Skip to navigation – Site map

HomeNuméros45Symposium on law and coercionConceptual analysis and the Coerc...

Symposium on law and coercion

Conceptual analysis and the Coercion Thesis

Frank Jackson


In chapter two of Coercion and the Nature of Law, Kenneth Himma sets his defence of the Coercion Thesis – the thesis that it is a conceptual truth, and not merely a practical necessity, that a legal system contains sanctions – within an account of the nature of conceptual analysis. I discuss his account, and the nature of conceptual analysis more widely, explaining why I follow Himma in accepting the Coercion Thesis. I conclude with a short discussion of where I think he misunderstands the distinction between modest conceptual and immodest conceptual analysis.

Top of page

Full text

1It used to be a convention on bicycle paths near our house that one moved left when approaching another bicycle rider. It later became a law. What changed when that happened? When it was a convention, bad things could happen when one violated the convention – angry words perhaps, and the chance of having an accident went up. When moving to the left became a law, a further bad thing entered the picture: one became liable to a coercive sanction administered through a legal system.

  • 1 Himma 2020.

2The existence of coercive sanctions for the violation of a law is a pervasive feature of legal systems. Of course, there are laws which are, so to speak, laws in name only; there is an understanding that they will not be enforced. And some laws have “periods of grace” that sometimes get indefinitely extended. But the fact that these cases are worthy of special note suggests that it is an essential feature of a legal system that it comes with coercive sanctions. Kenneth Himma1 expresses this idea by saying that it is a conceptual truth that there are such sanctions: coercive enforcement mechanisms are a conceptually necessary feature of legal systems. He calls this the Coercion Thesis.

3I am no philosopher of law, but he convinced me (a task made easier by the fact that I was antecedently disposed to accept something like the Coercion Thesis). My remit here is to discuss the way he argues for the Coercion Thesis in chapter two, “Methodology and the Nature of Law,” via claims about conceptual analysis.

  • 2 Jackson 1998.

4Himma develops his argument by drawing on the distinction between modest conceptual analysis (MCA) and immodest conceptual analysis (ICA) I made in 1998.2 As it happens, he somewhat misunderstands the distinction I had in mind but, in my view, this does not materially affect the interest of his line of argument for the Coercion Thesis via conceptual analysis. I will relegate my discussion of the misunderstanding to a comment at the end. What is important, though, is what is suggested in a passage like the following (in reading this passage, one needs to know that Himma uses “C” for what a concept C refers to; MCA can be read as conceptual analysis as Himma thinks it should be conceived of; its relationship to what I had in mind is by the way for the moment).

We can apprehend the metaphysical nature of things only as they appear to us mediated through the concepts we deploy to organize and make sense of our experience; we have no reliable way to apprehend things as they are utterly independent of the concepts through which we organize the materials of our experience. It is an exercise in futility to attempt to understand C as it really is independent of the empirically contingent linguistic practices that enable us to talk about Cs to begin with.

  • 3 Himma 2020: 37–38, his emphasis.

The only epistemically viable approach to conceptual analysis is MCA. MCA is epistemically viable because it is concerned with explicating the metaphysical nature of an entity of interest as it is determined, in part, by our empirically contingent linguistic practices; … . MCA purports to tell us something about how our conceptual activities structure the world – and not about how the world is structured independent of these conceptual activities that enable us to talk about something and hence to understand it.3

5The wordings in this passage (and they aren’t atypical) suggest views which many will find highly controversial. How can empirically contingent linguistic practices restrict our understanding of what our concepts refer to, and if they can, shouldn’t we reform those practices and shouldn’t we deploy the kinds of experimental methods to be found in the sciences to get around the restrictions? Isn’t the metaphysical nature of an entity one thing and our talk about it another? Is it really plausible that our conceptual activities structure the world? Surely our thoughts seek to ascertain the structure to be found in the world; they do not create that structure.

