Political Understanding

Abstract

Public opinion research has shown that voters accept many falsehoods about politics. This observation is widely considered troubling for democracy – and especially participatory ideals of democracy. I argue that this influential narrative is nevertheless flawed because it misunderstands the nature of political understanding. Drawing on philosophical examinations of scientific modelling, I demonstrate that accepting falsehoods within one's model of political reality is compatible with – and indeed can positively enhance – one's understanding of that reality. Thus, the observation that voters accept many political falsehoods does not necessarily establish that they lack political understanding. I then address three worries: that voters cannot generally engage in such political modelling; that political modelling obscures facts that are crucial to political understanding; and that successful political modelling would require knowing that one's model contains falsehoods. My responses reveal how, going forward, we should measure political ignorance, and they highlight the standing importance of participatory democracy.

Introduction

When members of the public are polled about politics, usually by means of multiple-choice questions, they often appear ignorant about political matters. In other words, they often accept falsehoods regarding what the current state of politics is, which policies are in place, who is responsible for enacting those policies or indeed how political institutions function (Achen and Bartels 2016, 21–51; Delli Carpini and Keeter 1996; Somin 2016, 17–46).

This observation is perhaps the most consistent result of modern public opinion research. For over fifty years, public opinion researchers have lamented ‘the pervasiveness of ignorance about a wide range of political issues’ (Somin 2016, 17), the ‘paucity of information most people possess about politics’ (Ferejohn 1990, 3) or, similarly, the fact that ‘large numbers of Americans are woefully underinformed’ about politics (Delli Carpini and Keeter 1996, 270). ‘That the public is overwhelmingly ignorant when it comes to politics’, Jeffrey Friedman (1998, 397, emphasis in original) therefore concludes, ‘is a discovery that has been replicated unfailingly by political scientists; indeed, it is one of the strongest findings that have been produced by any social science – possibly the strongest.’

This pessimistic picture of the ordinary voter may seem troubling for democracy. The democratic ideal, as it has standardly been conceived, relies on participation by the electorate to hold policy makers accountable. However, if voters are profoundly ignorant about politics, they may be unable to achieve this task. For some, the solution is to retreat to forms of democracy that require little participation by the people at large (Green 2010; Parvin 2015; Somin 2016). Others, more boldly, take political ignorance as a reason to experiment with alternatives to democracy (Bell 2015; Brennan 2016; Guerrero 2014; López-Guerra 2014).

Yet, for all of its influence, this narrative of political ignorance is by no means uncontested. A different strand of political research, which relies less on public opinion polling and more on ethnographic methods involving extended fieldwork and interviews, has consistently generated a contrasting picture of ordinary voters. Katherine Cramer's (2016) path-breaking fieldwork, in which she documents rural support for political conservatives, is a case in point. Reflecting on the years she spent speaking to predominantly white rural communities in Wisconsin, Cramer (2016, 144) observes the following:

Often when people try to explain why members of the white working class vote for Republicans, they explain it as a product of ignorance or, perhaps, a lack of sophistication. But there is another way to read these conversations. These understandings [the understandings of rural life expressed by her interlocutors], whether or not one agrees with them, have roots and reasons behind them.

She later continues:

One can view as misinformation or ignorance the perception among rural folks that they are the victims of distributive injustice, but the conclusion that people vote the way they do because they are stupid is itself pretty shallow. It overlooks that much of political understanding is not about facts. It is about how we see those facts. Support for less government among lower-income people is often derided as the opinions of people who have been duped. But the stands taken in favor of small government delineated in this book are rooted in views of the world that carry a great deal of meaning for the people that hold them. Listening in on these conversations, it is hard to conclude that the people I studied believe what they do because they have been hoodwinked. (Cramer 2016, 209–10)

Cramer's point is not that her interlocutors do not make false factual claims about politics; she elsewhere acknowledges that they often do (Cramer 2016, 90–3). However, even so, Cramer insists on the understanding and insight involved in the overall political perspective they articulate – what she calls their ‘rural consciousness’.

Cramer's observations are not isolated. Listening in to the political talk that goes on in black churches, barbershops and other predominantly black public spaces, Melissa Victoria Harris-Lacewell (2004, xxiii) famously reports that black Americans employ sophisticated political ideologies that allow them ‘to understand the complexity of the political world’.¹ In particular, these ideologies allow black voters ‘to understand persistent black economic inequality, to identify the significance of race in that inequality, to determine the role of whites in perpetuating that inequality, and to devise strategies for overcoming that inequality’ (Harris-Lacewell 2004, 21). Harris-Lacewell (2004, 2) therefore insists that black Americans are ‘able to make sense of political issues about which [by the lights of public opinion research] they appear to have little information’.

