Autonomy and Manipulation: Refining the Argument Against Persuasive Advertising

As I explain below, there are two additional conditions: the self-reflection on the process of development must be minimally rational and uninhibited.

This suggestion—that one’s autonomy could be enhanced through external manipulation—is not new. Sneddon (2013) provides an example in which a benevolent neuroscientist helps an agent act in accordance with her commitment to a project of drinking more tea (37). This is precisely the sort of benevolent manipulation that I will defend here.

Although they approach the possibility differently, the possibility of autonomy-enhancing advertising is raised by Anker et al. (2010). They adhere to the Dworkin/Frankfurt model of autonomy, but they claim that “divergent marketing” is at odds with autonomy while “convergent marketing” is not.

Throughout the paper I use the term “persuasive advertising” in keeping with Crisp’s usage. Crisp uses this label to distinguish persuasive advertising from informative advertising. Phillips (1994), following Beauchamp (1984), prefers the term “manipulative advertising” because this leaves room for the possibility of advertising that makes use of rational persuasion, which no one takes to be problematic. Although I use the term “persuasive advertising,” I am typically talking about manipulative advertising, not ads that work by means of rational persuasion.

One would expect a Kantian objection to pertain to autonomy as well. Villarán is indeed concerned with autonomy but only in the strict Kantian sense of being a moral agent who acts out of respect for the moral law. Villarán argues that advertising threatens to make us heteronomous by directing us to obey self-interested desires that are at odds with Kant’s notion of duty. This is certainly about autonomy, but it is distinct enough from the standard manipulation objection that I do not discuss it at length here. Villarán and Kant are principally concerned with moral autonomy, but the views I am discussing deal with personal autonomy. For more on this distinction and an attempt to generate a Kantian view of personal autonomy, see Taylor (2005).

Nevertheless, if Villarán is right about advertising, that would cast doubt on my conclusion about the moral permissibility of manipulative advertising. Although I cannot give a full response here, I believe that my conclusion could be defended from the Kantian critique. First, Villarán presents examples of advertising that encourage obeying desires even when this means violating one’s moral duty (10). I would agree that such advertising is morally problematic, but I do not think that all advertising is guilty of this charge. Advertising might lead me to act on a desire in a way that is consistent with my moral duty. If I act on an inclination that was induced by advertising, then the heteronomous action could be “in conformity with duty” even if it was not done “from the motive of duty.” For instance, an ad might show me pictures of starving children and evoke my compassion and sympathy, prompting me to donate money to famine relief. Such an action conforms with the duty of beneficence even if the advertising-induced inclination was the determining ground of the will. Second, Villarán suggests that a Kantian account of manipulative advertising must begin by “denouncing that it implies treating humanity merely as means” (11), but this is not necessarily true of the kind of advertising that I defend here. In the cases like Ricardo’s, the persuasive ads respect the agents’ capacity to set ends for themselves, so it could be argued that they are not treated as a mere means.

It is possible that Kantians will my find this response to Villarán’s criticism unsatisfying. They might persist in their belief that manipulative advertising (even construed along the lines I have suggested) is at odds with the Kant’s ideas about autonomy and respecting rational agency. If that is the case, then I would defer to Crisp’s point that Kant’s “standards are too high” (414). Kant’s view of moral autonomy ultimately requires us to conceive of an agent that is “entirely external to the causal nexus found in the ordinary empirical world” (Crisp 1987, p. 414) and this is not the common sense notion of personal autonomy at stake in most discussions of advertising. Acting from the motive of duty is a noble moral principle, but requiring advertising to be consistent with this kind of moral autonomy sets an unreasonably high standard. Because of the arguments I defend throughout the paper, I think we ought to lower the bar given that circumventing rational faculties appears to be morally permissible in some cases. As Hyman (2009) points out, if we believe in the idea of responsible ads at all, we must accept that “non-cognitive appeals can be used responsibly” (201). Cf Maciejewski (2004).

The account of autonomy defended by Dworkin 1988 requires procedural independence. The account I favor in this paper—that of Christman (1991)—also requires procedural independence. Some have argued that this does not go far enough and that there must be substantive independence in addition to procedural (see, for example, Stoljar 2000; Benson 2005).

Cunningham’s objection to Sneddon might be something of a straw man argument. First, she is treating autonomy like an all-or-nothing affair, but an agent’s autonomy comes in degrees. Sneddon’s argument is pointing out how the capacity for strong evaluation can make an agent more autonomous, and he argues that the bombardment of consumerist advertising diminishes agents’ capacity for such evaluation. Second, Sneddon does not claim that autonomy requires that agents engage in strong evaluation; autonomy might simply require the capacity for strong evaluation. Cunningham makes it sound as if Sneddon limits autonomy to the very few individuals who actively question all of their most basic commitments, but that is not the case on Sneddon’s view. Third, the comparison with Catholicism overstates Sneddon’s point. Adherence to an ideology does not undermine autonomy; the question is whether or not one’s adherence to the ideology is autonomous in the first place. I am grateful to an anonymous reviewer for raising these concerns.

Christman (1991) worries that Dworkin’s view might have an infinite regress problem. We use second-order desires to validate the autonomy of our first-order desires, but then it seems that we must find a third-order desire to sign off on the second-order desire. And so on. See also Thalberg (1978). Another concern is that the time-slice view might endorse problematic manipulation. It is not enough to require second-order approval of a first-order desire; we must consider the possibility that an external manipulator has created both the first-order desire and the second-order approval of it. See Sneddon (2013, p. 34). These problems are interesting, but it is not my aim to address them in depth here. My interest in adopting Christman’s view of autonomy is not avoiding such problems; rather, my aim is to accommodate differing intuitions about cases like Ricardo’s.

Even Crisp mentions this experiment, although he says that it was ice cream rather than soda and popcorn. This story continues to circulate even though all attempts to replicate the experiment have failed and the initial researcher, James Vicary, has been charged with falsifying the data: “Nobody has ever replicated Vicary’s findings and his study was a hoax” (Karremans et al. 2006). See also Rogers (2001).

I do not intend to license paternalistic manipulation of desires.

As I note in this section, my argument refines Crisp’s, but his position leaves room for such an adjustment. It is not as if he objects to all forms of manipulation. He says that “Often, we are manipulated by others without our knowledge, but for a good reason, and one that we can accept” (Crisp 1987, p. 414). He gives an example of consenting to have our emotions manipulated by a skillful actor.

It is also possible that Ricardo would reject the manipulation. Perhaps he is the sort of agent who finds all external manipulation repugnant. Christman’s account allows for this kind of agent relativity. The key question is whether or not the agent accepts the process through which the desire was formed. Reasonable people may differ on this issue. My aim here is to show that some manipulation through advertising could be accepted by some agents. This qualification is important to avoid overstating the conclusion.

This is according to a 2017 Pew Research Center poll. See, Elisa Shearer and Jeffrey Gottfried, “News Use Across Social Media Platforms 2017”.

This term was coined by Eli Pariser in his book The Filter Bubble: What the Internet Is Hiding from You. Cf., Bozdag (2013).

For a discussion of filter bubbles, new media, and many other components of the “attention economy,” see Castro and Pham (2020). Among other things, Castro and Pham discuss how filter bubbles undermine our agency.

Although I believe that this point addresses Waide’s concern about advertisers desensitizing themselves, there are several other facets of the virtue ethics critique that deserve consideration. Waide worries that advertising sells us the message that consumption of market goods will make us happy, but virtue ethics tells us that this is mistaken. On the Aristotelian view, the good life is one that is achieved through proper cultivation and exercise of virtues, but when associative advertising supplants this pursuit with materialistic desires, it yields “the result that we become worse and, quite likely, less happy persons” (Waide 1987, p. 73). I would agree on all these counts, and I think the virtue ethics critique provides a powerful argument against advertising that simply promotes consumerism. But as Villarán (2017) points out, on Waide’s view “advertising that does not promote materialism is free of blame” (4). So Waide is not condemning all manipulative advertising per se. He is condemning it insofar as it promotes vapid consumerism, and this is consistent with my view. There could be a manipulative advertisement that does not promote consumerism (such as the Oxfam ad that I suggested in note 5), and this would be immune to the virtue ethics objection but would still be subject to the manipulation objection that I rebut in this paper.

Social media sites can manipulate more than users’ media consumption. In 2012, Facebook engaged in an experiment on roughly 700,000 users to see if they could alter their moods by filtering the content they saw. The results confirmed the hypothesis. Altering users’ news feeds had an appreciable effect on their emotional states. See Kramer et al. (2014).

The empirical psychological research on this subject is vast. See, for example, Eyal (2014) for a discussion of how people become “hooked” on these technologies. Software developers have been immensely successful at deploying such psychological research in order to get people to spend more time looking at screens. I am not, by any means, suggesting that this situation is good. I argue elsewhere that it is, in fact, quite harmful. I am, however, suggesting that these manipulative methods could theoretically be harnessed for good. Thanks to Clinton Castro for suggesting this example.

Eyal (2014) discusses dopamine in particular. Its manipulative function is not merely that it brings us pleasure; it actually suppresses our critical faculties: “Research shows that levels of the neurotransmitter dopamine surge when the brain is expecting reward […] which suppresses the areas of the brain associated with judgement and reason while activating the parts associated with wanting and desire” (Eyal 2014, p. 7; emphasis added). So when I return to the exercise app because it has successfully triggered my dopamine response, it is certainly manipulating me through non-rational means. But I can (on reflection) consent to this kind of manipulation given that (1) it is for an end that I want to pursue and (2) it is done in a way that I would endorse.

It would be difficult to provide an exhaustive list. For a review of the literature in psychology, see De-Sola Gutiérrez et al. (2016), Cf. Lee et al. (2014), Rotondi et al. (2017), Rosen et al. (2013) and Thomée et al. (2011).

This desire could be seen as alien on Dworkin’s view given that the first-order desire to smoke is inconsistent with Samantha’s second-order desires. But it would also be alien on Christman’s view if Samantha rejects the process that led to creating the desire to smoke. In the folly of her youth, Samantha did not think that she would become addicted, and regrets ever starting.

I am grateful to Amy Sepinwall for suggesting that I include this example in the paper.

The tobacco companies also argued that the labels violated their right to free speech.