Levels of reasons why and answers to why questions By: Insa Lawler Lawler, I. (2019). Levels of reasons why and answers to why questions. Philosophy of Science 86(1), 168-177. Made available courtesy of University of Chicago Press: https://doi.org/10.1086/701073 ***© 2019 Philosophy of Science Association. Reprinted with permission. No further reproduction is authorized without written permission from University of Chicago Press. This version of the document is not the version of record. Figures and/or pictures may be missing from this format of the document. *** Abstract: According to Skow, correct answers to why questions cite only causes or grounds, but not nonaccidental regularities. Accounts that cite nonaccidental regularities typically confuse second-level reasons with first-level reasons. Only causes and grounds are first-level reasons why. Nonaccidental regularities are second-level reasons why. I first show that Skow’s arguments for the accusation of confusion depend on the independent thesis that only citations of first-level reasons why are (parts of) answers to why questions. Then I argue that this thesis is false. Consequently, the claim that correct answers to why questions cite only causes or grounds is refuted as well. Keywords: why questions | reasons why | causes and grounds | nonaccidental regularities Article: ***Note: Full text of article below http://libres.uncg.edu/ir/uncg/clist.aspx?id=20624 https://doi.org/10.1086/701073 Levels of Reasons Why and Answers to Why Questions Insa Lawler*y According to Skow, correct answers to why questions cite only causes or grounds, but not nonaccidental regularities. Accounts that cite nonaccidental regularities typically con- fuse second-level reasons with first-level reasons. Only causes and grounds are first-level reasons why. Nonaccidental regularities are second-level reasons why. I first show that Skow’s arguments for the accusation of confusion depend on the independent thesis that only citations of first-level reasons why are (parts of ) answers to why questions. Then I argue that this thesis is false. Consequently, the claim that correct answers to why ques- tions cite only causes or grounds is refuted as well. 1. Introduction. Skow’s theory of reasons why (2016, 2017) is framed in terms of a theory of answers to why questions regarding the “occurrence of a concrete event” (2016, 27). A central claim is that correct answers to such why questions cite only causes or (partial) grounds of the event in question, but not nonaccidental regularities (2016, 3; 2017, 907). This claim is based on a distinction between different levels of reasons why. Only causes and grounds are ‘first-level reasons why’. Nonaccidental regularities are ‘second- level reasons why’. Skow argues that accounts that cite nonaccidental regu- larities in answers to why questions typically confuse second-level reasons with first-level reasons (2016, chap. 4; 2017). This accusation is crucial to Skow’s rebuttals of (putative) noncausal explanations, such as mathematical explanations. *To contact the author, please write to: Institute for Philosophy II, Ruhr University Bo- chum, Bochum, Germany; e-mail: insa.lawler@rub.de. yI thank Raphael van Riel for discussing parts of this paper with me, as well as two anon- ymous reviewers for their constructive criticisms and suggestions. Support for this re- search by the Volkswagen Foundation for the project “A Study in Explanatory Power” and by the OeAD for an Ernst Mach Scholarship is gratefully acknowledged. Received November 2017; revised March 2018. Philosophy of Science, 86 (January 2019) pp. 168–177. 0031-8248/2019/8601-0008$10.00 Copyright 2019 by the Philosophy of Science Association. All rights reserved. 168 DISCUSSION 169 The level distinction plays an important role in Skow’s account. Unsur- prisingly, it has been disputed (e.g., Baumgartner 2017; Pincock 2017; Lange 2018, 36). However, in what follows, I argue that even if the distinction were granted and only causes and grounds were first-level reasons why, Skow’s ac- cusation of confusion and his central claim that correct answers to why ques- tions cite only causes or grounds can be refuted. I first show that Skow’s argu- ments for the accusation as well as his arguments for the central claim crucially depend on the independent thesis that only citations of first-level reasons why are (parts of) answers to why questions. Then I argue that this thesis is false. Skow’s arguments against the claim that some answers to why questions are conjunctions of citations of a first-level reason why and its second-level reason why can be rebutted, and there are plausible arguments in favor of that claim. Like Skow, I am exclusively concerned with why questions concerning particular events. 