6With one exception to be noted later, I think that these worries, though understandable given the wordings, are mistaken. Himma’s methodology in defence of the Coercion Thesis is just fine, or so I will argue. I will spell this out by offering a kind of reconstruction of his line of argument for the Coercion Thesis. (It will be interesting to read his thoughts on my reconstruction.)

7We could not make sense of the world we inhabit without seeing – sometimes literally, but often in the sense of thinking or conceiving of – the items in it as falling into categories. We see the chairs we encounter as alike and as different from the tables and the gaps between them. If we didn’t, we would have trouble crossing the room, and even more trouble complying with a request like, “Please move the chairs further away from the tables.” We know that round objects roll more smoothly than square ones, and this bit of knowledge is about the behaviour of items that fall into the similarity classes we speakers of English use “round” and “square,” respectively, for. We seek understand how the economy is travelling by classifying it in terms of inflation, GDP, level of unemployment, etc., and we English speakers use “inflation,” etc., to report on our endeavours. We could not read text if we did not see various tokens of “a” as alike in the relevant way and as different from various tokens of “b,” “c,” etc., despite the fact that different tokens of “a” vary substantially depending on whether they are in Times, Courier, or one’s doctor’s handwriting. Understanding, predicting, and explaining the world we occupy involves a massive exercise in conceiving of the items in it as being alike and unalike in a myriad of ways; it involves a massive exercise in classification, if you like, along with our ability to identify those ways – that is, our ability to have reliable beliefs about when items are alike and unalike in those ways – and to talk and write about them. We need the last in order to share what we believe about them with others, and with ourselves in the future (which is why we take notes at meetings, for example).

8There is a sense in which we create the classifications and a sense in which we do not. The items to be found in our world are alike and unalike in their many ways, independently of us, or so we realists insist. Yes, it may take training, sometimes a good deal of training, to see items as alike in the relevant ways. Seeing and hearing sentences as grammatical or ungrammatical was something we all had to learn, but what we learnt to grasp is something that exists independently of us. We are acquiring an ability to spot the relevant similarities and differences; we aren’t creating them. What we create are the classifications we choose, implicitly or explicitly, to employ in making sense of the world, and not whether or not some item falls under some classification. Here is a simple example to clarify the issue. Anthropologists decided that it would be good to classify people together in terms of whether or not they have a parent in common, co-opting the term “sibling” to do the job in language: x and y are siblings just if they share a parent. They created the term and the classification, but they did not thereby create any siblings. Parents do that.

9Now if we create the classifications, and the words we use to report on them, it is perfectly reasonable for us to appeal to our intuitions about whether or not some item falls under some given classification, as captured in the words we use for that classification, in deciding whether or not the item does indeed fall under the classification. We are, in fact, all familiar with this kind of appeal to intuition when we wonder whether or not we have spelled a word correctly, or whether or not some sentence or other is or is not grammatical. Sometimes we find out whether or not we have spelt a word correctly by writing down the various versions we are undecided and seeing which one “looks” right. This is less common these days now that we have spell checkers in our word processing programs, but used to be a common occurrence for those of us with shaky spelling. The same happens with sentences. Some sentences sound or look “off” – “we is happy” is an example – and that’s how we find out that they are ungrammatical. Of course, we can make the same discovery using a grammar book, but (i) we do not need to (no one reading this essay needed a grammar book to tell themselves that “we is happy” is ungrammatical), and (ii) the grammar books get written by people who draw on their own intuitions and those of the people they consult. Anyone working in theoretical linguistics knows about placing an asterisk before a sentence to indicate ungrammaticality: the asterisks get assigned on the basis of intuition, on whether or not the sentences sound or look “off.”

  • 4 Himma 2020.

10I have just said friendly things about appealing to intuitions in settling whether or not some item falls under some given classification, as captured by some term in natural language. How does this bear on Himma’s methodology? I think he is doing exactly what I have just been describing, but for the topic of special interest to him: the classification we make when we use the term “legal system.” He is appealing to our intuitions about what does and does not count as a legal system, as we understand the term “legal system” and the concept of a legal system – where this is understood as a classification we are imposing – and urging that in order to count as a legal system, a system must contain laws that come with punishments for breaking them. And when he appeals to what we find in dictionaries, that’s fine by me. In doing that, he is reassuring us that his intuitions are shared by others, namely, those who wrote the dictionaries. And why do I find the Coercion Thesis so appealing? I share his intuition about what happens when one removes all constraints from the laws in some given legal system. Let me put the point in my own terms, but I take what I am about to say to be consonant with what Himma says in detail in chapter ten of his book.4

11It is one thing to classify some body of rules and regulations as a legal system, and quite another to classify it as a set of norms of behaviour designed, or intended, to make things go better in one way or another. We can imagine a society with a functioning legal system but where no one in that society ever gets punished because no one breaks a law and no one gets punished by mistake: there are no lawbreakers and no law abiders punished in error in this society. This extremely unlikely possibility would not be a counterexample to the Coercion Thesis. For it would still be true in that society that had certain laws been broken, punishments would have been meted out. Next, imagine that this last feature is removed from our imagined society: it is not true that had “laws” been broken, punishments would have been meted out. Now there is nothing to make the difference between a law and a norm of behaviour. We have turned our imagined society into one without a legal system but with an elaborate set of behavioural norms. Subtracting constraints from a legal system turns it into a set of norms of behaviour. The constraints are not optional extras for a legal system; they are an essential part of what makes the difference between a legal system and a set of norms of behaviour.

12I have talked of the classifications we make and why they are so important. But there is of course a distinction between the classifications we make and the classifications it would be good to make for one or another purpose. I would not be doing justice to Himma’s book and its guiding methodology if I did not highlight the fact that his claim that the concept of a legal system is such that the Coercion Thesis is true is more than a merely descriptive one. His thesis is also a normative one: it would be good to make the possibility of punishment for breaking its laws a necessary condition for being a legal system, for it is part of the essential rationale for having legal systems. This theme appears in many places, and I will restrict myself to quoting a passage from late in the book.

  • 5 Himma 2020: 240.

All that stands between civilisation and the state of nature is an institutional normative system that backs mandatory norms prohibiting certain assaults on persons or property with the threat of severe detriment; this is why every existing legal system in our world ... backs mandatory norms prohibiting acts likely to lead to breaches of the peace with the threat of incarceration – or worse.5

13As we might put it: had the Coercion Thesis not been an integral part of the concept of a legal system, it would have been sensible to change the concept to make it such a part. (Having two exact divisors was once not part of the concept of being a prime number. It was sensible to make it part of the concept because it was sensible to exclude the number one from the class of prime numbers.)

  • 6 Jackson 1998.

14I now turn to his misunderstanding of the distinction I had in mind when, in From Metaphysics to Ethics,6 I spoke of modest versus immodest conceptual analysis.

  • 7 Jackson 1998.

15We have been talking about the ways we classify or categorise things, in a very inclusive sense of “things” that encompasses anything we can sensibly think of placing into a category, and are thinking of conceptual analysis as the business of articulating the classifications we make (and sometimes of those it would be good to make). We are not talking about words as such. We are talking about the classifications themselves, but cannot talk or write about these classifications without using words, obviously. Now, for any classification, we need to distinguish (i) what it takes to fall under that classification, from (ii) whether or not some given thing, or indeed anything at all, in fact falls under that classification. Or, to say it in the language of concepts, one question is what it takes for a concept to apply to something, the nature of the concept, and another is whether or not the concept applies to some given thing or indeed anything. What I mean by immodest conceptual analysis is a product of failing to be suitably sensitive to the distinction between the two questions when drawing on a favoured conceptual analysis en route to a thesis about the nature of the world. Sometimes we want to argue as follows: there are Ks (obvious fact, for some K); a necessary condition for being a K is alpha (from reflections on the concept of being K, perhaps expressed in terms of what it takes to be a K); therefore, there are alphas. I am sure that it is sometimes fine to argue in this way. I am also sure that sometimes it is a mistake to argue in this way. What I mean by immodest conceptual analysis is arguing in this way when it is a mistake to do so. The example I give in From Metaphysics to Ethics7 was the attempt that some have made to refute four-dimensionalism by consulting the concept of change. As Himma cites this example, I will give a different example here, one that often comes up in debates in ethics – sometimes implicitly, sometimes explicitly.

