Telling Propaganda from Legitimate Political Persuasion

Abstract

How does propaganda differ from the legitimate persuasive practices that animate a healthy democracy? The question is especially salient as digital technologies facilitate new modes of political persuasion and the public square saturates with information factual and fabricated alike. In answer, we propose a typology based on the rhetorical strategies that propaganda and its legitimate counterpart each employ. We argue that the point of contrast between the phenomena turns on two key features: whether the rhetorical strategy sufficiently engages our deliberative capacities, and whether it runs counter to our epistemic interests. While in practice the boundary between the concepts is not always sharp, the account identifies a set of conceptual tools that help better frame and come to grips with propaganda and legitimate political persuasion in an information-dense and increasingly complex media landscape.

Introduction

Persuasion is a key component of the public debate that's vital to the health of democracies. Propaganda, by contrast, is a persuasive practice usually thought to compromise healthy public discourse. Though historically associated with totalitarian regimes, there is increasing recognition that propaganda is a feature of liberal democracies.¹ The rise of populism in many democratic countries has benefitted from techniques of targeted propaganda, as revealed by the spread of political strategies like “microtargeting” and the Cambridge Analytica scandal (Beckman 2018). Propagandistic interferences in democratic elections and the risks they present to democratic processes have been the subject of considerable attention in recent years, and a global pandemic has led to an information environment ripe for propagandistic exploitation. Yet today's media landscape presents challenges for telling propaganda from legitimate persuasion. Not only is there an abundance of information factual and fabricated alike, digital platforms facilitate new modes of manipulation. In this context, a conceptual distinction between persuasion that is legitimate and not is pressing. The question we aim to address in this essay is: what sets these persuasive practices apart? In answer, we propose a typology for identifying and distinguishing the concepts that is based on the rhetorical strategies each employs.

The paper proceeds as follows. We start with a few clarificatory points and lay out the conceptual tools on which our accounts rely (§1). In §2 we present our account of legitimate persuasion that is based on two rhetorical strategies: rational and non-rational persuasion. Next, we canvass the philosophical literature on propaganda and argue that the definitions on offer are inadequate (§3). In §4 we claim that propaganda necessarily involves at least one of two rhetorical strategies we term rational manipulative and irrational persuasion, in combination with non-rational persuasion, explaining what each entails. In §5 we identify an additional constraint on propaganda. We discuss the explanatory and descriptive adequacy of our accounts in §6 and §7, address potential objections and consider the sometimes-fuzzy boundary between propaganda and legitimate persuasion.

1. Preliminary Remarks

First, the scope of our interest is restricted to discourse that unfolds within the public sphere. While social media platforms blur the private-public boundary, their inclusion under the rubric of public discourse is both warranted and key to our accounts, given that these platforms are widely used as venues for getting news, debating policy, joining movements and campaigning, i.e., for political persuasion legitimate and not.

Given our focus on the political realm, our use of ‘propaganda' excludes public relations and advertising activities aimed at advancing private or corporate interests.² A campaign run by a toilet paper producer urging its customers to “share a square” amid the shortage in the early days of the coronavirus outbreak could thus not be a candidate for propaganda. An ad in which a political figure condemns an opponent's handling of the pandemic, however, could. We are nevertheless aware that the line between corporate, private and political interests is not always sharp and that there is considerable overlap in persuasion techniques used across these areas.

Here is an example of the type of case we have in mind when we use the term ‘propaganda'. In 2020, English-language Chinese state media ran a Facebook campaign touting the government's transparency in sharing data about the pandemic (Molter 2020). The posts push the narrative that China was a leader in pandemic response, enabling the world to battle the virus effectively. Yet not only was initial reporting of the outbreak repressed,³ official criteria for counting cases were revised eight times between January and April 2020,⁴ and the country has repeatedly stymied the WHO's efforts to collect critical data (Crossley 2021).

Here is a second example. Throughout 2021, pro-Kremlin media outlets published articles warning of the Pfizer vaccine's deadly effects, claiming that it might be a biological weapon, drawing a causal link between it and recorded deaths (for which there is no evidence), or referring to it as the “Shot of Death” (EUvsDisinfo 2021). These cases share three core features we think are essential to propaganda. Their content involves selected truths or outright falsity, they seek to influence their audience, and they aim to do so to advance their purveyor's interests. We return to this in §4.

Next we present three concepts on which our accounts rely. The first is the notion of perlocutionary effects. Political persuasion often aims to shape the doxastic commitments of its targets. Impressing a doxastic attitude upon an audience may be its ultimate end, or persuaders may be interested in their audience's beliefs only insofar as they are prerequisite for performing or refraining from some act. Whether used to mould beliefs or behaviour, speakers engage in political persuasion to secure some form of uptake. One way to capture this feature of persuasion is to use the tools of speech act theory. To say “the Pfizer vaccine could be a biological weapon” is to perform a locutionary act, the act of making a meaningful utterance. In performing a locutionary act, one also performs an illocutionary act. This describes the function of the utterance, say, to assert. If hearers take up the belief that the vaccine could be a biological weapon or make decisions on this basis, these are the speech act's perlocutionary effects.⁵

The concept of perlocutionary effects is a fitting way of referring to the effects that political persuasion pursues, and we consider these effects through the lens of practical and epistemic interests. Persuasive contributions to public discourse typically aim at securing perlocutionary effects that further speakers’ practical (e.g., political) interests. Yet whether these contributions are consistent with hearers’ epistemic interests (i.e., their interest in acquiring true beliefs) is a separate matter. This distinction between the types of interests pursued or undermined by political persuasion acts as a key point of contrast between our accounts of legitimate persuasion and propaganda.

A third concept central to our accounts is the notion of legitimacy. By legitimate persuasion we mean persuasion that doesn't infringe on intellectual autonomy. For a practice to be democratically legitimate, there must be broad public acceptance of it. We assume that persuasive practices that don't respect intellectual autonomy are the kinds of practices that most take exception to.

Following others, we take intellectual autonomy to refer not to the ideal of self-reliance (which seems impractical, given how much of our knowledge is outsourced),⁶ but to a kind of non-interference in one's capacity for intellectual self-direction. Contra the strict requirements of classical thinkers such as Kant, intellectual autonomy is perceived by many as a virtue of the believer that is compatible with epistemic dependence. Not relying on others would harm our capacity to acquire knowledge, thus implying a non-virtuous epistemic attitude. Intellectual autonomy is compatible with epistemic dependence when dependence does not interfere with self-direction (Carter 2020). Our cognitive processes are, in turn, self-directed when, as Carter notes, “A subject's intellectual agency is not disconnected from the way she acquires and maintains beliefs”.

Intellectual self-direction is a function of one's intellectual agency being connected to how one acquires and maintains beliefs. Agential disconnect occurs when, for example, a subject believes she is directing her cognitive affairs, yet she is forming beliefs in ways that don't engage (or insufficiently engage) her cognitive capacities. Her capacity for self-direction is undermined because she is disconnected from the way in which she has acquired the beliefs in question. When agential disconnect occurs, intellectual autonomy is not retained. This form of intellectual autonomy is also a way of maintaining self-trust in our cognitive capacities (Jones 2012).

A persuasive practice is thus legitimate insofar as it doesn't impede a subject's ability to intellectually self-direct, and this, in turn, is a matter of whether the rhetorical strategies a practice employs interfere with her capacity to connect with how she acquires and maintains her beliefs.

We propose a typology for distinguishing between legitimate persuasion and propaganda that's based on the rhetorical strategies each employs. The scheme of classification – rational, rational manipulative, irrational and non-rational persuasion – is based on ideal types to provide a basis for comparison. As we define them, these categories describe paradigmatic states of affairs. While the typology is idealised and some among its normative implications may thus be unrealistic, it identifies a set of conceptual tools with which we can better frame and come to terms with non-ideal discourse. We return to this in our concluding remarks.

Finally, throughout the paper we present real-world cases to draw out the features characteristic of and that differentiate political persuasion from propaganda, and to make more tangible the application of these two categories to public discourse. We restrict the scope of these cases to those connected with the coronavirus pandemic. We think this focus is interesting for three reasons. Not only is it topical, but on the heels of the pandemic came an “infodemic”. A portmanteau of “information” and “epidemic”, the term refers to an overload of information, online and offline, some accurate and some counterfeit (WHO 2020). Bad information can cost lives so the stakes in this context are high, highlighting the importance of being well equipped to tell legitimate from illegitimate persuasion.

2. The Rhetorical Strategies of Legitimate Political Persuasion

We define legitimate political persuasion as public discourse that uses either rational or a combination of non-rational and rational persuasive means. These means are aimed at securing perlocutionary effects that advance either speakers’ or receivers’ practical interests, and are never such that they undermine hearers’ epistemic interests.

2.1. Rational persuasion

An important tradition in Western political thought is the notion that rational persuasion, henceforth RP, ought to play a fundamental role in public debate. This is the species of political persuasion that Aristotle refers to as logos, to influence by appeal to facts, evidence and reasons. Its ideal is to persuade on the basis of sufficient grounds, grounds that make accepting a speaker's claim rational. Note that it doesn't require neutrality or exhaustivity. To rationally persuade, one needn't present both sides of an argument, or cite all evidence relevant to one's claims. One must present the facts in a sufficiently fair way, avoiding misleading suggestions or implications. Since Plato's works, RP is considered a non-coercive means of influencing behaviour, of overcoming disagreement and bringing about civic reform.

The TousAntiCovid application was crafted by the French government to rationally persuade the public to adopt its use and share data about their health by providing daily updates on epidemiological trends, and alerting users when in close proximity to others testing positive for Covid-19. As more people use the app, the more useful it becomes. Relying on a sort of reciprocity argument – share data about yourself, in turn be alerted to potential risks to your health – the app provides actionable evidence that appeals to users’ capacity to act rationally in their own interest.

When vaccination campaigns stress that inoculation helps prevent illness and interrupts chains of transmission, this too is RP. Public health officials don't need to cite all available data about vaccine effectiveness, or point out the extremely low risk of developing, for example, a blood clot, so long as their arguments accurately reflect the balance of existing evidence.

RP provides adequate grounds for its claims. It appeals to our deliberative capacities by offering reasons its audience can evaluate. It doesn't impede our ability to intellectually self-direct and so is consistent with intellectual autonomy. This is not to say that it's necessarily consistent with hearers’ practical interests. A politician who uses RP to win votes pursues her own interests. But while RP often pursues its speaker's practical interests, it does so without jeopardising hearers’ epistemic interests (i.e., their interest in acquiring quality information). As we'll discuss in §4, this is a key point of contrast with propaganda, which we claim necessarily aims at advancing speakers’ practical interests at the expense of hearers’ epistemic interests.

2.2. Non-rational and rational persuasion

Non-rational persuasion (NRP) solicits hearers’ adherence not on the basis of reasons (logos) but by appeal to emotions (pathos) or to speakers’ virtues (ethos). The use of music and images, appeals to authority, to cognitive biases or instinctive responses are means of NRP. Both legitimate and illegitimate political persuasion make use of such methods. Many of the well-known propaganda techniques identified by the Institute of Propaganda Analysis⁷ in the 1940s involve non-rational persuasive means.⁸ One such device is the transfer technique in which the persuader projects certain qualities associated with some person or idea to another for the purpose of promoting or undermining it. Another is the use of a testimonial in which the persuader, a celebrity, extols the virtues of some person or idea to lend credibility to it.

For NRP to be consistent with intellectual autonomy, it must be used in tandem with RP. A French ad campaign launched in 2020⁹ showed a family ignoring physical distancing requirements to celebrate the birthday of its elderly matriarch, only for her to be hospitalised days later. Exploiting evocative music and images, the ad creates a narrative that provokes empathy and nudges viewers into awareness of what matters to them.

Emotions can help hearers track reasons. Here, emotional arousal is used to make certain facts more salient to the viewer. The ad relies on non-rational means to persuade viewers to comply with physical distancing requirements but this is not to say that it hinders viewers’ engagement with reasons. Bringing to viewers’ attention the potential repercussions of overlooking the policy functions instead to facilitate engagement with the facts. Legitimate political persuasion (i.e., persuasive means consistent with intellectual autonomy) can thus take the form either of RP, or of a second rhetorical strategy that involves the coupling of non-rational and rational persuasion (henceforth NRRP).

2.3. Paternalistic persuasion

Like purely rational means of persuasion, NRRP doesn't run counter to hearers’ epistemic interests though it may or may not be consistent with hearers’ practical interests. Some attempts to persuade, however, are pursued specifically for the benefit of the hearer. This is paternalistic persuasion (PP). A practice is paternalistic if it interferes with someone's choices or actions to promote their good without their consent. Say public officials deliberately present the rationale for gradually reopening the economy following a nationwide lockdown in terms of potential gains (“a gradual reopening will save lives”) as opposed to losses (“an abrupt reopening will cost lives”). Framing the policy this way is designed to make it more likely that the public will react in the desired way.

This kind of persuasion is paternalistic in that it aims at advancing its targets’ interests, while distrust in their ability to appropriately weigh the evidence underlies the strategic use of the framing effect. Notice that the attempt to persuade is rational in that it points to facts, but that it has a non-rational component: it elicits a cognitive bias that causes one to be more risk averse when a positive frame is presented, so more inclined to accept the policy.

A second example of PP is to pre-emptively expose people to misinformation to build resilience to it. In a recent study, participants were exposed to weakened doses of manipulation techniques in a controlled learning environment that involved a simulated social media platform (Roozenbeek et al. 2020). The study's results suggest that much like the practice of viral inoculation, inoculation theory can be effectively used to confer cognitive resistance to manipulation strategies (as opposed to manipulated information). Such practices aim to improve an audience's epistemic¹⁰ and practical position, and they employ a combination of non-rational and rational means. Certain pre-reflective features of our cognitive processing (i.e., the fact that we can better resist misinformation attempts when we have already been exposed to weakened versions of the strategies it employs) are exploited and combined with factual refutations of the misinformation's content to achieve the desired effect.

Although paternalistic interference is often thought to involve means other than RP (such as coercion), it's consistent with rational means of persuasion.¹¹ The cases discussed also involve an important non-rational component. They appeal not only to facts, but to cognitive biases or other pre-reflective features of how we process information. PP is thus a species of NRRP that has the additional, distinctive feature of being alter-centred, i.e., of pursuing the practical and epistemic interests of its subjects.

On our account, both RP and NRRP are legitimate forms of political persuasion. A subset of the latter rhetorical strategy is PP, which pursues the practical and/or epistemic benefit of its subjects.¹² Non-paternalistic legitimate persuasion likewise aims at advancing speakers’ practical interests, and never does so at the expense of hearers’ epistemic interests.

To the extent that NRRP involves non-rational appeals, one might wonder whether and how this rhetorical strategy differs from a practice like manipulation. After all, manipulation can be construed as an attempt to influence that insufficiently engages one's rational faculties. Do NRRP and PP not do just this by guiding targets’ deliberation to the desired outcome?

It bears repeating that attempts at NRRP and PP are rational in substance. While they frame information so as to leverage cognitive biases or appeal to other non-rational sensibilities, arguments and reasons are provided. PP is by definition alter- as opposed to ego-centred and in this sense is at odds with manipulation, which furthers the interests of the manipulator, usually at the expense of the manipulated.

Still, some have argued that these forms of PP, sometimes called nudges (in the sense of Thaler and Sunstein 2008), are problematic because they infringe on intellectual autonomy (Riley 2017). But as McKenna argues (2020), nudges don't impose coercion on choice, but rather help frame the epistemic environment to facilitate decision making. After all, an intellectually autonomous thinker need not be a solitary thinker, and rhetorical practices that offer relevant information to assist with decision making are legitimate supports to their choice.

Notice that NRRP and PP are not geared toward interfering with their subjects’ ability to engage with the facts. These practices function by putting a spotlight on what the evidence is and prompting subjects to overcome the biases and blind spots that obstruct their engagement with it (McKenna 2020: 16). The use of the framing technique discussed above was designed to make targets more receptive to the evidence, while the inoculation strategy reduces vulnerability to future attempts at (irrational) persuasion by exposing subjects to relevant techniques and facts. Despite their partial reliance on NRP, these strategies don't hamper critical engagement with reasons, they bolster it. So it seems right to think that NRRP is a legitimate means of persuasion in that it doesn't infringe on intellectual autonomy.¹³

3. Defining Propaganda: State of the Art

A vast literature on propaganda was produced in the twentieth century issuing mostly from the social sciences, with comparatively few philosophers weighing in. Unsurprisingly, disagreement about the term abounds. In the philosophical scholarship the concept has been defined as epistemically defective (Cunningham 2002; Ross 2002), as a kind of ad populum argument (Walton 1997), as “the employment of a political ideal against itself” (Stanley 2015) and as an appeal to either ideals, affect or identity (Cassam 2021). Each definition picks up on key features. In what follows, we discuss these accounts in turn and argue that each is wanting in that it's either too vague, too narrow or otherwise ill-suited to present purposes.

Before proceeding, however, we note that there are many more insightful accounts in the propaganda scholarship than could be feasibly discussed here. Ellul (1965), for example, describes propaganda as a socio-psychological phenomenon that aims at bringing about mass conformity. Herman and Chomsky's (1988) model of propaganda explains how corporate news agendas are set by political and financial factors that serve the interests of the wealthy and powerful, and Marlin (1989) formulates a definition that identifies the bypassing of subjects’ reflective judgement as among propaganda's distinctive features, to name but a few. The accounts discussed in what follows have been chosen as a function of their relevance to present aims. First, as we aim to provide a philosophical treatment of propaganda, we restrict the scope of our focus to philosophical contributions to the subject.¹⁴ Second, as we are concerned with political propaganda, we omit accounts that conceive of the account in broader terms, i.e., as a phenomenon that permeates the areas of advertising, public relations, education and the arts.

Cunningham's (2002) and Ross's (2002) accounts turn on the notion of epistemic defectiveness. Roughly, this is the idea that the quality of the content propaganda transmits is flawed. Cunningham claims that although it can't ultimately be defined, propaganda is structured by a litany of epistemic disservices (125). On Ross's account, propaganda is “an epistemically defective message used with the intention to persuade a socially significant group of people on behalf of a political institution, organization or cause” (24).

Defining the phenomenon in terms of its epistemological features is compelling because this taps into the intuition that its contents are flawed. But spelling out “epistemic defectiveness” with sufficient specificity to meaningfully refine the concept of propaganda appears challenging.

Cunningham doesn't explain what epistemic defectiveness entails, but offers examples, including instrumentalizing truth (also unexplained), bypassing reason, and giving primacy to credibility and belief over knowledge and understanding. Ross explains that a message “is epistemically defective if it is either false, inappropriate, or connected to other beliefs in ways that are inapt, misleading or unwarranted” (2002: 23). The first condition is clear enough, but the reader is left in the dark as to what precisely is implied by conditions two and three. This paints a vague picture of epistemic defectiveness. Many kinds of non-propagandistic messages could be described as such suggesting that propaganda's distinctiveness may lie elsewhere. A related point is that by Cunningham and Ross' accounts, propaganda, political lies¹⁵ and misinformation are one and the same, as the latter two concepts conform to the conditions laid out for the first.

For Walton (1997: 390), propaganda is a kind of ad populum argument because it persuades a mass audience to accept a conclusion based on widely accepted premises, typically working by arousing the audience's emotions and enthusiasms. Walton registers an additional ten criteria that characterise propaganda, including the notions that it aims at eliciting certain conduct from its targets, that it's a kind of advocacy dialogue, and that it's eristic, taking a quarrel-like form.

But propaganda need not necessarily follow the formula of an argument that appeals to popularly accepted premises. Consider the following case. The Trump administration aired a video at a press briefing in the early days of the Covid-19 outbreak. The opening slide read “The media minimized the risk from the start” and went on to play selective quotes of news anchors ostensibly downplaying the virus, with the next slide noting “While President Trump took decisive action” against a background of triumphant music. The notion that the media downplayed the gravity of the crisis was at that point not a common view. In fact, Trump himself had made numerous statements about the harmlessness of most cases, stressing that it would disappear with warmer weather and so on. Rather than relying on widely held views to win support for its conclusion – the notion that the Trump administration was competently responding to the crisis – it seems plausible to think that this functioned to popularise a narrative that was at the time only subscribed to by a few.

While we agree both that this kind of speech aims at securing some form of uptake and that it's one-sided, neither point helps distinguish propaganda from legitimate political persuasion. Consider a government's efforts to persuade the public to get a Covid-19 vaccine by providing evidence about its benefits, and debunking misconceptions about its risks. The goal of this kind of persuasion is to influence hearers’ behaviour or beliefs. While it relies on a fair and balanced presentation of the facts, it doesn't provide an overview of the other side of the argument by e.g., noting that it's possible, though unlikely, that some may develop allergic reactions to the vaccine.

A thorough review of each of Walton's criteria would take us too far astray, but many share this feature.¹⁶ We think the above is sufficient to conclude that the definition is, on the one hand, too narrow – for not all propaganda relies on widely accepted ideas to win support for its conclusions – and on the other too broad to meaningfully distinguish propaganda from legitimate persuasive practices.

In How Propaganda Works Stanley (2015) defines propaganda in terms of ideals. Undermining propaganda is a contribution to public discourse presented as embodying a cherished ideal but which tends to erode it, while supporting propaganda is presented as embodying an ideal and contributes to its realisation via emotional or non-rational means (53). An example of the former is the ideal of political equality (‘one man, one vote’) put in play to motivate restrictive US laws on voter identification (69). An example of the latter is the use of a national flag to boost patriotic sentiment (58).

In “Bullshit, Post-Truth and Propaganda”, Cassam (2021) builds on Stanley's conception. While some propaganda appeals to ideals, Cassam claims that others work by manipulating emotions (affective propaganda) or by appealing to an identity narrative (identity propaganda). This seems correct; propaganda can target affect, or sow divisions between social groups by pitting one group (‘us’) against another (‘them’).

But more needs to be said about what makes the manipulation of emotions propagandistic, because this alone seems insufficient for a case to count as one of propaganda. Indeed, legitimate political persuasion can rely on similar tools. Think of political leaders urging youth in the UK not to “kill granny”. The slogan aims at eliciting caution in youth to secure compliance with physical distancing measures and to safeguard the health of older, more vulnerable generations.

Cassam's tripartite distinction identifies three modes of the phenomenon's functioning, but as described, the account is ill-suited to present purposes as it makes no distinction between propagandistic and legitimate appeals to emotion in attempts at public persuasion.

4. The Rhetorical Strategies of Propaganda

We suggest that propaganda is distinctive in that it employs at least one of two rhetorical strategies we term rational manipulative persuasion (RMP) and irrational persuasion (IP), either or both of which are combined with non-rational persuasion (NRP). In the following we detail what these strategies involve and identify an additional constraint on the concept.

4.1. Rational manipulative persuasion

Recall that rational persuasion (RP) presents an adequately balanced set of reasons in support of a proposition. Some persuasive practices likewise appeal to facts and evidence, but in a disingenuous way. They express truths while communicating something false, relying on omissions, cherry-picked data and the like. We term this type of RP rational manipulative persuasion (RMP). It's manipulative because in presenting facts and evidence in a distorted light, it insufficiently engages people's deliberative capacities¹⁷ and subverts autonomous rational processes.¹⁸

Consider the following case. When, championing his administration's response to the pandemic, Trump asserted that despite a recent rise in cases nationwide fatalities were at half the level of an earlier peak, he misleadingly suggested that the rise in cases had not been accompanied by a rise in deaths (Qiu 2020). The selective use of facts diverted from the reality that at that point fatalities were double those of the preceding month. Much propaganda takes this form: it has a veneer of rationality but rather than sustaining the rational process, it subverts it by relaying the facts in an insufficiently balanced light, seeing to it that the target only has access to certain evidence. The reasons provided undermine rather than advance the target's rational reflection.

4.2. Irrational persuasion

Where RP presents a sufficiently balanced set of reasons in favour of some position, irrational persuasion (IP) doesn't appeal to facts or evidence at all. Cogent, fact-based arguments are supplanted by falsities or fallacies such as false dilemmas, scapegoating and ad hominem attacks. Propaganda often relies on this approach. In the context of the pandemic, consider the claim “In the choice between the loss of our way of life as Americans and the loss of life, we have to always choose the latter.”¹⁹ Framing policy responses to the coronavirus in dichotomous terms misleadingly suggests that a choice must be made between lives on the one hand and livelihoods on the other, when in fact protecting public health and reopening economies go hand in hand. Similarly, references to the virus as the “kung flu” or “China virus” are examples of (propagandistic) scapegoating. This tactic appeals to emotions and serves to deflect blame, inducing its targets to redirect their anger toward an out-group.

IP also makes use of outright falsity. When used propagandistically, falsity at times masquerades as truth. For example, early in the global outbreak, Russian state-funded media outlet Sputnik News published a story claiming that the pandemic was orchestrated by global elites to install Western hegemony.²⁰ Other times, falsities seem meant to be recognised as such and play a core role in a broader strategy that Steve Bannon has referred to as “flooding the zone”. Rather than building consensus around a single narrative, the approach aims to undermine the possibility of consensus by saturating media ecosystems with false and conflicting information. The contradictory messages – “there is no evidence of a Covid-19 pandemic”, “the virus is a deadly combination of coronavirus and HIV”, “the virus is actually safe and can be cured with saline solution” – may be designed to induce a post-truth posture in respondents that is valuational (i.e., the truth ultimately has no value), epistemological (i.e., not wanting to know the truth) or metaphysical (i.e., the truth can't be known).²¹ Or, the deluge of conflicting information may simply function to make the task of telling the sources and discourses that are trustworthy and fact-based apart from those that are not more difficult.

What makes persuasion irrational may not just be that its contents are false or logically flawed, but that the perpetrator or receiver of IP accepts the material despite relevant counter-evidence. This implies that an attempt at persuasion may at one point in time be rational, but as pertinent evidence accrues, it becomes irrational. In January 2020 it may have been perfectly rational to maintain that there was no evidence of a global pandemic, but by April wholly irrational.

It's worth noting that should the otherwise rational persuader make a logical misstep and unwittingly use a fallacious argument, she is best thought of as engaging in unreasonable as opposed to irrational persuasion (Blair 2012: 75). Take the persuader who forgets to appropriately qualify her argument (by appropriately restricting or limiting its scope), yet who in all other respects is engaging in a rational discourse. While she makes a move that is unreasonable, the move is not on that basis alone irrational. Indeed, that we are prone to error doesn't imply that we are necessarily irrational. There is thus an important difference between the persistently or wilfully irrational persuader on the one hand, and the rational persuader for whom an occasional logical slip is the exception and not the rule, on the other.

As IP relies on fallacious arguments and untruths rather than on reasons, like RMP, it subverts autonomous rational processes. To deceive with falsities or to mislead with fallacious reasoning is to insufficiently engage people's reflective capacities, which are either inappropriately directed or circumvented entirely. In this sense, IP is a kind of manipulation. This seems right. The above examples are consistent with the kinds of spurious, underhanded tactics that come to mind when we think of manipulative influence aiming to mould behaviour or beliefs. This also sits well with the intuition that propaganda involves a manipulative component.

IP and RMP often work in tandem in propagandistic discourse. Take a pro-Trump presidential campaign ad claiming first that Joe Biden had no plan to deal with the coronavirus, and that Biden condemned Trump's decision to impose restrictions on travel to China, playing a recording of Biden saying “hysterical xenophobia”. The first claim is false.²² The second claim is partially true. At a campaign stop around the time Trump announced the travel restriction, Biden said “This is no time for Donald Trump's record of hysteria, xenophobia – hysterical xenophobia – and fear-mongering to lead the way instead of science” (Valverde 2020). Biden uttered those remarks but whether they ultimately referred to the travel ban is unclear. Propaganda may be most effective when it combines RMP with IP, for by stating something that is at least partially consistent with reality and recognised as true, it may be easier to impart some other belief that is false.²³

4.3. Illegitimate non-rational persuasion

As we argued above, NRP is consistent with intellectual autonomy if and only if it's used in tandem with RP. Propaganda by contrast involves the use of RMP or IP (or both), combined with NRP. Let's revisit the cases outlined above to clarify the role this rhetorical strategy plays in propaganda. Recall Trump's selective use of evidence of nationwide coronavirus fatalities. The case is one of RMP, for truth is used to convey a falsehood. But the case also involves the use of NRP, as Trump aims to transfer the status or weight of empirical data to the view he promotes, i.e., he aims to make credible the notion that his administration's mitigation efforts were working well. The Chinese Communist Party's efforts to present itself as a bastion of transparency in the context of Covid-19 is likewise an example of cherry-picking. This use of RMP too was coupled with NRP: the social media campaign made repeated reference to the WHO's “objective acknowledgements” of China's contribution to maintaining global public health, a classic use of a testimonial, and posts were accompanied by images of Chinese medical experts and supplies sent to particularly hard-hit countries.

Next, recall the case of the congressman who set up a false dilemma between saving lives and livelihoods. By noting that the position he promotes is that of the US government, the congressman makes a non-rational appeal to listeners’ patriotism, and to their trust in their government's authority. Finally, describing Covid-19 as the “China virus” or “kung flu” is not only an example of scapegoating, it's also a classic case of transfer in which the frightening features of the virus are projected on to a group to stigmatise or vilify them.

That NRP is a core component of propaganda seems right, as it's paradigmatically discourse that exploits cognitive biases and involves emotional appeals. When used in tandem with and to amplify the effects of a deceptive rhetorical strategy such as RMP or IP, it's illegitimate because it compromises intellectual autonomy. On the other hand, when coupled with RP, it can be perfectly legitimate and even help sustain rational processes.

5. Perlocutionary Effects and Hearers’ Interests

Propaganda is purveyed to secure perlocutionary effects, to get the addressee to take on some posture or behaviour. It's of course not unique in this way. Speakers mean for their utterances to achieve an effect in hearers, usually one that advances speakers’ own interests. Hearers’ interests are usually taken into account in a communicative exchange as well. Say I tell you “The Covid-19 curfew is in effect between 6 pm and 6 am”. Saying so is typically not only in my interest, it is presumably pertinent to your interests too (or else why address the utterance to you?). In a communicative exchange, a speaker usually (a) intends for her utterance to have an impact of some kind that (b) furthers her interests, and her utterance (c) usually takes her hearer's interests into account as well.

So when a speaker makes an utterance, there is typically a default presumption that her hearer's interests have been factored into the communicative act. Say a public official explains to her audience that compliance with stay-at-home orders will help suppress or slow the spread of the coronavirus. In doing so, she undoubtedly intends to secure certain perlocutionary effects that promote her interests. But her communicative act also takes into account those she addresses. Grasping the probable effects of physical distancing is material to her audience's ability to make informed decisions about compliance with the order, i.e., it's material to their interests. Much discourse is this way. Propaganda is not, because its contents deceive.

A point of contrast with certain other communicative acts is thus that propaganda aims to impact its targets to promote the interests of its purveyor without regard for the hearer's interests. Take the following claim about Covid-19: “if I contracted Covid-19, because of my athletic background, I wouldn't feel anything or at most the symptoms of a gentle flu” (Ricard and Medeiros 2020). The misleading remarks made by Bolsonaro minimise the seriousness of the virus, as well as the risks it can pose to the health of anyone who contracts it. To assert that one's athleticism is a determining factor in the symptoms one experiences is to pay no heed to one's audience's interests, because the assertion is false. Propaganda disregards the interests of those it targets despite often seeming to do the opposite.

Given the account sketched above, we suggest that the perlocutionary effects sought by the purveyor of propaganda aim to advance her practical interests without regard for – indeed, to the potential detriment of – hearers’ epistemic interests. Among our epistemic aims is to acquire true beliefs, thus propaganda runs counter to our epistemic interests to the extent that it's characterised by deceptive content. The point of contrast with legitimate persuasion is that while it too might aim at advancing speakers’ political interests (often, but not always), it doesn't do so at the expense of hearers’ epistemic interests. We return to this point in §5.

Notice that this is not to make a claim about how propaganda ultimately impacts hearers’ interests. That it chiefly aims at promoting a speaker's political interests while running counter to hearers’ epistemic interests isn't to say that it runs counter to hearers’ practical interests too. The practical interests of speaker and hearer could be aligned,²⁴ or it may not bear on hearers’ interests at all.²⁵ Nor does it imply that propaganda invariably has the effect of undermining hearers’ epistemic interests for it may be the case that hearers are ultimately not duped.

It's also worth noting that the picture sketched of propaganda so far is consistent with the notion that it can be sincerely believed by its practitioners. That one's political goals are furthered by misleading claims doesn't preclude the possibility that one might genuinely believe them.

The notion of bad faith may be helpful here. In everyday use, the notion refers to a kind of intentional deceit: feigning one set of feelings or attitudes while acting as if influenced by others (e.g., the employer who promises a promotion with no intention of delivering). If Bolsonaro was aware of the misleading nature of his assertions on Covid-19, he feigned belief in a set of propositions he knew to be false, i.e., he made the assertions in bad faith. Bad faith has also been described as a sort of self-deception to avoid one's own freedom and responsibility. Say Bolsonaro knows his assertions are false, but engages in a kind of self-deceit as to this fact. His utterances are made in bad faith, but consistent with sincerity to the extent that he partially believes what he says to be true. We think both types of bad faith are characteristic of propaganda. Those who purvey this discourse do so in bad faith with the intention to deceive either their audience or themselves.

6. Discussion

Political persuasion is legitimate only insofar as it respects intellectual autonomy. The distortions, omissions and deception characteristic of RMP and IP undermine deliberation. Propaganda is an illegitimate persuasive practice because subverting people's autonomous rational processes is synonymous with infringing on their intellectual autonomy. While we set legitimate political persuasion in opposition to propaganda, we don't mean to suggest that all political persuasion that's not legitimate is necessarily propaganda. A practice such as coercion, where one is persuaded by use of force or threats, is both distinct from propaganda and illegitimate. The category of illegitimate political persuasion thus extends beyond that of propaganda, but an exhaustive survey of non-propagandistic illegitimate persuasion is beyond the scope of present purposes.

In the paper we refer to a series of cases to draw out propaganda's features. A concern might be that some among these seem too mild to count as propaganda. It may be that when one thinks of propaganda, the sorts of cases that come to mind are more striking, like the propaganda of totalitarian regimes. Recall, however, that we start from the premise that propaganda is a feature of liberal democracies today. This view is fairly uncontroversial. While the propaganda of, say, Nazi Germany, may be more conspicuous than cases we cite here, we believe they share a set of core features and indeed the definition we propose is meant to account for all propaganda, irrespective of the political structure in which it's disseminated.

There is one caveat. We have outlined a view of propaganda on which it involves either the use of truth to express falsehood in the case of RMP, or the use of outright falsehood in the case of IP. We characterise these as rhetorical strategies, so it should be clear that our definition is suited to making sense of propaganda expressed in language. But certain paradigmatic cases of propaganda are expressed primarily in image form. Posters from the second world war that depict, for instance, a grotesque caricature of Hitler menacingly bearing a SS-emblazoned dagger and warning the reader “Don't kid yourself … it's up to you to stop him!”, come to mind.

The rhetorical strategies of RMP and IP are inapt in cases where propaganda involves a visual component that does at least as much work as its linguistic component, as in the case described above.²⁶ The poster doesn't appeal to rational capacities, it functions by eliciting strong emotion in the viewer. Thus when imagery plays a central role in propaganda, it's perhaps best explained in terms of NRP. But as our primary concern is with propaganda expressed in discourse, we've set aside the question of propaganda as art or image, and the definition we propose doesn't account for it.

We characterise propaganda as illegitimate public discourse as its methods are inconsistent with intellectual autonomy. Our conception is thus not normatively neutral. It is, however, compatible with the notion that propaganda's illegitimate means could be used for worthy ends. Propaganda has acquired a negative connotation in that it's usually associated with objectionable ends, but this need not be so. Anti-Nazi propaganda, for example, sought to further the Allies’ war effort and secure their ultimate victory. We venture that cases in which propaganda serves a laudable goal are nevertheless likely to be among a minority, as only under singular circumstances is deceptive (as opposed to rational) persuasion best tailored to achieving a worthwhile goal.

Readers may have noticed that among the cases we cite is at least one which appears to be a clear case of a conspiracy theory. Published by Sputnik News, this is the claim that Covid-19 was engineered by global elites to install Western hegemony. Cassam (2019) defends a view of conspiracy theories on which they are theories about conspiracies that are speculative and contrarian, and which serve the basic function of advancing a political or ideological objective. In other words, by Cassam's view, “conspiracy theories are first and foremost forms of political propaganda” (2019: 7). We find this view plausible. While some conspiracy theories are not political in nature (e.g., those about the death of Elvis), those that are meet each criterion we have listed here, with the additional feature of being theories about a secret plot. The claim that the coronavirus was faked by ruling elites to secure Western hegemony is public discourse that is deceptive (a form of irrational persuasion, given the substantial amounts of relevant counter-evidence) and that seeks to secure certain perlocutionary effects without regard for its targets’ interests. The theory serves to undermine the West and its democratic institutions, whether it's sincerely believed by those who purvey it or not. Conspiracy theories whose content is political are thus a particular species of propaganda whose peculiarity lies in their identification of a set of purported conspirators conducting nefarious undercover political activities.

As we have identified propaganda chiefly in terms of the rhetorical strategies it employs, one might wonder whether all cases of IP or RMP in public discourse are necessarily cases of propaganda. Specifically, one might worry that our definition fails to make a meaningful distinction between propaganda and mere lies. If lies involve (knowingly) asserting falsehoods, they fall under the umbrella of irrational persuasion. But surely there can be a difference between telling a (political) lie and purveying propaganda.

While telling a lie is to irrationally persuade, it's distinct from propaganda when it isn't coupled with non-rational persuasion. Consider the infamous 1998 lie told by Bill Clinton, “I did not have sexual relations with that woman, Ms. Lewinsky.” This is irrational persuasion used to secure certain perlocutionary effects that promote the speaker's interests without regard for hearers’ interests. It clearly doesn't involve non-rational persuasion, a key ingredient in propaganda. Therein lies the difference between these distinct but related concepts.

Much of what is said in the political sphere appeals to audiences’ emotions, so involves a component of non-rational persuasion. So long as this discourse is also rational, it is legitimate. Political lies that appeal to emotions, however, are on our account necessarily propagandistic. Considering the preponderance of emotive political discourse and of political lies, the reader may worry that this implies that on our characterisation, propaganda is very widespread.²⁷ Yet, while political speech is often designed to elicit emotion, it's far from being the case that this is always so. Consider Joe Biden's false remarks about his support for the Iraq war,²⁸ or Donald Trump's bizarre lie about hurricane Dorian being forecast to hit Alabama in 2019.²⁹ Or, trying to backtrack on his promise that if Americans liked their health plans they could keep them, Obama's lie that “what we said was, you can keep [your plan] if it hasn't changed since the [Obamacare] law passed.” Compare these to political speech that similarly involves falsehood, but is combined with an emotional appeal. For example, Obama's 2012 claim that if elected, Mitt Romney would “fire Big Bird” (Gill 2012), a beloved children's character. In trying to portray his opponent as heartless, the claim combined falsehood³⁰ with an appeal to fear or outrage. And while some political discourse looks this way (indeed, a worrying amount), we think it is plausible to maintain that the class of political discourse that involves neither an appeal to affect nor a falsehood is quite broad. Not all political lies are propagandistic; a great many of them are simple lies that don't involve an attempt at non-rational persuasion.

Finally, we saw earlier that paternalism is alter-centred in that it's motivated by a claim that the individuals interfered with will benefit from the interference. Propaganda often purports to be in service of its targets, while it's actually ego-centred persuasion, aiming to advance the interests of those purveying it and, we have argued, undermining the epistemic interests of those it targets. That propaganda may often pose as alter-centred persuasion makes it a good candidate for confusion with PP. Paternalistic persuasive tactics may also be mistaken for propaganda given their appeal to non-rational features of our cognitive processing, which may resemble propagandistic ploys to manipulate the public to those either wary of paternalistic interference, or in the grip of false belief.

7. Conclusion

We have tried to circumscribe the concept of propaganda to clarify the boundary between it and legitimate political persuasion. Telling propaganda from mere persuasion in practice is of course not always straightforward. Some cases are clear-cut, others are not. Whether an utterance can be identified as irrational persuasion, for instance, depends on the evidence one has access to and the extent to which the available evidence supports or refutes the relevant claim. In a 2020 national address, French president Emmanuel Macron announced severe restrictions on movement, repeating five times that the nation was at war with an invisible enemy. For some unacquainted with the growing evidence at the time, the war metaphor seemed propagandistic, inapt, like inappropriate hyperbole designed to legitimise another prolongation of the country's state of emergency. Yet in light of the evidence, it seems reasonable to think that the rhetoric was meant to persuade the public to take seriously the facts: that the virus was potentially lethal, spreading quickly, and that solidarity in respecting the restrictions would be key to slowing its spread.

The boundary between a sufficiently and insufficiently fair and balanced presentation of the facts – between RP and RMP – can likewise be fuzzy. During her 2019 speech about climate change at the UN General Assembly Greta Thunberg asserted “People are suffering. People are dying. … And all you can talk about is money and fairytales of eternal economic growth. How dare you!” One might argue that to assert that people are suffering and dying misleadingly suggests that there is an unambiguous and direct causal relationship between climate change and human mortality rates, which is difficult to establish.³¹ Whether one reads this kind of case more or less charitably is likely to be coloured by one's political views. But the question is ultimately whether Thunberg's rhetoric advocates an irrational level of alarm about climate change. While her speech aims to elicit shame and fear in its audience, belief in the substance of her message – that there are numerous and pressing reasons to implement policies addressing climate change – is ultimately rational given the evidence.

There are many cases in which it's difficult to tell whether we are facing propaganda or legitimate persuasion. The degree of fuzziness is not only in the message, but in the interpretive strategies used by receivers. When we face a message that is potentially propagandistic, we may either accept it or try to resist it by attributing second-order intentions to the purveyors of the message. If we are not ideologically aligned with the source of the message, we tend to find a way of interpreting the message that makes it look propagandistic. Whereas if we are aligned with the values and political orientation of the source, we tend to be more charitable in our interpretation. An example that Stanley discusses in his book (2015: 42) is debatable in this respect. He says that the phrase “There are Muslims among us” pronounced by a non-Muslim American politician is propagandistic even if it is true, because it raises the presence of Muslims to the attention of her audience to “sow fear about Muslims”. But out of this context, it is difficult to attribute a propagandistic intention to the speaker, because the sentence is true and doesn't use illegitimate means of persuasion. The degree of second-order intentions beyond the communicative and the perlocutionary intentions of the speakers we may attribute depends on our own views on the subject matter, and on the context of the utterance. In general, a way of determining the propagandistic character of a contribution is to check whether it is propagandistic at the level of its communicative intention and its perlocutionary intention. Other levels’ intentions depend on our own views and political orientations. The burden is on hearers to not let their political views and motivations cloud their interpretations of what they hear, that is, not to over- or under-read into what is said.

As we noted at the outset, some of the normative implications of our account may be unrealistic given the somewhat idealised nature of the categories of rhetorical strategy we define. The picture we paint of legitimate persuasion, for instance, may be such that it resembles only a portion of what we would intuitively deem legitimate persuasion in practice. As we note above, there might be more ambiguity and middle ground between the two concepts than our typology suggests. We think it's nevertheless useful: the typology identifies tools sharp enough to be, on the whole, explanatory and predictive, and to make a meaningful distinction between legitimate persuasion and propaganda.

A related worry is that the categories as we've defined them may not be entirely mutually exclusive. It's possible that there are special cases in which RMP combined with NRP is consistent with intellectual autonomy, or in which PP is inconsistent with intellectual autonomy. It seems that any attempt to define as amorphous a concept in terms of the conditions which instantiate it has the potential to run into problems of this kind. The contested nature of propaganda spells trouble for an account that aims to identify a set of necessary and sufficient conditions: accounting for the plethora of ways in which a speech act can be propagandistic while avoiding over- and under-inclusivity is tricky. So it may be helpful to conclude by considering our account from a different angle, in terms of what it tells us about the effects that propaganda functions to serve.

We have argued that propaganda's core distinguishing feature is its speciousness. It communicates falsehood, manipulating the cognitive processes of its audience for purveyors’ interests. On our account, a functional description of the concept would go as follows. Propaganda works to produce unjustified doxastic attitudes in its audience – unjustified in that they fall short of the epistemic norm of truth – via means that are objectionable because these means tend to subvert deliberative faculties. Propaganda functions by using unacceptable means to impress faulty beliefs on its targets. This is why, in an information-dense society, it's important to be armed with conceptual tools to tell it apart from legitimate persuasion.