Between Logic and the World, by Bernhard Nickel

Between Logic and the World, by Bernhard Nickel In Between Logic and the World, Bernhard Nickel distinguishes two tasks in understanding generics. The first task is to give a compositional semantics—ideally, one that coheres with independent theories of semantic phenomena like plurality and conjunction. Between Logic and the World undoubtedly makes a substantial contribution to this task. Nickel argues that his proposed semantics allows us to understand logically complex generics as well as generics containing gradable terms. The second task is to give a theory of metaphysical genericity. Nickel explains this with an analogy (p. 13). Just as there is a metaphysical phenomenon underlying the meaning of modals—modality—there is one underlying the meaning of generics—genericity. Just as a semantic theory of modals and a metaphysical theory of modality will constrain and illuminate each other, so too will a semantic theory of generics and a metaphysical theory of genericity. Nickel’s ability to thoughtfully connect issues in compositional semantics with those in philosophy of language and metaphysics makes the book an exciting read, and one to be recommended to any reader interested in those interfaces, not solely those interested in generics. Of course, for anybody interested in generics and genericity, the book is essential and insightful reading. I will focus on Nickel’s second task—giving a theory of metaphysical genericity. What is the target of such a theory? Nickel is concerned with characterizing sentences that contain bare plurals in the sentence-initial position, like (1) (pp. 23-6). These can be distinguished from kind-predications like (2) and capacity-ascriptions like (3): (1) Ravens are black; (2) Baseball was invented in 1839; (3) Frigidaire fridges hold 20 gallons of milk. We can schematically represent bare-plural characterizing generics as ‘Fs are G’. For Nickel, a theory of metaphysical genericity answers the following question: What relation must obtain between (an arbitrary) kind F and (an arbitrary) property G such that the generic ‘Fs are G’ is true? This question is apt to produce puzzlement. If there's one thing we know about generics, it is that they are impressively diverse. Some generics seem to require that all, or almost all, the members of the kind instantiate the ascribed property (for example, ‘Numbers are abstract’), while others seem to require this of relatively few, or even none (for example, ‘Unicorns have horns’). Some seem to answer to cultural conventions (for example, ‘Touchdowns are worth 7 points’) while others seem mind-and-language independent (for example, ‘Ducks lay eggs’). Scepticism is a natural reaction to this diversity. It is the reaction I've defended elsewhere (Liebesman 2011). Nickel's non-sceptical position is ambitious. He aims to provide a theory of genericity that unifies a seemingly disparate class. This ambition is one of the most exciting features of the book. So what is Nickel's answer to the question? It can be easily (if uninformatively) summarized as follows: A generic ‘F's are G’ (where ‘F’ picks out a kind) is true if and only if the property of being G is characteristic of F-kind.Characteristicness is the relation that must obtain between a kind and a property, such that the corresponding generic is true. Importantly for Nickel, not all generics are about kinds, and the truth-conditions for non-kind generics are more complex. (I will return to this.) Here's Nickel's definition of characteristicness: Property P is characteristic for kind K relative to explanatory strategies S if and only if it is possible to explain why P is present among Ks by the most general invocation of one or more of the strategies in S. There's a lot to unpack in the definition of characteristicness and, with one minor exception, the rest of my discussion will be dedicated to unpacking and challenging it. I'll begin with some explication. Focus on the left-hand side of the definitional biconditional. Note there that Nickel hasn't provided us with a straightforward dyadic relation between kinds and properties, as was suggested by my initial gloss of Nickel’s answer to the problem of metaphysical genericity. Rather, he's provided us with a triadic relation between kinds, properties, and explanatory strategies. One reason that Nickel invokes the strategies is to account for the context sensitivity of generics. ‘Dobermans have pointy ears’, he claims, is false relative to a context in which we are discussing dog-breeding but true relative to a context in which we discuss dog-showing (p. 181). This is because when we discuss dog-breeding we focus on the genetic endowment of Dobermans, while when we discuss dog-showing we discuss their actual appearance. As a matter of genetics, Dobermans are disposed to have floppy ears, but as a matter of dog-show aesthetics, their ears are made pointy. Another reason that Nickel invokes explanatory strategies is to try to avoid overgenerating true generics. I will return to this. Still focusing on the left-hand side of the biconditional, the next natural question is ‘What are explanatory strategies?’ For now, some examples will suffice. We can explain the behaviour of baseball players, at least partly, in terms of the rules of the game. Why do batters vacate the box after the third strike? Because they are out. We can explain the colour of London moths in terms of adaptive biology. Why are London moths black? Because they have adapted to match the colour of soot. The explanatory strategies invoked in evaluating a generic can be perfectly objective or derivative on conventions, and encompass a variety of explanations that can be given using similar theoretical tools—for example, those of adaptive biology or the rules of baseball. The fact that explanatory strategies can be either mind-and-language independent or parasitic on our practices goes some way towards showing how Nickel's theory attempts to unify the disparate class: the diversity of explanatory strategies is reflected in the diversity of true generics. The right-hand side of the definition contains the unfamiliar notions of ‘being present in’ and ‘the most general invocation’ of some explanatory strategies. It also contains a possibility modal and the notion of explanation. There are lots of varieties of possibility and Nickel doesn't explicitly tell us which he utilizes. I'll assume a fairly expansive notion. He similarly doesn't explicitly tie himself to any particular type or theory of explanation, though he does focus on causal explanations. What about ‘being present in’? We know that there can be true generics even when no member of the kind instantiates the predicated property. As Nickel writes ‘Lions have four legs can be true even if all lions happen to have lost a leg in accidents or by the designs of a madman’ (p. 182). To combat this problem Nickel claims that a property can be present in a kind in virtue of past or future instances instantiating that property (pp. 182-3). Nickel also explicitly argues that merely possible instantiation isn't enough to make a property present in a kind, because he is worried about too much presence. Finally, we can consider the notion of being the most general invocation of some explanatory strategies. Nickel uses this notion to combat a worry about overgeneration. If we can explain, using some explanatory strategy S, why people pursue their goals, then it seems that we can use the same strategy to explain why it is that people pursue some very specific goal, for example, peanut butter. However, this yields a problem for Nickel: the generic ‘People pursue peanut butter’ is false, even though we can explain why some people do pursue peanut butter (p. 194). In order to preserve the truth of ‘People pursue their goals’ while rejecting ‘People pursue peanut butter’, Nickel argues that explanatory strategies must be invoked in the most general manner. Nickel's idea is that if we can explain why a property P is present by invoking some explanatory factor f, and we can explain why a property P′ is present in that same kind, using only determinates of f, then our explanation of P is more general than our explanation of P′ (p. 194). This is a bit confusing given that Nickel doesn't explicate his notion of an explanatory factor, nor is it obvious what it is for a factor to have a determinable. The most straightforward view is to take explanatory factors to be properties invoked in explanations. The idea would be that when we explain the pursuit of goals, we use some explanations that invoke given properties. When we explain the pursuit of peanut butter we use only determinates of those properties. With Nickel’s account of characteristicness explicated, we can now turn to objections. Nickel's main concern is that his account will overgenerate: it will treat some false generics as true. It is easy to see why overgeneration would be a worry. Almost every fact is explainable in some way or another. The worry is that just about whenever a property is present in some kind, we'll be able to explain its presence. This, in turn, risks making just about any property present in a kind characteristic of that kind. Later I'll discuss a few versions of this worry. Before doing so, I'll turn to a worry that Nickel doesn't consider in the book, namely, that his account undergenerates true generics. There are three sorts of cases in which this might happen. The first sort of case stems from Nickel’s notion of being present in. There are true generics ‘Fs are G’, even when no past, present, or future F is G. Imagine that the engineers at Toyota create a new SUV model: the Aggressor. After production has started, but before there are any completed SUVs, it is true that Aggressors have four wheels. Now imagine that there's an apocalypse before any Aggressors can be completed. The generic was true before the apocalypse despite the fact that no Aggressor ever had, has, or will have four wheels. Nickel's restriction on being present in gives the wrong result: having four wheels is not present in Aggressor-kind, so the property is not characteristic of the kind, so the generic is predicted to be false. By Nickel's account of being present in, the property of being G is not present in F-kind in such examples. This means that the property cannot be characteristic of the kind, and therefore that, on Nickel's view, the generic is false. Something has to be done to account for these examples, though perhaps tinkering with the notion of being present in may do the trick. (In Chapter 3 Nickel does discuss the modal import of generics and he adds a counterfactual element to his official normality-incorporating truth-conditions. However, he never does anything similar for characteristicness, so we're left with a worry that his normality-based account and his theory of characteristicness do not match perfectly with one another.) The second sort of case involves fundamental facts themselves. No matter which type of explanation we consider, it seems that there will be some unexplained facts. Perhaps these involve fundamental physics. Given a few relevant unexplained facts, we may be able to construct a true generic. ‘Quarks have charge’ is a plausible example. By hypothesis, this is unexplained so having charge cannot be characteristic of Quark-kind. However, the generic is true. This suggests that not all true generics require the existence of explanations and that Nickel’s link between characteristicness and the truth-conditions of generics undergenerates true generics. The third sort of case involves what Nickel refers to as the ‘distinctive grain’ of generics. We've already discussed this by contrasting ‘People pursue their goals’ with ‘People pursue peanut butter’. The former is true while the latter is not. The challenge for Nickel was that—assuming some people pursue peanut butter—it seems as if the same sorts of mechanisms that explain the pursuit of goals could also explain the pursuit of peanut butter. Nickel's response was to constrain characteristicness to depend on the most general invocation of a set of explanatory strategies. The upshot is that if the most general invocation of some strategies explains the presence of goal-pursuit in people, we can't use those same strategies to explain some determinable of goal-pursuit, for example, peanut-butter-pursuit. In general, Nickel predicts that when we have a true generic of the form ‘Fs are G’, we can't get true generics ‘Fs are G′’, where G′ is a determinate of G, or where G′ is a determinable of G. Both assumptions are incompatible with the strategies underlying ‘Fs are G’ being the most general. If G′ is a determinate of G, then using some strategies to explain the presence of G′ will be less general than using those same strategies to explain the presence of G. After all, each of the invoked factors will either be the same, or more specific versions of the same. If, on the other hand, G′ is a determinate of G, then we can't use S to explain G′ on pain of that explanation’s subsuming our explanation of G. In other words, the true generic ‘Fs are G’ rules out the truth of ‘Fs are G′’, where G is a determinate of G′. The problem is that it does look as if there are such generics. Consider ‘Dogs are mammals’ and ‘Dogs are animals’. Both are true generics, though being a mammal is a determinable of being an animal. At least holding the strategies fixed, Nickel cannot treat them both as true. As I stressed earlier, Nickel's main concern in his book is with overgenerating true generics, rather than undergenerating them. The worry is pressing for his account, given that just about every fact has some sort of explanation, in his sense of ‘explanation’. When it comes to dealing with potential overgeneration, Nickel utilizes two strategies: (i) to embrace it; and (ii) to block it. His embrace consists of treating as true a variety of generics that have often been thought to be false. His block consists of separating out a class of generics as derivative and deserving of special treatment. I'm sceptical of the way he pursues each of these strategies. To understand his embracing strategy in outline, consider (4), and (5): (4) Dogs yodel; (5) Basenjis yodel. It seems as if Nickel is forced to treat (4) as true. Here's why. (5) is true. Given the truth of (5) it is possible to explain why yodelling is present among dogs by the most general invocation of whatever strategy allows us to explain the truth of (5). In other words, given that yodelling is characteristic of Basenjis, it seems that it must also be characteristic of dogs. Most theorists are inclined to reject (4) as well as related generics like ‘Books are paperbacks’ and ‘Birds waddle’. Nickel embraces all these. He argues that our reluctance to endorse them comes not from their falsity, but, rather, from the fact that in most utterances they give rise to a dubious scalar implicature. If all this is true, then Nickel's theory isn't overgenerating true generics: it is just generating them. Nickel's embracing strategy goes hand in hand with one of the most exciting and controversial claims in the book: there's a non-trivial (and non-non-monotonic) logic of generics. In particular, he claims that the following principle, which he calls ‘kind percolation’, holds for generics: if As and Bs are kinds, and the As are a subkind of the Bs, then As are F entails Bs are F. So, from the fact that penguins waddle, and that penguins are a subkind of birds, it follows that birds waddle. Kind percolation is controversial (and rejected by most), for reasons captured by our examples. Nickel's defence of kind percolation—and his embracing strategy—consists of two components. The first component appeals to negation data to motivate the claim that generics like ‘Dogs yodel’ are true. The second component appeals to scalar implicature to explain why we often hesitate to assert and endorse them. I’ll focus my remarks on this second component. According to Nickel, generics like ‘Dogs yodel’ give rise to false quantity implicatures, and this is why we hesitate to assert or endorse them. In particular, ‘Dogs yodel’, according to Nickel, gives rise to the implicature that yodelling is the only means of vocalization for dogs. Nickel stresses that such scalar implicatures are relatively common. ‘John cleaned the living room’, for instance, gives rise to the implicature that John didn’t clean any other room. There's a problem with this strategy: it would apply equally to uncontroversially true generics. ‘Birds fly’ is true and normal speakers don't hesitate to endorse or assert it. Nickel's view doesn't predict that: if he's correct then ‘Birds fly’ should have the same status as ‘Birds waddle’. It will give rise to the implicature that flying is the only means of locomotion for birds, and according to Nickel that's false because birds waddle. The lesson isn't that generics fail to give rise to quantity implicatures of the sort Nickel invokes. Rather, the lesson is that one can't both hypothesize that they give rise to quantity implicatures and endorse generics like ‘Birds waddle’ on pain of predicting, of myriads of familiar generics we're happy to assert, that they should seem just as odd as ‘Birds waddle’. We’ve now seen that when faced with some purported overgeneration, one of Nickel’s strategies is to claim that the counterintuitive generics really are true. Now let’s turn to his second strategy: attempting to block such overgeneration. To understand Nickel's blocking strategy, consider (6) and (7): (6) Ravens are white; (7) Albino ravens are white. There's a worry that, given the truth of (7), Nickel is forced to treat (6) as true. In his own words: … assume for the purpose of reductio that the mapping from characteristic properties to mechanisms and normality held here, as well. The semantics of generics deliver that, since [(7)] is true, being white is one way of being a normally colored albino raven. The mapping connecting normality and characteristic properties entails that because being white is a way of being a normally colored albino raven, the presence of whiteness among at least some albino ravens can be explained by invoking the explanatory strategies of evolutionary biology. But then we’ve just explained the presence of whiteness in at least some ravens by invoking the explanatory strategies of evolutionary biology, since albino ravens are, after all, ravens. Hence, being white is a characteristic property of ravens. Hence, being white is one way of being a normally colored raven. Hence, [(7)] entails [(6)]. (p. 204) Nickel goes on to claim that he has only two options. The first option is to claim that we shift explanatory strategies when moving between (6) and (7). He finds this option implausible. The second option is to reject the claim that the mapping between characteristicness and normality holds for (7). This is his preferred option. The idea is that (7) is a derivative generic, and its truth doesn't allow us to immediately conclude anything about normality for ravens. So how does Nickel account for the truth of (7)? He takes it that the truth of (7), along with other generics concerning non-kinds, is derivative on a corresponding generic concerning a kind (p. 205). The idea is that the non-kind term ‘Albino ravens’ is derived from a kind-term (‘ravens’). We then identify a true generic involving that kind-term and the same dimension as the derivative generic—in this case, colour. The relevant true generic is ‘Ravens are black’. (7) is true just in case its corresponding kind-involving generic (‘Ravens are black’) is true in virtue of some mechanism m, and there is another mechanism m′, just like m except that it operates in the derived non-kind, and it ensures that the members of the non-kind have the colour in question—in this case they are white. The result of this is that Nickel can avoid holding that (7) entails (6), since the truth-conditions of (7) don’t ensure that being white is characteristic of raven-kind, relative to familiar biological explanatory strategies. There’s a problem, though. Even if Nickel avoids holding that (7) entails(6), he doesn’t avoid holding that there is a true reading of (6). After all, whiteness is present in ravens, and it is possible to explain why using the most general invocation of whatever strategy it is which explains why whiteness is present in albino ravens—let’s call it ‘S’. S isn’t the explanatory strategy that Nickel likes to focus on, but the strategy exists nonetheless, and it seems that we can happily focus on it and generate a true reading of (6). To avoid generating a true reading of (6), Nickel must place some constraints on which explanatory strategies are available. Here, Nickel relies on deference. Roughly, the idea is that we defer to biologists when discussing ravens, and that deference will ensure that the relevant explanatory strategies concern ravens in general, and not albino ravens specifically. Nickel acknowledges the metasemantic mysteries involving such deference, but claims that we are perfectly willing to embrace a deference-based metasemantics for names and kinds terms, so we shouldn't worry so much about it in the case of explanatory strategies (p. 186). I'm less sanguine than Nickel about his reliance on deference. As he notes elsewhere, generics seem to have relatively stable interpretations. On Nickel's view, this means that deference must secure strategies with relative stability. How could this be? We have an idea of how it could work with proper names or natural kind terms. In the case of ‘water’ we have some sample, along with a relation (same liquid as) that determines its extension. The relation, in turn, is determined by deference to experts, the world, or both. On such a model, ‘water’ is not context sensitive. We could try to generalize this to generics. Perhaps there is a single explanation that serves as an anchor and a relation (same type as) that determines a set of explanatory strategies. However, Nickel doesn't think that the sort of deference relevant to interpreting generics works like this. It’s not as if there is a single set of explanatory strategies associated with each generic across all contexts—for Nickel, generics are context sensitive and that sensitivity is sourced to a contextual shift in explanatory strategies. So, what work is deference doing for Nickel? Is it assigning a value to the context sensitive strategies variable in each particular context, or do we pick out paradigmatic explanations in each context and deference then uses those explanations to secure strategies? Or, perhaps deference is utilized in both of these roles. Either way, Nickel needs to tell some story about how S is ruled out as an available explanatory strategy for ‘Ravens are black’. The problem doesn't stop here, though. After all, it seems as if almost every fact we can think of is explainable in one way or another. If, in principle, we are allowed to select any of these strategies, then it seems as if whenever some F is G, and that fact is explainable, then, on Nickel's theory, there is a true reading of ‘Fs are G’. Here's an argument to that effect: G is characteristic of F-kind if and only if Fs are G (assumption from Nickel's view, for reductio); Some F-member (call it a) is G, and that a is G is not a fundamental/uncaused fact (assumption); There is some metaphysical/causal explanation of a's being G, call it ME (from 2); ME is an instance of the most general invocation of some explanatory strategy S; G is present among Fs (from 2); It is possible to explain 5 by the most general invocation of S (from 3 and 4); G is characteristic of F-kind (from 5 and Nickel's account of characteristicness); Fs are G relative to S (from 6 and 1). 4 is the controversial premise in this argument, but it is hard to see how Nickel could block it. We can begin to see why 4 is plausible by thinking about the notion of an explanatory strategy. Earlier I introduced the notion through examples, and Nickel does the same. However, he does give us another hint about his ontology of explanatory strategies: … we can think of sets of explanatory strategies as sets of explanations instantiating them, unified by a suitable coherence relation … (p. 187) Sets are pretty easy to come by, and on the ontology of explanatory strategies suggested in this passage, strategies are easy to come by as well. Consider the singleton of ME: that's a strategy with just one instance. Since ME is fairly specific, we cannot use it to explain why any determinable or determinate of G is present in F-kind. Therefore, relative to {ME}, it follows that Fs are G. I see two options for Nickel. One option is to exclude {ME} as an explanatory strategy. If he chooses this option then it is incumbent upon him to tell us exactly what qualifications must be met by a set of explanations (or a set of sets of explanations) to qualify as an explanatory strategy. The other option is for Nickel to embrace the argument presented above, but try to diminish its importance. On this view, Nickel admits that whenever some member of a kind K instantiates a property P, there is a true reading of ‘Ks are P’. However, he can insist that the true reading is extremely hard, perhaps even practically impossible, to come by. So, insofar as we are interested in capturing actual judgments about generics, the possibility of far-flung readings is irrelevant. Here again, Nickel must rely on the magic of deference to rule out the unintended readings. I’m generally sceptical about whether deference can play such a role, and I’m specifically sceptical about whether invoking the deference involved in interpreting names or natural kind terms helps. I opened this review by observing that Nickel is optimistic. He thinks that there's a relation—characteristicness—that holds between kinds and properties whenever the corresponding generic is true and that we can give an informative metaphysical theory of that relation. I'm less optimistic about this project. On my view, looking for an informative metaphysical theory of characteristicness is just as promising as looking for an informative metaphysical theory of the conditions under which a complex object instantiates a property that happens to be instantiated by one of its parts. Such property inheritance, on my view, is unsystematic and highly dependent upon the particular property and object in question. On Nickel's view, characteristicness is not so unsystematic—it’s dependent on the existence of some relevant sorts of explanations. I've criticized this from two angles. On the one hand, there may not be enough explanations. The properties of quarks may not be explainable even though there are true generics about quarks. On the other hand, there may be too many explanations. The vast majority of familiar facts are not causally or metaphysically fundamental, so the explanations exist. Nickel could argue that those explanations aren't available, or aren't of the correct sort, but I don't see how he can make this plausible. Objections aside, Nickel’s view is the clearest and most comprehensive extant defence of optimism about the metaphysics of genericity. Anybody with an interest in that project must read Between Logic and the World, and all subsequent discussion of metaphysical genericity should start with his work. Whether or not one endorses Nickel’s views, his discussion contains a multitude of important insights about generics and genericity. © Mind Association 2017 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Mind Oxford University Press

Between Logic and the World, by Bernhard Nickel

Mind , Volume 127 (505) – Jan 1, 2018

Loading next page...
 
/lp/ou_press/between-logic-and-the-world-by-bernhard-nickel-B7DYG00Ajh
Publisher
Oxford University Press
Copyright
© Mind Association 2017
ISSN
0026-4423
eISSN
1460-2113
D.O.I.
10.1093/mind/fzx024
Publisher site
See Article on Publisher Site

Abstract

In Between Logic and the World, Bernhard Nickel distinguishes two tasks in understanding generics. The first task is to give a compositional semantics—ideally, one that coheres with independent theories of semantic phenomena like plurality and conjunction. Between Logic and the World undoubtedly makes a substantial contribution to this task. Nickel argues that his proposed semantics allows us to understand logically complex generics as well as generics containing gradable terms. The second task is to give a theory of metaphysical genericity. Nickel explains this with an analogy (p. 13). Just as there is a metaphysical phenomenon underlying the meaning of modals—modality—there is one underlying the meaning of generics—genericity. Just as a semantic theory of modals and a metaphysical theory of modality will constrain and illuminate each other, so too will a semantic theory of generics and a metaphysical theory of genericity. Nickel’s ability to thoughtfully connect issues in compositional semantics with those in philosophy of language and metaphysics makes the book an exciting read, and one to be recommended to any reader interested in those interfaces, not solely those interested in generics. Of course, for anybody interested in generics and genericity, the book is essential and insightful reading. I will focus on Nickel’s second task—giving a theory of metaphysical genericity. What is the target of such a theory? Nickel is concerned with characterizing sentences that contain bare plurals in the sentence-initial position, like (1) (pp. 23-6). These can be distinguished from kind-predications like (2) and capacity-ascriptions like (3): (1) Ravens are black; (2) Baseball was invented in 1839; (3) Frigidaire fridges hold 20 gallons of milk. We can schematically represent bare-plural characterizing generics as ‘Fs are G’. For Nickel, a theory of metaphysical genericity answers the following question: What relation must obtain between (an arbitrary) kind F and (an arbitrary) property G such that the generic ‘Fs are G’ is true? This question is apt to produce puzzlement. If there's one thing we know about generics, it is that they are impressively diverse. Some generics seem to require that all, or almost all, the members of the kind instantiate the ascribed property (for example, ‘Numbers are abstract’), while others seem to require this of relatively few, or even none (for example, ‘Unicorns have horns’). Some seem to answer to cultural conventions (for example, ‘Touchdowns are worth 7 points’) while others seem mind-and-language independent (for example, ‘Ducks lay eggs’). Scepticism is a natural reaction to this diversity. It is the reaction I've defended elsewhere (Liebesman 2011). Nickel's non-sceptical position is ambitious. He aims to provide a theory of genericity that unifies a seemingly disparate class. This ambition is one of the most exciting features of the book. So what is Nickel's answer to the question? It can be easily (if uninformatively) summarized as follows: A generic ‘F's are G’ (where ‘F’ picks out a kind) is true if and only if the property of being G is characteristic of F-kind.Characteristicness is the relation that must obtain between a kind and a property, such that the corresponding generic is true. Importantly for Nickel, not all generics are about kinds, and the truth-conditions for non-kind generics are more complex. (I will return to this.) Here's Nickel's definition of characteristicness: Property P is characteristic for kind K relative to explanatory strategies S if and only if it is possible to explain why P is present among Ks by the most general invocation of one or more of the strategies in S. There's a lot to unpack in the definition of characteristicness and, with one minor exception, the rest of my discussion will be dedicated to unpacking and challenging it. I'll begin with some explication. Focus on the left-hand side of the definitional biconditional. Note there that Nickel hasn't provided us with a straightforward dyadic relation between kinds and properties, as was suggested by my initial gloss of Nickel’s answer to the problem of metaphysical genericity. Rather, he's provided us with a triadic relation between kinds, properties, and explanatory strategies. One reason that Nickel invokes the strategies is to account for the context sensitivity of generics. ‘Dobermans have pointy ears’, he claims, is false relative to a context in which we are discussing dog-breeding but true relative to a context in which we discuss dog-showing (p. 181). This is because when we discuss dog-breeding we focus on the genetic endowment of Dobermans, while when we discuss dog-showing we discuss their actual appearance. As a matter of genetics, Dobermans are disposed to have floppy ears, but as a matter of dog-show aesthetics, their ears are made pointy. Another reason that Nickel invokes explanatory strategies is to try to avoid overgenerating true generics. I will return to this. Still focusing on the left-hand side of the biconditional, the next natural question is ‘What are explanatory strategies?’ For now, some examples will suffice. We can explain the behaviour of baseball players, at least partly, in terms of the rules of the game. Why do batters vacate the box after the third strike? Because they are out. We can explain the colour of London moths in terms of adaptive biology. Why are London moths black? Because they have adapted to match the colour of soot. The explanatory strategies invoked in evaluating a generic can be perfectly objective or derivative on conventions, and encompass a variety of explanations that can be given using similar theoretical tools—for example, those of adaptive biology or the rules of baseball. The fact that explanatory strategies can be either mind-and-language independent or parasitic on our practices goes some way towards showing how Nickel's theory attempts to unify the disparate class: the diversity of explanatory strategies is reflected in the diversity of true generics. The right-hand side of the definition contains the unfamiliar notions of ‘being present in’ and ‘the most general invocation’ of some explanatory strategies. It also contains a possibility modal and the notion of explanation. There are lots of varieties of possibility and Nickel doesn't explicitly tell us which he utilizes. I'll assume a fairly expansive notion. He similarly doesn't explicitly tie himself to any particular type or theory of explanation, though he does focus on causal explanations. What about ‘being present in’? We know that there can be true generics even when no member of the kind instantiates the predicated property. As Nickel writes ‘Lions have four legs can be true even if all lions happen to have lost a leg in accidents or by the designs of a madman’ (p. 182). To combat this problem Nickel claims that a property can be present in a kind in virtue of past or future instances instantiating that property (pp. 182-3). Nickel also explicitly argues that merely possible instantiation isn't enough to make a property present in a kind, because he is worried about too much presence. Finally, we can consider the notion of being the most general invocation of some explanatory strategies. Nickel uses this notion to combat a worry about overgeneration. If we can explain, using some explanatory strategy S, why people pursue their goals, then it seems that we can use the same strategy to explain why it is that people pursue some very specific goal, for example, peanut butter. However, this yields a problem for Nickel: the generic ‘People pursue peanut butter’ is false, even though we can explain why some people do pursue peanut butter (p. 194). In order to preserve the truth of ‘People pursue their goals’ while rejecting ‘People pursue peanut butter’, Nickel argues that explanatory strategies must be invoked in the most general manner. Nickel's idea is that if we can explain why a property P is present by invoking some explanatory factor f, and we can explain why a property P′ is present in that same kind, using only determinates of f, then our explanation of P is more general than our explanation of P′ (p. 194). This is a bit confusing given that Nickel doesn't explicate his notion of an explanatory factor, nor is it obvious what it is for a factor to have a determinable. The most straightforward view is to take explanatory factors to be properties invoked in explanations. The idea would be that when we explain the pursuit of goals, we use some explanations that invoke given properties. When we explain the pursuit of peanut butter we use only determinates of those properties. With Nickel’s account of characteristicness explicated, we can now turn to objections. Nickel's main concern is that his account will overgenerate: it will treat some false generics as true. It is easy to see why overgeneration would be a worry. Almost every fact is explainable in some way or another. The worry is that just about whenever a property is present in some kind, we'll be able to explain its presence. This, in turn, risks making just about any property present in a kind characteristic of that kind. Later I'll discuss a few versions of this worry. Before doing so, I'll turn to a worry that Nickel doesn't consider in the book, namely, that his account undergenerates true generics. There are three sorts of cases in which this might happen. The first sort of case stems from Nickel’s notion of being present in. There are true generics ‘Fs are G’, even when no past, present, or future F is G. Imagine that the engineers at Toyota create a new SUV model: the Aggressor. After production has started, but before there are any completed SUVs, it is true that Aggressors have four wheels. Now imagine that there's an apocalypse before any Aggressors can be completed. The generic was true before the apocalypse despite the fact that no Aggressor ever had, has, or will have four wheels. Nickel's restriction on being present in gives the wrong result: having four wheels is not present in Aggressor-kind, so the property is not characteristic of the kind, so the generic is predicted to be false. By Nickel's account of being present in, the property of being G is not present in F-kind in such examples. This means that the property cannot be characteristic of the kind, and therefore that, on Nickel's view, the generic is false. Something has to be done to account for these examples, though perhaps tinkering with the notion of being present in may do the trick. (In Chapter 3 Nickel does discuss the modal import of generics and he adds a counterfactual element to his official normality-incorporating truth-conditions. However, he never does anything similar for characteristicness, so we're left with a worry that his normality-based account and his theory of characteristicness do not match perfectly with one another.) The second sort of case involves fundamental facts themselves. No matter which type of explanation we consider, it seems that there will be some unexplained facts. Perhaps these involve fundamental physics. Given a few relevant unexplained facts, we may be able to construct a true generic. ‘Quarks have charge’ is a plausible example. By hypothesis, this is unexplained so having charge cannot be characteristic of Quark-kind. However, the generic is true. This suggests that not all true generics require the existence of explanations and that Nickel’s link between characteristicness and the truth-conditions of generics undergenerates true generics. The third sort of case involves what Nickel refers to as the ‘distinctive grain’ of generics. We've already discussed this by contrasting ‘People pursue their goals’ with ‘People pursue peanut butter’. The former is true while the latter is not. The challenge for Nickel was that—assuming some people pursue peanut butter—it seems as if the same sorts of mechanisms that explain the pursuit of goals could also explain the pursuit of peanut butter. Nickel's response was to constrain characteristicness to depend on the most general invocation of a set of explanatory strategies. The upshot is that if the most general invocation of some strategies explains the presence of goal-pursuit in people, we can't use those same strategies to explain some determinable of goal-pursuit, for example, peanut-butter-pursuit. In general, Nickel predicts that when we have a true generic of the form ‘Fs are G’, we can't get true generics ‘Fs are G′’, where G′ is a determinate of G, or where G′ is a determinable of G. Both assumptions are incompatible with the strategies underlying ‘Fs are G’ being the most general. If G′ is a determinate of G, then using some strategies to explain the presence of G′ will be less general than using those same strategies to explain the presence of G. After all, each of the invoked factors will either be the same, or more specific versions of the same. If, on the other hand, G′ is a determinate of G, then we can't use S to explain G′ on pain of that explanation’s subsuming our explanation of G. In other words, the true generic ‘Fs are G’ rules out the truth of ‘Fs are G′’, where G is a determinate of G′. The problem is that it does look as if there are such generics. Consider ‘Dogs are mammals’ and ‘Dogs are animals’. Both are true generics, though being a mammal is a determinable of being an animal. At least holding the strategies fixed, Nickel cannot treat them both as true. As I stressed earlier, Nickel's main concern in his book is with overgenerating true generics, rather than undergenerating them. The worry is pressing for his account, given that just about every fact has some sort of explanation, in his sense of ‘explanation’. When it comes to dealing with potential overgeneration, Nickel utilizes two strategies: (i) to embrace it; and (ii) to block it. His embrace consists of treating as true a variety of generics that have often been thought to be false. His block consists of separating out a class of generics as derivative and deserving of special treatment. I'm sceptical of the way he pursues each of these strategies. To understand his embracing strategy in outline, consider (4), and (5): (4) Dogs yodel; (5) Basenjis yodel. It seems as if Nickel is forced to treat (4) as true. Here's why. (5) is true. Given the truth of (5) it is possible to explain why yodelling is present among dogs by the most general invocation of whatever strategy allows us to explain the truth of (5). In other words, given that yodelling is characteristic of Basenjis, it seems that it must also be characteristic of dogs. Most theorists are inclined to reject (4) as well as related generics like ‘Books are paperbacks’ and ‘Birds waddle’. Nickel embraces all these. He argues that our reluctance to endorse them comes not from their falsity, but, rather, from the fact that in most utterances they give rise to a dubious scalar implicature. If all this is true, then Nickel's theory isn't overgenerating true generics: it is just generating them. Nickel's embracing strategy goes hand in hand with one of the most exciting and controversial claims in the book: there's a non-trivial (and non-non-monotonic) logic of generics. In particular, he claims that the following principle, which he calls ‘kind percolation’, holds for generics: if As and Bs are kinds, and the As are a subkind of the Bs, then As are F entails Bs are F. So, from the fact that penguins waddle, and that penguins are a subkind of birds, it follows that birds waddle. Kind percolation is controversial (and rejected by most), for reasons captured by our examples. Nickel's defence of kind percolation—and his embracing strategy—consists of two components. The first component appeals to negation data to motivate the claim that generics like ‘Dogs yodel’ are true. The second component appeals to scalar implicature to explain why we often hesitate to assert and endorse them. I’ll focus my remarks on this second component. According to Nickel, generics like ‘Dogs yodel’ give rise to false quantity implicatures, and this is why we hesitate to assert or endorse them. In particular, ‘Dogs yodel’, according to Nickel, gives rise to the implicature that yodelling is the only means of vocalization for dogs. Nickel stresses that such scalar implicatures are relatively common. ‘John cleaned the living room’, for instance, gives rise to the implicature that John didn’t clean any other room. There's a problem with this strategy: it would apply equally to uncontroversially true generics. ‘Birds fly’ is true and normal speakers don't hesitate to endorse or assert it. Nickel's view doesn't predict that: if he's correct then ‘Birds fly’ should have the same status as ‘Birds waddle’. It will give rise to the implicature that flying is the only means of locomotion for birds, and according to Nickel that's false because birds waddle. The lesson isn't that generics fail to give rise to quantity implicatures of the sort Nickel invokes. Rather, the lesson is that one can't both hypothesize that they give rise to quantity implicatures and endorse generics like ‘Birds waddle’ on pain of predicting, of myriads of familiar generics we're happy to assert, that they should seem just as odd as ‘Birds waddle’. We’ve now seen that when faced with some purported overgeneration, one of Nickel’s strategies is to claim that the counterintuitive generics really are true. Now let’s turn to his second strategy: attempting to block such overgeneration. To understand Nickel's blocking strategy, consider (6) and (7): (6) Ravens are white; (7) Albino ravens are white. There's a worry that, given the truth of (7), Nickel is forced to treat (6) as true. In his own words: … assume for the purpose of reductio that the mapping from characteristic properties to mechanisms and normality held here, as well. The semantics of generics deliver that, since [(7)] is true, being white is one way of being a normally colored albino raven. The mapping connecting normality and characteristic properties entails that because being white is a way of being a normally colored albino raven, the presence of whiteness among at least some albino ravens can be explained by invoking the explanatory strategies of evolutionary biology. But then we’ve just explained the presence of whiteness in at least some ravens by invoking the explanatory strategies of evolutionary biology, since albino ravens are, after all, ravens. Hence, being white is a characteristic property of ravens. Hence, being white is one way of being a normally colored raven. Hence, [(7)] entails [(6)]. (p. 204) Nickel goes on to claim that he has only two options. The first option is to claim that we shift explanatory strategies when moving between (6) and (7). He finds this option implausible. The second option is to reject the claim that the mapping between characteristicness and normality holds for (7). This is his preferred option. The idea is that (7) is a derivative generic, and its truth doesn't allow us to immediately conclude anything about normality for ravens. So how does Nickel account for the truth of (7)? He takes it that the truth of (7), along with other generics concerning non-kinds, is derivative on a corresponding generic concerning a kind (p. 205). The idea is that the non-kind term ‘Albino ravens’ is derived from a kind-term (‘ravens’). We then identify a true generic involving that kind-term and the same dimension as the derivative generic—in this case, colour. The relevant true generic is ‘Ravens are black’. (7) is true just in case its corresponding kind-involving generic (‘Ravens are black’) is true in virtue of some mechanism m, and there is another mechanism m′, just like m except that it operates in the derived non-kind, and it ensures that the members of the non-kind have the colour in question—in this case they are white. The result of this is that Nickel can avoid holding that (7) entails (6), since the truth-conditions of (7) don’t ensure that being white is characteristic of raven-kind, relative to familiar biological explanatory strategies. There’s a problem, though. Even if Nickel avoids holding that (7) entails(6), he doesn’t avoid holding that there is a true reading of (6). After all, whiteness is present in ravens, and it is possible to explain why using the most general invocation of whatever strategy it is which explains why whiteness is present in albino ravens—let’s call it ‘S’. S isn’t the explanatory strategy that Nickel likes to focus on, but the strategy exists nonetheless, and it seems that we can happily focus on it and generate a true reading of (6). To avoid generating a true reading of (6), Nickel must place some constraints on which explanatory strategies are available. Here, Nickel relies on deference. Roughly, the idea is that we defer to biologists when discussing ravens, and that deference will ensure that the relevant explanatory strategies concern ravens in general, and not albino ravens specifically. Nickel acknowledges the metasemantic mysteries involving such deference, but claims that we are perfectly willing to embrace a deference-based metasemantics for names and kinds terms, so we shouldn't worry so much about it in the case of explanatory strategies (p. 186). I'm less sanguine than Nickel about his reliance on deference. As he notes elsewhere, generics seem to have relatively stable interpretations. On Nickel's view, this means that deference must secure strategies with relative stability. How could this be? We have an idea of how it could work with proper names or natural kind terms. In the case of ‘water’ we have some sample, along with a relation (same liquid as) that determines its extension. The relation, in turn, is determined by deference to experts, the world, or both. On such a model, ‘water’ is not context sensitive. We could try to generalize this to generics. Perhaps there is a single explanation that serves as an anchor and a relation (same type as) that determines a set of explanatory strategies. However, Nickel doesn't think that the sort of deference relevant to interpreting generics works like this. It’s not as if there is a single set of explanatory strategies associated with each generic across all contexts—for Nickel, generics are context sensitive and that sensitivity is sourced to a contextual shift in explanatory strategies. So, what work is deference doing for Nickel? Is it assigning a value to the context sensitive strategies variable in each particular context, or do we pick out paradigmatic explanations in each context and deference then uses those explanations to secure strategies? Or, perhaps deference is utilized in both of these roles. Either way, Nickel needs to tell some story about how S is ruled out as an available explanatory strategy for ‘Ravens are black’. The problem doesn't stop here, though. After all, it seems as if almost every fact we can think of is explainable in one way or another. If, in principle, we are allowed to select any of these strategies, then it seems as if whenever some F is G, and that fact is explainable, then, on Nickel's theory, there is a true reading of ‘Fs are G’. Here's an argument to that effect: G is characteristic of F-kind if and only if Fs are G (assumption from Nickel's view, for reductio); Some F-member (call it a) is G, and that a is G is not a fundamental/uncaused fact (assumption); There is some metaphysical/causal explanation of a's being G, call it ME (from 2); ME is an instance of the most general invocation of some explanatory strategy S; G is present among Fs (from 2); It is possible to explain 5 by the most general invocation of S (from 3 and 4); G is characteristic of F-kind (from 5 and Nickel's account of characteristicness); Fs are G relative to S (from 6 and 1). 4 is the controversial premise in this argument, but it is hard to see how Nickel could block it. We can begin to see why 4 is plausible by thinking about the notion of an explanatory strategy. Earlier I introduced the notion through examples, and Nickel does the same. However, he does give us another hint about his ontology of explanatory strategies: … we can think of sets of explanatory strategies as sets of explanations instantiating them, unified by a suitable coherence relation … (p. 187) Sets are pretty easy to come by, and on the ontology of explanatory strategies suggested in this passage, strategies are easy to come by as well. Consider the singleton of ME: that's a strategy with just one instance. Since ME is fairly specific, we cannot use it to explain why any determinable or determinate of G is present in F-kind. Therefore, relative to {ME}, it follows that Fs are G. I see two options for Nickel. One option is to exclude {ME} as an explanatory strategy. If he chooses this option then it is incumbent upon him to tell us exactly what qualifications must be met by a set of explanations (or a set of sets of explanations) to qualify as an explanatory strategy. The other option is for Nickel to embrace the argument presented above, but try to diminish its importance. On this view, Nickel admits that whenever some member of a kind K instantiates a property P, there is a true reading of ‘Ks are P’. However, he can insist that the true reading is extremely hard, perhaps even practically impossible, to come by. So, insofar as we are interested in capturing actual judgments about generics, the possibility of far-flung readings is irrelevant. Here again, Nickel must rely on the magic of deference to rule out the unintended readings. I’m generally sceptical about whether deference can play such a role, and I’m specifically sceptical about whether invoking the deference involved in interpreting names or natural kind terms helps. I opened this review by observing that Nickel is optimistic. He thinks that there's a relation—characteristicness—that holds between kinds and properties whenever the corresponding generic is true and that we can give an informative metaphysical theory of that relation. I'm less optimistic about this project. On my view, looking for an informative metaphysical theory of characteristicness is just as promising as looking for an informative metaphysical theory of the conditions under which a complex object instantiates a property that happens to be instantiated by one of its parts. Such property inheritance, on my view, is unsystematic and highly dependent upon the particular property and object in question. On Nickel's view, characteristicness is not so unsystematic—it’s dependent on the existence of some relevant sorts of explanations. I've criticized this from two angles. On the one hand, there may not be enough explanations. The properties of quarks may not be explainable even though there are true generics about quarks. On the other hand, there may be too many explanations. The vast majority of familiar facts are not causally or metaphysically fundamental, so the explanations exist. Nickel could argue that those explanations aren't available, or aren't of the correct sort, but I don't see how he can make this plausible. Objections aside, Nickel’s view is the clearest and most comprehensive extant defence of optimism about the metaphysics of genericity. Anybody with an interest in that project must read Between Logic and the World, and all subsequent discussion of metaphysical genericity should start with his work. Whether or not one endorses Nickel’s views, his discussion contains a multitude of important insights about generics and genericity. © Mind Association 2017

Journal

MindOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial