are distinct from those introduced by the right side. In pa rticular, the left side introduces being a bachelor (and the entity contributed by for all x ), while the right side introduces a complex entity that has as constituents being unmarried being an adult and being male and represents the property of being a bachelor. The analysis of being a bachelor then, is a proposition that has as parts the property of being a bachelor and a second entity that represents this property and has as its parts

PAGE 85

85 properties that are also parts of being a bachelor Again, the relation between the parts of this second entity is spelled out in terms of the rela tion between lexical items (the relation tin virtue of which they make a sentence ) and the relation between the le xical items and their semantic values. This is supposed to solve the paradox of analysis because, contrary to appearances, the right side of an analysis senten ce does not introduce the same enti ty as the left side. In the sentence above, the left side does not introduce the property of being a bachelor It introduces a complex entity that has as its parts the parts of being a bachelor and the relation between these parts is spelled out in terms of the lexical items that make up the right side of the analysis sentence (King 1998, p. 162). The trouble with accounts of this sort, I believe, is that they cannot be taken literally. The natural place for talk of parts, components, etc. is talk of concrete particulars. To extend such talk to abstracta is to invoke a metaphor. While the metaphor is surely appropria te, it difficult to see either properties or concepts as things that literally have constituents, components, parts, and the like. As we will see in the final, the account I offer suggests a way to eliminate this metaphor, assuming it is one. I will now spell out the second puzzle about analysis and say what I think is the correct way to try to dissolve it. Then I will sa y why it dissolves the first paradox as well. 5.3 A Second Puzzle I will illustr ate the paradox us ing both the bi-conditional formulation of an analysis and the formulation that makes direct a ppeal to talk of analysis. Le t us begin with a bi-conditional formulation. Consider the following analysis: x is a brother iff x is a male sibling The right side of the bi-conditional is suppos ed to state the necessary and sufficient conditions for being a brother (or, perhaps, for the application of the concept of a brother). If the

PAGE 86

86 analysis is correct, however, is a brother and is a male sibling express the same concept. As a result, we can substitute one for the other and generate the following equivalent claim: x is a brother iff x is a brother While the first formulation may look like an analysis, it cannot be, given that is equivalent to something that could not be an analysis clai m (much less an analysis). Although, the generated claim is true given the truth of the initial bi-conditional, it cannot be an analysis claim because, as we have noted, analysis claims are suppos ed capture facts about grounding. Grounding is a irreflexive notion, so we should not say things to the effect that A grounds A But, since the biconditional formulation commits us to saying that ev ery analysis is equivale nt to claim of this sort, it commits us to saying that every analys is is equivalent to a non-analysis claim. There is a second problem for the bi-conditi onal formulation. The notion introduced by iff is symmetric. As a result, there is nothing about iff that indicates any sort of explanatory or conceptual priority. For a ny analysis to the effect that A iff B we have no more reason to read it as saying that A grounds B than we do to read it as saying that B grounds A The second problem, then, is that the bi-conditional formulation commits us to saying that every analysis is equivalent to a claim which, though true, coul d not be an analysis, due to the symmetric character of the ke y notion introduced. One might try to avoid these probl ems by making a direct appeal to talk of analysis. Let us replace the bi-conditional formulation with this: Being a male sibling is the analysis of being a brother. While the bi-conditional intr oduces a symmetric notion, the appeal to analysis does not, for this formulation does not entitle us to the following claim: Being a brother is the analysis of being a male sibling.

PAGE 87

87 Clearly, this claim is false, so direct appeal to talk of analysis captures something important that the bi-conditional does not, namely, the fact that grounding is asymmetric. This is one advantage of the direct appeal to analysis talk. However, if the analysis is correct, being a brother and being a male sibling are interchangeable. Prima facie at least, this applies to analysis contexts as well. So, the analysis, if true, is equivalent to this: Being a brother is the analysis of being a brother. This claim is false, given that analysis is a form of grounding. Unlike the bi-conditional, then, analysis introduces an irreflex ive notion. So, although the direct appeal to analysis gives us an analysis claim, that claim, if true, commits us to saying that it is equivalent to something which, though an analysis claim, is false. The problems with these two form ulations of analysis claims, then, are slightly different. On the one hand, the bi-conditional formulation (because it treats analysis as a notion that is both reflexive and symmetric) commits us to saying that every analysis is equiva lent to a (true) nonanalysis claim. On the other hand, the direct appeal to talk of analysis (although it introduces a notion that is neither symmetric nor reflexive) commits us to saying that every analysis is equivalent to claim which, though an analysis claim, is false. In brief, the bi-conditional formulation does not introduce the notion of analysis, and the dire ct appeal to analysis, though it introduces the notion of analysis, commits us to treating claims with different truth values as being nevertheless equivalent. 5.4 Equivalence and Substitution An obvious line of response to these worries is to look for reasons to reject the principle of substitution. One m ight argue that E2 cannot always replace E1 salva veritate Someone might argue, for example, as follows. If two expressions ar e interchangeable in this way, then the two

PAGE 88

88 expressions must have the same sense. But if tw o expressions they have the same sense, then, whenever a person has a belief that is correctly ascribed using the first expression, that person has a belief that is correctly as cribed using the second expression. In brief, if two expressions have the same sense, then we can substitute one for the other in belief contexts salva veritate So, if we can find a reason to think that we can not carry out the substitu tion in belief contexts, we will have found a reason to doubt that the expre ssions have the same sense, thus blocking the substitution in analysis contexts. Why think that we cannot substitute analysans and analysandum in belief contexts? One reason that might be offered is that to carry ou t the substitution is to attribute unrealistically sophisticated beliefs to ordinary speakers. The principle of substitution commits us to saying, for example, that, if someone makes a judg ment about how many people in the room know what time it is, that person makes a judgment about how many people in the room have a suitably justified true belief about what time it is. The objection, ag ain, would be that it is implausible that our day-to-day judgments about knowledge involve a level of sophistication that calls for our day-to-day belief attr ibutions to be sensitive to such a subtle distinction as that between a belief which is justified and a belief which is ju stified in the way that philosophers write about when they consider Gettier cases. Someone might offer a different, though relate d, reason for challenging the principle of substitution. Ordinary speakers are in many cases willing to assent to sentences that contain the predicate that expresses the analysandum while being unwilling (or at le ast hesitant) to assent to sentences that invoke the pr edicate that expresses the analysans. In ordinary contexts, people would be willing to assent to a sentence Jack knows how many people are in the room, while being unwilling (or at least more hesitant) assent to the sentence, Jack has a suitably justified

PAGE 89

89 true belief about how many people are in the room Again, the objection to the principle of substitution is that the facts about which sent ences speakers would assent to are evidence for what beliefs they have, and the discrepancy be tween the unwillingness of ordinary speakers to assent to sentences that introduce the analysans suggests that it is wrong to attribute beliefs using sentences that introduce the analysans It is possible, I think, to give a reasonable response to each of these objections. Consider the first. Do ordinary speakers have the level of sophistication that motivates the first objection? I want to say that they do, but the sophistication is not of the sort that th e objection mentions. Having a concept does not require ev er having entertained the analys is of that concept. Nor does it even require seeing that the analysis is co rrect upon entertaining it. Having the concept requires a certain degree of facility with respect to deploying the concept. To have the concept of a bachelor, for example, does not require havi ng entertained a proposition to the effect that being an eligible unmarried male is the analysis of being a bachelor But it does require one to have a sophisticated disposition, which includes, for example, being disposed not to apply the concept if faced with a case of an unmarried male that it ineligible for marriage, say, due to age. Presumably, if someone has the concept of a bachel or, there is something in her history that puts her in a position not to apply the co ncept to this sort of case. This disposition is surely one that we all would attribute to ordina ry speakers who have the concep t of a bachelor. Despite having this disposition, however, an ordinary speaker w ith the concept of a bachelor would not, when asked to give an analysis of being a bachelor be likely to think of the eligibility condition (without a great deal of effort). So, to say that the analysandum and the analysans are the same concept does attribute a significant degree of so phistication to ordinary speakers, but the sophistication comes in the form of a comple x disposition (to make certain judgments under

PAGE 90

90 certain conditions) rather than knowledge of the analysis (or even being disposed to recognize the truth of the analysis upon entertaining it). The mistake in th e first challenge, then, is to suppose that the principle require s us to attribute to ordinary speakers a certain level of sophistication with respect to which propositions they have entertained whereas it commits us only to attributing knowle dge in the form of dispositionsdispositions we already agree ordinary speakers have. We can respond to the second objection in a similar way. Having the concept of knowledge may require me to have sophisticated di spositions, (e.g., the dis position to assent to sentences containing the expres sion knows only when, among other things, someone has a suitably justified true belief). But being so disposed does not require one to have the same disposition with respect to sentences containing the expression has a suitably justified true belief. An ordinary speakers linguistic history includes a good deal of exposure to sentences containing the expressions knows and very little, if any, exposure to sentences containing the expression has a suitably justified true belief. Someones history might include a good deal of exposure to sentences containing justified. true, et c., but having facility with the simpler expressions that make up a complex expression (in th is case, has a suitably justified true belief) does not amount to facility with the complex expr ession they make up. So, there is nothing in the linguistic history of an ordi nary speaker who has the concep t of knowledge to suggest that she has the same disposition with respect to kno ws and has a suitably justified true belief. The mistake in the second challenge to the prin ciple of substitution, the n, is that it presupposes that, if one does not have the same assent-related dispositions with respect to two sentences, then the sentences cannot be equivalent, and that, as a result, belief at tribution which make use of the key expressions cannot be equivalent either.

PAGE 91

91 These are not intended as decisive arguments against trying to block the substitution. But they do suggest that, even upon refl ection, it is not clear that the prima facie plausibility of the principle is misleading. For the rest of the discussion, I will assume that the principle of substation is true, though I grant that the matter calls for a more careful discussion. 5.5 The Proposed Solution We have noted som e problems with the bi-cond itional formulation and the direct appeal to analysis. In both cases, the principle of subs titution committed the formulation to a claim that introduced a reflexive notion. We got either A iff A or A is the analysis of A, both of which are unacceptable. I want to argue that the problem is due not to the principle of substitution but to the expressions iff and is the analysis of. In brief, I want to sugge st that, if we formulate analyses in terms of the in virtue of rather than iff and analysis, we can grant the principle of substitution and avoid the paradox. The solution I want to suggest is the sort of solution suggest ed by Black. The key lesson in his paper, I believe, is that we may find a way around the paradox if we look for a formulation of analyses which does not, like the iff, analysis, and is iden tical with formulations, commit us to a formulation that (given the principle of substitution) allows us to generate a claim (such as A is the analysis of A ) that could be true only if analysis were a reflexive notion. According to the account of analysis I want to suggest, analysis claims are grounding claims in the vicinity of claims to th e effect that one feature is always had in virtue of certain other features. For example, the analysis of being a brother is in the vicinity of the following claim: For any x, if x is a brother, x is a brother in virtue of being a male sibling As it stands, this proposal cannot be correct, for it says, in effect, that anything with a certain feature has that feature in virtue of having that very feature. Like analysis, in virtue of

PAGE 92

92 introduces an irreflexive notion, so we cannot say things to the effect that A in virtue of A The suggestion, however, is in the right ballpark. I believe that we can avoid the reflexivity problem by removing being a male sibling from the anal ysis and replacing that expression with is a male and is a sibling. On the f ace of it, it is true that anything th at is a brother is a brother in virtue of being a male sibling. But, as we noted in the second chapter, the following claims have some plausibility as well: For any x, if x is a brother, x is a brother in virtue of being male For any x, if x is a brother, x is a brother in virtue of being a sibling Black suggests that the analyses of being a brother introduces a three-term relation that holds between being a brother being a male and being a sibling My suggestion is slightly different. The in virtue of claim connecting being a brother with being a male sibling should be replaced with one that connects being a brother individually with being a male and being a sibling. The analysis of being a brother is thus along the lines of For any x, if x is a brother, then i) x is a brother in virtue of being male; and ii) x is a brother in virtue of being a sibling. Something to notice here is that the these partial analysis claims could both be true even if there were some further true in vi rtue of that helps explain why x is a brother. For example, if something is a brother in virtue of being male, and there is a featur e in virtue of which things are male, then there is feature, in addition to being male and being a sibling in virtue of which things are brothers. For a complete explanation of being a brother therefore, we need a condition that functions as a so rt of thats all clause. For example, we could add this: There is no further property P such that x is a brot her in virtue of having P.

PAGE 93

93 This condition, however, will not work. If something is a brother in virtue of being male and is a male in virtue of having some furt her feature, then, it is a brother in virtue of having that further feature. But our condition rules th is out, because it says that there is no further property in virtue of which x is a brother. Here is a slightly be tter candidate that allows for a further property in virtue of which x is a brother and still makes for a complete analysis: If there is some further property P such that x is a brot her in virtue of having P then either x is a male in virtue of having P or x is a sibling in virtue of having P This condition rules out there being any propert ies missing from the explanation by specifying the way in which any other relevant feature woul d be related to the ones that are explicitly mentioned. Condition iii) does not say that if there is a further property P that figures in the analysis of being a brother, then it is either part of the analysis of being male or part of the analysis of being a sibling That is, it does not say the following: If there is some further property P such that x is a brot her in virtue of having P then either, for any x, if x is a male, x is a male in virtue of having P or, for any x, if x is a sibling, x is a sibling in virtue of having P Although it is plausible that, if A has B in its analysis, and B has C in its analysis, then A has C in its analysis, to read iii) in this way would be too restrictive. Take the concept of a red marker. If something is a red marker, it so in virtue of being red and in virtue of being a marker. But being red is a determinable feature, s o, there is some shade of red S such that, there are possible cases where some x has the property of being a red marker in virtue of having S. But S is not part of the analysis of being red Condition iii) covers this kind of case as well as cases where the additional feature is part of the analysis of one of the features introduced in the analysis. A better way to express the analysis of being a brother, then, is as follows:

PAGE 94

94 For any x, if x is a brother, then i) x is a brother in virtue of being male; ii) x is a brother in virtue of being a sibling; and iii) if there is some further property P such that x is a brot her in virtue of having P then either x is male in virtue of having P or x is a sibling in virtue of having P. The analysis of the concept of a brother is a series of true in virtue of claims each of which introduces a condition that is insufficient for being a brother And adding the appropriate thats all clause is supposed to supply the suffi ciency needed for a complete analysis. Why think this is an adequate formulation of the analysis of the concept of a brother? One advantage it has over the other formul ations we have looked at is th at it is consistent with the principle of substitution. The principle of substitution cause d trouble for the bi-conditional account because it committed us to saying that ever y analysis is equivalent to a non-analysis. This commitment was the result of the symmetric and reflexive nature of the bi-conditional. If analysis is a type of grounding, then it can be neither symmetric nor reflexive. Like the direct appeal to analysis, the in virt ue of formulation introduces a notion that is both asymmetric and irreflexive, but, unlike the analysis formulation, it is consistent with the principle of substitution. We can substitute every occurrence of is a brother with is a male sibling without generating a claim which could be true only if analysis we re a reflexive notion. We can carry out the substitution because we have eliminated is a male sibling from the analysis by replacing it with being male and being a sibling. Every occurrence of in virtue of in this formulation is followed by an expression that introduces a featur e, but in no case is that expression one that is interchangeable with being a br other. As a result, we can, in line with the principle of substitution, replace any occurren ce of is a brother with is a ma le sibling in the analysis, without changing its truth-value. Consider condition i). If we car ry out the substitution, we get,

PAGE 95

95 x is a male sibling in virtue of being male, which is true. Thus the analysis is consistent with the principle of substitution. Another merit is that is that it comes cl ose to capturing the necessary and sufficient conditions we are after, namely, the explanator y ones. The proper formulation will capture the necessary conditions because it says th at if something is a brother it is both male and a sibling, and the thats all clause should introduce any other necessary conditions there may be, because these necessary conditions will be a one in virtue of which the concept applies. The correct formulation will capture conditions that are sufficient for being a bachelor because, due to the thats all clause, it introduces every feature in virtue of which the concept applies. The relevant (i.e., explanatory) conditions, we can say, are al l of the ones which are both necessary and the features in virtue of which the analysandum applies. This is an advantage over the bi-conditional formulation, because the latter, being strictly modal, will introduce any necessary condition whatsoever, explanatory or not. There is, however, a counter example to th is formulation. Let us change the example slightly. What is the analysis of being a Martian brother ? If the current account is correct, we should say the following is the correct analysis: For any x, if x is a Martian brother, then i) x is a Martian brother in virtue of being Martian; ii) x is a Martian brother in virtue of being a brother; and iii) if there is some further property P such that x is a Martia n brother in virtue of having P then either x is Mart ian in virtue of having P or x is a brother in virtue of having P But this cannot be correct. If something is a Mart ian brother, then it is so in virtue of being a Martian male But it is not the case that x is a Martian in virtue of being a Martian male, and it is

PAGE 96

96 not the case that x is a brother in virtue of being a Martian male. The thats all clause is too restrictive. There is a further property ( being a Martian male ) such that x is a Martian brother in virtue of having that property, and x do es not either have that property of being a Martian in virtue of having that property or have that property of being a brother in virtue of having that property. However, we can modify the thats all clause in a way that enables it to include being a martial male in the analysis as a feature that itsel f is had in virtue of the other feature mentioned. Returning to the analysis of being a brother I suggest the following formulation: For any x, if x is a brother, then i) x is a brother in virtue of being male; ii) x is a brother in virtue of being a sibling; and iii) if there is some further property P such that x is a brot her in virtue of having P then either 1. x is male in virtue of having P or x is a sibling in virtue of having P ; or 2. x has P in virtue of being a male or x has P in virtue of being a sibling. This form handles the brother case in the same way as before, for all we have done is add condition 2. Notice th at it handles the Martian brother case. If something is a Martian brother, then it is so in virtue of being a Martian male The addition of condition 2 captures this because if something is a Martial male (the further pr operty mentioned by iii)), it is, in accordance with conditiona 2, a Martian male in virtue of being male. We noted earlier that the claim th at anything that is a brother is a brother in virtue of being a male sibling seems true. The prosed account, ho wever, says that it is false. The proposed account says that anything that is a brother is a brother in in virtue of being male and in virtue of being a sibling (plus the thats all clause). The intuitive character of the claim that anyting that

PAGE 97

97 is a brother is so in virtue of being a male si bling is, on the proposed view, explainable in terms of the overlap in vocabulary with respect to th e false analysis claim and the proposed one. The similarity in vocabulary enabales us to direct our listeners atten tion to the correct analysis though pragmatic implicature, even though the sentence we use to introduce the analysis is, strictly speaking, false. It may be necessary to refine this account furthe r, but I will not try to do that here. Instead, I will say how this solution helps dissolve the epistemic puzzle with which we began. 5.6 Revisiting the Epistemic Puzzle I have spelled out two versions of a p uzzl e about grounding. In each case, the problem was that any proposed analysis turn s out to be equivalent to a clai m that cannot be an analysis. The bi-conditional formulation led us to say that every analysis is equivalent to a non-analyses, while the direct appeal to analy sis led us to say that an anal ysis is equivalent to something false. Our grounds for denying the equivalence when it came to the bi-conditional formulation was that the analysis was seen to be equivalent to a claim that introduced a notion which is both symmetric and non-reflexive. The initial puzzle was about the informative ch aracter of analyses. If an analysis is equivalent to a trivial claim, then it cannot be as informative as it seems to be. The proposed solution denies that analyses are equivalent to trivial claims. The initial version of the paradox involves substituting one expression for another, which results in trivial claim that is the same in content as the analysis. But the account I spell ou t, casts analyses in a way that eliminates the presence of synonymous expressions in the analysis As a result, it does not represent analyses in a way that is open to the line of re asoning that generate s the first puzzle. However, there is a related puzzle that is also called the paradox of analyses, and the proposal in this chapter does not point to a soluti on to that puzzle. What I have in mind is a

PAGE 98

98 puzzle that can be traced back to the Socratic fallacy. The basic puzzle is this: Analysis claims are supposed to be the sort of claim that can inform us of something we did not already know. But, for any claim, to understand it (and to learn something as the result of understanding it) one must have all of the relevant c oncepts. But to have the concept is to have the information the analysis is supposed to supply. In brief, being in a position to entertai n the analysis presupposes knowing the truth of the analysis, so it is impossible to learn someth ing as the result the analysis cannot inform us of anything. These are two different puzzles. The first a ssumes that analyses are claims about which concepts are identical with which. The puzzle is about how these iden tity claims can be informative. The second puzzle is about concep t acquisition. An analysis is supposed to be a claim such that someone could, in coming to understand it, acquire ne w (non-dispositional) knowledge about a concept. But understanding an analysis requires one to deploy the concept one is supposed to learn about by coming to under stand the analysis, and being in a position to deploy the concept requires one already to have the information one is supposed to acquire by coming to learn the analysis. Some of the consid erations about dispositi ons in the discussion of the principle of substitution may help us get around this puzzle. Arguably, the knowledge needed to understand an analysis claim is knowle dge in the form of a disposition. This line of thought has been discussed in some detail, for example, by Fumerton (1983). I will not pursue the matter any further here. I will conclude by discussing some apparent exam ples of analysis which do not seem to be handled by the account I am suggesting. 5.7 Some Problematic Cases The above account is intended as a theory of an alysis. It is supposed to tell us the form that analyses should take. In this section, I want to discuss som e apparent counterexamples. The

PAGE 99

99 proposed account has some intuitive appeal because it treats analysis as a form of decomposition. There are, of course, strong candidates for analys es that do not appear to be decompositional in this way. Consider, for example, a functional noti on such as that of a laundry detergent. We analyze this notion in terms of a causal role. A (very simple) an alysis would look something like this: If x is a laundry detergent, then there is some property P or other such that i) x has P; ii) P takes the dirt out of clothing, bedding, etc. This is not decompositional analysis in the sense that the analysis of being a brother is. Of course, the consequent contains conjuncts. But notice that, while the left-hand side has the form x is F the right-hand side has the fo rm There is some property P such that . In the cases we have been focusing on, the antecedent contains a predicate, while the consequent contains several predicates and a thats al l clause. In the current case, the left-hand side contains a predicate, while the right-hand si de (qua functional analysis) makes crucial use of an indefinite quantifier expression that characterizes a property. So, apparen tly, this analysis does not have a form that my suggested account captures A plausible response to this worry is that x is a laundry detergent does not introduce a property in the same sense that x is a brother does. It is a pr edicate that is shorthand for a quantifier expression. That is to say, x is a laundry detergent is used in place of There is a property P such that . So, strictly speaking, the above is not an analysis of the concept of a laundry detergent, or, if it is an an alysis, it is what we might call a corrective analysis rather than an explicatory analysis. According to this suggestion, when I say x is a laundry detergent, I am saying, There is a property P such that ., and none of the predicates in the analysis (predicates which characterize P ) expresses the concep t of a laundry detergent. This does not

PAGE 100

100 commit us to saying that there is no concept of a laundry detergent. It is just that the surface features of laundry detergent claims are mislead ing. Proposition about la undry detergent are not predicative but quantif icational. So, the suggested analysis tells us that we have a defective use of a predicate. Of course, one might think th at as long as we can st ate the conditions under which a predicate applies, the predicate is not de fective. But this is not an option for someone who thinks that a predicate x is F is deficient unless it attributes a property that corresponds to the entire concept expressed by is F, something that is in a laundry detergent does not do, for the property it attributes is the realizer of being a laundry detergent A second response one might consider is that the following transformation of the proposed analysis is unproblematic: If x is a laundry detergent, then x has some property P such that i) x has P; ii) P takes the dirt out of clothing, bedding, etc. Then the proposed analysis would be the following: For any x, if x is a laundry detergent, x is a laundry detergent in virtue of having some property P such that i) x has P; ii) P takes the dirt out of clothing, bedding, etc. This is not a viable option, for it leads directly to the paradox of analysis. The property of being a laundry detergent just is the property of having some property P such that x has P and P takes the dirt out of clothing, bedding, etc If this formulation seems plausible, I suggest that we explain the apparent plausibil ity by noting the overlap in voca bulary between this formulation and the proposed analysis. Both refer to dirt, bedding, and clothing, so th is false analysis claim can be used to direct our atten tion on the notions that figure in the correct and, to that extent,

PAGE 101

101 direct our attention on the correct analysis itself. In this wa y, we can appeal to pragmatic implicature to explain why the transformation seems unproblematic. There are plenty of predicat es that raise the same worry for the proposed account, for example, is red and is a limit bear the same sort of analysis as is a laundry detergent. When I say, x is a limit, I am saying something like There is a function whose arguments yield values that approach x but never reach it. And when I say that so mething is red, I am (plausibly) saying something about the existence of a surface feature that has certain visual effects under certain conditions. The important feature these no tions all share is that they are higher-order notions. They are claims about the property of having some property or other, so if the quantificational readi ng applies to claims about laundry detergent, then it is plausible that these other notions can be given the same treatment, making them instances of corrective analysis rather than analysis in the form of explication, which is the sort of analysis that has been the focus of this chapter.

PAGE 102

102 CHAPTER 6 PROSPECTS FOR FUTURE APPLICATIONS In the dissertation, I have ar gued that grounding explanations are central to philosophy and that uncovering the proper form of grounding claims can shed light on su bstantive philosophical issues. In Chapter 3, we situated grounding explan ations within a framework that seems to apply to several sorts of explana tion, including nomic explanati ons, causal explanations. The conclusions I drew with respect to what form grounding explanations take, when they are made fully explicit, supplied the resources for maki ng progress with respect to truth-making and analysis. In this final chapter, I explore the prospects for future applications of the findings in earlier chapters. In particular, I spell out the beginnings of a theo ry of essence that is suggested by what I say about analysis in Chapter 5. I also return to the discussi on of talk of parts and wholes, when applied to properties. In Chapte r 5, I noted that such talk strikes me as metaphorical. In this final chapte r, I explain why I do not believe properties to have parts. The main purpose of this chapter is to introduce some ideas that can serve as a foundation for future applications of the account of gr ounding explanations I have spelled out in earlier chapters. For this reason, the final chapter is more speculative than the earlier chapters. 6.1 Essence Let us begin with the notion of essence. In Chapter 5, the focus was not on the notion of analysis that is at work in day-to-day contexts (and in philosophy when we speak of conceptual analysis ). Analysis in this sense is a kind of inte llectu al activity. In Chapter 4, the focus was on analysis in the sense in which it occurs in philo sophy when it is appropriate to say that one has offered an analysis An analysis in this sense is a type of grounding claim. (Such claims are discovered as a result of analysis in the firs t sense, which may explain why these different notions bear the same name.) In this final chap ter, I explore the possibility that, while the term

PAGE 103

103 analysis is not used this way in ordinary talk, analysis nevert heless has a counterpart in day-today talk, namely, talk of essences and essential features The basic suggestion is that to speak of somethings essence, and to say that a feature is e ssential, is to advance an analysis. This is not to say that talk of essences a nd essential features is about cl aims; rather, it is to say that essentialist claims have the sa me form as analysis claims. Let us begin by looking at some theories of e ssence. Consider a simple modal conception of essence. According to this concep tion, if it is necessary that x has P then x has P essentially. We can characterize this view as follows: x has P essentially just in case i) x has P ; and ii) it is necessary that x has P It is part of the essence of Socrates (let us say) that he is human. According to the simple modal conception, then, to say that Socrates is essentially human is to say that it is necessary that Socrates has the property of bei ng human. One problem for this th eory is that it is not possible for an object to have a feature unless that object exists, so if it is necessary that Socrates is human, it is necessary that Socrates exists. But it is not necessary that Socrates exists, since he exists only contingently. As a result, the simple modal conception is too strong. We can try to refine the simple modal conception by adding another condition, namely, that is necessary that the individual has the property if that individual exists Thus if it is necessary that the individual has the property if that individual exists, then the property is essential to that individual. This gives us the following: x has P essentially just in case i) x has P ; and ii) necessarily, if x exists, x has P. This variant is also too weak. Cons ider the singleton set . It is essential to this set that it contain Socrates as its only member. But, intuitiv ely, it is not essential to Socrates that he be a member of this set. Surely, assuming sets exist, it is necessary that if Socrates exists, he is a

PAGE 104

104 member of Singleton Socrates, but it is not part of his essence. But this conception of essence commits us to saying that it is part of his e ssence to be a member of , so the conception is too weak. Both versions of th e modal conception of e ssence are too weak, it seems, because they count some features as essen tial in cases where they do not play any sort of explanatory role. In other words, the correct theory of essence require s a constraint on which features count as essential. Th at constraint is an explanatory one. Essen tial features, not doubt, are necessary features, but they are also explanatory. While essences are (presumably) explanatory, there is a question about what they explain. One option is that they explain a things ident ity, or the conditions under which something is identical with a given thing. Some conceptions of essence, then, spell this suggestion out in terms of the conditions under which nominally distinct en tities would be identical. Thomas J. McKay characterizes one such view as follows: These are principles that say that the having of certain properties is sufficient to secure the identity of an individual. In other words, only that particular individual could have those properties. An example would be this: If, at possible world w1, a certain table b1 is made from a chunk of wood m, then if a table b2 is made in the same way from m at w2, then b2 = b1. (McKay 1986, p. 295) We can state the gist of this account as follows: If x has P essentially, then x has P and, necessary, for any y, if y has P y is identical with x McKays informal characterization, we saw, is in terms of properties that are sufficient to secure the identity of an individual. If we read secure in a purely modal way, then we have too weak a definition. Let P be the property of being the sole member of Singleton Socrates. Socrates has P and it necessary that, if something has P it is identical with Socrates. Socrates having P secures the truth of the identity claim in this purely modal sense, but this does not explain anything about Socrates. Intuitively, it do es not tell us anything informative about what

PAGE 105

105 Socrates is All it tells us is that anything that is the sole member of Singleton Socrates is identical with Socrates. There are versions of this acc ount that appeal to the expr ession makes, and people who advance these formulations appear to have in virtue of in mind. Robert C. Coburn, for example, explains the need for an explanatory set of properties if we are going to appeal to identity conditions in giving an account of essence: After all, something about me makes me the i ndividual I am, we want to say. But if we give up the view that it is some feature or features I possess that individuate me, we seem forced to the idea that what makes me the individual I am is just the fact that the possessor of the properties I instantiate is what it is quite independent ly of any of the features it exhibits an idea that seems on the face of it unintelligible. (Coburn 1986, p. 165) Coburn talks of something that makes him the indi vidual he is. This is a stronger relation than simple entailment, being at least a cousin of the in virtue of. We can spell this conception out as follows: If x has P essentially, then x has P ; and xs having P makes x x Consider a statue that is also a hunk of clay. Being this hunk of clay is a property had by this statue, and having that property makes in some (constitutional) sense the statue the statue that it is. But we may not want to say that being this hunk of clay is part of th e essence of the statue, since the statue could have been ma de of a different hunk of clay. If P s being necessary to an object (if it exists) is a conditi on on being essentially had, then simply making something the thing that it is is insufficient for being part of the essence of that thing, for if x makes y, x is sufficient for y, though it may not be necessary for y. We can strengthen this account by adding the condition that not only does P make x x, it is the only property that can do that, so something must have P if it is x. This gives us the following account: If x has P essentially, then x has P xs having P makes x x, and, neces sarily, if x exists, x has P

PAGE 106

106 This might be a better formulation of the pr oposal, but both versions say that xs having P makes x x. It is not clear at all wh at this is supposed to mean. Sometimes it is put in terms of P s making x the particular x is, which really is not helpful. One reading of the claim that something makes x x says that it makes x numerica lly identical with x, th at is, it makes x selfidentical. But being self-identical is not, it s eems, a fact about something that that can be explained in terms of other facts nothing is self-identi cal in virtue of having some further set of properties. Even if there were pr operties in virtue of which Socr ates is self-identical, it does not seem to make sense to say that the properties that make Socrates self-iden tical are different from the properties that make other things self-identical and pr esumably, Socrates essence is different from the essence of anything else. There are, of course, different accounts of what identity involves. Any account that appeals to absolute identity says that to give the essence of So crates is to state the conditions in virtue of which something is num erically identical with Socrates. Appealing to this notion enables us to avoid saying things to the effect that the essence of Socrates is the set of features that make Socrates the thing that he is. What is wrong with appealing to absolute identity? As we just saw, it seems that the no tion of absolute identity either does not call for an explanation, or if it does, it does not call for an explanation that invo lves any details about the relata of the identity relation. If a = b, then there are no further facts about a and b such that we can say that a = b in virtue of those further facts. They are just identical. Either identity is a basic notion or it is not. If it is not, then we can give an account of identity and say, a = b in virtue of . But this is explanation will not capture the essence of the relata It will give us an account of the identity relation, not an account of, for example, Socrates.

PAGE 107

107 I want to suggest that it may be useful to appe al to the controversial idea that there is such a thing as relative identity When we appeal to this notion, we do not simply use two referring expressions and make an ident ity claim to the effect that a = b. Instead, we specify a feature F and say of a and b (or of x and y, if we want to use variables) that they are the same F We are still appealing to numerical identit y, but we discuss identity relative to a feature. Might it be that to state the essence of Socrates is to state the conditions in virt ue of which, for any x, x is the same as Socrates, for some Of course, one would need to say something about what should be plugged in for if the account is to have any plau sibility. Notice that we cannot let be, say, person. What does it take to be the same person as Socrates? Intu itively, it is the same thing that it takes to be the same person as Plat osomething in the vicinity of participate in the same mental life. This formulation, though it men tions Socrates, gives us an instance of a more general question about the nature of personal identity. What we n eed, if this account of essence is to work, is a formulation that invokes some that captures the releva nt details about Socrates, not just any person. Here is a suggestion I think is worth exploring. Perhaps, in the Socrates case, we should substitute individual for The term individual is not m eant to introduce the notion of a person. All it is meant to introdu ce the category that enables us to draw the distinction between individuals and universals. The basi c idea is that an individuals es sential features are the ones in virtue of which nominally distinct entities are identical relative to th e category of being an individual. When it comes to universals, we substitute universal for With respect to Socrates, then, we say the following. The expression being Socrates should be read as being the same individual as Socrates. Let us assume that among Socrates essential features are being a person, having a as a parent and having b as a parent These

PAGE 108

108 features are essential to being Socrates because they figure in a generalization to the following effect: For any x, if x is Socrates, then (i) x is Socrates in virtue of being human; (ii) x is Socrates in virtue of being a person; (iii) x is Socrates in virtue of having a as a parent; (iv) x is Socrates in virtue of having b as a parent; etc. The suggestion is that there is a set of features that only Socrates can have and are such that, if something is the same individual as Socrates, then it is so in virtue of having these features. If something is the same individual as Socrates in virtue of having these features, then the features are sufficient for being Socrates, given the appropriat e thats all clause. This is what makes the features explanatory and appropria te for being deemed essential. Once we add the appropriate thats all clause, we will have something that looks very much like an analysis of being Socrates (i.e., an analysis of being the same individual as Socrates ). Can this reading of is Socrates capture Socr ates essence? Someone might argue that our handle on the proper name Socrates generates our intuitions about which features do the explaining here. The objections is that be ing in a position to unde rstand claims about being the same individual as Socrates puts us in a position to make esse ntialist claims about Socrates only insofar as understanding the claims about Socrates does. For example, it is our handle on the name Socratesnot our handle on the expression is the same individual as Socratesthat puts us in a position to know that it is true that Socrates is essentially a person. Someone could well argue in this way. To do the objection justice would require a detailed discussion of the semantics of proper names, something I cannot do here. For now, I want to suggest a different pictur e, one that seems at least as pl ausible at first blush. According to this picture, it is true that our ability to unde rstand claims about what it takes to be the same

PAGE 109

109 individual as Socrates presupposes an ability to understand occurrences of the name Socrates and to be in a position to make claims about Socrates (thereby attributing proper ties to him). But the latter abilities to do not re quire an ability to make essentialist claims about Socrates. It is plausible that we are first in a position to attribut e properties to Socrates (a s in Socrates is wise, Socrates is human, and Socrates is a person), wh ile not being in a position to say which of the properties attributed are essential to Socrates. The way to dete rmine which ones are essential to Socrates is to take the claims indivi dually and ask whether, for instance, being wise is a feature in virtue of which it is true of something that it is the same individual as Socrates. Being wise is not such a feature, while being a person is. In this way, our handle on the name Socrates gets us started when it comes to making essentialist cl aims, since it enables us to attribute properties to Socrates, but this ability in itself does not require us to know which features are essential to Socrates. After all, it is possible to be competent with a term without having anything like an explicit analysis of the notion it introduces. What are the merits of such a view? Let us consider a distincti on between two ways of understanding the expression is Socrates. The r eading I favored above take s it to be short for being the same individual as Socrates. The othe r reading takes it to attribute a property that does not involve identity, a pr operty analogous to, say, being a bachelor. According to this conception of being Socrates, the latter property may be complex, for we can make generalizations that indicate consti tutive connections with other prope rties. For example, just as we can make claims to the effect that if someth ing is a bachelor, it is unmarried, we can say that if something is Socrates, it is a person. There are, I think, independent reasons fo r favoring the first reading. One noteworthy difference between being Socrates and being a bachelor concerns the number of things that can

PAGE 110

110 instantiate these properties. Ma ny things can instantiate the pr operty of being a bachelor, but only one thing can instantiate the property of bei ng Socrates. Of course, there are features like being the tallest person in the world Only one person at a time can in stantiate this feature, but if someone instantiates this feature at a time t there are nevertheless othe r possible times at which other persons instantiate the feature, and there are other possible worlds in which someone else instantiates the feature at t But when it comes to being Socrates, there is, for any time whatsoever, only one object (namely, Socrates) that can instantiate this feature, and there are no possible worlds in which something other than Socr ates instantiates the feature. What explains this difference between being Socrates and being a bachelor? If we take being Socrates to mean being the same individual as Socrates, we have a ready answer. Objects that are themselves distinct cannot all be identical with Socrates. Only one object can be identical with Socrates, namely, Socrates. But there is no obvious answer if we accept the reading that treats is Socrates as being analogous with being a bach elor. If being Socrat es is a property that cannot be defined in terms of identity, then why is there a restriction on wh at can instantiate it? Surely, there is something that can be said in defense of this view, but I will not pursue the matter any further here. I just want to note that it is a merit of the identity-based reading of is Socrates that it supplies a ready answer to th e question of why only one things can have the property of being Socrates. I should note, however, that someone might thi nk that more than one thing can instantiate the property of being Socrates. Someone might argue that, while at one time a certain physical object instantiates th e property of being Socrates, at a later time a distinct physical object instantiates the property of being Socrates. When a physical object instantiates the property of being Socrates, we call that object Socrates body, and Socrates body, we all know, changes

PAGE 111

111 from moment to moment because Socrates both and gains some of his micro-constituents. According to this view, then, it is false to say that Socrates instantiates the property of being Socrates (if we give is Socrates the non-identity-based reading). This account still leaves us with a mystery, however. If it is not Socrates that instantiates the pr operty of being Socrates but instead some physical objectwhat reason do we have to believe that only one physical object at a time can instantiate being Socrates ? If we treat being Socrates as analogous to being a bachelor, then we do not have an obvi ous answer to this question. Extending the project carried out in this disserta tion so as to include an adequate treatment of the notion of essence, then, would involve a closer look at the literature on essentialism and the semantics of proper names and other referring expressions. It would also be useful to look more closely at the notion of relative identity. The above considerations are intended to show that grounding explanations are cent ral to the notion of essence and that future research on the topic of essence should be carried out with this centrality in mind. 6.2 Structural Universals In Chapter F our, I expressed doubt about the cla im that universals are st ructured entities. I suggested that to say that universals have parts ca nnot be taken literally. I will now return to that topic. The metaphor, no doubt, is appropriate. When we discuss universals, we normally try to find necessary connections between them. But the connections are not only necessary. They are, in some sense, constitutive connections. One might argue as follows. Take the property of being a bachelor. Anything that is a b achelor is a male, so there must be something about the universal that explains this pattern of instantiati on, and the most natural explanation is that being male is a part of being a bachelor Whenever something is a bachelor, then, it is also a male because the property of being male is literally a pa rt of the property of being a bachelor.

PAGE 112

112 David Lewis (1986) is critical of this con ception of universals. He lays out several variations of the picture and rej ects all of them. However, consid erations in the earlier chapters suggest that some of his reasoni ng is faulty. I will discuss hi s account of what he calls the pictorial and magical c onceptions of universals. According to the pictorial con ception of universals, they are isomorphic to their instances (see Lewis 1986, pp. 33-4). Consider the univer sal methane. Its instances are methane molecules. These have as their parts a car bon atom and four hydrogen atoms. Now if the universal is isomorphic to its instances, then it has the universals carbon and hydrogen as parts. The problem with this suggestion, Lewis says, is that it requires us to say that methane has hydrogen as a part four times over Any methane molecule has four distinct parts that are all hydrogen atoms, but the universal has the very same thing as a part four times over. But it does not make sense to say of something that it has an object as a part four times over. A car, for example, can have four wheels, but it ca nnot have the same wheel four times over. Lewis points out that we cannot simply drop th e commitment to isomorphism, for there are other molecules that have as their parts carbon atoms and hydrogen atoms (pp36-7). The butane molecule, for example, has as its parts four hydr ogen atoms and ten carbon atoms. As a result, we cannot drop the isomorphism and say simply that methane and butane have the universals hydrogen and carbon as their parts. For distinct entities cannot have the same parts. In other words, merely dropping the isomorphism commit s us to saying that methane and butane are identical universals. A possible response, Lewis notes, is that me thane and butane have the same parts, but those parts are arranged differently (see Lewis 1986, p. 35). There is, in addition to mereological composition, a form of non-mereological compos ition between universals, so we do after all

PAGE 113

113 have a way of distinguishing methane from butan e. Lewis does not believe in non-mereological composition. Mereology, he thinks, captures composition in its full generality. Armstrong (1986) argues that we can make sense of non-me reological composition. Consider possible cases where we have an object a, and object b, and a relation R where R is non-symmetrical. We can have a case where aRb, and we can have cases where bRa. We have just identified two distinct states of affairs which have the same parts. The difference is in how the parts are arranged (see Armstrong 1986, p. 85). In response, Lewis says something to the effect that th is possibility is beside the point. It is a description of possible instances of universal s, but it does not tell us anything about the structure of the universals themselves. I disc uss non-mereological composition in a bit more detail below. There is also the magical conception of universals. Accord ing to this picture, universals have no parts at all. Mereologically speaking, they are simple. The problem with this view, he thinks, is that it leaves us unable to explain necessary connect ions between universals. If being a bachelor has being male as a part, then we have something in the way of an explanation for the fact that that when something instantiated the property of being a bachelor it also instantiates the property of being male, since being male is a part of being a bachelor But if the two universals are not connected mereologically, then there is no explanation available. That the universals are instantiated in the combinations they are is inex plicable. In this sense, we have a magical conception of universals. I am inclined to say that treating universals as mereologically simple does not leave us with a mysterious picture of universals. The reas on is that there are ways of talking about the nature of universals that do not appeal to pa rts and wholes and, plau sibly, explain why the

PAGE 114

114 metaphor is appropriate (assuming it is a metaphor). It seems clear that it is essential to properties that they have the po ssible patterns of instantiation they do. Intuitively speaking, an analysis tells us which properties are instantiated in terms of which. When it comes to properties like being Socrates I suggested that the property in question is being the same individual as x where x should be replaced by an individual name When it comes to properties, there is a distinction to be drawn. For the expression being F has two readings. We can read it as having property F, and we can read it as b eing identical with property F, where the latter is to be understood in terms of relative iden tity. According to the former reading, to give the essence of P is to make a claim whose meaning is explained by the following conditional: If having property Q is essential to having property P then for any x, if x has P then x has Q and x has P in virtue of having Q Being unmarried, for example, is essential to being a bachelor, because, necessarily, anything that is a bachelor is a bachelor in virtue of being unmarrie d. To say that having Q is essential to having P is, in this case, just to st ate part of the analysis of being a bachelor On the other hand, to say what being a bachelor is (to say what it takes to be the same property as being a bachelor) would include saying that being a bachelor is instantiated in virtue of instantiating being unmarried The property of being a bachelor then, is essentially abstract, and it is essentially such that anything that inst antiates it does so in virtue of instantiating the property of being unmarried That is, anything th at is identical with being a bachelor is so in virtue of being abstract and being such that an ything that instantiates it does so in virtue of instantiating the property of being unmarried etc. The statement of s propertys e ssence will involve claims that require us to invoke both of these readings. When we want to say of something x that it is the same property as P (when we want to state the essence of P ), we will say things to the effect that x is the same property as P in

PAGE 115

115 virtue of being abstract (for example), but we will also sa y things to the e ffect that x is instantiated in the same way as P. In this sense, to state the essence of P (to say what it takes to be identical with P ) involves saying what it is to have P. So, Lewis argument that treating properties as mereologically simple commits us to a mysterious view about the connections between properties presupposes th at we can state the essence of a property without spec ifying its possible patterns of instantiation. There is nothing mysterious about the fact that being a bachelor is always co-instantiated with being unmarried That it is part of its essence. It is this app eal to essence that makes the patterns of instantiation something other than the sort of brute m odal fact that Lewis complains about. Given the conception of essence I have intr oduced here, we can see why the metaphor of parts and wholes is appropriate by noting that the role of indivi dual IVO claim and the presence of a thats all clause. I am inclined to say that we can eliminate th e metaphor by appealing to the in virtue of locution. Rather than saying, as I did in the discussion of analysis, that, if x has P in virtue of having Q and Q does not suffice for P then Q is part of something that is sufficient for P, we should say the following: If x has P in virtue of having Q and having Q does not suffice for having P then there is some property Q* such that i) x has P in virtue of Q* ; ii) x does not have Q in virtue of having Q* ; and iii) if there is some further property R such that x has P in virtue of having R then either 1. x has R in virtue of having Q or x has R in virtue of having Q* ; or 2. x has Q in virtue of having R or x has Q* in virtue of having R. Rather than saying that a requirem ent on having P in virtue of having Q is that Q and some Q* are both a part of something that suffices for P we make a generalizatio n about the instantiation

PAGE 116

116 of P in terms of the instantiation of Q and Q* I will spell out the function of each clause in terms of the part/whole metaphor. The first clau se introduces a second part of the thing that suffices for P (the other part being Q ). The second clause guarantees that Q* is not a part of Q we need this condition because, if Q* were a part of Q then having Q might be suffice for having P (making it not a part of so mething that suffices for having P ). The third clause guarantees that having Q and Q* suffices for having P (that Q and Q* make up a whole that suffices for P ). Since this account of the essence of properties treats them as mereologically simple, I want to address the charge that it is a version of th e magical conception of universals. The magical conception says that, if universals do not have parts, then the nece ssary patterns of instantiation I am appealing to in order to make essentialist clai ms about universals are inexplicable. Consider the following claim: Necessarily, for any x, if x is a bachelor, x is unmarried Here is a claim that grounds it: Necessarily, for any x, if x is a bachelor, x a bachelor in virtue of being unmarried This second claim is equivale nt to the following claim: Being unmarried is essen tial to being a bachelor. In light of our discussion of the connection be tween determination, sufficiency, and explanatory notions such as causation and IVO, we are in a position to say that the second of these claims, because it invokes the in virtue of locution not only entails the first but expl ains it as well. If something is a bachelor, it has to be un married because, as the second claim says, being unmarried is one of bachelorhood s grounding features. Of course the second claim introduces the notion of necessity, so it cannot be appealed to in order to give an a ccount of necessity. In

PAGE 117

117 other words, being equivalent to the third claim, it is an essentialist claim, the latter being a claim that introduces determination, which is defined in terms of the notion of sufficiency. So, the view I am defending in this chapter does not se em to supply the resources for giving an account of modality in terms of anything like explanatory links between properties. Whether this is a drawback is something that would require a more detailed discussion than I can provide here. One would need to determine whether it is appropriate to endorse a conception of modality that does not explain it in terms of constitutive connecti ons between properties. I cannot do that here. 6.3 Non-mereological Modes of Composition I want to conclude this chapter with a brief discussion of non-m ereological composition. Armstrong believes this to be a relation that brings together objects and th eir features to form states of affairs. The state of affairs of the balls being red for example, is explainable in terms of the ball, being red, and the instantiation relatio n, so the state of affairs is not basic. We can say things like The state of affairs of the balls being red at time t obtains because the ball is red at t One might think that considerations in Chapte r 1 suggest that there is a puzzle here. For because is ambiguous between grounding and causa tion, and since we certainly do not, in this case, mean to say what caused the state of affa irs, we must have an IVO claim in mind. But notice that when we try to state this claim in terms of in virtue of, we end up saying The state of affairs of the balls being red at time t obtains in virtue of the balls being red at t. But this claim is false because it tries to explain the occurrence of a state of affairs in terms of the occurrence of that very state of affairsnote the two occurrences of the balls being red at t). So, there is a puzzle. We want to say that the state of affairs of the balls being red obtains because the ball is red. But it seem s that the explanatory link betw een this state of affairs, the

PAGE 118

118 ball, and the property of being red is non-causal, and, as a result, we need to be able to recast this claim in terms of in virtue of (or one of its cousins). I suggest that we appeal to essentiality. If we do this, we will not giving an account of the relation between states of affairs and other entiti es in terms of composition. Instead, we will be dissolving a puzzle that is due to th inking of states of affairs as th ings that have parts. The basic essentialist picture I want to suggest is this. The state of affairs of the balls being red at time t is essentially an instance of the balls being red (note the absence of a time element) where the balls being red is a kind of complex universal. If this is correct, then we have a true grounding claim with respect to the state of affairs. And it follows from this grounding claim that the state of affairs occurred becaus e the ball is red. Surel y, if a state of affairs is grounded in a universal xs being F, it follows that the state of affa irs obtained because x instantiated F In this way, we give a partial explanation of a certain the state of affairs (which is an individual) in terms of the type of state of affairs it is. There is a corresponding claim to the effect that the balls being red at t essentially occurred at t. I am assuming for the sake of discussion that specifying the time element secures reference to a particular state of affairs. But this may not be correct. Specifying the time element in a state of affairs does not necessarily direct our attention on a token rather than a type. My giving someone a gift at time t is a property instance. There are tw o possible scenarios, one in which I give a gift in virtue of giving a tie and one in which I give a gift in virtue of giving a watch. Arguably, we have two instances of my giving so meone a gift, even though in each case it is appropriate to refer to my giving someone a gift at time t So, the basic argument I am advancing here hinges on the proper conception of states of affairs. I do not have the space to do justice to the subtleties.

PAGE 119

119 LIST OF REFERENCES Ar mstrong, D. M. 1986: In Defense of Structural Universals. Australasian Journal of Philosophy Vol. 64, No. 1, pp. 85-8 : A World of States of Affairs Cambridge: Cambridge University Press. : Difficult Cases for Truthmaking. The Monist 83(1), pp. 150-60. : Truths and Truthmakers, in R. Schantz (ed.) What is Truth Berlin: de Gruyter, 2002, pp. 27-37. : Truth and Truthmakers Cambridge, U.K.; New York: Cambridge University Press. Austin, J.L. 1962: How to Do Things with Words (J.O. Urmson, ed.). Cambridge: Harvard University Press. Beebee H. and Dodd J. (eds.) 2005: Truthmakers: The Contemporary Debate. Oxford: Clarendon Press. Bennett, Jonathan 1988: Events and Their Names Indianapolis, Indiana: Hacket. Bigelow, John 1988: The Reality of Numbers New York: Oxford University Press. Black, Max 1944: The Paradox of Analysis. Mind, New Series Vol. 53, No. 211, pp. 263-7. Cheyne, C. and Pigden, C. 2006: Negative Truths from Positive Facts. Australasian Journal of Philosophy Vol. 84, No. 2, pp. 249-65. Cobb, Jeffrey 2001: Problems for Linguistic Solutions to the Paradox of Analysis. Metaphilosophy Vol. 32, No. 4, pp. 419-26. Coburn, R.C. 1986: Individual Essences and Po ssible worlds, in French, Uehling and Wettstein 1986, pp. 165-83. Davidson, D. 1971: Agency, in Essays on Actions and Events Oxford: Clarendon Press, 1980, pp. 43-61. Originally published in 1971 in R. Binkley, R. Bronaugh and A. Marras (eds) Agent, Action, and Reason, Toronto: University of Toronto Press, 1971. : True to the Facts, in Inquiries into Truth and Interpretation Oxford: Clarendon Press, 1984, pp. 37-55. Originally published in 1969 in Journal of Philosophy 66, pp. 748-64. Frege, G. 1892: On Sense and Reference, in P. Geach and M. Black (eds.) Translations from the Writings of Gottlob Frege Oxford: Basil Blackwell, 1960, pp. 56-78. Originally published in German in 1892. French, P. A., Uehling, T. E., Jr. and Wettstein, H. K. (eds) 1986: Midwest Studies in Philosophy XI Minneapolis: University of Minnesota Press.

PAGE 120

120 Fumerton, Richard A. 1983: The Paradox of Analysis. Philosophy and Phenomenological Research Vol. 43, No. 4. pp. 477-97. Hornsby, J. 2005: Truth without Truthmaking Entities, in H. Beebee and J. Dodd (eds.) Truthmakers: The Contemporary Debate, Oxford: Clarendon Press, 2005, pp. 33-47. Kim, J. 1974: Noncausal Connections, in Kim 1993, pp. 22-32. Originally published in 1974 in Nous 8, pp. 41-52. : Supervenience as a Philosophical C oncept, in Kim 1993, pp. 131-60. Originally published in 1990 in Metaphilosophy 20, pp. 1-27. : Supervenience and Mind: Sele cted Philosophical Essays New York: Cambridge University Press. King, J. 1998: What is a Philosophical Analysis?. Philosophical Studies 90, pp. 155-79. Langton, R. and Lewis, D. 1998: Defining Intrinsic. Philosophy and Phenomenological Research Vol. viii, No. 2, pp. 333-45. Lewis, D. 1986: Against Structural Universals. Australasian Journal of Philosophy Vol. 64, No. 1, pp. 25-46. : Critical Notice of D. M. Armstrong, A Combinatorial Theory of Possibility. Australasian Journal of Philosophy Vol. 70, No. 2, pp. 211-24. : Things Qua Truthmakers, in H. Lill ehammer and G. Rodriguez-Pereyra 2003, pp. 25-39. Lewis, D. and Rosen, G. 2003: Postscript to Things Qua Truthmaker, in H. Lillehammer and G. Rodriguez-Pereyra 2003, pp. 39-42. Liggins, D. 2005: Truthmakers and Explanation, in H. Beebee and J. Dodd 2005, pp.105-15. Lillehammer H. and Rodriguez-Pereyra G. (eds.) 2003: Real Metaphysics: Essays in Honour of D.H. Mellor London and New York: Routledge. Mackie, J.L. (1965). Causes and Conditions, in E. Sosa and M. Tooley (eds.) Causation. Oxford: Oxford University Press, 19 93, pp. 33-55. Originally published in 1965 in American Philosophical Quarterly Vol. 2, No. 4, pp. 245-55 and 261-4. McKay, Thomas J. 1986: Against Constitutional Sufficiency Principles, in French, Uehling and Wettstein 1986, pp. 295-304. Mclaughlin, B. 1989: Type Epiphenomenalism, Type Dualism, and the Causal Priority of the Physical. Philosophical Perspectives Vol. 3, pp. 109-35. Melia, J. 2005: Truthmaking without Truthmak ers, in H. Beebee and J. Dodd 2005, pp. 67-84.

PAGE 121

121 Molnar, G. 2000: Truthmakers for Negative Truths. Australasian Journal of Philosophy Vol. 78, No. 1, pp. 72-86. Moore, G. E. 1942: A Reply to my Cr itics, in Paul Ar thur Schilpp (ed.) The Library of Living Philosophers, Volume IV: The Philosophy of G. E. Moore, Evanston and Chicago: Northwestern University Press, pp. 533-677. Parsons, J. 2006: Discussion Note: Negative Truths from Positive Facts. Australasian Journal of Philosophy Vol. 84, No. 4, pp. 591-602. Pollock, J. 1987: Epistemic Norms, in J. Kim and E. Sosa (eds.) Epistemology: An Anthology MA: Blackwell Publishing, 2000, pp. 192-225. Or iginally published in 1987 in Synthese 71, pp. 61-95. Quine, W.V. 1951: Two Dogmas of Empiricism, in P.K. Moser (ed.) A Priori Knowledge, New York: Oxford University Press, 1987, pp. 42-67. Originally published in 1951 in The Philosophical Review 60, pp. 20-43. White, Morton G. 1945: Analysis and Identity: A Rejoinder. Mind, New Series Vol. 54, No. 216, pp. 357-61.

PAGE 122

122 BIOGRAPHICAL SKETCH W illiam Butchard received a BA in philosophy at the University of Florida (1994), an ME in education (1997), and an MA (2004), and a PhD in Philosophy (2008) at the same university. Upon graduation, he intends to pursue a career in teaching philosophy.