2. A Putative Confusion of Levels of Reasons Why. According to sev- eral accounts, nonaccidental regularities (such as laws) need to be or could be cited in correct answers to why questions.1 The deductive-nomological (D-N) account (Hempel and Oppenheim 1948), Friedman’s account (1974), Railton’s account (1978), and Kitcher’s account (1989) include law-like state- ments. Salmon and Lewis allow for citing causal regularities (Salmon 1984, 262, 274; Lewis 1986). Woodward and Hitchcock (2003) include invariant generalizations. According to Lange (2013), some answers to why questions appeal to necessities. An explanation for someone’s not being able to cross all of Königsberg’s bridges exactly once cites both that the bridge arrangement has a particular property and that it is necessary that if a bridge arrangement has this property, one cannot cross all of the bridges exactly once (488–91). According to Skow, all these accounts are mistaken. Correct answers to why questions cite only causes or grounds, but not nonaccidental regulari- ties (2016, chap. 4.3; 2017).2 He argues that the mistake is due to a confu- sion of different levels of reasons why (2017). For instance, Woodward’s account “rests on a confusion of levels of reasons why in much the same way the DN model did forty years earlier” (2016, 95). Before I turn to eval- uating this accusation, I introduce the different levels of reasons why. 2.1. Reasons Why. Skow follows accounts that tie why questions to reasons why (2016, 26–27; examples: Achinstein 1975, 29–30; Brogaard 1. For the following discussion, it does not make a difference whether we consider to be decisive the regularities’ nonaccidentalness or the fact that they obtain. What does matter is whether any of them are citable in answers to why questions regarding particular events. 2. This claim is similar to remarks by Scriven (1959), as Skow points out (2016, 84–85). 170 INSA LAWLER 2009, 461; Stanley 2011, 209). The basic form of answers to why questions is “One reason why p is that q.” Reasons why are particular facts (Skow 2016, 35). For instance, that her diaphragm was stimulated is a reason why Juliana has the hiccups. Skow dis- tinguishes between different levels of reasons why (2016, chap. 4.2; 2017, 907): First-Level Reasons Why. A fact F1 is a first-level reason why event E occurred iff F1 is a reason why E occurred. Second-Level Reasons Why. A fact F2 is a second-level reason why iff F2 is a reason why F1 is a reason why E occurred. Citing second-level reasons answers the ‘follow-up’ question “What does F1 have to do with E?” or “Why is F1 a reason why E occurred?” (Skow 2016, 75, 80). Causal background conditions are examples of second-level reasons. Take the question “Why did the match light?” (2016, chap. 4.2; 2017, 907): That someone struck the match is a first-level reason why the match lit. That oxygen was present is a second-level reason why; it is a rea- son why that someone struck the match is a reason why the match lit. Other examples include nonaccidental regularities, such as laws (2016, chap. 4.2; 2017, 908): Skow illustrates this with the question “Why did the rock hit the ground at a speed of 4.4 meters/second?” A first-level reason is that the rock was dropped from a height of 1 meter. A second-level reason is the law s 5 ffiffiffiffiffiffiffiffi 2dg p (s 5 impact speed, d 5 drop height, g 5 gravitational accelera- tion near the earth’s surface ≈ 9.8, 4:4 ≈ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 � 1 � 9:8 p ). Second-level reasons why can be first-level reasons why. Take causal background conditions: That oxygen was present is also a reason why the match lit. However, according to Skow, nonaccidental regularities are not first-level reasons why: Nonaccidental Regularities. Nonaccidental regularities are second-level reasons why, but not first-level reasons why. For instance, that s 5 ffiffiffiffiffiffiffiffi 2dg p is not a reason why the impact speed was 4.4 meters/second (2017, 908). Skow does not argue for Nonaccidental Reg- ularities. He likes “to put the burden on the other side: why think [laws] are [first-level reasons why]?” (2017, 909). Skow further claims that only causes and grounds are first-level reasons why (2016, chap. 3; 2017, 907–9): Causes and Grounds. Only causes and (partial) grounds are first-level reasons why. DISCUSSION 171 His argumentative strategy is to take Causes and Grounds as initially plausible and to defend it against putative counterexamples. I do not discuss Nonaccidental Regularities and Causes and Grounds, but take them for granted (for some discussion, see, e.g., Baumgartner [2017]; Pincock [2017]; Lange [2018, 36]). Instead, in what follows, I challenge Skow’s claims about second- level reasons why and answers to why questions. 2.2. The Accusation of Confusion Revisited. Let us get back to the ac- cusation of confusion. Skow accuses Hempel of “inaugurat[ing] a long tradi- tion of confusing the two levels of reasons” (2016, 81). The alleged confusion underlying the citing of nonaccidental regularities in answers to why ques- tions is to take nonaccidental regularities to be first-level reasons why. Ac- cording to Skow, this confusion is due to particular pragmatic effects (2017, 909): We think that laws are first-level reasons because they are often cited in good responses to an asked why question. Skow illustrates his point as fol- lows: “she asks me why [the rock] hit the ground at 4.4 meters/second. I re- spond, ‘Well, I dropped it from 1 meter up, and impact speed s is related to drop height d by the law s 5 ffiffiffiffiffiffiffiffi 2dg p (and of course ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 � 1 � 9:8 p ≈ 4:4)’” (909). This is a fine response. However, this does not mean that the law is a first-level reason why: My response is a good one, but it does not follow that every part of my response is part of an answer to the question asked. In my view, the first part of my response—“I dropped it from 1 meter”—is an answer to the explicit question (“why did the rock hit the ground at 4.4 meters/sec- ond?”), but the second part, the law, is not. It, instead, is an answer to an unasked follow-up why question, a follow-up question I can anticipate would be asked immediately if I only answered the explicit question. The follow-up is, why is the fact that I dropped it from 1 meter up a reason why it hit the ground at 4.4 meters/second? (Skow 2017, 910–11) So Skow’s stance is that citing neither the law nor its application is part of the answer to the question “Why did the rock hit the ground at 4.4 me- ters/second?” They are just part of a good response to it. The law citation is an answer to an anticipated follow-up question. The confusion of different levels of reasons why is a confusion of good responses and proper answers. The same holds true in the case of the Königsberg bridges example. Skow suggests that Lange mistakenly takes a second-level reason (the necessity of not being able to cross all of the bridge exactly once) to be part of the an- swer to the relevant why question (2016, chap. 5.3; 2017, 913–14). 2.3. Why There Is Probably No Confusion of Levels of Reasons Why. The accusation that so many of Skow’s rival accounts make such an ele- 172 INSA LAWLER mentary confusion is puzzling. In fact, none of those accounts is committed to the claim that nonaccidental regularities are first-level reasons why.3 Here is why: Taking citations of second-level reasons to be parts of answers why p does not mean that one classifies second-level reasons as first-level rea- sons. How one classifies reasons is one thing and what one counts as part of an answer is another. These issues are independent of each other. Claim- ing that correct answers to why questions (could) cite nonaccidental regu- larities is compatible with agreeing that such regularities are not first-level reasons why. Skow’s accusation becomes less puzzling when one realizes that he makes the following assumption (2017, 907): 3. Fo could Why Questions. Correct answers to why questions cite only first-level reasons why. Skow endorses Why Questions at the outset of his account. According to Skow, complete answers as to why p are conjunctions of elements of the form “One reason why p is that q” (2016, 42, 51); that is, they are conjunc- tions of citations of first-level reasons why. Skow is so convinced of Why Questions that he interprets his opponents’ accounts as sharing it. This leads to descriptions of their accounts as clas- sifying all facts that are cited in answers to why questions as first-level rea- sons. For instance, he claims that the D-N account “take[s] certain second- level reasons, laws of nature, to also be first-level reasons” (2017, 908) and that “Salmon’s view . . . is that all the other facts . . . are really part of the answer to the question (in my preferred terms, they really are [first-level] reasons why)” (2016, 93). However, these descriptions are likely to be false. The defenders of rival accounts are not committed to accepting Why Ques- tions. Skow neglects the vivid possibility that his opponents simply reject Why Questions. Skow’s accusation of confusion as well as his claim that correct answers to why questions cite only causes or grounds essentially depend on Why Questions. If second-level reasons why could be cited in answers to why questions, there would be no basis for accusing others of conflating differ- ent levels of reasons why, and the claim that answers to why questions cite only causes or grounds would be refuted as well. So let us turn to examining the plausibility of Why Questions. 