  • 8 Mackie 1977.
  • 9 The “traditionally” is to signal the fact I am talking about the non-cognitivist, no-property, vers (...)
  • 10 See, e. g., Jackson 1998: 139–162.

16The example runs as follows. Some actions have the property of being morally good (obvious fact, although one may wish there were more of them); being morally good is distinct from the kinds of properties that figure in naturalistic accounts of the world (from reflections on the concept of being good that, as it might be, make appeal to the open-question argument or the impossibility of natural properties being necessarily motivating); therefore, some actions have properties distinct from any that figure in naturalistic accounts of the world, and naturalism is false. There are, that is, properties in the sense of instantiated properties that do not figure in naturalistic accounts of what our world is like. I think naturalists should resist any attempt to refute the metaphysics they like along these lines. They should insist that there are two ways to go in response to this line of argument. The first is to hold that it teaches us that there are no actions with the property of being morally good, either because an error theory is true, as Mackie says8 – there are no moral properties in the sense that none are instantiated – or else because talk of moral properties is a confusion, as expressivists traditionally hold.9 The second way to go is to hold that the conceptual claim is a mistake; being good is not inconsistent with being some natural property or other, as moral functionalists, for example, hold.10 The one thing one should not say, I insist, is that we have here an easy – too easy, surely – way to make exciting discoveries about the kind of world we inhabit.

  • 11 Himma, however, believes in objective truth, as he makes clear in Himma 2020: ch. 2, § 6.

17How does this relate to what Himma says about MCA and ICA? The short answer is that what he means by MCA is close to what I mean by conceptual analysis, as I say near the beginning of this essay. But there is one feature he associates with MCA that I reject, and here we come to the point of disagreement heralded earlier. He thinks that being suitably modest requires giving our concepts a role in structuring the world.11 I think, as against this, that being suitably modest only requires being sensitive to the issue canvassed in the immediately preceding paragraphs – that we should not give conceptual analysis too large a role in metaphysics.

Top of page


Himma, K.E. (2020). Coercion and the Nature of Law. Oxford University Press.

Jackson, F. (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford University Press.

Mackie, J.L. (1977). Ethics: Inventing Right and Wrong. Penguin.

Top of page


1 Himma 2020.

2 Jackson 1998.

3 Himma 2020: 37–38, his emphasis.

4 Himma 2020.

5 Himma 2020: 240.

6 Jackson 1998.

7 Jackson 1998.

8 Mackie 1977.

9 The “traditionally” is to signal the fact I am talking about the non-cognitivist, no-property, version expressivists first advanced.

10 See, e. g., Jackson 1998: 139–162.

11 Himma, however, believes in objective truth, as he makes clear in Himma 2020: ch. 2, § 6.

Top of page


Electronic reference

Frank Jackson, “Conceptual analysis and the Coercion Thesis”Revus [Online], 45 | 2021, Online since 11 November 2021, connection on 27 March 2023. URL:; DOI:

Top of page

About the author

Frank Jackson

Emeritus Professor, The Australian National University, School of Philosophy

Address: 146 Ellery Crescent, Australian National University, Canberra, ACT 0200


Top of page


All rights reserved

Top of page
  • OpenEdition Journals
Search OpenEdition Search

You will be redirected to OpenEdition Search