Recently, a panel of the U.S. Court of Appeals for the Ninth Circuit heard oral argument in what are bound to be two of this year’s most controversial cases. Both are challenges to a California law that prohibits licensed psychologists from offering therapy that tries to change a minor patient’s sexual orientation. The practice—so-called "reparative therapy"—has come under heavy criticism from psychologists and activists alike, who point not only to the mounting evidence that changing sexual orientation is impossible, but also to the psychological trauma inherent in treating gay people as if they were broken and in need of "repair."

As if the facts of these cases weren’t provocative enough, the challenges to California’s law are complicated by the fact that the plaintiffs, practicing psychologists who are licensed in California, simply engage in "talk therapy" with their patients. As a result, these cases raise an important, unresolved constitutional question: Is the kind of one-on-one counseling and advice at issue in talk therapy "free speech" protected by the First Amendment?

Based on the U.S. Supreme Court’s case law, the answer to this question should be obvious. Whatever one’s view of the merits or evils of "reparative" talk therapy, it consists entirely of spoken communication. That is enough to bring it presumptively within the scope of the First Amendment’s protection.

Unfortunately, there is currently deep confusion among federal courts about when government regulation of an occupation, such as talk therapy, crosses the line into impermissible regulation of speech. And the confusion in this area is being exploited by government officials in areas well beyond the realm of psychology and sexual orientation. In North Carolina, for example, state officials have shut down a website run by blogger Steve Cooksey, who ran a "Dear Abby" style column advocating a high-protein "paleo" diet. The executive director of the state’s Board of Dietetics/Nutrition sent Steve a formal warning, including printouts of his website on which she had highlighted in red pen the things Steve was and was not allowed to say. The state’s justification? Person-to-person advice about what one should eat is not speech—it is "nutritional counseling," and may only be conducted by licensed nutritionists.

The recent arguments in the Ninth Circuit did little to resolve this confusion. While the panel—comprising circuit judges Alex Kozinski, Susan Graber, and Morgan Christen—asked probing questions, they received few clear answers. According to the attorney defending California’s law, the answer turns on a metaphysical distinction between "speech" on the one hand and "therapy" (which is conducted purely through speech) on the other. One of the plaintiffs’ attorneys, by contrast, suggested simply jettisoning the distinction between "speech" and "conduct" altogether.

Neither of these approaches is correct; indeed, they will only further confuse this area of law.

The core question in the California cases is straightforward: Is a restriction on face-to-face advice and counseling a restriction on free speech subject to heightened scrutiny, or is it simply a restriction on conduct requiring much more deferential review? And the answer is surprisingly simple. In 2010′s Holder v. Humanitarian Law Project, the U.S. Supreme Court considered the constitutionality of a federal law that, among other things, prohibited attorneys from giving expert legal advice to designated terrorist groups. The government defended the law by arguing this kind of legal advice was conduct, not speech. But the Supreme Court rejected that argument, and in doing so laid out a test for whether a law triggers heightened First Amendment scrutiny: If a law’s prohibitions are "triggered" by the content of the communication, then heightened scrutiny applies; if they are not, it does not.

Viewing California’s law through the lens of Holder makes the legal analysis straightforward. To the extent the law applies to simple talk therapy, its prohibitions are triggered by speech—literally coming into play based exclusively on what a therapist says—and must be analyzed under First Amendment. This does not mean, of course, that all reparative therapy is speech; a law prohibiting psychologists from using electric shocks or nauseating drugs to "cure" people of homosexuality would be a restriction on conduct, not speech. It does mean, however, that a law that functions exclusively to prohibit particular people from saying particular things is a content-based restriction on speech.

Holder is also helpful because it puts the focus where it should be: on what, precisely, the government is trying to prevent. All communication necessarily involves conduct—writing this article, for example, has required us to hit our keyboards with our fingers. But rather than lose ourselves in metaphysical contemplation of whether typing is "speech" or "conduct," Holder simply asks whether the government is trying to regulate speech because of the information, ideas or advice that it communicates to the listener.

This simple, subtle distinction explains a lot. It explains why (as the Ninth Circuit has previously held) the government cannot prohibit a doctor from advising her patients about the benefits of medical marijuana, but can prohibit a doctor from prescribing marijuana, even though both acts involve the spoken or written word. In the first case, government would be imposing a punishment triggered by speech—literally just by the words "medical marijuana has benefits." In the second, the government is regulating the legal effect of speech—providing someone with a piece of paper that gives them the legal right to purchase a controlled substance.

This doesn’t mean, of course, that people can never be punished for their speech. For example, it is obviously true, as Kozinski noted at the oral argument, that doctors are routinely held liable for giving bad advice in medical malpractice suits. But, as the Supreme Court reminded us just last year when it struck down the Stolen Valor Act in U.S. v. Alvarez, there is a significant difference between imposing liability on someone for specific harms caused to a specific individual (as in malpractice or defamation suits) and laws that impose broad prophylactic restrictions on saying things without regard for context. Nothing about First Amendment protection for one-on-one advice forbids that kind of after-the-fact liability, any more than First Amendment protection for emails or newspapers forbids lawsuits for defamation via emails or newspapers.

Moreover, concluding that the First Amendment applies to one-on-one advice is not the end of the constitutional analysis; it is just the beginning. Even under strict scrutiny, restrictions on speech can be upheld if the government is able to demonstrate that the law is narrowly tailored to serve a compelling government interest. This is a high bar, but it is not insurmountable, as is demonstrated by Holder itself, in which the high court ultimately upheld the federal prohibition on providing legal advice to designated terrorist groups. It may be the case that, after discovery, California will be able to demonstrate that reparative therapy is so invariably harmful to minors that nothing less than a flat prohibition on the practice will suffice. But it is the duty of an engaged judiciary to require that California make this showing.

Tempers will surely continue to run hot over the merits of "reparative" therapy, and rightly so. But it is important not to let hard cases (or unsympathetic plaintiffs) make bad law. The federal courts’ proud tradition of protecting speech—all speech, whether paid or unpaid, face-to-face or over the Internet, popular or reviled—demands respect, and so do the voices of the bloggers, counselors and thousands of other people who speak for a living whose rights also hang in the balance here.

Robert McNamara and Paul Sherman are attorneys at the Institute for Justice, which litigates occupational-speech cases nationwide.