3. Second-Level Reasons Why and Answers to Why Questions. As we have seen, Skow assumes that answers to why questions cite only first-level r a different rebuttal of the accusation, see Pincock (2017). He argues that there be a disagreement about the nature of first-level reasons why. DISCUSSION 173 reasons why. He considers this assumption to be uncontroversial (2016, 51). According to him, the only thing that should be controversial is what first-level reasons why are (25). As I have argued, this is not the case. In- stead, it seems to be controversial what kinds of facts are citable in answers to why questions. Although Skow does not give explicit arguments for ex- cluding citations of second-level reasons, his accusation of confusion re- veals two arguments for it. In what follows, I first rebut these arguments. Then I offer arguments against excluding citations of second-level reasons. 3.1. Arguments for Excluding Second-Level Reasons Why. According to Skow, citations of second-level reasons are only parts of a good response to an asked why question, but not proper parts of the answer to it. His rea- soning reveals two arguments for excluding second-level reasons: (i) Citing second-level reasons answers the closely related follow-up question “What does F1 have to do with E?” (2017, 907). (ii) Citations of second-level rea- sons why, taken by themselves, do not answer the question “Why did E oc- cur?” For instance, s 5 ffiffiffiffiffiffiffiffi 2dg p , taken by itself, does not answer why the im- pact speed was 4.4 meters/second. Yet both arguments can be rebutted. The first argument is not compelling because an answer to a question can clearly have parts that are also answers to closely related questions. Take the question “Where did you park your car?” and the reply “I parked it next to the building which is Juliana’s house and the only building in the street with a blue roof.” The second part of the which clause also expresses an answer to “How could one identify Juliana’s house?” But this does not exclude it from being part of the answer to the initial question. The answer simply specifies the location where the car is parked in two complementary ways. The second argument looks more promising. If we accept that non- accidental regularities are not first-level reasons why p, it seems difficult to argue that a citation of them, taken by itself, answers the why question. However, even if so, the second argument can be rebutted. That a proposi- tion, taken by itself, is not an answer to a question does not mean that it could not be a part of an answer. Take the question “How did you rescue her?” and as a reply “I jumped into the water, swam 5 meters, grabbed her, and brought her ashore.” “I swam 5 meters,” taken by itself, is not an answer to the question. But it is nonetheless a part of the conjunction, and the latter is an answer to the question. A third argument Skow might want to offer is that including citations of second-level reasons why would amount to giving up his simple account of answers to why questions. Arguably, the answer “The drop height is 1 meter and s 5 ffiffiffiffiffiffiffiffi 2dg p ” could not be stated in Skow’s preferred form “One reason why p is that q.” A simple account of answers might have some intrinsic value. But the alternative ‘because’ form works fine when one cites both 174 INSA LAWLER first-level and second-level reasons why, for example, “The impact speed was 4.4 meters/second because the drop height is 1 meter and s 5 ffiffiffiffiffiffiffiffi 2dg p .” As Skow himself notes, “‘reasons why’ answers and corresponding ‘be- cause’ answers are often equivalent, in some sense” (2016, 23). If so, there is room for an alternative simple account that features “p because . . .” prop- ositions as the basic unit of answers to why questions. 3.2. Arguments against Excluding Second-Level Reasons Why. In or- der to argue against Why Questions, one might argue that all answers to why questions need to cite nonaccidental regularities, as Hempel would ar- gue (1965, 349). One has to answer the follow-up question “What does F1 have to do with E?” in order to have an answer to the why question. How- ever, it suffices to argue for the weaker claim that some answers to a why question are a conjunction of a citation of a first-level reason why and a ci- tation of its second-level reason why. So let us turn to arguments for such conjunctions. My first argument is concerned with the nature of complete answers to why questions. Questions are not understood as speech acts here, but as the content that is expressed by interrogative sentences (Skow 2016, 26– 27; see also, e.g., van Fraassen 1980, 137–38). Answers are understood as propositions. An important issue discussed is whether answers have to be determined relative to some subject. Are answers always answers for some subject? If they were, one could rebut Why Questions as follows: In at least some cases, what Skow considers to be the complete answer is not a com- plete answer for every subject who lacks additional knowledge. Take the fol- lowing example by Lange (2013): A mother has three children and 23 straw- berries. According to Lange, the facts that jointly explain why she failed to distribute the strawberries equally among them are that she has three chil- dren and 23 strawberries and that three does not divide into 23. According to Skow, the only facts that explain are the first two (2016, 112–13; 2018, 15). So the complete answer to the question “Why did Jane fail to distribute the strawberries equally?” is “There are exactly two reasons why she failed: She has three children and she has 23 strawberries.” As Skow notes himself, “including only the reasons why Jane failed is apt to leave one’s audience unsatisfied” (2016, 114). This is understated: Arguably, the putative com- plete answer is unsatisfying for everyone who does not know (or recognize the relevance of the fact) that three does not divide into 23. But if knowledge or recognition of the latter is necessary for considering an answer to the question satisfying, it seems plausible that the fact that three does not divide into 23 is cited in the complete answer as to why Jane failed to distribute the strawberries equally. Something similar could be true in the impact speed ex- ample. Just stating that the drop height was 1 meter is not a complete answer for everyone who does not know how drop height and impact speed are re- DISCUSSION 175 lated. A complete answer to this question cites the law or some other relevant nonaccidental regularity. (Again, this does not mean that the law is a first- level reason why.) This argument is not a pragmatic argument; it is based on an assumption about the semantics of answers. What if answers were not determined relative to some subject? In this case, one could rebut Why Questions as follows: Every question has a sub- ject matter, that is, a queried property or relation. For instance, the subject matter of “Who robbed the bank?” is the property of being someone who robbed the bank in question. A complete answer to this question specifies all entities who fulfill this property. What is the subject matter of a why ques- tion? One proposal that is general enough to capture the variety of why ques- tions is that the subject matter of a why question is the property of being something the event in question (explanatorily) depends on (adopted from Kim [1994]). First-level reasons why fulfill this property. However, this also holds true for second-level reasons why. That Jane failed to distribute the strawberries equally among her children also depends on the fact that three does not divide into 23 (irrespective of whether the latter is a necessary or a contingent fact).4 That the impact speed was 4.4 meters/second also depends on the fact that s 5 ffiffiffiffiffiffiffiffi 2dg p , and so on. That is why complete answers can cite nonaccidental regularities in addition to citing first-level reasons why. My second argument is concerned with the relation between the answer to the follow-up question and the answer to the why question. Skow argues only that the answer to the follow-up question, taken by itself, does not answer the why question. However, one could argue that the answer to a follow-up question such as “Why does F1 have to do with E?” complements an answer as to why p. The subject matter offollow-up questions is the property of being a link between the event occurrence in question and the cited fact. This link can be a further fact or a regularity, such as the gravitational law that con- nects drop height and impact speed. Importantly, the same could not be said regarding the cases that motivate Skow’s pragmatic good-response argument. Take his example “Is Connor coming to the party?” (2016, 72) and as a re- sponse “No, he is sick.” Clearly, “He is sick” is not part of an answer to the yes-or-no question. Instead, it is an answer to the expected follow-up question “Why isn’t he coming?” In this case, despite its relatedness, the answer to the follow-up question does not complement an answer to the first question. The first question is just a matter of yes or no. Follow-up questions that are con- cerned with reasons why are different. Their answers can complement an- swers as to why p. (This may not hold for all why questions.) 4. Note that my argument does not endorse Lange’s claim that the mathematical impos- sibility of dividing 23 by three is essential to the explanation (for this issue see Skow [2018, 13–16]). 176 INSA LAWLER Altogether, it seems plausible that in some cases (at least complete) an- swers to why questions cite second-level reasons why. 4. Concluding Remarks. Skow attaches great weight to the different lev- els of reasons why. However, I have shown that the distinction does not bear decisive weight for two claims that are at the heart of Skow’s account. Even if one granted the distinction, his claim that correct answers to why ques- tions cite only causes or grounds and his claim that accounts that include citations of nonaccidental regularities in answers to why questions typically confuse second-level reasons with first-level reasons can be refuted. The question whether nonaccidental regularities and causes are reasons of the same level is independent of the question what kinds of reasons are citable in answers to why questions. I have argued that Skow’s accusation of con- fusion is unwarranted. The true object of disagreement is what kinds of rea- sons why are citable in answers to why questions. I have further rebutted Skow’s argument for excluding citations of second-level reasons why in an- swers to why questions, and I have argued that such citations can be included. At the very least, some complete answers to why questions cite nonaccidental regularities in addition to citing first-level reasons why. Thus, Skow’s central claim that correct answers to why questions cite only causes or (partial) grounds is refuted as well. REFERENCES Achinstein, P. 1975. “The Object of Explanation.” In Explanation, ed. S. Körner, 1–45. Oxford: Blackwell. Baumgartner, M. 2017. “Reasons without Argument.” Metascience 26 (3): 511–14. Brogaard, B. 2009. “What Mary Did Yesterday: Reflections on Knowledge-wh.” Philosophy and Phenomenological Research 78 (2): 439–67. Friedman, M. 1974. “Explanation and Scientific Understanding.” Journal of Philosophy 71 (1): 5–19. Hempel, C. 1965. Aspects of Scientific Explanation, and Other Essays in the Philosophy of Science. New York: Free Press. Hempel, C., and P. Oppenheim. 1948. “Studies in the Logic of Explanation.” Philosophy of Science 15 (2): 135–75. Kim, J. 1994. “Explanatory Knowledge and Metaphysical Dependence.” Philosophical Issues 5:51–69. Kitcher, P. 1989. “Explanatory Unification and the Causal Structure of the World.” In Scientific Explanation, ed. P. Kitcher and W. Salmon, 410–505. Minneapolis: University of Minnesota Press. Lange, M. 2013. “What Makes a Scientific Explanation Distinctively Mathematical?” British Jour- nal for the Philosophy of Science 64:485–511. ———. 2018. “Reply to My Critics: On Explanations by Constraint.” Metascience 27 (1): 27–36. Lewis, D. 1986. “Causal Explanation.” In Philosophical Papers, vol. 2, ed. D. Lewis, 214–40. Ox- ford: Oxford University Press. Pincock, C. 2017. Review of Reasons Why, by Bradford Skow. Notre Dame Philosophical Re- views. https://ndpr.nd.edu/news/reasons-why/. Railton, P. 1978. “A Deductive-Nomological Model of Probabilistic Explanation.” Philosophy of Science 45 (2): 206–26. https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1111%2Fj.1933-1592.2009.00250.x&citationId=p_7 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1111%2Fj.1933-1592.2009.00250.x&citationId=p_7 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.2307%2F1522873&citationId=p_11 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1093%2Fbjps%2Faxs012&citationId=p_13 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1093%2Fbjps%2Faxs012&citationId=p_13 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1007%2Fs11016-017-0230-x&citationId=p_6 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&system=10.1086%2F286983&citationId=p_10 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&system=10.1086%2F288797&citationId=p_17 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&system=10.1086%2F288797&citationId=p_17 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.2307%2F2024924&citationId=p_8 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1007%2Fs11016-017-0252-4&citationId=p_14 DISCUSSION 177 Salmon, W. 1984. Scientific Explanation and the Causal Structure of the World. Princeton, NJ: Princeton University Press. Scriven, M. 1959. “Truisms as the Grounds for Historical Explanation.” In Theories of History, ed. H. Gardiner, 443–75. New York: Free Press. Skow, B. 2016. Reasons Why. Oxford: Oxford University Press. ———. 2017. “Levels of Reasons and Causal Explanation.” Philosophy of Science 84 (Proceed- ings): 905–15. ———. 2018. “Of Strawberries and Energy Conservation.” Metascience 27 (1): 11–18. Stanley, J. 2011. Knowing (How).” Noûs 45 (2): 207–38. van Fraassen, B. 1980. The Scientific Image. Oxford: Oxford University Press. Woodward, J., and C. Hitchcock. 2003. “Explanatory Generalizations, Part I: A Counterfactual Ac- count.” Noûs 37 (1): 1–24. https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&system=10.1086%2F693966&citationId=p_21 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1111%2Fj.1468-0068.2010.00758.x&citationId=p_23 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1111%2F1468-0068.00426&citationId=p_25 https://www.journals.uchicago.edu/action/showLinks?doi=10.1086%2F701073&crossref=10.1007%2Fs11016-017-0250-6&citationId=p_22 I_Lawler_Levels_2019.pdf Levels of Reasons.pdf Cit p_7:1: Cit p_7:2: Cit p_11:1: Cit p_13:1: Cit p_13:2: Cit p_6:1: Cit p_10:1: Cit p_17:1: Cit p_17:2: Cit p_8:1: Cit p_14:1: Cit p_21:1: Cit p_23:1: Cit p_25:1: Cit p_22:1: