Emory Law Journal

On the Distinction Between Speech and Action
Frederick Schauer David and Mary Harrison Distinguished Professor of Law, University of Virginia. This Article was prepared for the Emory University School of Law’s Randolph W. Thrower Symposium, February 5, 2015, on “The New Age of Communication: Freedom of Speech in the 21st Century.” Earlier versions were presented at the University of California, Berkeley, Boalt Hall School of Law; the Harvard Kennedy School; the University of Chicago Law School; the New York University Colloquium on Political Philosophy; and the University of Lucerne. I have profited from reactions and questions on those occasions, as well as from comments on earlier drafts from Mary Anne Case, Aziz Huq, Leslie Kendrick, Liam Murphy, Micah Schwartzman, Samuel Scheffler, David Strauss, John Tasioulas, and Alex Tsesis.

Does the First Amendment rest on a mistake? More specifically, is the First Amendment’s necessary distinction between speech and action fundamentally unsustainable?

The First Amendment, in relevant part, protects “the freedom of speech.” 1U.S. Const. amend. I. By contrast, the First Amendment does not protect simply “freedom,” or “liberty.” And thus it appears, initially and obviously, that the First Amendment’s protections extend to some acts or events or behaviors but not others. Indeed, not only the First Amendment but also any coherent principle of freedom of speech presupposes a meaningful distinction between the activities encompassed by the principle and those that are not. Loosely and preliminarily, we can label the former “speech” and the latter “action.” 2By way of emphasizing the preliminary nature of the terminology, I do not at this point in the analysis want anything to turn on the word “action,” especially because all speech is a form of action (or conduct). But the basic idea is that of a distinction between speech and non-speech behavior, or between the behavior that we designate as “speech” and the behavior that we do not. And because any non-vacuous account of a free speech principle is premised on the idea that being an act of speech in the relevant sense grants to that act a degree of protection from restriction not granted to non-speech acts causing equivalent consequences, 3See Frederick Schauer, Free Speech: A Philosophical Enquiry 3–14 (1982) (analyzing the necessary conditions for a distinct Free Speech Principle); Thomas Scanlon, A Theory of Freedom of Expression, 1 Phil. & Pub. Aff. 204, 204 (1972) [hereinafter Scanlon, A Theory of Freedom of Expression] (“The doctrine of freedom of expression is generally thought to single out a class of ‘protected acts’ which it holds to be immune from restrictions to which other acts are subject.”). Although Scanlon now believes that the influential account he offered in 1972 is “mistaken in important respects,” T.M. Scanlon, Comment on Baker’s Autonomy and Free Speech, 27 Const. Comment. 319, 319 (2011) [hereinafter Scanlon, Comment on Baker’s Autonomy]; see also T.M. Scanlon, Jr., Freedom of Expression and Categories of Expression, 40 U. Pitt. L. Rev. 519, 530–34 (1979) [hereinafter Scanlon, Freedom of Expression and Categories of Expression], the subsequent narrowing of his earlier position, even if relevant to taking the earlier article as a self-standing contribution usable or not on its own terms, does not touch the basic analytic point about the structure of a free speech principle.Although the most obvious application of the distinction set out in the previous paragraph is with respect to a differential immunity from restriction for speech and non-speech behavior causing equivalent harm or other negative consequences, the distinction between speech and non-speech behavior, as articulated in the text, could also arise in the context of positive rather than negative consequences. See Scanlon, A Theory of Freedom of Expression, supra, at 204. For example, if there were an affirmative obligation on the part of government to subsidize speech-relevant activities, a free speech principle would generate a greater obligation to subsidize or otherwise support speech than to subsidize or support non-speech activities bringing equivalent benefits. What is key is the differential, and not whether the differential attaches to the restrictive as opposed to the supportive activities of the agent against whom the free speech claim is offered. See Leslie Kendrick, Free Speech as a Special Right (unpublished manuscript) (on file with author). some sort of distinction between speech and action is a necessary condition for a viable principle of freedom of speech. But unless there are free-speech-relevant attributes that are possessed by speech but not by action, the distinction between speech and action, at least as a matter of free speech theory, cannot do the work that appears to be required of it.

Although some kind of free-speech-relevant distinction between speech and action is thus a necessary condition for a meaningful free speech principle, it is by no means clear that such a distinction can be maintained. There is, to be sure, a difference between an actual fire and shouting fire in a crowded theater, 4Cf. Schenck v. United States, 249 U.S. 47, 52 (1919) (“The most stringent protection of free speech would not protect a man in falsely shouting fire in a theatre and causing a panic.”). just as there is a difference among a pipe, a picture of a pipe, and a verbal description of a pipe. 5René Magritte, The Treachery of Images (1928–1929) (“Ceci n’est pas une pipe”). In some contexts, distinctions between words and things and between speech and action plainly exist. But the existence of such a distinction in some contexts does not entail the conclusion that the everyday distinction between speech and action will mark anything of free speech significance, nor that the distinction can carry the weight that any meaningful principle of free speech must demand of it.

Controversies over the existence (or not) of a distinction between speech and action have occasionally appeared as weapons in contemporary debates about hate speech and pornography, with proponents of regulation questioning the distinction 6See, e.g., Rae Langton, Sexual Solipsism: Philosophical Essays on Pornography and Objectification 27–37, 103–16 (2009) (drawing on speech-act theory to consider speech as action rather than distinct from it); Catharine A. MacKinnon, Only Words 29–41 (1993) (characterizing the speech–action distinction as legalistic); Susan J. Brison, Speech, Harm, and the Mind-Body Problem in First Amendment Jurisprudence, 4 L. Theory 39, 56–61 (1998) (challenging the distinction between the verbal and the physical); Mary Ellen Gale, Reimagining the First Amendment: Racist Speech and Equal Liberty, 65 St. John’s L. Rev. 119, 149 (1991) (questioning the distinctions between speech and action and speech and conduct); Alon Harel, Hate Speech and Comprehensive Forms of Life, in The Content and Context of Hate Speech: Rethinking Regulation and Responses 306, 308–10 (Michael Herz & Peter Molnar eds., 2012) (same). Stanley Fish also questions the distinction, but, given his skeptical stance about the ontology of distinctions in general, it is difficult to know what to make of his claims about the distinction between speech and action. Stanley Fish, There’s No Such Thing as Free Speech, and It’s a Good Thing, Too 106 (1994) (noting the “general difficulty of separating speech from action”). and opponents accusing their adversaries of failing to grasp a distinction lying at the foundation of the idea of rationality, and, indeed, of what it is to be human. 7See, e.g., Amy Adler, Inverting the First Amendment, 149 U. Pa. L. Rev. 921, 927, 970–85 (2001) (arguing that the speech–action distinction plays a “central role in the First Amendment”); Nadine Strossen, Regulating Racist Speech on Campus: A Modest Proposal?, 1990 Duke L.J. 484, 492, 494, 532–33, 541–44 (objecting to the “abrogation of the traditional distinctions between speech and conduct”); Kathleen M. Sullivan, Discrimination, Distribution and Free Speech, 37 Ariz. L. Rev. 439, 442, 444 (1995) (responding to attacks on the distinction between speech and conduct); Kathleen M. Sullivan, Free Speech Wars, 48 SMU L. Rev. 203, 206, 209–10 (1994) (same). Yet these debates have taken place in such a narrow (and too often tendentious) context that they have avoided confronting the most important foundational issues about freedom of speech. In fact, it is not uncommon for the defenders of a speech–action distinction to take the existence of a free speech principle as a given and thus as the premise for the necessity of accepting the distinction. 8See, e.g., Charles W. Collier, Hate Speech and the Mind-Body Problem: A Critique of Postmodern Censorship Theory, 7 L. Theory 203, 204 & n.7 (2001) (assuming that there is a “constitutionally significant difference between speech and . . . action”); Stephen E. Gottlieb, The Speech Clause and the Limits of Neutrality, 51 Alb. L. Rev. 19, 23 (1986) (noting that much of First Amendment doctrine is “predicated on the speech/action distinction”). As a matter of positive law or political rhetoric such a strategy may well be defensible, but for engaging in a deeper exploration of the foundations of the very idea of free speech it is plainly unacceptable. In some contexts examining with an open mind whether a free speech principle is itself sound is an important task, 9Such contexts would include not only philosophical inquiry for its own sake, but also institutional and constitutional design in domains in which free speech principles are not yet accepted. One example would be countries with rudimentary free speech protection, and as to which foundational questions might thus be asked about how much free speech, if any, should be permitted. Another would be non-governmental settings (corporations and private colleges and universities, for example) in which free speech is considered in the context of institutional design decisions about who should be allowed or encouraged to say what, even apart from questions of positive law. And of course the groundings of the idea of free speech are plainly relevant to questions arising in the interpretation of the First Amendment itself, as is apparent both from the fact that much of existing free speech theory has been developed in the context of the First Amendment and from the frequency with which the Supreme Court makes reference to foundational principles in deciding First Amendment cases. E.g., United States v. Alvarez, 132 S. Ct. 2537, 2547–50 (2012) (relying on the foundational premise that truth is not to be authoritatively determined by government); Cohen v. California, 403 U.S. 15, 24–26 (1971) (basing decision on self-expression rationale for the First Amendment); Stanley v. Georgia, 394 U.S. 557, 564–65 (1969) (stressing freedom of belief and freedom of thought); Vincent Blasi, Ideas of the First Amendment vii-viii (2d ed. 2012); Paul Horwitz, First Amendment Institutions 8–9 (2013); Robert C. Post, Constitutional Domains: Democracy, Community, Management 15–16 (1995); Steven H. Shiffrin, The First Amendment, Democracy, and Romance 2–6 (1990). and for that task we cannot simply assume the conclusion of the inquiry. Rather, when we inquire into whether there can be a sound free speech principle at all, we must subject to critical analysis just what it means to draw a distinction between speech and action, whether the distinction can actually be drawn, and whether the distinction, even if it can be drawn, can provide the basis for a principle of freedom of speech.

Thus, unburdened by any assumptions from existing positive law or political history—including but not limited to the law and history of the First Amendment—I examine here whether the kind of distinction between speech and action that is necessary to any principle of free speech can in fact be sustained. Much of the focus will be on issues of autonomy and freedom of thought, and even more particularly on arguments grounded in respect for the decision-making capacities of autonomous agents whose volitional decisions are often thought to be a necessary mediating step between speech and harmful action. But my goal is broader than that, for in questioning the viability of a speech–action distinction in this context I hope to raise questions about the speech–action distinction in other free speech contexts as well, and thus ultimately about the deep soundness of any free speech principle at all.

I. On the Structure of a Free Speech Principle

Although my inquiry is pre-constitutional and pre-doctrinal, the structure of American constitutional doctrine illuminates the basic idea and the principal problem. And according to that doctrine, government regulation of most of the vast universe of human behavior need only satisfy the minimal scrutiny of the rational basis test. 10See, e.g., Ferguson v. Skrupa, 372 U.S. 726, 730–32 (1963) (holding that the wisdom of a law restricting the practice of debt adjustment to lawyers was for the legislature and not for the courts to decide); Williamson v. Lee Optical of Okla., Inc., 348 U.S. 483, 487–90 (1955) (upholding on rational basis grounds the requirement that opticians could not fit eyeglasses without receiving a prescription from an optometrist or ophthalmologist); W. Coast Hotel Co. v. Parrish, 300 U.S. 379, 391, 398 (1937) (holding that regulation for purposes of health, safety, morals, and welfare satisfied requirements of due process as long as it bore a “reasonable relation to a proper legislative purpose, and [was] neither arbitrary nor discriminatory” (quoting Nebbia v. New York, 291 U.S. 502, 537 (1934))); Nebbia v. New York, 291 U.S. 502, 537 (1934) (applying reasonableness test to regulation of business); see also United States v. Carolene Prods. Co., 304 U.S. 144, 152–53 (1938) (announcing that “rational basis” is the standard to be employed in evaluating the constitutionality of social and economic regulation). We can call this the “baseline rule,” in the sense of it being the default standard applicable to all governmental action. In practice, the American baseline rule is a test of virtually no stringency, 11See also N.Y.C. Transit Auth. v. Beazer, 440 U.S. 568, 594 (1979) (concluding that courts applying rationality review should not interfere with policy decisions “[n]o matter how unwise” they may be); John F. Manning, The Necessary and Proper Clause and Its Legal Antecedents, 92 B.U. L. Rev. 1349, 1349 (2012) (describing rationality review as “very forgiving”); Cass R. Sunstein, Naked Preferences and the Constitution, 84 Colum. L. Rev. 1689, 1713 (1984) (concluding that rationality review almost always results in validation); Shoshana Zimmerman, Pushing the Boundaries?: Equal Protection, Rational Basis, and Rational Decision Making by District Courts in Cases Challenging Legislative Classifications on the Basis of Sexual Orientation, 21 S. Cal. Interdisc. L.J. 727, 733 (2012) (same). See generally Scott H. Bice, Rationality Analysis in Constitutional Law, 65 Minn. L. Rev. 1, 3–4 (1980). although one can imagine tests more stringent than the existing rational basis test that would still operate in this baseline fashion. 12And thus the same analytic point applies even if, as is occasionally the case, rational basis review has (or is argued to have) slightly more “bite.” See Logan v. Zimmerman Brush Co., 455 U.S. 422, 442 (1982) (concluding the rationality review requires “something more than the exercise of a strained imagination”); U.S. R.R. Ret. Bd. v. Fritz, 449 U.S. 166, 180 (1980) (Stevens, J., concurring) (arguing that rationality review should require more than simply a “conceivable” or “plausible” basis); U.S. Dep’t of Agric. v. Moreno, 413 U.S. 528, 535–38 (1973) (invalidating food stamp qualification criterion on rational basis grounds). But under the baseline rule that actually exists, the government may, as long as it meets extremely minimal standards of rationality, regulate most aspects of personal 13See Washington v. Glucksberg, 521 U.S. 702, 735 (1997) (rejecting right to assisted suicide on rational basis grounds); Kelley v. Johnson, 425 U.S. 238, 248 (1976) (upholding rationality and thus constitutionality of personal appearance regulation for police officers). At one time the rational basis standard applied to sexual conduct, but no longer. Lawrence v. Texas, 539 U.S. 558, 578 (2003) (overruling Bowers v. Hardwick, but not specifying the standard of review); Bowers v. Hardwick, 478 U.S. 186, 196 (1986) (finding a moral basis for a law sufficient to satisfy the rationality standard). or business 14See supra note 10 and accompanying text. behavior. And thus, continuing to adhere to the preliminary terminology noted above, we can say that existing doctrine establishes that the government may regulate action subject only to the negligible scrutiny of the rational basis standard.

When the state seeks to regulate speech, however, it must show something more. We might characterize this “something more” as a “compelling interest,” 15Although commonly associated with equal protection or due process, the compelling interest formulation has occasionally surfaced in First Amendment doctrine. See New York v. Ferber, 458 U.S. 747, 756–57 (1982) (recognizing “compelling” interest, even as against First Amendment concerns, in protecting the well-being of minors); Roe v. Wade, 410 U.S. 113, 163 (1973) (“compelling”); Loving v. Virginia, 388 U.S. 1, 9 (1967) (“The fact of equal application does not immunize the statute from the very heavy burden of justification which the Fourteenth Amendment has traditionally required of state statutes . . . .”); Korematsu v. United States, 323 U.S. 214, 216 (1944) (“It is to say that courts must subject [restrictions] to the most rigid scrutiny.”). or instead employ a different formulation of the exceptionally heavy burden on the state to justify its regulation, 16It is more than plausible to characterize the clear and present danger principle in this way, given that normally the state may, if constrained only by the rational basis standard, deal with dangers that are neither clear nor present. Schenck v. United States, 249 U.S. 47, 52 (1919). Much the same can be said about the existing crystallization of the clear and present danger standard in Brandenburg v. Ohio. 395 U.S. 444, 447 (1969) (holding that advocacy of illegal action may be prohibited if it is directed to producing imminent lawless action and is likely to produce that result). Brandenburg uses the standards of imminence and likelihood to embody its stringent protection of speech, but Brandenburg’s implicit assumption is that, when dealing with non-speech behavior, the state may address problems that are neither “imminent” nor even “likely.” Frederick Schauer, Is it Better to Be Safe than Sorry?: Free Speech and the Precautionary Principle, 36 Pepp. L. Rev. 301, 307 (2009). More general examinations of the strength of the Brandenburg standard include, notably, Hans A. Linde, “Clear and Present Danger” Reexamined: Dissonance in the Brandenburg Concerto, 22 Stan. L. Rev. 1163, 1185 (1970); Frank R. Strong, Fifty Years of “Clear and Present Danger”: From Schenck to Brandenburg—And Beyond, 1969 Sup. Ct. Rev. 41, 52. but the basic idea of requiring much more of a showing of necessity than is required by the rational basis standard is the same. Indeed, even when the heightened burden of justification embodies a degree of scrutiny less stringent than the compelling interest test, as for example with the so-called intermediate scrutiny applied to commercial speech, 17Cent. Hudson Gas & Elec. Corp. v. Pub. Serv. Comm’n of N.Y., 447 U.S. 557, 566 (1980) (setting out the standard applicable to commercial advertising). The Central Hudson test remains the applicable standard today. See, e.g., Sorrell v. IMS Health Inc., 131 S. Ct. 2653, 2667–68 (2011); Lorillard Tobacco Co. v. Reilly, 533 U.S. 525, 553–55 (2001). the basic structure remains the same, for the fact that the object of regulation 18Or, for that matter, the goal of the regulation, as would be the case under the so-called O’Brien analysis, according to which heightened First Amendment scrutiny is triggered not by the nature of the activity regulated but instead by the nature of the state’s interest in regulating. United States v. O’Brien, 391 U.S. 367, 376–77 (1968); see Laurence H. Tribe, American Constitutional Law 792 (2d ed. 1988); John Hart Ely, Flag Desecration: A Case Study in the Roles of Categorization and Balancing in First Amendment Analysis, 88 Harv. L. Rev. 1482, 1496–501 (1975). The roles of government purpose and motive in First Amendment adjudication are analyzed comprehensively in Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amendment Doctrine, 63 U. Chi. L. Rev. 413 (1996), and Jed Rubenfeld, The First Amendment’s Purpose, 53 Stan. L. Rev. 767, 793–95 (2001). is “speech,” again to put it loosely and preliminarily, is what causes the regulation to be measured against a standard at least somewhat more stringent than that of mere rationality. 19See supra note 16. For present purposes the size of the gap between rational basis scrutiny of “action” and heightened scrutiny of “speech” can be set aside, but the very idea of there being a right to free speech presupposes at least some gap. 20See Ashcroft v. Free Speech Coal., 535 U.S. 234, 253 (2002) (observing that First Amendment doctrine rests on “vital distinctions between words and deeds” and between “ideas and conduct”). Without the gap between free speech scrutiny and the scrutiny of some larger or other category, free speech would be merely an instance of some larger category, and it would be a conceptual error to think that there was a right to free speech in any meaningful sense. 21On the conditions necessary to conceive of the right to free speech (or any other right) as a distinct or independent right, see Frederick Schauer, Free Speech on Tuesdays, 34 L. & Phil. 119 (2015).

Implicit in the foregoing analysis and conclusion is a conception of rights as entities or principles that raise the standard of justification for restriction of the activities covered by the right above what it would otherwise be under some baseline standard of justification. For example, in the United States the justification for regulating the activity of operating a pushcart need only satisfy the rational basis baseline standard. 22New Orleans v. Dukes, 427 U.S. 297, 303–05 (1976). According to the conception of rights offered here, therefore, we can say that there is no constitutional right to operate a pushcart. 23See, e.g., id. at 305. But if the justification for regulating pushcarts were required to be different from and higher than the standard for regulating everything else, we could then conclude that there was a right to operate a pushcart. And so too with speech. Because the standard for regulating speech, unlike the standard for regulating pushcarts, is indeed higher than the baseline now associated with rational basis scrutiny, there exists a constitutional right to speak in a way that there is not, as a matter of existing constitutional doctrine, a right to operate a pushcart. 24This approach to rights appears to be implicit in Ronald Dworkin, Taking Rights Seriously 266–67 (1977) (finding the idea of a general right to liberty “absurd”). The argument for such a right to liberty, developed as an objection to Dworkin, is set out in Douglas N. Husak, Ronald Dworkin and the Right to Liberty, 90 Ethics 123 (1979) (“[T]here is no reason to believe that the right to liberty cannot conflict with the right to equality.”).

The structure of American constitutional doctrine can thus illuminate this conception of just what it is for a right to exist, but this conception of rights is by no means limited to the United States, or even to the rights created by positive law at all. Consequently, even outside the domain of American constitutional law, the structure of the putative right to freedom of speech is the same. As a pre-constitutional or extra-constitutional question of moral or political philosophy, for example, the idea of a right to free speech—or a free speech principle—similarly rests on the existence of a difference between what happens when the right or principle applies and when it does not. Indeed, even if, contra Ronald Dworkin, 25See Dworkin, supra note 24. there is a general right to liberty, 26See Randy E. Barnett, Restoring the Lost Constitution: The Presumption of Liberty 4 (rev. ed. 2014); Randy E. Barnett, The Structure of Liberty: Justice and the Rule of Law 14–15 (2d ed. 2014) (arguing for a natural rights presumption of liberty). it is still the case that when we speak of other and more specific rights and liberties we are not simply listing the instantiations of the general right. Rather, when we refer to a right to free speech we are designating something structurally different from, and stronger than, the myriad forms of behavior that would be subsumed by a right to liberty simpliciter. And because that differential strength manifests itself primarily in the way in which the right to free speech encompasses behavior whose arguable negative consequences would otherwise justify intervention or restriction even under a general right to liberty, 27See supra note 3 and accompanying text. the structure of a right to free speech necessarily presupposes something different between what is encompassed by the right and what is encompassed by any general right to liberty we may happen to possess. Accordingly, if we designate that category of activities encompassed by liberty in general as “action,” and the category of activities covered by the putative right to free speech as “speech,” then one way of understanding the distinction between speech and action is as the linchpin of the very right to free speech in the first place.

II. A False Start with a Revealing Premise

Among the first attempts to grapple with the distinction between speech and action was one that was also, and notoriously so, among the least successful. In Toward a General Theory of the First Amendment, 28Thomas I. Emerson, Toward a General Theory of the First Amendment, 72 Yale L.J. 877, 914–15 (1963). and then in The System of Freedom of Expression, 29Thomas I. Emerson, The System of Freedom of Expression 6–20 (1970). Thomas Emerson attempted to work out a system of absolute but bounded speech—a principle of free speech structured such that it covered only a small portion of the universe of communicative or expressive activity but which granted absolute protection to that which it covered. 30Id. On the distinction between what the First Amendment covers and the degree of protection it offers to what it does cover, see Frederick Schauer, The Boundaries of the First Amendment: A Preliminary Exploration of Constitutional Salience, 117 Harv. L. Rev. 1765, 1769–77 (2004). Although the Supreme Court has often conflated questions of coverage and questions about the degree of protection, it has more recently appeared to recognize the distinction and its importance. See United States v. Stevens, 559 U.S. 460, 468, 470 (2010) (announcing a presumption against creating new categories of uncovered speech); see also Brown v. Entm’t Merchants Ass’n, 131 S. Ct. 2729, 2733–35 (2011) (same, but using the word “protection” to designate coverage). Although everything that counted as “expression” should be absolutely protected, Emerson argued, the protection of the First Amendment should not be understood to extend to those acts, some of which happen to be verbal or linguistic or even communicative, which were not expression but instead were to be considered “action” or “conduct.” 31Emerson, supra note 29, at 8–9, 17–18. Accordingly, insisted Emerson, and paralleling Justice Douglas’s idea that “speech brigaded with action” was not covered by Douglas’s conception of an absolute First Amendment, 32Pittsburgh Press Co. v. Pittsburgh Comm’n on Human Relations, 413 U.S. 376, 398–99 (1973) (Douglas, J., dissenting) (“There comes a time, of course, when speech and action are so closely brigaded that they are really one.”); see also Parker v. Levy, 417 U.S. 733, 768 (1974) (Rehnquist, J., dissenting) (“A command is a speech brigaded with action . . . .”); DeFunis v. Odegaard, 416 U.S. 312, 343 (1974) (Douglas, J., dissenting) (“Speech is closely brigaded with action when it triggers a fight, as shouting ‘fire’ in a crowded theater triggers a riot.” (citation omitted)). most criminal solicitation, 33Emerson, supra note 29, at 403–05. some incitement, 34Id. some picketing, 35Id. at 356–59, 444–47. some commercial advertising, 36Id. at 414–17. and some espionage, 37Id. at 58–59. among others, were “action” and not “conduct,” thus being entitled to no protection under the First Amendment, whereas most forms of advocacy, 38Id. at 124–26. most varieties of protest, 39Id. at 356–59. and all literature, 40Id. at 3, 495. for example, were primarily “expression” and thus entitled to full (absolute) protection.

As was quickly obvious to his critics, 41See Tribe, supra note 18, at 789 & n.28 (criticizing the adoption of “an artificial dichotomy between (protected) speech-related conduct in which ‘expression’ predominates and (unprotected) conduct in which ‘action’ is dominant”); Ely, supra note 18, at 1494–96, 1495 n.53 (“I simply do not think this distinction can be made to work.”); Louis Henkin, Foreword: On Drawing Lines, 82 Harv. L. Rev. 63, 77–80 (1968) (criticizing the distinction between speech and conduct). For slightly more recent expressions of the same concern, see Barnes v. Glen Theatre, Inc., 501 U.S. 560, 576 (1991) (Scalia, J., concurring in judgment) (questioning the distinction between speech and conduct by observing that all laws restrict conduct); Rubenfeld, supra note 18, at 783–84 (2001) (noting that speech is conduct and that many non-speech actions have a communicative function). Emerson’s distinction was question-begging in the extreme. Rather than employing some sort of distinction between the properties of expression and the properties of action as an analytic device to determine which behaviors were encompassed by the First Amendment and which were not, Emerson drew the distinction between expression and action on grounds that appeared, at best, obscure, and then proceeded to apply the label “expression” to those behaviors he found protected 42Because Emerson defined the coverage of the First Amendment in a way that made the degree of protection absolute for all covered speech, he forced issues into the decision about coverage—issues that for others would sometimes be about coverage and sometimes about the degree of protection to be given to certain forms of covered speech. Emerson, supra note 29. On the importance of keeping the questions separate, see Kent Greenawalt, Speech, Crime, and the Uses of Language 5 (1989) (describing as “absurd” the attempt to conflate coverage and protection). and the labels “action” or “conduct” to those he found unprotected, all the while saying little about the actual factors that would distinguish the one from the other. 43See Robert C. Post, Racist Speech, Democracy, and the First Amendment, 32 Wm. & Mary L. Rev. 267, 285 (1991) (noting that the distinction between speech and action is what we say to mark a distinction drawn on other grounds). Indeed, the elusive nature of the distinction between speech and conduct facilitates labeling (or perceiving) an act as one or the other depending on the outcome preferences of the labeler. See Dan M. Kahan et al., “They Saw a Protest”: Cognitive Illiberalism and the Speech-Conduct Distinction, 64 Stan. L. Rev. 851, 854, 859, 861–62, 896, 900 (2012) (demonstrating experimentally that the phenomenon of motivated reasoning influences subjects’ determination whether an act is speech or conduct).

Although Emerson’s distinction was thus employed more to label outcomes than to generate or justify them, it was nevertheless premised on the sound idea that the First Amendment could make sense only if there were some distinction between speech and action, and only if that distinction related in a meaningful way to the point of the First Amendment. 44See Schauer, supra note 3, at 3–14; Schauer, supra note 21. And thus if we examine Emerson’s own views about the point of what he called the distinction between expression and conduct, we observe two things. First, and of lesser importance to the analysis here, is Emerson’s catalog of positive justifications for a principle of freedom of expression, a catalog that included the conventional appeals to the search for truth, to the role of discourse in democratic governance, and to the virtues of individual self-expression, 45Emerson, supra note 28, at 878–79. as well as to the less conventional (at the time) idea that various political, sociological, and psychological factors, what Emerson called the “dynamics” of the limitation, 46Emerson, supra note 29, at 9–11; Emerson, supra note 28, at 887–93. made speech especially vulnerable to restriction. Second, and more importantly, Emerson appeared to recognize that the positive values of searching for truth, facilitating democratic governance, and fostering self-expression could also be served by non-expressive (as Emerson understood it) conduct. 47Thus, Emerson acknowledged that the distinction between expression and action may not always be “clear,” may at times be “obscure,” and may seem at times “artificial.” Emerson, supra note 29, at 18. Moreover, Emerson might even be charitably interpreted to have acknowledged that even the regulation of conduct might be plagued by the same pathologies that affected the regulation of expression. 48Id. As a result, he understood that it was both difficult yet necessary to offer a distinction between expression and conduct, a distinction implicit in the idea of freedom of speech and thus in the First Amendment itself.

To meet this challenge, Emerson relied on the conclusion that “expression is normally conceived as doing less injury to other social goals than action. It generally has less immediate consequences, is less irremediable in its impact.” 49Emerson, supra note 29, at 9. To oversimplify, Emerson constructed much of his view about the importance of the distinction between expression (speech) and conduct (action) on the premise that expression, compared to conduct, was normally more self-regarding 50The philosophers’ traditional distinction between self-regarding and other-regarding acts is not congruent with the distinction between harmless and harmful acts, because an act can be other-regarding and beneficial, and an act can be self-regarding and harmful to the actor. See Jovan Babić, Self-Regarding/Other-Regarding Acts: Some Remarks, 5 Prolegomena 193, 198–200 (2006); C.L. Ten, Mill on Self-Regarding Actions, 43 Phil. 29, 31–32, 34, 37 (1968). Still, other-regardingness is a necessary condition for harm to someone other than the actor. and therefore less harmful. 51Emerson, supra note 29, at 9; Emerson, supra note 28, at 889–90.

Emerson’s reliance on what is “normally” the case makes clear that it is unfair to accuse him of making the implausible claim that speech is always or necessarily more self-regarding than action. Rather, Emerson argued that on average, or in the aggregate, the category of speech, qua category, was less other-regarding than the category of action, 52To the same effect, see Michael D. Bayles, Mid-Level Principles and Justification, in Justification: Nomos XXVIII 49, 54 (J. Roland Pennock & John W. Chapman eds., 1986) (arguing that the exercise of freedom of speech “is less likely to interfere with the exercise of other liberties than is, say, liberty of action”). Similarly, Martin Redish maintains that “it is almost certainly true in the overwhelming majority of cases that speech is less immediately dangerous than conduct.” Martin H. Redish, Freedom of Expression: A Critical Analysis 19 n.48 (1984). Redish also maintains that “the simple reality is that, for the most part, expression is less acutely harmful [than conduct] either to other individuals or to society.” Martin H. Redish, Fear, Loathing, and the First Amendment: Optimistic Skepticism and the Theory of Free Expression, 76 Ohio St. L.J. 691 (700) (2015); see also Franklyn S. Haiman, “Speech Actsand the First Amendment 85 (1993) (understanding speech as a largely harmless activity). And, perhaps most prominently, Ronald Dworkin, especially in Dworkin, supra note 24, at 200–03 (using freedom of speech as an exemplar of a broadly conceived personal liberty based on a right to equal concern and respect). and accordingly that the category of speech could be differentially protected relative to the category of action, even assuming equal positive benefits from the two categories, with less detrimental effect on society’s ability to deal with the negative consequences caused by both speech and action. By assuming that the category of speech is less harmful than the category of action, 53See supra note 49 and accompanying text. Emerson could argue that protecting speech but not action would have only minimal effect on the state’s ability to regulate harmful activities.

Not only did Emerson acknowledge that these relative determinations of other-regardingness or harm-producing capacities were based on the tendencies of categories rather than truths about every case, but he also made clear that in shifting from the term “speech” to that of “expression” he was not advocating that the principle of freedom of speech could or should be expanded to include the full range of behavior that might be considered self-expressive. 54Compare C. Edwin Baker, Human Liberty and Freedom of Speech (1989), with C. Edwin Baker, Scope of the First Amendment Freedom of Speech, 25 UCLA L. Rev. 964, 996, 1019 (1978), which both go pretty far in exactly this direction. For an iconic critique of this move, see Robert H. Bork, Neutral Principles and Some First Amendment Problems, 47 Ind. L.J. 1, 25 (1971) (observing that theories of this variety “do not distinguish speech from any other human activity”). Rather, Emerson’s terminological shift was designed to make clear that the freedom of speech encompassed forms of communicative behavior that would not count as “speech” according to the ordinary language meaning of that term—flag burning (or flag waving), painting, sculpture, music, armband-wearing, uniform-wearing, picketing, and parading, for example, to say nothing of writing and printing—and also to emphasize that there were forms of language—“speech” in the ordinary language sense—that did not fall under his heading of expression. Behind Emerson’s distinction was the belief that some forms of human communication operate in a way that is more reflective than reflexive, as when I actually think about your argument rather than when I unthinkingly react to your surprising me by shouting “Boo!” For Emerson, the difference between the reflective and the reflexive was the keystone of distinguishing speech from action, and thus the foundation for the differential protection of speech that lies at the heart of the First Amendment. 55See Emerson, supra note 29, at 500 (describing the “shock effect” of some of the communications he labeled as “action”).

Emerson’s ultimately question-begging approach nonetheless contained two important (although not necessarily sound) ideas. First, the category of speech is better understood as the category of communication. 56 See Larry Alexander, Is There a Right of Freedom of Expression? 7–8 (2005) (equating freedom of speech with freedom of communication). Thus, what we may at times understand as a distinction between speech and action is better conceived as a distinction between communicative and non-communicative conduct. 57Id. at 8. And, second, the category of communication, according to Emerson, is, as a category, less likely to produce negative consequences than the category of non-communicative conduct. 58See supra note 50 and accompanying text. It is these two ideas—that communication as a category is different from non-communicative conduct in a free-speech-relevant way, and that communication as a category is less harmful than non-communicative conduct—to which we must now turn.

III. Is Thinking Different from Doing?

Emerson’s terminological shift from “speech” to “expression” reveals his belief that the distinction between communication and action arises out of a fundamental and natural distinction between thinking and doing. For Emerson, there is an important distinction between thinking or contemplating or reflecting on something, on the one hand, and actually taking an action, on the other. But is the distinction sound, and, even if it is, what does it say, if anything, about the idea of freedom of speech?

The place to start in examining the question is with the seemingly straightforward difference between a person’s contemplation of engaging in some action and her actually engaging in it, with free speech protecting only the former and not the latter. Thus, what might make communication relevantly different from conduct, different in a way that connects with the First Amendment’s ideas and ideals, is the way in which communication about some action—whether it is to describe, commend, or condemn it—is conceptually distinct from the action itself. If I urge you to shoot someone, that act is different from my shooting someone, and it is different from you shooting someone. And what might make this undeniable conceptual separation important is not only the possibility that the communication, even if advocating action, might not lead to the action advocated, but also that the link between the communication and the action requires an additional act of volition on the part of the recipient of the communication. 59This is most obvious with respect to advocacy, but also is applicable to praising and blaming an action, because there is still a non-necessary connection between the commendation or condemnation on the part of a communicator and the recipient of the communication adopting some attitude towards the action that is commended or condemned. If Clarence Brandenburg’s advocacy of acts of “revengeance” against African Americans and Jews 60Brandenburg v. Ohio, 395 U.S. 444, 446–47 (1969). Mr. Brandenburg did not use the word “African Americans.” Id. had actually inspired, produced, or caused such an act of revengeance, an intentional decision by the person committing the act would also have been necessary.

That a volitional act on the part of the recipient of a communication is necessary to convert a communication into an action reveals the importance of the role that autonomy plays in the development of a free-speech-relevant distinction between speech and action. 61See Scanlon, A Theory of Freedom of Expression, supra note 3, at 206–07, 216–17; David A. Strauss, Persuasion, Autonomy, and Freedom of Expression, 91 Colum. L. Rev. 334, 353–60 (1991) (defending the “persuasion principle” by reference to the importance of the autonomy of the hearer). But see Scanlon, Comment on Baker’s Autonomy, supra note 3, at 319–20 (questioning the value of focusing on autonomy). Because the gap between receiving a communication pertinent to action and the action itself is filled by a volitional act on the part of the recipient, the recipient’s autonomy of action is accordingly a necessary condition for action. Advocacy, for example, cannot produce action without conscious intervention by the recipient of the communication. And thus, so it is said, 62Scanlon, Freedom of Expression and Categories of Expression, supra note 3, at 531; Strauss, supra note 61; see also Gerald Dworkin, Paternalism: Some Second Thoughts, in Paternalism 105, 106–07 (Rolf Sartorius ed., 1983) (arguing that autonomy is denied when there is manipulation of someone’s “information set”); Richard Moon, The Constitutional Protection of Freedom of Expression 21 (2000) (arguing that full autonomy can only “emerge in the social realm”). any attempt on the part of government (or, for that matter, anyone else) to impede the flow of communication prerequisite to this decision would amount to a lack of respect for the recipient’s autonomy. If we are not able to decide for ourselves what to do, including deciding to do or not do things for which the state will punish us if we do them, then we have lost, so the argument goes, an irreducible element of what it is to be human. And since it would not be plausible to take the value of autonomy as generating or justifying a total right (as a Hohfeldian privilege, 63Wesley Newcomb Hohfeld, Some Fundamental Legal Conceptions as Applied in Judicial Reasoning, 23 Yale L.J. 16, 32 (1913). or liberty 64On the use of the word “liberty” as a synonym for what Hohfeld called a “privilege,” see Leif Wenar, Rights, Stan. Encyclopedia Phil., (last updated July 2, 2011). ) to liberty—the right simply to do whatever one wishes—the right that autonomy generates is typically restricted to the inputs to autonomous decision-making as opposed to the outputs. 65See Roger A. Shiner, Freedom of Commercial Expression 167–68, 234–38 (2003) (analyzing differences between speaker-focused and hearer-focused autonomy). Freedom of communication, therefore, is said to be the freedom of a decision-maker to unimpeded access to those arguments and information necessary for her to make the best decision she can about what actually to do. Under this account, freedom of speech is in the final analysis about freedom of decision, and thus necessarily about freedom of thought.

Yet although freedom of thought certainly seems like a good thing, we need to further sharpen the inquiry. So let us begin by dividing a person’s thoughts into four categories. First are thoughts as to which rightness or wrongness is beside the point, such as thinking that vanilla ice cream is better than chocolate, or thinking about any of the other topics that we tend to designate as tastes. Second are thoughts that are simply correct, such as the thoughts that skin color is irrelevant to moral worth, or that Louisville is not the capital of Kentucky. Third are thoughts that are actually wrong but generally harmlessly so, as with the beliefs of astrology or the beliefs in the existence of Santa Claus or the Easter Bunny. And fourth are those thoughts that are both wrong and harmful, the category about which much more now needs to be said.

We know, of course, that much of the history of the development of freedom of thought as an idea has stemmed from the frequency with which people, and especially people in power, placed in the fourth category thoughts that more appropriately belonged to the first, second, or third. We know that those in power attempted to prohibit people from denying (or, occasionally, affirming) that Catholicism represented the One True Faith; 66See Thomas More’s Trial by Jury: A Procedural and Legal Review with a Collection of Documents (Henry A. Kelly, Louis W. Karlin & Gerard B. Wegener eds., 2011) (1980); Andrew P. Roach & Maja Angelovska-Panova, Punishment of Heretics: Comparisons and Contrasts Between Western and Eastern Christianity in the Middle Ages, 47 J. Hist. 145, 154 (2012). tried to prohibit Copernicus and Galileo from thinking that the Earth revolved around the Sun; 67See The Trial of Galileo: 1612–1633, at 1–2 (Thomas F. Mayer ed., 2012). and enacted obscenity laws in an effort to prevent people from thinking about extra-marital sex and teenage boys from thinking about sex at all. 68See Alec Craig, Suppressed Books: A History of the Conception of Literary Obscenity 40–53 (1963); Noel Perrin, Dr. Bowdler’s Legacy: A History of Expurgated Books in England and America 163–70 (1969); C.H. Rolph, Books in the Dock 40–61 (1969). Yet although such mistakes have been frequent throughout history, their frequency is a contingent empirical fact—that is, their frequency and consequences will vary with time, place, and culture. Moreover, the frequency of mistakes of this kind is still not inconsistent with the existence of the fourth category of genuinely harmful thoughts. Perhaps the frequency of mistakes, especially by the state, in distinguishing the fourth category from the other three should lead to a principle disempowering the state from drawing the distinction between harmful and harmless speech at all, but that would be a possible outcome of the analysis rather than a premise. 69Moreover, the existence of dangerous mistakes in branding harmless (or beneficial) speech as harmful is only one dimension of a more complex decision-theoretic calculus. Although there are negative consequences to designating harmless or beneficial speech as harmful, so too will there be negative consequences in designating harmful speech as harmless. The full analysis would therefore have to weigh the expected (in the expected value sense of the product of the likelihood of the error multiplied by its consequences) costs of erroneously designating harmless speech as harmful against the expected costs of erroneously designating harmful speech as harmless. When Justice Powell, in Gertz v. Robert Welch, Inc., 70418 U.S. 323 (1974). announced that “under the First Amendment there is no such thing as a false idea,” 71Id. at 339. he was making a (somewhat exaggerated) descriptive claim about First Amendment doctrine, and is thus most charitably understood not as asserting the implausible conceptual or meta-ethical claim that ideas do not have truth value, or that the truth-value of some ideas, even moral ones, 72To be clear, claims about inherent racial superiority or the evil of homosexuality, for example, are erroneous moral claims, but still moral claims. cannot be negative. Rather, Justice Powell is best understood as insisting only that the principles of First Amendment doctrine do not authorize governmental inquiry into the truth of ideas, even ideas that are plainly false.

So let us then turn to the fourth category—the category we can call harmful thoughts. 73Cf. Meir Dan-Cohen, Harmful Thoughts: Essays on Law, Self, and Morality (2002). To put the matter directly, we want to ask whether there can be harmful thoughts, and not just whether under the First Amendment there can be harmful thoughts. Consider as an example, an example inspired by United States v. Stevens, 74559 U.S. 460 (2010) (invalidating on First Amendment grounds a federal law prohibiting the sale and distribution of images of animal cruelty). the thought that it is permissible and maybe even positively desirable for animals to be tortured for the (non-nutritional) gratification of humans. And then let us assume what I hope that most of us would accept—although obviously not all of us, or else Stevens would not even have arisen—that the content of the idea is not just different, and not merely a matter of taste, but simply wrong. We should hold off for a moment considering the objection that the thought will not necessarily lead to actual torture, for this is what we are about to address. But first assume simply that the activity—torturing animals for human gratification—that the thought is about is an activity that is both wrong and harmful.

Initially, we can posit that people who think that animals ought to be tortured are more likely to torture animals than people who do not have that thought. And as so put, the claim seems plainly true, even as we recognize that the percentages may be low. That is, the claim—that people who have the thought that animals ought to be tortured are more likely to torture animals than people who do not have the thought—is a claim that is consistent with many or most people who have the thought not actually committing the act. But as a matter of conditional probability, or relevance in the evidentiary sense, 75Thus, Rule 401 of the Federal Rules of Evidence makes a piece of evidence relevant if it “has any tendency to make a fact more or less probable than it would be without the evidence and the fact is of consequence in determining the action.” Fed. R. Evid. 401. it would be difficult to deny that the probability of a person torturing animals is higher if the person has the thought that torturing animals is a good thing than if the same person does not have that thought. Relatedly, for most animal torturers having the thought that it would be good (or desirable, or necessary, or something of that sort) to torture this animal at this time would be a necessary condition for their engaging in the act of torture. 76I say “most” only in order to keep open the logical but empirically unlikely possibility of an instinctive or reactive act of animal torture. This is decidedly not to say that having the thought is a sufficient condition for a person engaging in the act. It is only to say that in the normal case having the thought is a necessary condition, in addition to it being a probabilistic indicator of undetermined size of the likelihood of animal torture.

None of this, of course, is inconsistent with there being many people (or even most people) who think that torturing animals is a good thing but who still do not torture them. Yet it seems a pretty safe empirical bet that people who do have the thought that torturing animals is a good thing are statistically more likely to torture animals than people who do not have that thought.

Now let us suppose that we as a society believe, and as our laws against animal cruelty reflect, that torturing animals is wrong. And if we believe that torturing animals is wrong, then we ought to believe that the thought that torturing animals is good should be placed in the fourth category—the category of harmful thoughts. The content of the thought is wrong, the behavior that the thought is about is harmful, and the fact that people having the thought is statistically likely to increase the incidence of the harmful behavior itself 77Note that “statistically likely to increase the probability” is not the same as the behavior being likely to occur. Sending text messages while operating an automobile is statistically likely to increase the probability of accidents even though most driver texting does not produce accidents. —more animals are going to wind up being tortured when more people have the thought that torturing them is a good thing than when fewer people have that thought. The thought is a harmful thought, therefore, not simply because of its content, but because having a thought of this content will increase the likelihood of the harmful behavior. The thought that torturing animals is good therefore properly belongs in the fourth category of thoughts that are both wrong and harmful.

The task before us is then to consider how we as a people and as a collective political institution should react to people who have harmful thoughts. We could (and do) respect the autonomy and their freedom to think as they wish by doing nothing—even if in an increasingly likely technological world we could. 78This is not an essay in neuroscience, or, more specifically, in what fMRI scans and other techniques of modern neuroscience now or in the foreseeable future will enable people to find out about what other people are thinking. On the actual or potential application of these techniques to various legal questions, see Michael S. Pardo & Dennis Patterson, Minds, Brains, and Law: The Conceptual Foundations of Law and Neuroscience (2013). Nevertheless, some of these possibilities, however remote, and however large the moral issues they raise, make inquiring into the topic of freedom of thought (as opposed to the external manifestations of that thought) more important now than would have been the case a generation ago. After all, having the thought is not itself harmful, and we are now calling it a harmful thought only because the thought, harmless in itself, increases the likelihood of harmful action.

But if the possessor of harmful thoughts ought to be allowed to indulge those thoughts—to keep having them, and not to be punished for having them—because his autonomy of thinking might not (and probably will not) produce a harmful action, then, similarly, how are we to think about the full range of actions preparatory to, and probabilistically indicative of, the commission of a crime. As we know, many of these preparatory actions are designated as independent crimes—preparatory offenses. 79See Dan Bein, Preparatory Offences, 27 Isr. L. Rev. 185 (1993); Kimberly Kessler Ferzan, Inchoate Crimes at the Prevention/Punishment Divide, 48 San Diego L. Rev. 1273, 1282–85 (2011). Possession of burglar tools is an independent crime, even though the burglar tools possessor might not actually burgle anything. 80An example of the common criminalization of possession of burglar tools is N.Y. Penal Law §140.35 (McKinney 2010). Similarly, possession of a hand grenade is a crime although the hand grenade owner may not try to blow anyone up. 81On the crime of possession of hand grenades, see United States v. Freed, 401 U.S. 601, 609 (1971). And so too with the assault rifle owner who might not shoot anyone, 82See Va. Code Ann. § 18.2-94 (West 2012). Of course the Second Amendment might be relevant here after District of Columbia v. Heller, 554 U.S. 570, 595 (2008), and McDonald v. City of Chicago, 561 U.S. 742, 791 (2010), but exploring that question, including the interesting parallels here between the First and the Second Amendments, would take us too far afield. the pit bull owner who would not allow his pit bulls to go unrestrained, 83On pit bull prohibitions, see Am. Dog Owners Ass’n v. City of Yakima, 777 P.2d 1046 (Wash. 1989) (en banc). and so on.

In cases such as the ones just listed we often criminalize the preparatory act even though the possessor or preparer in the exercise of his autonomy might not ultimately go ahead and actually do what it is we are really worried about. But in most of these cases we nevertheless do not allow respect for an agent’s putative autonomy to lead us to refrain from restraining conduct (or arguing for such restraint) that is preparatory to and probabilistically related to intrinsically harmful actions. And then the question is whether these examples are any different from the example of the animal-torture thinker who might not actually go ahead and torture animals. If we can prosecute the burglar tools owner without a burglary actually taking place, can we prosecute the animal torture thinker without animal torture actual taking place?

Indeed, the basic point extends beyond the narrow domain of preparatory offenses, because the universe of prohibitions broader than the evil against which they are directed is vast. In the service of the goal of reducing automobile accidents we require people to drive below a designated speed, even if in the exercise of their autonomy they might well compensate for the increased risks coming from higher speeds by driving with greater care. Nor do we permit people to possess heroin, even though it is (remotely) possible that they will neither use nor sell it. And in addition, the existence of vicarious liability in numerous domains again shows that the necessity of a mediating volitional act in order for the tort or crime to be effectuated is often thought not to preclude responsibility. That the seller of alcohol, 84See generally Richard A. Smith, Note, A Comparative Analysis of Dramshop Liability and a Proposal for Uniform Legislation, 25 J. Corp. L. 553, 558–63 (2000). the seller of guns, 85Vicarious liability for those who sell the guns subsequently used for unlawful purposes is now largely precluded by the Protection of Lawful Commerce in Arms Act, Pub. L. No. 109-92, 119 Stat. 2095 (2005) (codified at 15 U.S.C. §§ 7901–7903), but the issues are still alive. See Neal S. Shechter, Note, After Newtown: Reconsidering Kelley v. R.G. Industries and the Radical Idea of Product-Category Liability for Manufacturers of Unreasonably Dangerous Firearms, 102 Geo. L.J. 551, 577–78 (2014). the seller of cigarettes, 86See generally Frank J. Vandall, Reallocating the Costs of Smoking: The Application of Absolute Liability to Cigarette Manufacturers, 52 Ohio St. L.J. 405, 406 (1991). or even the employer of a contract killer 87That the employer or even the aider and abettor of a contract killer is liable for murder is plain. United States v. Hardwick, 523 F.3d 94, 97 (2d Cir. 2008); United States v. Dorman, 108 Fed. App’x 228, 231 (6th Cir. 2004). Whether under some circumstances those who provide instructions for contract killing can be liable in tort, or whether instead such instructions are shielded by the First Amendment, remains a contested issue. See Rice v. Paladin Enters., 128 F.3d 233, 242–43, 267, 355–56 (4th Cir. 1997) (rejecting a First Amendment defense for aiders and abettors); Wilson v. Paladin Enters., Inc., 186 F. Supp. 2d 1140, 1144 (D. Or. 2001) (same). is under some circumstances liable in addition to (and, importantly, not instead of, showing that even the existence of a legally responsible volitional intermediary does not preclude vicarious liability) the principal shows once again that respect for autonomy is rarely understood to generate doctrines immunizing anyone other than the final agent from liability. The entire domain of vicarious liability, as well as tort liability for foreseeable misuse, 88See, e.g., Osorio v. One World Techs., Inc., 659 F.3d 81 (1st Cir. 2011) (upholding judgment against manufacturer of a table saw although proper use of the blade guard would have prevented the injury); LeBouef v. Goodyear Tire & Rubber Co., 451 F. Supp. 253, 257 (W.D. La. 1978), aff’d, 623 F.2d 985, 988–89 (5th Cir. 1980) (upholding liability for tire manufacturer when tire user’s intentional disregard of normal safety precautions would have prevented the tire failure that caused the accident). all reject the notion that the existence of an autonomous agent whose autonomous decision causes a harm will serve to immunize from liability those other agents whose actions might have contributed to the autonomous agent’s harmful acts.

Although the examples that undercut the argument from autonomy thus exist even outside the realm of preparatory offenses, let us stay on the safest ground by considering only the category of harmful preparatory acts. As just noted, such acts, with little controversy, are routinely and generally both criminalized and taken to provide the basis for tort liability. It turns out, therefore, that there is no general existing legal principle immunizing intrinsically harmless acts from legal sanction if they are probabilistically related to actual harmful acts. And because we have seen that we can characterize harmful thoughts in much the same way, the question of how to conceive of harmful thoughts is importantly similar to the question of how to conceive of harmful preparatory acts.

Identifying the similarity between harmful thoughts and harmful preparatory acts enables us to sharpen the inquiry even further. Let us say that having thought T increases the probability of harm H(1) with a probability of P(1), and thus that the expected harm—EH(1)—is P(1) x H(1). But the same structure applies to acts as well, so that we can also say that engaging in action A (buying or possessing burglar tools, say) increases the probability of harm H(2) by P(2), and thus that the expected harm—EH(2)—is P(2) x H(2). The important question, therefore, is whether we should treat EH(1) differently—less restrictively—from how we would (and do) treat EH(2), when their values are equivalent, or when EH(1) is greater than EH(2). To put it differently, should we immunize harm-producing thoughts from restriction more than we immunize harm-producing actions when the expected harms are the same, or when the expected harms of the thought are greater than the expected harms of the action. In other words, should there be a principle of freedom of thought?

The alert reader will have noticed that I have just translated into freedom of thought terms the same formulation that Scanlon and others 89See supra notes 3, 61–62. have used to characterize a free speech principle. If a principle of freedom of speech just means that speech enjoys a degree of immunity from government action greater than that possessed by non-speech action having the same or equivalent negative consequences, then a principle of freedom of thought must similarly mean that thoughts have a degree of immunity from government action greater than that enjoyed by non-thought conduct again having the same or equivalent negative consequences. But if this is what a robust or genuine principle of freedom of thought must mean, we still must address whether as a normative matter such a principle ought to be accepted. 90I again bracket the question of the physical inevitability of freedom of thought, in part because the physical inevitability captured by the concentration camp slogan, “Die Gedanken sind frei,” may in our modern neuroscientific world no longer be so inevitable. See supra note 78 and accompanying text.

In addressing the question whether there should be a principle immunizing harmful thoughts in a way that harmful preparatory acts having equivalent expected harm are not immunized, it is necessary to return to the argument from autonomy. 91See supra text accompanying notes 61–65. It seems initially obvious that respect for someone’s autonomy requires that we respect her thought processes. 92See Seana Valentine Shiffrin, Methodology in Free Speech Theory, 97 Va. L. Rev. 549, 554 (2011) (emphasizing the importance of “control over one’s thoughts”). But then we must still ask why we should respect someone’s right (and the existence of a right in the sense explicated in Part I above is what is now under consideration) to have a wrong and harmful thought?

One reason for respecting what appear to be harmful thoughts is the possibility that the possessor of the thought is right, that we (or the state) are wrong, and that consequently the thought is not in fact harmful. This Millian perspective—it was John Stuart Mill who most famously connected the liberty of (expressed) opinion to epistemic fallibilism and thus to the possibility that the received opinion is mistaken and the suppressed opinion correct 93John Stuart Mill, On Liberty (David Spitz ed., W. W. Norton & Co., Inc. 1975) (1859). On Mill’s fallibilism, see Stephen Holmes, John Stuart Mill: Fallibilism, Expertise, and the Politics-Science Analogy, in Knowledge and Politics 125, 126–27 (Marcelo Dascal & Ora Gruengard eds., 1989); John Skorupski, John Stuart Mill 8–12 (1989). —is certainly a wise caution before acting on any received belief and before restricting any action thought harmful.

Although humility is almost always wise, it is not yet apparent that a principle of humility can generate the differential principle we are seeking. For although I might be wrong in thinking that that you are wrong in what you are thinking, I might also be wrong in thinking that you are wrong in what you are doing. Humility is in general a good thing, yet it is far from clear that there is reason to believe that my (over)confidence about my assessment of your thoughts ought to be tempered by a stronger principle of humility than that which should properly temper my assessment of your actions, especially your preparatory actions. Yes, I might be wrong, but it is hard to see why that possibility should be any less applicable to my empirical probabilistic assessments of the consequences of your preparatory or other accessory actions than it is to my empirical probabilistic assessment of the consequences of your thoughts. 94See Scanlon, Comment on Baker’s Autonomy and Free Speech, supra note 3, at 322 n.11 (noting with respect to autonomy that the “same is true of paternalistic legislation restricting behavior other than expression”).

Of course we know, as discussed above, 95See supra notes 73–77 and accompanying text. that thoughts might well not produce actions. But this rejoinder is unavailing. It is true that thoughts might not produce actions consistent with those thoughts, but harmless (as stipulated) actions that are probabilistically causally related to harmful actions might also not produce those harmful actions. If respect for some agent’s ability to change her mind before committing an actually harmful act leads to a principle of moral, constitutional, or legal protection, then it ought logically to lead to a principle protecting harmless acts just as it protects harmless thoughts. Or, to be more precise, there remains no reason to reject a principle protecting acts whose expected harm is no greater than the expected harm of what we have been calling a harmful thought.

Even more importantly, it is not clear whether the idea of autonomy can do the work that many theorists appear to be requiring of it. 96On the connection between freedom of thought and autonomy, see especially Shiffrin, supra note 92, at 554; Seana Valentine Shiffrin, Reply to Critics, 27 Const. Comment. 417, 418–26 (2011); and Seana Valentine Shiffrin, A Thinker-Based Approach to Freedom of Speech, 27 Const. Comment. 283 (2011). Consider, for example, the 1964 incident involving Kitty Genovese, who was brutally attacked and murdered while multiple onlookers allegedly did nothing at all. 97The accuracy of the standard story, according to which thirty-eight people heard her screams and did nothing, is the source of some controversy. See Kevin Cook, Kitty Genovese: The Murder, the Bystander, the Crime that Changed America (2014). Now one (deliberately offensive and tendentious) way of describing the behavior of the non-intervening onlookers is as respecting the autonomy of the attacker. But of course this characterization is absurd, and it is absurd precisely because we take the principle of autonomy to be limited by, or to exclude, an agent’s autonomous decision to harm others. My freedom to swing my arm ends at the tip of your nose, as it is said. 98Zechariah Chafee, Jr., Freedom of Speech in Wartime, 32 Harv. L. Rev. 932, 957 (1919) (“Your right to swing your arms ends just where the other man’s nose begins.”). Chafee may have gotten the line from Holmes, or it may even have been around for longer. And once we see this, we can understand that autonomy is in an important way asymmetric. Once we engage in the discounting commended by the principle of humility, we have no reason to respect a person’s autonomous other-regarding and harm-producing actions, even as she, presumably (or hopefully) applying the same principle of humility to discount her own assessment of her own proposed action, preserves her obligation to do what she thinks best. Just as you have an obligation to do what you think it is best to do, so too do I have an obligation to try to keep you from doing so when my assessment is that what you think it is best to do will be both other-regarding and harm-producing. 99See Larry Alexander, Can Law Survive the Asymmetry of Authority?, 19 Quinnipiac L. Rev. 463, 468 (2000); Larry Alexander, “With Me, It’s All er Nuthin’”: Formalism in Law and Morality, 66 U. Chi. L. Rev. 530, 553 (1999).

The foregoing argument does not work, of course, for any instance in which a preparatory thought has a lower expected harm than a preparatory action. But we have stipulated an equivalence, and with that stipulation in place the asymmetry of autonomy—your obligation to follow your autonomy where it leads on the basis of your best judgment that it is for the good does not entail my obligation to refrain from attempting to limit that autonomy when in my best judgment your (mistaken) exercise of your (mistaken) best judgment will produce harm to third parties 100And if we reject anti-paternalism as a distinct principle, the same argument applies to harms to the agent herself. For a forceful and recent rejection of anti-paternalism, see Sarah Conly, Against Autonomy: Justifying Coercive Paternalism (2013). —produces the conclusion that a principle of autonomy cannot generate a distinct principle of freedom of thought. Of course the stipulation of equivalence may be empirically unsound, and the class of harmful thoughts (a more plausible way of understanding the issue as opposed to making these assessments in each individual case) may be a class less harmful, in the expected harm sense, than the class of harmful preparatory actions. Accordingly, we need to relax the stipulation of equivalence and take this possibility seriously. But before doing so, it will be important to shift the analysis from freedom of thought to freedom of speech.

IV. From Thoughts to Speech

I have up to now framed the issue in terms of freedom of thought rather than freedom of speech. And there are two reasons for this: First, freedom of thought appears to have an even tighter connection to autonomy than does freedom of speech. What, after all, is more mine than my thoughts? Second, and partly as a consequence of the first reason, the argument for there being no principle of freedom of thought appears initially to be even stronger than the argument for there being no principle of freedom of speech. As a result, if the more difficult argument against a free thought principle is sound, then so too, a fortiori, will be the less difficult argument against a free speech principle—at least an autonomy-based or otherwise individualistic argument against a free speech principle. 101In recent years freedom of thought has been influentially urged as a basis for a principle of free speech. See Charles Fried, The New First Amendment Jurisprudence: A Threat to Liberty, 59 U. Chi. L. Rev. 225, 233 (1992); Shiffrin, A Thinker-Based Approach to Freedom of Speech, supra note 96; Shiffrin, supra note 92, at 554; Shiffrin, Reply to Critics, supra note 96, at 418–26. But as Shiffrin makes clear, these efforts have started with the premise that free speech exists as an independent principle, but one which remains in need of justification. See Shiffrin, supra note 92, at 550 (describing her project as one of developing an “ideal theory of freedom of speech”). That is an important enterprise, but justifying a principle of free speech that is stipulated (in an argument, or in an item of positive law such as the First Amendment) is different from the project, which is mine here, of not assuming such a principle at the outset, but, rather, of exploring whether what others take as their premise can itself be justified. See id.

So if we now examine directly the question of free speech, we can run the same analysis. Initially, we should put aside three different classes of speech, for reasons that will become apparent. First is speech that by its very utterance causes injury, to use the Chaplinsky formulation, 102Chaplinsky v. New Hampshire, 315 U.S. 568, 572 (1942) (describing fighting words as words “which by their very utterance inflict injury”). for here the harm from speech resembles the harm from a punch, a kick, or, perhaps even more relevantly, a spit. 103Spitting is interesting in this context because whatever harm is produced by being intentionally spat upon, just like the harm of being slapped in the face, has little to do with pain, and even less to do with a wound or physical injury. And thus it seems incumbent on defenders of a speech–action distinction to explain why the plainly communicative and non-physically-harmful act of spitting on another ought to be excluded from the realm of free-speech-relevant behavior. Second is the class in which the causal relationship simply does not exist, as in the case of the often alleged but never proved causal relationship between non-violent, non-degrading, and non-exploitative sexually explicit material and sexual violence. 104See Frederick Schauer, Causation Theory and the Causes of Sexual Violence, 1987 Am. B. Found. Res. J. 737, 737–38. And the third class is comprised of those instances in which there may be a causal relationship between some category of speech and some category of consequences, but where the consequences are not in fact harmful, as with the nineteenth century worry that exposure to sexually explicit materials would cause teenage boys to masturbate more than would otherwise have been the case. 105See Rolph, supra note 68, at 53. This worry appears to be at the heart of the case most important in establishing nineteenth century obscenity doctrine, Regina v. Hicklin, [1868] 3 LRQB 360, at 361, 363 (Eng.), although the court in that case puts the matter in somewhat less transparent terminology.

Having bracketed these three classes, we are left with the task of examining speech that is not harmful in itself but which is genuinely causally likely to increase the incidence of genuinely harmful actions. And we can examine this speech in the same way that we have examined freedom of thought. 106See supra Part III. Thus, if we assume equivalent causal contributions to some harmful consequence by some act of speech and by some act of non-speech action, or if we simply assume equivalent expected harm, is there any reason to immunize preparatory speech acts to a greater extent than we immunize preparatory non speech conduct? There may be, but, as we have just seen, if there is such a reason, it is a not a reason that can be derived from conceptions of autonomy. 107See text accompanying notes 59–65. If the principle of autonomy protects neither autonomous harm producing actions nor the actions that would increase the likelihood of subsequent and consequent harm, then there seems no reason, just as in the case of thoughts whose consequences have the same structure, to treat speech differently from the way in which we treat action.

At this point it is necessary to return to Emerson. Once we understand that intrinsically harmless preparatory actions are routinely prohibited when they are probabilistically related to harmful acts, even though a volitional act is necessary to link the harmless act to the harmful one, 108See supra note 83–88 and accompanying text. the existence of such a volitional link can no longer constitute a sound basis for distinguishing speech from action. Rather, the case for speech being relevantly different from action would have to rest on the determination that the category of speech that is causally productive of harmful action is, as a category, less efficacious than is the category of harmless actions that are causally productive of harmful actions.

Although Emerson and others have simply asserted this to be the case, 109See supra note 50. it turns out that this is a virtually impossible proposition to establish. Once we understand that the relevant comparison is not between speech and harmful actions but rather between speech and those actions, not harmful in themselves, that might produce harmful actions, then the existence of a free speech principle that would immunize the former but not the latter from control must rest on this extremely problematic empirical claim. We could of course say that under circumstances of empirical uncertainty the default rule ought to be freedom, but we could say this about action as easily as about speech. And thus the putative default rule would still not provide a sound basis for the distinction between speech and action or for an autonomy-based principle of freedom of speech. This approach may be effective as a matter of existing positive law, but as a matter of pre-positive-law analysis we appear to be left with the view that any autonomy-based or individualistic conception of freedom of speech rests far more on an unestablished and likely false conclusion about the harmlessness of speech than is typically recognized.

To be more specific, if the claim that speech is harmless rests on its inability to be causally related to intrinsically harmful acts, then we have far too many counter-examples in our experience to be comfortable with this formulation. 110See, e.g., supra notes 102–03 and accompanying text. But if the claim instead rests on the view that any harm that is (probabilistically) caused by a communicative act requires the mediation of any additional volitional act on the part of the recipient of the communication, then accepting this conception of harmlessness would unsettle far more of existing tort and other non-speech law than we have previously thought. And with both of these alternatives aside, therefore, what remains is a view of speech that maintains that speech, as a category, is differentially harmless, a view that is in need of far more empirical evidence than it has thus far received. It is of course true that most speech is harmless. But it is also true that most action is harmless. Recognizing this seemingly obvious but often ignored fact has led many people to make claims about the categorial or differential harmlessness of speech 111See supra note 50. that may not stand up to close analysis or empirical testing.

Because it is true both that most speech will not produce harmful action and that most action will not produce harmful action, the question before us is not whether a society should permit the regulation of harmless speech. Rather, it is the question whether the proper line between permissible and impermissible state intervention is a line between speech and action, or instead, and seemingly preferably, between causally inert (of harm) behavior, some of which is speech and some of which is action, and causally efficacious (of harm) behavior, some of which is speech and some of which is action. Under the latter principle, some speech now protected under existing First Amendment doctrine would be regulable, but if applied accurately the differences between this principle and existing doctrine might well be small. Under non-ideal conditions the difference might of course be greater, but the question then is still whether guarding against the consequences of non-ideal application should be a strategy limited to speech or rather one of more pervasive application.

Conclusion

The basic idea of this Article is that it is hardly clear that respect for an agent’s autonomy ought to lead other agents, or the state, to tolerate autonomous communicative actions that are determined to be likely to cause harm to third parties any more than they should tolerate autonomous non-communicative actions whose consequences are equivalent. If the principle of freedom to engage in autonomous actions is one that is limited to cases of harm to others, then, at the very least, this limitation has more impact on standard autonomy views about freedom of speech than has commonly been appreciated. If there is a reason to protect autonomous speech that does not apply as well to autonomous action, then, as we have seen, it cannot be a reason derived from the idea of autonomy itself, and instead needs to be based on a much greater empirical showing of the differential consequences of speech and action than has to date been provided.

All of this might be irrelevant to the very existence of a free speech principle if it turns out that speech, or a subset of speech, or, most accurately, a subset of communication, has as a category positive attributes not possessed by the category of non-speech action. And thus even if the harm-producing capacities of the two categories were equivalent, differential benefit-producing capacities might still be sufficient to justify a sound free speech principle. And although dealing with that possibility must remain for another day, the analysis here may suggest that even that task—and thus the task of distinguishing speech from action for purposes of explaining, say, the value of speech in the search for truth 112For a collection of the most important works in the “search for truth” or “marketplace of ideas” free speech tradition, see Vincent Blasi, Ideas of the First Amendment 318–689 (2d ed. 2012). —is considerably more difficult than has traditionally been assumed. 113On the empirical difficulties, see Daniel E. Ho & Frederick Schauer, Testing the Marketplace of Ideas, 90 N.Y.U. L. Rev. (forthcoming 2015).

In addition, this Article is situated almost entirely within the domain of individualistic or autonomy-based accounts of freedom of speech rather than democracy-based accounts. Some of the latter are about the methods of democratic deliberation and decision-making, 114E.g., Alexander Meiklejohn, Free Speech and Its Relation to Self-Government (1948); Robert Post, Citizens Divided: Campaign Finance Reform and the Constitution 157 (2014); Cass R. Sunstein, Democracy and the Problem of Free Speech (1993); Lillian BeVier, The First Amendment and Political Speech: An Inquiry into the Substance and Limits of Principle, 30 Stan. L. Rev. 299 (1978); Robert H. Bork, Neutral Principles and Some First Amendment Problems, 47 Ind. L.J. 1 (1971); Frank A. Morrow, Speech, Expression, and the Constitution, 85 Ethics 235 (1975). and some are based on the importance in a democracy of using speech and the press as checks on the possibility of governmental overreaching or other abuse of power. 115See Vincent Blasi, The Checking Value in First Amendment Theory, 1977 Am. B. Found. Res. J. 521, 524. But what most species of democracy-based accounts of free speech share is a non-reliance on the distinction between speech and action 116And that is possibly why such accounts might better be situated as components of a broader protection of democracy rather than as a self-standing right. See Austl. Cap. Television Pty. Ltd. v Commonwealth (1992) 177 CLR. 106, 244 (Austl.) (locating a freedom of political communication within the Australian Constitution’s protection of democracy); Bork, supra note 54. that is so central to individualistic or autonomy-based accounts, whether they be based on listener-autonomy or speaker-autonomy or both. Listeners may secure information valuable to their autonomy in countless ways, only some of which are based on the propositions uttered by others. And speakers may embody their autonomy also in countless ways, only some of which involve speaking in even the broadest sense. There may be good historical reasons for carving out speech from these broader categories, but if we set aside the history and the existing legal or constitutional doctrine, we will discover that the non-historical reasons for doing so—for distinguishing speech from action—do not stand up to close analytic scrutiny.

Footnotes

David and Mary Harrison Distinguished Professor of Law, University of Virginia. This Article was prepared for the Emory University School of Law’s Randolph W. Thrower Symposium, February 5, 2015, on “The New Age of Communication: Freedom of Speech in the 21st Century.” Earlier versions were presented at the University of California, Berkeley, Boalt Hall School of Law; the Harvard Kennedy School; the University of Chicago Law School; the New York University Colloquium on Political Philosophy; and the University of Lucerne. I have profited from reactions and questions on those occasions, as well as from comments on earlier drafts from Mary Anne Case, Aziz Huq, Leslie Kendrick, Liam Murphy, Micah Schwartzman, Samuel Scheffler, David Strauss, John Tasioulas, and Alex Tsesis.

1U.S. Const. amend. I.

2By way of emphasizing the preliminary nature of the terminology, I do not at this point in the analysis want anything to turn on the word “action,” especially because all speech is a form of action (or conduct). But the basic idea is that of a distinction between speech and non-speech behavior, or between the behavior that we designate as “speech” and the behavior that we do not.

3See Frederick Schauer, Free Speech: A Philosophical Enquiry 3–14 (1982) (analyzing the necessary conditions for a distinct Free Speech Principle); Thomas Scanlon, A Theory of Freedom of Expression, 1 Phil. & Pub. Aff. 204, 204 (1972) [hereinafter Scanlon, A Theory of Freedom of Expression] (“The doctrine of freedom of expression is generally thought to single out a class of ‘protected acts’ which it holds to be immune from restrictions to which other acts are subject.”). Although Scanlon now believes that the influential account he offered in 1972 is “mistaken in important respects,” T.M. Scanlon, Comment on Baker’s Autonomy and Free Speech, 27 Const. Comment. 319, 319 (2011) [hereinafter Scanlon, Comment on Baker’s Autonomy]; see also T.M. Scanlon, Jr., Freedom of Expression and Categories of Expression, 40 U. Pitt. L. Rev. 519, 530–34 (1979) [hereinafter Scanlon, Freedom of Expression and Categories of Expression], the subsequent narrowing of his earlier position, even if relevant to taking the earlier article as a self-standing contribution usable or not on its own terms, does not touch the basic analytic point about the structure of a free speech principle.Although the most obvious application of the distinction set out in the previous paragraph is with respect to a differential immunity from restriction for speech and non-speech behavior causing equivalent harm or other negative consequences, the distinction between speech and non-speech behavior, as articulated in the text, could also arise in the context of positive rather than negative consequences. See Scanlon, A Theory of Freedom of Expression, supra, at 204. For example, if there were an affirmative obligation on the part of government to subsidize speech-relevant activities, a free speech principle would generate a greater obligation to subsidize or otherwise support speech than to subsidize or support non-speech activities bringing equivalent benefits. What is key is the differential, and not whether the differential attaches to the restrictive as opposed to the supportive activities of the agent against whom the free speech claim is offered. See Leslie Kendrick, Free Speech as a Special Right (unpublished manuscript) (on file with author).

4Cf. Schenck v. United States, 249 U.S. 47, 52 (1919) (“The most stringent protection of free speech would not protect a man in falsely shouting fire in a theatre and causing a panic.”).

5René Magritte, The Treachery of Images (1928–1929) (“Ceci n’est pas une pipe”).

6See, e.g., Rae Langton, Sexual Solipsism: Philosophical Essays on Pornography and Objectification 27–37, 103–16 (2009) (drawing on speech-act theory to consider speech as action rather than distinct from it); Catharine A. MacKinnon, Only Words 29–41 (1993) (characterizing the speech–action distinction as legalistic); Susan J. Brison, Speech, Harm, and the Mind-Body Problem in First Amendment Jurisprudence, 4 L. Theory 39, 56–61 (1998) (challenging the distinction between the verbal and the physical); Mary Ellen Gale, Reimagining the First Amendment: Racist Speech and Equal Liberty, 65 St. John’s L. Rev. 119, 149 (1991) (questioning the distinctions between speech and action and speech and conduct); Alon Harel, Hate Speech and Comprehensive Forms of Life, in The Content and Context of Hate Speech: Rethinking Regulation and Responses 306, 308–10 (Michael Herz & Peter Molnar eds., 2012) (same). Stanley Fish also questions the distinction, but, given his skeptical stance about the ontology of distinctions in general, it is difficult to know what to make of his claims about the distinction between speech and action. Stanley Fish, There’s No Such Thing as Free Speech, and It’s a Good Thing, Too 106 (1994) (noting the “general difficulty of separating speech from action”).

7See, e.g., Amy Adler, Inverting the First Amendment, 149 U. Pa. L. Rev. 921, 927, 970–85 (2001) (arguing that the speech–action distinction plays a “central role in the First Amendment”); Nadine Strossen, Regulating Racist Speech on Campus: A Modest Proposal?, 1990 Duke L.J. 484, 492, 494, 532–33, 541–44 (objecting to the “abrogation of the traditional distinctions between speech and conduct”); Kathleen M. Sullivan, Discrimination, Distribution and Free Speech, 37 Ariz. L. Rev. 439, 442, 444 (1995) (responding to attacks on the distinction between speech and conduct); Kathleen M. Sullivan, Free Speech Wars, 48 SMU L. Rev. 203, 206, 209–10 (1994) (same).

8See, e.g., Charles W. Collier, Hate Speech and the Mind-Body Problem: A Critique of Postmodern Censorship Theory, 7 L. Theory 203, 204 & n.7 (2001) (assuming that there is a “constitutionally significant difference between speech and . . . action”); Stephen E. Gottlieb, The Speech Clause and the Limits of Neutrality, 51 Alb. L. Rev. 19, 23 (1986) (noting that much of First Amendment doctrine is “predicated on the speech/action distinction”).

9Such contexts would include not only philosophical inquiry for its own sake, but also institutional and constitutional design in domains in which free speech principles are not yet accepted. One example would be countries with rudimentary free speech protection, and as to which foundational questions might thus be asked about how much free speech, if any, should be permitted. Another would be non-governmental settings (corporations and private colleges and universities, for example) in which free speech is considered in the context of institutional design decisions about who should be allowed or encouraged to say what, even apart from questions of positive law. And of course the groundings of the idea of free speech are plainly relevant to questions arising in the interpretation of the First Amendment itself, as is apparent both from the fact that much of existing free speech theory has been developed in the context of the First Amendment and from the frequency with which the Supreme Court makes reference to foundational principles in deciding First Amendment cases. E.g., United States v. Alvarez, 132 S. Ct. 2537, 2547–50 (2012) (relying on the foundational premise that truth is not to be authoritatively determined by government); Cohen v. California, 403 U.S. 15, 24–26 (1971) (basing decision on self-expression rationale for the First Amendment); Stanley v. Georgia, 394 U.S. 557, 564–65 (1969) (stressing freedom of belief and freedom of thought); Vincent Blasi, Ideas of the First Amendment vii-viii (2d ed. 2012); Paul Horwitz, First Amendment Institutions 8–9 (2013); Robert C. Post, Constitutional Domains: Democracy, Community, Management 15–16 (1995); Steven H. Shiffrin, The First Amendment, Democracy, and Romance 2–6 (1990).

10See, e.g., Ferguson v. Skrupa, 372 U.S. 726, 730–32 (1963) (holding that the wisdom of a law restricting the practice of debt adjustment to lawyers was for the legislature and not for the courts to decide); Williamson v. Lee Optical of Okla., Inc., 348 U.S. 483, 487–90 (1955) (upholding on rational basis grounds the requirement that opticians could not fit eyeglasses without receiving a prescription from an optometrist or ophthalmologist); W. Coast Hotel Co. v. Parrish, 300 U.S. 379, 391, 398 (1937) (holding that regulation for purposes of health, safety, morals, and welfare satisfied requirements of due process as long as it bore a “reasonable relation to a proper legislative purpose, and [was] neither arbitrary nor discriminatory” (quoting Nebbia v. New York, 291 U.S. 502, 537 (1934))); Nebbia v. New York, 291 U.S. 502, 537 (1934) (applying reasonableness test to regulation of business); see also United States v. Carolene Prods. Co., 304 U.S. 144, 152–53 (1938) (announcing that “rational basis” is the standard to be employed in evaluating the constitutionality of social and economic regulation).

11See also N.Y.C. Transit Auth. v. Beazer, 440 U.S. 568, 594 (1979) (concluding that courts applying rationality review should not interfere with policy decisions “[n]o matter how unwise” they may be); John F. Manning, The Necessary and Proper Clause and Its Legal Antecedents, 92 B.U. L. Rev. 1349, 1349 (2012) (describing rationality review as “very forgiving”); Cass R. Sunstein, Naked Preferences and the Constitution, 84 Colum. L. Rev. 1689, 1713 (1984) (concluding that rationality review almost always results in validation); Shoshana Zimmerman, Pushing the Boundaries?: Equal Protection, Rational Basis, and Rational Decision Making by District Courts in Cases Challenging Legislative Classifications on the Basis of Sexual Orientation, 21 S. Cal. Interdisc. L.J. 727, 733 (2012) (same). See generally Scott H. Bice, Rationality Analysis in Constitutional Law, 65 Minn. L. Rev. 1, 3–4 (1980).

12And thus the same analytic point applies even if, as is occasionally the case, rational basis review has (or is argued to have) slightly more “bite.” See Logan v. Zimmerman Brush Co., 455 U.S. 422, 442 (1982) (concluding the rationality review requires “something more than the exercise of a strained imagination”); U.S. R.R. Ret. Bd. v. Fritz, 449 U.S. 166, 180 (1980) (Stevens, J., concurring) (arguing that rationality review should require more than simply a “conceivable” or “plausible” basis); U.S. Dep’t of Agric. v. Moreno, 413 U.S. 528, 535–38 (1973) (invalidating food stamp qualification criterion on rational basis grounds).

13See Washington v. Glucksberg, 521 U.S. 702, 735 (1997) (rejecting right to assisted suicide on rational basis grounds); Kelley v. Johnson, 425 U.S. 238, 248 (1976) (upholding rationality and thus constitutionality of personal appearance regulation for police officers). At one time the rational basis standard applied to sexual conduct, but no longer. Lawrence v. Texas, 539 U.S. 558, 578 (2003) (overruling Bowers v. Hardwick, but not specifying the standard of review); Bowers v. Hardwick, 478 U.S. 186, 196 (1986) (finding a moral basis for a law sufficient to satisfy the rationality standard).

14See supra note 10 and accompanying text.

15Although commonly associated with equal protection or due process, the compelling interest formulation has occasionally surfaced in First Amendment doctrine. See New York v. Ferber, 458 U.S. 747, 756–57 (1982) (recognizing “compelling” interest, even as against First Amendment concerns, in protecting the well-being of minors); Roe v. Wade, 410 U.S. 113, 163 (1973) (“compelling”); Loving v. Virginia, 388 U.S. 1, 9 (1967) (“The fact of equal application does not immunize the statute from the very heavy burden of justification which the Fourteenth Amendment has traditionally required of state statutes . . . .”); Korematsu v. United States, 323 U.S. 214, 216 (1944) (“It is to say that courts must subject [restrictions] to the most rigid scrutiny.”).

16It is more than plausible to characterize the clear and present danger principle in this way, given that normally the state may, if constrained only by the rational basis standard, deal with dangers that are neither clear nor present. Schenck v. United States, 249 U.S. 47, 52 (1919). Much the same can be said about the existing crystallization of the clear and present danger standard in Brandenburg v. Ohio. 395 U.S. 444, 447 (1969) (holding that advocacy of illegal action may be prohibited if it is directed to producing imminent lawless action and is likely to produce that result). Brandenburg uses the standards of imminence and likelihood to embody its stringent protection of speech, but Brandenburg’s implicit assumption is that, when dealing with non-speech behavior, the state may address problems that are neither “imminent” nor even “likely.” Frederick Schauer, Is it Better to Be Safe than Sorry?: Free Speech and the Precautionary Principle, 36 Pepp. L. Rev. 301, 307 (2009). More general examinations of the strength of the Brandenburg standard include, notably, Hans A. Linde, “Clear and Present Danger” Reexamined: Dissonance in the Brandenburg Concerto, 22 Stan. L. Rev. 1163, 1185 (1970); Frank R. Strong, Fifty Years of “Clear and Present Danger”: From Schenck to Brandenburg—And Beyond, 1969 Sup. Ct. Rev. 41, 52.

17Cent. Hudson Gas & Elec. Corp. v. Pub. Serv. Comm’n of N.Y., 447 U.S. 557, 566 (1980) (setting out the standard applicable to commercial advertising). The Central Hudson test remains the applicable standard today. See, e.g., Sorrell v. IMS Health Inc., 131 S. Ct. 2653, 2667–68 (2011); Lorillard Tobacco Co. v. Reilly, 533 U.S. 525, 553–55 (2001).

18Or, for that matter, the goal of the regulation, as would be the case under the so-called O’Brien analysis, according to which heightened First Amendment scrutiny is triggered not by the nature of the activity regulated but instead by the nature of the state’s interest in regulating. United States v. O’Brien, 391 U.S. 367, 376–77 (1968); see Laurence H. Tribe, American Constitutional Law 792 (2d ed. 1988); John Hart Ely, Flag Desecration: A Case Study in the Roles of Categorization and Balancing in First Amendment Analysis, 88 Harv. L. Rev. 1482, 1496–501 (1975). The roles of government purpose and motive in First Amendment adjudication are analyzed comprehensively in Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amendment Doctrine, 63 U. Chi. L. Rev. 413 (1996), and Jed Rubenfeld, The First Amendment’s Purpose, 53 Stan. L. Rev. 767, 793–95 (2001).

19See supra note 16.

20See Ashcroft v. Free Speech Coal., 535 U.S. 234, 253 (2002) (observing that First Amendment doctrine rests on “vital distinctions between words and deeds” and between “ideas and conduct”).

21On the conditions necessary to conceive of the right to free speech (or any other right) as a distinct or independent right, see Frederick Schauer, Free Speech on Tuesdays, 34 L. & Phil. 119 (2015).

22New Orleans v. Dukes, 427 U.S. 297, 303–05 (1976).

23See, e.g., id. at 305.

24This approach to rights appears to be implicit in Ronald Dworkin, Taking Rights Seriously 266–67 (1977) (finding the idea of a general right to liberty “absurd”). The argument for such a right to liberty, developed as an objection to Dworkin, is set out in Douglas N. Husak, Ronald Dworkin and the Right to Liberty, 90 Ethics 123 (1979) (“[T]here is no reason to believe that the right to liberty cannot conflict with the right to equality.”).

25See Dworkin, supra note 24.

26See Randy E. Barnett, Restoring the Lost Constitution: The Presumption of Liberty 4 (rev. ed. 2014); Randy E. Barnett, The Structure of Liberty: Justice and the Rule of Law 14–15 (2d ed. 2014) (arguing for a natural rights presumption of liberty).

27See supra note 3 and accompanying text.

28Thomas I. Emerson, Toward a General Theory of the First Amendment, 72 Yale L.J. 877, 914–15 (1963).

29Thomas I. Emerson, The System of Freedom of Expression 6–20 (1970).

30Id. On the distinction between what the First Amendment covers and the degree of protection it offers to what it does cover, see Frederick Schauer, The Boundaries of the First Amendment: A Preliminary Exploration of Constitutional Salience, 117 Harv. L. Rev. 1765, 1769–77 (2004). Although the Supreme Court has often conflated questions of coverage and questions about the degree of protection, it has more recently appeared to recognize the distinction and its importance. See United States v. Stevens, 559 U.S. 460, 468, 470 (2010) (announcing a presumption against creating new categories of uncovered speech); see also Brown v. Entm’t Merchants Ass’n, 131 S. Ct. 2729, 2733–35 (2011) (same, but using the word “protection” to designate coverage).

31Emerson, supra note 29, at 8–9, 17–18.

32Pittsburgh Press Co. v. Pittsburgh Comm’n on Human Relations, 413 U.S. 376, 398–99 (1973) (Douglas, J., dissenting) (“There comes a time, of course, when speech and action are so closely brigaded that they are really one.”); see also Parker v. Levy, 417 U.S. 733, 768 (1974) (Rehnquist, J., dissenting) (“A command is a speech brigaded with action . . . .”); DeFunis v. Odegaard, 416 U.S. 312, 343 (1974) (Douglas, J., dissenting) (“Speech is closely brigaded with action when it triggers a fight, as shouting ‘fire’ in a crowded theater triggers a riot.” (citation omitted)).

33Emerson, supra note 29, at 403–05.

34Id.

35Id. at 356–59, 444–47.

36Id. at 414–17.

37Id. at 58–59.

38Id. at 124–26.

39Id. at 356–59.

40Id. at 3, 495.

41See Tribe, supra note 18, at 789 & n.28 (criticizing the adoption of “an artificial dichotomy between (protected) speech-related conduct in which ‘expression’ predominates and (unprotected) conduct in which ‘action’ is dominant”); Ely, supra note 18, at 1494–96, 1495 n.53 (“I simply do not think this distinction can be made to work.”); Louis Henkin, Foreword: On Drawing Lines, 82 Harv. L. Rev. 63, 77–80 (1968) (criticizing the distinction between speech and conduct). For slightly more recent expressions of the same concern, see Barnes v. Glen Theatre, Inc., 501 U.S. 560, 576 (1991) (Scalia, J., concurring in judgment) (questioning the distinction between speech and conduct by observing that all laws restrict conduct); Rubenfeld, supra note 18, at 783–84 (2001) (noting that speech is conduct and that many non-speech actions have a communicative function).

42Because Emerson defined the coverage of the First Amendment in a way that made the degree of protection absolute for all covered speech, he forced issues into the decision about coverage—issues that for others would sometimes be about coverage and sometimes about the degree of protection to be given to certain forms of covered speech. Emerson, supra note 29. On the importance of keeping the questions separate, see Kent Greenawalt, Speech, Crime, and the Uses of Language 5 (1989) (describing as “absurd” the attempt to conflate coverage and protection).

43See Robert C. Post, Racist Speech, Democracy, and the First Amendment, 32 Wm. & Mary L. Rev. 267, 285 (1991) (noting that the distinction between speech and action is what we say to mark a distinction drawn on other grounds). Indeed, the elusive nature of the distinction between speech and conduct facilitates labeling (or perceiving) an act as one or the other depending on the outcome preferences of the labeler. See Dan M. Kahan et al., “They Saw a Protest”: Cognitive Illiberalism and the Speech-Conduct Distinction, 64 Stan. L. Rev. 851, 854, 859, 861–62, 896, 900 (2012) (demonstrating experimentally that the phenomenon of motivated reasoning influences subjects’ determination whether an act is speech or conduct).

44See Schauer, supra note 3, at 3–14; Schauer, supra note 21.

45Emerson, supra note 28, at 878–79.

46Emerson, supra note 29, at 9–11; Emerson, supra note 28, at 887–93.

47Thus, Emerson acknowledged that the distinction between expression and action may not always be “clear,” may at times be “obscure,” and may seem at times “artificial.” Emerson, supra note 29, at 18.

48Id.

49Emerson, supra note 29, at 9.

50The philosophers’ traditional distinction between self-regarding and other-regarding acts is not congruent with the distinction between harmless and harmful acts, because an act can be other-regarding and beneficial, and an act can be self-regarding and harmful to the actor. See Jovan Babić, Self-Regarding/Other-Regarding Acts: Some Remarks, 5 Prolegomena 193, 198–200 (2006); C.L. Ten, Mill on Self-Regarding Actions, 43 Phil. 29, 31–32, 34, 37 (1968). Still, other-regardingness is a necessary condition for harm to someone other than the actor.

51Emerson, supra note 29, at 9; Emerson, supra note 28, at 889–90.

52To the same effect, see Michael D. Bayles, Mid-Level Principles and Justification, in Justification: Nomos XXVIII 49, 54 (J. Roland Pennock & John W. Chapman eds., 1986) (arguing that the exercise of freedom of speech “is less likely to interfere with the exercise of other liberties than is, say, liberty of action”). Similarly, Martin Redish maintains that “it is almost certainly true in the overwhelming majority of cases that speech is less immediately dangerous than conduct.” Martin H. Redish, Freedom of Expression: A Critical Analysis 19 n.48 (1984). Redish also maintains that “the simple reality is that, for the most part, expression is less acutely harmful [than conduct] either to other individuals or to society.” Martin H. Redish, Fear, Loathing, and the First Amendment: Optimistic Skepticism and the Theory of Free Expression, 76 Ohio St. L.J. 691 (700) (2015); see also Franklyn S. Haiman, “Speech Actsand the First Amendment 85 (1993) (understanding speech as a largely harmless activity). And, perhaps most prominently, Ronald Dworkin, especially in Dworkin, supra note 24, at 200–03 (using freedom of speech as an exemplar of a broadly conceived personal liberty based on a right to equal concern and respect).

53See supra note 49 and accompanying text.

54Compare C. Edwin Baker, Human Liberty and Freedom of Speech (1989), with C. Edwin Baker, Scope of the First Amendment Freedom of Speech, 25 UCLA L. Rev. 964, 996, 1019 (1978), which both go pretty far in exactly this direction. For an iconic critique of this move, see Robert H. Bork, Neutral Principles and Some First Amendment Problems, 47 Ind. L.J. 1, 25 (1971) (observing that theories of this variety “do not distinguish speech from any other human activity”).

55See Emerson, supra note 29, at 500 (describing the “shock effect” of some of the communications he labeled as “action”).

56 See Larry Alexander, Is There a Right of Freedom of Expression? 7–8 (2005) (equating freedom of speech with freedom of communication).

57Id. at 8.

58See supra note 50 and accompanying text.

59This is most obvious with respect to advocacy, but also is applicable to praising and blaming an action, because there is still a non-necessary connection between the commendation or condemnation on the part of a communicator and the recipient of the communication adopting some attitude towards the action that is commended or condemned.

60Brandenburg v. Ohio, 395 U.S. 444, 446–47 (1969). Mr. Brandenburg did not use the word “African Americans.” Id.

61See Scanlon, A Theory of Freedom of Expression, supra note 3, at 206–07, 216–17; David A. Strauss, Persuasion, Autonomy, and Freedom of Expression, 91 Colum. L. Rev. 334, 353–60 (1991) (defending the “persuasion principle” by reference to the importance of the autonomy of the hearer). But see Scanlon, Comment on Baker’s Autonomy, supra note 3, at 319–20 (questioning the value of focusing on autonomy).

62Scanlon, Freedom of Expression and Categories of Expression, supra note 3, at 531; Strauss, supra note 61; see also Gerald Dworkin, Paternalism: Some Second Thoughts, in Paternalism 105, 106–07 (Rolf Sartorius ed., 1983) (arguing that autonomy is denied when there is manipulation of someone’s “information set”); Richard Moon, The Constitutional Protection of Freedom of Expression 21 (2000) (arguing that full autonomy can only “emerge in the social realm”).

63Wesley Newcomb Hohfeld, Some Fundamental Legal Conceptions as Applied in Judicial Reasoning, 23 Yale L.J. 16, 32 (1913).

64On the use of the word “liberty” as a synonym for what Hohfeld called a “privilege,” see Leif Wenar, Rights, Stan. Encyclopedia Phil., (last updated July 2, 2011).

65See Roger A. Shiner, Freedom of Commercial Expression 167–68, 234–38 (2003) (analyzing differences between speaker-focused and hearer-focused autonomy).

66See Thomas More’s Trial by Jury: A Procedural and Legal Review with a Collection of Documents (Henry A. Kelly, Louis W. Karlin & Gerard B. Wegener eds., 2011) (1980); Andrew P. Roach & Maja Angelovska-Panova, Punishment of Heretics: Comparisons and Contrasts Between Western and Eastern Christianity in the Middle Ages, 47 J. Hist. 145, 154 (2012).

67See The Trial of Galileo: 1612–1633, at 1–2 (Thomas F. Mayer ed., 2012).

68See Alec Craig, Suppressed Books: A History of the Conception of Literary Obscenity 40–53 (1963); Noel Perrin, Dr. Bowdler’s Legacy: A History of Expurgated Books in England and America 163–70 (1969); C.H. Rolph, Books in the Dock 40–61 (1969).

69Moreover, the existence of dangerous mistakes in branding harmless (or beneficial) speech as harmful is only one dimension of a more complex decision-theoretic calculus. Although there are negative consequences to designating harmless or beneficial speech as harmful, so too will there be negative consequences in designating harmful speech as harmless. The full analysis would therefore have to weigh the expected (in the expected value sense of the product of the likelihood of the error multiplied by its consequences) costs of erroneously designating harmless speech as harmful against the expected costs of erroneously designating harmful speech as harmless.

70418 U.S. 323 (1974).

71Id. at 339.

72To be clear, claims about inherent racial superiority or the evil of homosexuality, for example, are erroneous moral claims, but still moral claims.

73Cf. Meir Dan-Cohen, Harmful Thoughts: Essays on Law, Self, and Morality (2002).

74559 U.S. 460 (2010) (invalidating on First Amendment grounds a federal law prohibiting the sale and distribution of images of animal cruelty).

75Thus, Rule 401 of the Federal Rules of Evidence makes a piece of evidence relevant if it “has any tendency to make a fact more or less probable than it would be without the evidence and the fact is of consequence in determining the action.” Fed. R. Evid. 401.

76I say “most” only in order to keep open the logical but empirically unlikely possibility of an instinctive or reactive act of animal torture.

77Note that “statistically likely to increase the probability” is not the same as the behavior being likely to occur. Sending text messages while operating an automobile is statistically likely to increase the probability of accidents even though most driver texting does not produce accidents.

78This is not an essay in neuroscience, or, more specifically, in what fMRI scans and other techniques of modern neuroscience now or in the foreseeable future will enable people to find out about what other people are thinking. On the actual or potential application of these techniques to various legal questions, see Michael S. Pardo & Dennis Patterson, Minds, Brains, and Law: The Conceptual Foundations of Law and Neuroscience (2013). Nevertheless, some of these possibilities, however remote, and however large the moral issues they raise, make inquiring into the topic of freedom of thought (as opposed to the external manifestations of that thought) more important now than would have been the case a generation ago.

79See Dan Bein, Preparatory Offences, 27 Isr. L. Rev. 185 (1993); Kimberly Kessler Ferzan, Inchoate Crimes at the Prevention/Punishment Divide, 48 San Diego L. Rev. 1273, 1282–85 (2011).

80An example of the common criminalization of possession of burglar tools is N.Y. Penal Law §140.35 (McKinney 2010).

81On the crime of possession of hand grenades, see United States v. Freed, 401 U.S. 601, 609 (1971).

82See Va. Code Ann. § 18.2-94 (West 2012). Of course the Second Amendment might be relevant here after District of Columbia v. Heller, 554 U.S. 570, 595 (2008), and McDonald v. City of Chicago, 561 U.S. 742, 791 (2010), but exploring that question, including the interesting parallels here between the First and the Second Amendments, would take us too far afield.

83On pit bull prohibitions, see Am. Dog Owners Ass’n v. City of Yakima, 777 P.2d 1046 (Wash. 1989) (en banc).

84See generally Richard A. Smith, Note, A Comparative Analysis of Dramshop Liability and a Proposal for Uniform Legislation, 25 J. Corp. L. 553, 558–63 (2000).

85Vicarious liability for those who sell the guns subsequently used for unlawful purposes is now largely precluded by the Protection of Lawful Commerce in Arms Act, Pub. L. No. 109-92, 119 Stat. 2095 (2005) (codified at 15 U.S.C. §§ 7901–7903), but the issues are still alive. See Neal S. Shechter, Note, After Newtown: Reconsidering Kelley v. R.G. Industries and the Radical Idea of Product-Category Liability for Manufacturers of Unreasonably Dangerous Firearms, 102 Geo. L.J. 551, 577–78 (2014).

86See generally Frank J. Vandall, Reallocating the Costs of Smoking: The Application of Absolute Liability to Cigarette Manufacturers, 52 Ohio St. L.J. 405, 406 (1991).

87That the employer or even the aider and abettor of a contract killer is liable for murder is plain. United States v. Hardwick, 523 F.3d 94, 97 (2d Cir. 2008); United States v. Dorman, 108 Fed. App’x 228, 231 (6th Cir. 2004). Whether under some circumstances those who provide instructions for contract killing can be liable in tort, or whether instead such instructions are shielded by the First Amendment, remains a contested issue. See Rice v. Paladin Enters., 128 F.3d 233, 242–43, 267, 355–56 (4th Cir. 1997) (rejecting a First Amendment defense for aiders and abettors); Wilson v. Paladin Enters., Inc., 186 F. Supp. 2d 1140, 1144 (D. Or. 2001) (same).

88See, e.g., Osorio v. One World Techs., Inc., 659 F.3d 81 (1st Cir. 2011) (upholding judgment against manufacturer of a table saw although proper use of the blade guard would have prevented the injury); LeBouef v. Goodyear Tire & Rubber Co., 451 F. Supp. 253, 257 (W.D. La. 1978), aff’d, 623 F.2d 985, 988–89 (5th Cir. 1980) (upholding liability for tire manufacturer when tire user’s intentional disregard of normal safety precautions would have prevented the tire failure that caused the accident).

89See supra notes 3, 61–62.

90I again bracket the question of the physical inevitability of freedom of thought, in part because the physical inevitability captured by the concentration camp slogan, “Die Gedanken sind frei,” may in our modern neuroscientific world no longer be so inevitable. See supra note 78 and accompanying text.

91See supra text accompanying notes 61–65.

92See Seana Valentine Shiffrin, Methodology in Free Speech Theory, 97 Va. L. Rev. 549, 554 (2011) (emphasizing the importance of “control over one’s thoughts”).

93John Stuart Mill, On Liberty (David Spitz ed., W. W. Norton & Co., Inc. 1975) (1859). On Mill’s fallibilism, see Stephen Holmes, John Stuart Mill: Fallibilism, Expertise, and the Politics-Science Analogy, in Knowledge and Politics 125, 126–27 (Marcelo Dascal & Ora Gruengard eds., 1989); John Skorupski, John Stuart Mill 8–12 (1989).

94See Scanlon, Comment on Baker’s Autonomy and Free Speech, supra note 3, at 322 n.11 (noting with respect to autonomy that the “same is true of paternalistic legislation restricting behavior other than expression”).

95See supra notes 73–77 and accompanying text.

96On the connection between freedom of thought and autonomy, see especially Shiffrin, supra note 92, at 554; Seana Valentine Shiffrin, Reply to Critics, 27 Const. Comment. 417, 418–26 (2011); and Seana Valentine Shiffrin, A Thinker-Based Approach to Freedom of Speech, 27 Const. Comment. 283 (2011).

97The accuracy of the standard story, according to which thirty-eight people heard her screams and did nothing, is the source of some controversy. See Kevin Cook, Kitty Genovese: The Murder, the Bystander, the Crime that Changed America (2014).

98Zechariah Chafee, Jr., Freedom of Speech in Wartime, 32 Harv. L. Rev. 932, 957 (1919) (“Your right to swing your arms ends just where the other man’s nose begins.”). Chafee may have gotten the line from Holmes, or it may even have been around for longer.

99See Larry Alexander, Can Law Survive the Asymmetry of Authority?, 19 Quinnipiac L. Rev. 463, 468 (2000); Larry Alexander, “With Me, It’s All er Nuthin’”: Formalism in Law and Morality, 66 U. Chi. L. Rev. 530, 553 (1999).

100And if we reject anti-paternalism as a distinct principle, the same argument applies to harms to the agent herself. For a forceful and recent rejection of anti-paternalism, see Sarah Conly, Against Autonomy: Justifying Coercive Paternalism (2013).

101In recent years freedom of thought has been influentially urged as a basis for a principle of free speech. See Charles Fried, The New First Amendment Jurisprudence: A Threat to Liberty, 59 U. Chi. L. Rev. 225, 233 (1992); Shiffrin, A Thinker-Based Approach to Freedom of Speech, supra note 96; Shiffrin, supra note 92, at 554; Shiffrin, Reply to Critics, supra note 96, at 418–26. But as Shiffrin makes clear, these efforts have started with the premise that free speech exists as an independent principle, but one which remains in need of justification. See Shiffrin, supra note 92, at 550 (describing her project as one of developing an “ideal theory of freedom of speech”). That is an important enterprise, but justifying a principle of free speech that is stipulated (in an argument, or in an item of positive law such as the First Amendment) is different from the project, which is mine here, of not assuming such a principle at the outset, but, rather, of exploring whether what others take as their premise can itself be justified. See id.

102Chaplinsky v. New Hampshire, 315 U.S. 568, 572 (1942) (describing fighting words as words “which by their very utterance inflict injury”).

103Spitting is interesting in this context because whatever harm is produced by being intentionally spat upon, just like the harm of being slapped in the face, has little to do with pain, and even less to do with a wound or physical injury. And thus it seems incumbent on defenders of a speech–action distinction to explain why the plainly communicative and non-physically-harmful act of spitting on another ought to be excluded from the realm of free-speech-relevant behavior.

104See Frederick Schauer, Causation Theory and the Causes of Sexual Violence, 1987 Am. B. Found. Res. J. 737, 737–38.

105See Rolph, supra note 68, at 53. This worry appears to be at the heart of the case most important in establishing nineteenth century obscenity doctrine, Regina v. Hicklin, [1868] 3 LRQB 360, at 361, 363 (Eng.), although the court in that case puts the matter in somewhat less transparent terminology.

106See supra Part III.

107See text accompanying notes 59–65.

108See supra note 83–88 and accompanying text.

109See supra note 50.

110See, e.g., supra notes 102–03 and accompanying text.

111See supra note 50.

112For a collection of the most important works in the “search for truth” or “marketplace of ideas” free speech tradition, see Vincent Blasi, Ideas of the First Amendment 318–689 (2d ed. 2012).

113On the empirical difficulties, see Daniel E. Ho & Frederick Schauer, Testing the Marketplace of Ideas, 90 N.Y.U. L. Rev. (forthcoming 2015).

114E.g., Alexander Meiklejohn, Free Speech and Its Relation to Self-Government (1948); Robert Post, Citizens Divided: Campaign Finance Reform and the Constitution 157 (2014); Cass R. Sunstein, Democracy and the Problem of Free Speech (1993); Lillian BeVier, The First Amendment and Political Speech: An Inquiry into the Substance and Limits of Principle, 30 Stan. L. Rev. 299 (1978); Robert H. Bork, Neutral Principles and Some First Amendment Problems, 47 Ind. L.J. 1 (1971); Frank A. Morrow, Speech, Expression, and the Constitution, 85 Ethics 235 (1975).

115See Vincent Blasi, The Checking Value in First Amendment Theory, 1977 Am. B. Found. Res. J. 521, 524.

116And that is possibly why such accounts might better be situated as components of a broader protection of democracy rather than as a self-standing right. See Austl. Cap. Television Pty. Ltd. v Commonwealth (1992) 177 CLR. 106, 244 (Austl.) (locating a freedom of political communication within the Australian Constitution’s protection of democracy); Bork, supra note 54.