In sum, the following theme appears to emerge from close ethnographic engagement with voters: though members of the public are often uninformed or indeed misinformed about political issues, they nevertheless appear to make sense of the world in a relatively sophisticated way. On this evidence, people are not as politically ignorant as public opinion polls suggest; rather, their perspectives possess some epistemic value, some way of grasping the political world, that current measurements of political ignorance fail to capture.

Outline

I believe that there is something importantly right in this dissenting picture offered by political ethnographers. Yet it is puzzling how this could be. If people regularly accept falsehoods about political matters, how could their overarching perspectives on these matters be epistemically valuable? Perhaps we can learn something important about people by listening to their perspectives. We can discover, say, how the world feels to them. However, it is unclear how a representation of the world that is riddled with falsehoods could teach us something valuable about the political reality they are looking at. It is symptomatic of how puzzling this thought is that when Cramer attempts to articulate it, she veers dangerously close to relativism – suggesting, for instance, that ‘much of political understanding is not about facts’ (Cramer 2016, 209–10).

I wish to suggest that there is a coherent way of resolving this puzzle. It is no accident, I contend, that when trying to articulate the epistemic value realized by ordinary voters' perspectives, political ethnographers such as Cramer or Harris-Lacewell reach intuitively for the language of understanding. Understanding is a valuable – arguably the most valuable – epistemic aim. Yet the most remarkable feature of understanding is this: understanding, for all its epistemic value, can coexist with the acceptance of falsehoods. This is familiar from scientific practice. Scientific models are often riddled with falsehoods, and these falsehoods, far from impeding scientific understanding, play an indispensable role in facilitating that understanding.

I will argue that this phenomenon can carry over to the political context. The acceptance of falsehoods that simplify, idealize or otherwise distort political reality can, in principle, coexist with, and even enhance, voters' understanding of political reality (their ‘political understanding’). Thus, from the observation that voters regularly accept falsehoods about politically relevant matters, it does not necessarily follow that they lack understanding of these matters. In other words, even when voters are ignorant in the sense that they accept false propositions about politics, this does not necessarily entail that they are ignorant in the sense that they lack political understanding.

This argument has important methodological, empirical and normative upshots. The main methodological upshot concerns the measurement of political ignorance. In support of the claim that voters are politically ignorant, public opinion researchers have shown that voters often accept political falsehoods. However, this evidence is inconclusive. Falsehoods can, of course, sometimes undermine political understanding – indeed, this may even often be the case. Nevertheless, my argument shows that, in principle, falsehoods can also facilitate political understanding. I will therefore suggest that to ascertain when the acceptance of political falsehoods is a sign of political ignorance, and when it is not, we must consider political falsehoods more holistically: going forward, our attempts at measuring political ignorance must examine the function that particular falsehoods play within voters' broader system of cognitive commitments.

This first upshot only establishes that accepting political falsehoods can coexist with, and even facilitate, political understanding. However, I will also suggest that in at least some meaningful cases – in particular, cases unearthed by the aforementioned political ethnographers – this possibility is actualized. Thus, the philosophical framework I am offering contributes to reconciling the seemingly contradictory empirical pictures of the ordinary voter with which we started.

The normative upshot derives from these first two conclusions. As outlined earlier, the claim that voters are politically ignorant has motivated a shift away from participatory ideals of democracy, and towards political ideals that assign ordinary voters a more passive role. My argument offers two kinds of resources for resisting this trend. First, it exposes significant methodological and empirical difficulties with the trend's starting premise. Secondly, and more positively, the particular way in which models of political reality (and the falsehoods they contain) generate political understanding sheds light on why we have reason to favour participatory forms of democracy.

The rest of the article proceeds as follows. The second section introduces the idea of political understanding and explains why it is distinctively valuable for democratic citizens. The third section turns to the relationship between understanding and falsehood. Drawing on philosophical examinations of scientific modelling, I argue that accepting falsehoods within one's model of political reality (one's ‘political model’) can positively enhance one's understanding of that reality.

The fourth to sixth sections then address three worries relating to this argument: the fourth section holds that the public cannot generally, or even commonly, engage in such political modelling; the fifth section holds that this practice of political modelling actually obscures facts that are crucial to achieving political understanding; and the sixth section holds that, unlike scientists, members of the public typically do not know that the falsehoods in their models are false. Although these worries are not ultimately decisive, they remain instructive. The first reveals how, going forward, we should measure political ignorance. As for the second and third, they shed light on the standing importance of participatory democracy.

Political Understanding

Understanding

To understand something is, roughly, to grasp how the different aspects or elements of that thing relate to one another. Thus, to understand a subject matter or topic typically involves grasping how a set of propositions associated with that subject matter or topic relate to one another (for example, how they explain, support or otherwise depend on one another).² Grasping these relationships is widely believed to involve certain abilities as well, such as: the ability to infer one proposition from another; the ability to reason about what would happen if aspects of the subject matter changed; the ability to follow explanations relating to the subject matter; and perhaps the ability to offer such an explanation (Elgin 2017; Hills 2016).³

By way of illustration, consider what it means to understand or grasp planetary motion. To understand planetary motion notably involves awareness that: the solar system contains certain planets; these planets are subject to certain forces (for example, the force of gravity); and the force of gravity (among other things) determines or explains the different positions and trajectories of these planets. In addition, such understanding typically comes with certain abilities, such as the ability to predict planetary trajectories given a set of initial conditions and to determine how those trajectories would change if the initial conditions changed.

There are two important things to note about understanding, thus defined. The first concerns the vehicle of understanding: the mental platform that provides understanding. The nature of understanding – the fact that it involves grasping the interrelations between different aspects or elements of something – is mirrored in its vehicle. As Catherine Elgin (2017, 2) has shown, understanding is characteristically produced not by an isolated cognitive commitment (for example, a single belief or piece of knowledge), but rather by an ‘account’: a set of different cognitive commitments that are connected to, and mutually supportive of, one another.

Take, once more, understanding of planetary motion. Such understanding generally results from a large set of interconnected cognitive commitments: commitments regarding which planets exist; views about their physical properties; beliefs regarding the spatial relations that obtain between them; and beliefs about the force that explains these spatial relations. This integrated constellation of commitments, taken as a collective, gives rise to a representation of the planetary system that yields understanding of planetary motion.

The fact that the vehicle of understanding is an account has an important implication for assessing individual cognitive commitments. What matters, fundamentally, is whether the account as a whole produces a representation that helps us understand its subject matter. Hence, when considering whether to accept a particular cognitive commitment that is part of an account, we should not consider that commitment in isolation. Rather, we should consider how it fits into the overall account – and, specifically, whether it contributes positively or negatively to that account (Elgin 2017, 12–13).

The second thing to note is that understanding is valuable; indeed, it is perhaps the most valuable epistemic good there is. For one thing, understanding is the aim of our most successful forms of inquiry, such as scientific inquiry. Science, it is often said, aims to understand the world (De Regt and Gijsbers 2017; Lawler 2021). Moreover, understanding a phenomenon is generally considered preferable to merely having isolated pieces of knowledge, or isolated true beliefs, about that phenomenon – that is, pieces of knowledge or true beliefs that are not integrated within an account (Grimm 2012, 104; Hills 2016, 680).

Why does understanding seem so valuable? Why, specifically, does it seem more valuable than merely having isolated true beliefs or pieces of knowledge? The intuitive reason, Stephen Grimm (2012, 110) observes, is that understanding involves a particularly deep ‘mirroring’ of reality.⁴ Understanding requires not merely appreciating important features of something, but also grasping, via a constellation of mutually supportive cognitive commitments, how these features hang together: ‘A mind that grasps these relationships, or this structure’, Grimm (2012, 110) notes, ‘mirrors the world more profoundly than a mind that merely assents to propositions.’

To illustrate, consider someone who memorizes a long list of geometric axioms and theorems. They may end up having many true beliefs, or even knowing many things, about geometry. Yet they will be missing a crucial aspect of reality: they will not grasp how different axioms and theorems hang together – how they logically justify or derive from each other. That is what understanding, via an integrated set of cognitive commitments, captures.

The fact that understanding mirrors reality in this deep way has two practical benefits. First, it makes understanding fruitful. Having an account of how the different aspects or elements of a thing relate to each other makes it easier to predict what will happen if that thing changes slightly (Hills 2016, 678). If I grasp that planet B's gravitational pull explains planet A's trajectory, I am well positioned to predict how A's trajectory will change if B disappears.

Secondly, it means that understanding is stable. When a cognitive commitment is embedded in a system of interconnected commitments, it is steadier than if it were held in isolation from other commitments (Grimm 2012, 106–7). If I know how to calculate a planet's position from its initial conditions, I am less likely to be misled into accepting false claims about where it is currently located. This is because I can check. These general features of understanding – that it involves grasping how different elements of something hang together; that it typically stems from a constellation of interconnected cognitive commitments; and that it is a highly valuable epistemic achievement, in part, because it is distinctively fruitful and stable – also apply, as we will now see, to the political domain.

Understanding Political Reality

Just as one can understand planetary motion, so too can one understand political reality. To understand political reality – to have ‘political understanding’ – is to grasp, in the way described earlier, how important aspects of political reality hang together. Broadly speaking, political reality includes facts about social and economic circumstances, how those circumstances affect different social and political groups, what policies related to those circumstances exist, and who is responsible for enacting those policies.

Political understanding is greatly valuable to voters. Indeed, the practical values of understanding outlined earlier seem particularly useful in political contexts. Political circumstances change constantly, as new problems, policies, actors and decision-making rules arise. In this context, the ability to reason about new cases – which is characteristic of understanding's fruitfulness – seems crucial.

Furthermore, political discourse is rife with misinformation, which also makes political understanding desirable. Due to understanding's distinctive stability, having political understanding makes one less vulnerable to deception: I am less likely to abandon true political beliefs or pieces of political knowledge if these are embedded within a system of mutually supportive commitments than if I merely hold these in isolation from any such system.⁵

Thus, political understanding seems highly valuable to voters, insofar as they aim to form and maintain reliable views about political matters. Moreover, as with understanding more generally, the value of political understanding seems to surpass the value of merely possessing unintegrated true political beliefs or unintegrated pieces of political knowledge. These provide neither the stability nor the fruitfulness that political understanding provides. Accordingly, in his recent analysis of political competence, Matthias Brinkmann (2018, 169) describes political understanding, not true political belief or political knowledge, as what is fundamentally needed for democratic citizens competently to vote and reason about politics.

The comparative value of political understanding is crucial given my main purpose here. That purpose, recall, is to make sense of the recurrent suggestion that people's perspectives can generate epistemically valuable representations of political reality despite containing many political falsehoods. And, I will do so by arguing that political understanding can coexist, to a significant degree, with the acceptance of political falsehoods. If this is correct, and if political understanding is more epistemically valuable than merely having true political beliefs, then it need not always be troubling that people accept many political falsehoods. Political understanding is the greater epistemic achievement.

The core question that remains, however, is whether this thesis is correct. How could political understanding coexist with, let alone be facilitated by, the acceptance of political falsehoods?

Political Understanding amid Falsehood

Falsehoods in Scientific Understanding

To answer this question, we should first consider how it is possible for scientific understanding to cohere with, and indeed be facilitated by, falsehoods. The first part of the answer appeals to the vehicle of understanding. As discussed earlier, the fact that understanding is provided by an account – a set of interconnected cognitive commitments that collectively constitute a representation of something – has an important upshot. From the standpoint of understanding, whether an individual cognitive commitment is epistemically acceptable does not depend primarily on its independent merits (for example, whether it is true or false, or accurate or inaccurate). Rather, it depends primarily on how the commitment contributes to the broader account it is embedded in.

If, in turn, falsehoods about x can contribute positively to an account of x, then we will have an answer to our question: accepting falsehoods can be consistent with, and sometimes even desirable for, understanding. Why, though, should, we accept this further claim?

The strongest reason comes from scientific modelling. Science is the paradigmatic example of a practice that provides understanding of the world. Yet scientific models are usually rife with falsehoods (Elgin 2017, 23–32). This, moreover, is often no accident: scientists often deliberately incorporate blatant falsehoods – idealizations, simplifications, distortions and so on – within their models. And they incorporate those falsehoods precisely because doing so often helps achieve greater understanding. According to Elgin (2017, 1–2), many falsehoods involved in science ‘are ineliminable and epistemically valuable components of the understanding science supplies’. They are, in other words, ‘felicitous falsehoods’.⁶

To illustrate, consider a few scientific models. The ideal gas law, which models the relationship between the pressure, volume and temperature of a gas, represents gas molecules as infinitely small, as spherical, as never colliding with one another and as exerting no long-range forces on each other. These stipulations are all entirely false. Yet the ideal gas law remains central to thermodynamics (Elgin 2017, 15). Similarly, astronomical models routinely represent planets as point masses when this, of course, is not even approximately true (Elgin 2017, 27–8). Moreover, although it is widely acknowledged that energy is not a fluid, physicists often model energy as a fluid streaming from object to object (De Regt and Gijsbers 2017, 69–72).

So, scientists often deliberately incorporate blatant falsehoods within their models. How is this supposed to facilitate understanding? The answer has to do with exemplification. Scientific models are accounts that represent something (the target) in order to provide understanding of that thing. They provide this understanding by exemplifying important properties of the target – by instantiating important properties of the target in a way that renders them more salient than they would otherwise be (Elgin 2017, 183–203). For example, the ideal gas law (the model) represents certain gases (the target) in a way that exemplifies, and so highlights, the mathematical relationship between their volume, temperature and pressure. By rendering this property salient, the ideal gas law helps us grasp how different features of gases are related. This, in turn, is what understanding consists in.

Incorporating falsehoods within a model can contribute to exemplification in at least two ways. First, falsehoods can remove or filter out features of the target that are not immediately relevant to the property being exemplified. By removing information that may otherwise have distracted or overwhelmed us, screening off extraneous features in this way makes the property at issue more clearly visible (Strevens 2017, 38). For instance, by screening off the dimensions, shape, collisions and long-range forces of gas molecules, the ideal gas law yields a less cluttered representation of gases, in which the relationship between their volume, temperature and pressure stands out more clearly.

Second, falsehoods can distort features of the target on which the property being exemplified does depend so as to make that property more epistemically accessible. In some cases, the features of the target that ground the property at issue are highly complex or unfamiliar. Given our cognitive limitations, we might therefore struggle to perceive the property they ground. Simplifying, exaggerating or otherwise distorting those features can make it more visible. For De Regt and Gijsbers (2017, 70–1), this is how the fluid model of energy operates. The fluid model helps to exemplify the conservation of energy: the fact that the total energy of an isolated system remains constant. This conservation is easier to visualize when we – incorrectly – imagine energy as a fluid because we are all readily familiar with the way a fluid, such as water, can flow from place to place in a sealed circuit without its mass ever changing.

Thus, the presence of falsehoods within an account can facilitate understanding in at least two ways. Yet an important qualification is needed here: this does not mean that truth is irrelevant to understanding. While some falsehoods can facilitate understanding, others can clearly hinder understanding. For a falsehood to facilitate rather than hinder understanding, some truth-related constraints must be satisfied.

First and foremost, for a falsehood to be felicitous, the property it renders salient within the account must truly obtain in the account's target. Put differently, the account, with its constituent falsehoods, must help exemplify something that is really there.

This first truth-related constraint is broadly accepted. Elgin herself, whose theory of understanding is more accommodating of falsehood than most, acknowledges that an account cannot yield understanding unless it exemplifies properties that truly obtain in the target. This is partly why, in her estimation, astrology does not yield scientific understanding: the predictive properties that astrology ascribes to celestial bodies, and that it aims to render visible, are not really there (Elgin 2017, 45, 184, 260–1).

Yet disagreement persists as to whether there should be further truth-related constraints on understanding. Factivists typically suggest that there should be. In their view, it is not enough for an account to exemplify properties that truly obtain in the target; in addition, a certain proportion of the account's cognitive commitments must be true. This position comes in degrees. A strong version of factivism (call it ‘strong factivism’) might say that all of an account's cognitive commitments must be true (Lawler 2021). However, more moderate versions exist. For example, a moderate factivist might instead say that most of an account's cognitive commitments must be true, as well as all of its central cognitive commitments (Kvanvig 2009; Mizrahi 2012; Strevens 2017).⁷

Non-factivists disagree with this assessment. For Elgin (2017, 57–61), even moderate forms of factivism are too restrictive. In her eyes, an account can produce understanding even if most of its commitments are false, and even if this includes some of its central commitments. To illustrate this last point, she notably observes that although the ideal gas model is strictly false, it remains central to thermodynamics.

How far does my argument in this section hinge on this dispute? I have been arguing that falsehoods can play an important role in facilitating scientific understanding. This may seem to fit most naturally with non-factivism, and, indeed, I have drawn extensively on Elgin's theory of felicitous falsehoods.

However, on closer inspection, factivists too can accept the core of my argument. This is clearest with moderate factivism. As presently formulated, moderate factivism is compatible with thinking that a scientific account can provide understanding even if many of its cognitive commitments are false (though not the majority, and not central ones). In fact, moderate factivists also commonly recognize that some false commitments actively improve scientific understanding.⁸

The case of strong factivism is prima facie more challenging. If all of an account's cognitive commitments must be true for that account to yield understanding, that might seem to leave no room for falsehoods to play a role in facilitating scientific understanding. Yet, Insa Lawler (2021) has forcefully argued that this is not the case. Lawler (2021, 6882–3) wholeheartedly agrees with Elgin that scientific models often contain falsehoods and that these falsehoods can facilitate scientific understanding by exemplifying important properties of target phenomena. However, she points out that scientists usually regard these falsehoods instrumentally: they regard them as useful tools or heuristics, which they know to be false.⁹ Thus, Lawler (2021, 6872–3) continues, the presence of falsehoods within models does not imply that scientists' understanding is constituted by any false beliefs (see also Hannon 2020b, 278).

Crucially, then, whatever position one takes on the issue of understanding's factivity, it seems possible to accept my core conclusion in this section: that the presence of falsehoods within a model of reality need not destroy our understanding of that reality. On the contrary, such falsehoods can sometimes contribute positively to our understanding, in the two ways I have outlined.

Falsehoods in Political Understanding

Scientific practice suggests that the presence of falsehoods within a model or account does not necessarily impede understanding – even when these falsehoods are numerous, and even when they deviate significantly from the truth. We should apply this insight to political understanding.

As discussed earlier, public opinion researchers have shown, typically via multiple-choice questioning, that voters often accept falsehoods about politically relevant matters. From this observation, however, we cannot directly infer that voters lack political understanding. As in the scientific case, it is at least conceivable that some of the political falsehoods accepted by voters enhance their broader accounts of political reality by helping to exemplify important properties of that reality.

This first conclusion yields an important methodological upshot. Documenting the political falsehoods that people accept cannot by itself tell us whether those falsehoods are epistemically problematic overall. Hence, there is only so much we can learn from the multiple-choice questions public opinion researchers have traditionally asked voters. To determine with precision to what degree voters may be politically ignorant, we must identify the specific role that particular falsehoods play within voters' broader accounts or models of political reality.

This highlights the importance of ethnographic methods. Ethnographic investigations invite voters to explain their political perspectives in their own words. Accordingly, they help reveal how different cognitive commitments – including false commitments – hang together within those perspectives. Harris-Lacewell (2004, ch. 3), recall, explicitly documents how the political attitudes of speakers in black public spaces come together to form integrated structures, which she calls ‘political ideologies’. Likewise, in her fieldwork in rural Wisconsin, Cramer (2016) examines how her interviewees' beliefs and perceptions collectively constitute an overarching ‘rural consciousness’. Because they bring into view people's broader models of political reality, as well as the role of particular commitments within those models, ethnographic methods are needed to assess when the falsehoods documented by public opinion researchers are signs of ignorance (in the sense of lacking understanding) and when they may instead contribute to understanding.

Still, even if political falsehoods can, in principle, promote political understanding, one might remain sceptical that this abstract possibility is ever actualized. After all, many of the political falsehoods that voters accept are clearly detrimental to political understanding. The false beliefs that the 2020 US election was stolen or that vaccines are part of an elite mind-control conspiracy do not help exemplify important properties of the political environment. They simply distort political reality. If, in turn, none of the falsehoods voters actually accept are felicitous falsehoods, then the methodological recommendation articulated earlier might seem unnecessary.

Yet this is not merely an abstract possibility. Existing research in political ethnography suggests that this possibility is instantiated in at least some meaningful cases. The ‘rural consciousness’ and ‘black political ideologies’ documented by, respectively, Cramer and Harris-Lacewell contain numerous idealizing, simplifying or otherwise distorting falsehoods, and at least some of these falsehoods arguably help exemplify important properties of political reality.

Consider rural consciousness. Rural consciousness involves a set of interconnected beliefs and attitudes that collectively represent rural life. Put differently, it is an account or model whose target is rural life. For Cramer (2016, 12), rural consciousness yields understanding of rural life by exemplifying three of its properties: ‘that rural areas are ignored by decision-makers, including policy-makers’; ‘that rural areas do not get their fair share of resources’; and that ‘rural folks have fundamentally distinct values and lifestyles that are misunderstood and disrespected by city folks’.

Now, some of the commitments involved in rural consciousness, and that purport to explain these properties, are false. Take the proposition that rural areas do not receive their fair share of resources. In justifying this proposition, Cramer's interlocutors routinely maintained that rural areas receive fewer tax dollars per capita than urban areas (Cramer 2016, 77). However, rural areas actually receive about as many tax dollars per capita as urban areas (Cramer 2016, 90–3). Strictly speaking, Cramer's interlocutors were wrong.

Nevertheless, this falsehood may help represent a real problem. Rural sociologists have argued that, for a number of complex reasons, translating tax dollars into public goods is often more difficult in rural areas than in urban areas. One reason is that, due to economies of scale, it is comparatively more expensive, per capita, to provide such goods as broadband and schooling in sparsely populated areas (Johnson et al. 1995, 386). Another is that, because of the brain drain from rural to urban areas, many rural communities have few experienced public planners. Hence, they are frequently unable to administer tax-funded public policies as efficiently as urban communities (Dewees, Lobao and Swanson 2003, 195).

For Cramer, the upshot is that things end up feeling as if rural areas received fewer tax dollars. For the reasons mentioned earlier (among others¹⁰), rural areas can do fewer things with the tax dollars they receive. Therefore, although the proposition that rural areas receive fewer tax dollars is false, it nonetheless stands for, and helps to represent, something real. By replacing a complex reality with a more widely accessible proposition, this falsehood exemplifies an important property of rural life: the distributive disadvantage experienced by rural areas relative to urban areas.¹¹

In this, the false proposition that rural areas receive fewer tax dollars resembles the fluid theory of energy. The fluid theory, recall, distorts complex scientific facts (about the nature of energy) to make a property that emerges from those facts (the conservation of energy) more salient to scientists. Given the cognitive constraints human beings normally face, scientists might otherwise struggle to mentally represent those complex facts. Analogously, the false tax claim distorts a complex socio-economic reality (about taxes and the provision of public goods) to make a property that emerges from that reality (the distributive disadvantage of rural areas relative to urban areas) more salient to voters. This matters, in turn, because ordinary voters might otherwise struggle to represent that complex reality.

Let us consider why this is so. As explained earlier, the distributive disadvantage that rural areas face relative to urban areas results, in part, from facts about economies of scale, population densities, rural–urban migration and human capital. In a large-scale society, where different people must develop different skills and specializations, not everyone can be expected to have this esoteric knowledge. This generates an epistemic problem. To someone who is unaware of these esoteric socio-economic factors, it might be unclear that current distributive patterns – whereby rural areas receive roughly equal tax dollars per capita as urban areas – place rural areas at any disadvantage relative to urban areas. In this context, the tax claim's distortion plays a useful epistemic purpose: it allows voters who lack socio-economic expertise nevertheless to appreciate an important property of their environment.

This case illustrates one way that political falsehoods can help exemplify important properties of political reality, namely, by simplifying, exaggerating or otherwise distorting features of an object that ground the property being exemplified. However, there are other ways. As discussed earlier, falsehoods can also contribute to exemplification by screening off features of the object that might overshadow or distract from the property being exemplified. The ideal gas law, for instance, ignores the shape, dimensions, collisions and long-range forces of molecules, and it does so to bring out more clearly the relationship between gases' volume, pressure and temperature.

The falsehoods embedded in political models can perform this second role too. When observing political talk in black public spaces, Harris-Lacewell (2004, 17–19) notes that her interlocutors' models of political reality help ‘to reduce complexity by working as a perceptual screen’: they ‘filter information in the political world’, which helps to provide a clearer ‘roadmap for navigating the political world’.

For example, the perspectives Harris-Lacewell (2004, 181) documents often screen off, or give less attention to, social factors such as gender or class. Screening off such factors is an important epistemic cost – a point to which I will return later. Yet de-emphasizing some factors can also help bring out other political factors. In particular, Harris-Lacewell (2004, 20–1) suggests, doing so helps focus her interlocutors' attention on the relationship between race and inequality in the US: how racial hierarchy shapes social, economic and political institutions; what part white people play in perpetuating this hierarchy; and what might be the best strategies for overcoming this problem.

These features of political reality might be less visible in a model that does not screen off other features. The existence of racial inequality is, of course, obvious – particularly to those on the receiving end. However, in a complex social context involving innumerable other axes of inequality (gender, geography, class, sexual orientation and so on), its precise extent, its ultimate roots and its remedies may nonetheless be difficult to grasp. This difficulty is often amplified by ideological forces that actively obscure the nature of racial inequality. For instance, as Charles Mills (1997, 1–2) has argued, the self-proclaimed commitment to ideals of universal equality of the contemporary United States tends to conceal the central role that racial hierarchy plays in structuring its politics – so much so that the fact that racial hierarchy continues to operate as an implicit political system often goes unrecognized.

In this context, screening off certain social factors is epistemically useful: doing so helps exemplify properties of racial inequality that might otherwise be crowded out or concealed. Just as the ideal gas law's abstractions improve epistemic access to an important property of gases, so the abstractions documented by Harris-Lacewell improve epistemic access to important properties of political reality.

Let us take stock. I have argued that political falsehoods can facilitate as well as undermine political understanding. Thus, even if public opinion researchers are right that voters often accept political falsehoods, it need not follow that they lack political understanding. We can put this point in the language of ignorance: even if voters are often politically ignorant in one sense (namely, in the sense that they accept false propositions about politics), it does not necessarily follow that they are ignorant in a second, and arguably more important, sense (namely, in the sense of lacking political understanding). This is more than just a hypothetical suggestion. Existing research in political ethnography – which is crucial to identifying when falsehoods do and do not impair political understanding – suggests that, in at least some meaningful cases, falsehoods may contribute to understanding important features of political reality.

Still, this argument leaves three important worries unresolved. The first concerns generality. Even if political falsehoods can and sometimes do play a positive role within some people's political models, one might worry that the public nonetheless cannot generally – or even commonly – engage in fruitful political modelling. The second and third worries go further by disputing the value of falsehoods within political models. According to the second worry, the presence of falsehoods in political models seems, even in the cases I describe as felicitous, to obscure facts that are necessary for adequate political understanding. Therefore, by contrast with the scientific case, even the most felicitous political falsehoods seem all-things-considered problematic for understanding. The third worry focuses instead on how ordinary voters relate to their political models. Unlike scientists, who typically know that the falsehoods in their models are false, most voters are actually committed to their political models and to the falsehoods these contain. Hence, one might doubt that voters can derive any understanding – let alone all-things-considered understanding – from political falsehoods. I will consider these worries in turn.

The Generality of Political Modelling

Even if political falsehoods can and sometimes do play a positive epistemic role, one might doubt that they could generally, or even commonly, do so. I have suggested that political falsehoods can play a positive epistemic role within a broader political account or model – a set of cognitive commitments about political reality that hang together more or less closely. The problem, one might reply, is that most people simply do not have a broader political model.

The strongest motivation for this concern stems from Philip Converse's pioneering research on belief systems. Converse famously reports three findings, which have since received significant corroboration (Achen and Bartels 2016, 32–6; Converse 1964; Kinder and Kalmoe 2017). First, when asked about parties, policies and candidates via open-ended questions, the vast majority of voters seem unable to grasp, let alone spontaneously employ, core political concepts, such as ‘liberalism’ or ‘conservatism’. Secondly, voters' policy commitments generally do not hang together in either liberal or conservative ways. For example, having liberal views on one issue is only weakly predictive of having liberal views on another issue. Finally, voters' policy views are often unstable, changing significantly over time.

This, at first sight, suggests that voters do not engage meaningfully in political modelling. The first and second observations suggest that most voters do not structure their views according to the two most prominent models for thinking about politics – liberalism and conservatism. The third is even more concerning: it suggests that voters do not structure their views according to any political model. As explained earlier, having an account or model – a system of cognitive commitments that hang together more or less closely – makes one's commitments relatively stable. The instability of people's political commitments therefore suggests that they are not embedded in a broader political model (Hannon and de Ridder 2021).

This challenge goes to the heart of my argument. Unlike multiple-choice questions, Converse's studies come much closer to directly measuring political understanding. They examine not merely the truth or falsehood of people's political commitments, but how they fit together within a broader system. And the result is pessimistic. Voters seem not to integrate their political commitments within a broader account. If correct, this precludes the possibility that voters might have political understanding despite accepting political falsehoods.

However, this objection ultimately fails to establish that voters generally do not engage in political modelling. The problem is that Converse, and more recent corroborations of his studies, consider an insufficiently broad range of possible political models or accounts. Liberalism and conservatism are two important systems for organizing one's commitments about politics. But they are only two – elite-driven – models among many.

Indeed, we have already seen that the political outlooks of some non-elite voters are structured not so much by partisan labels (for example, liberalism or conservatism) as by their social group identities (for example, race, class or regional identity). Rural consciousness, for example, is a perspective that organizes and assesses various aspects of political reality (such as policies or politicians) based on their relation to the challenges and aspirations of rural life. Relatedly, Harris-Lacewell's (2004, 14) interviews suggest that for many black Americans, race is the most central category for organizing political views.

Once we adopt a broader interpretation of what might count as a political account or model, where this includes group-based perspectives, Converse's results actually support the view that voters organize their political commitments within broader models. First, when questioned about their policy views, people readily give explanations in terms of social group concerns – for example, concerns relating to their class, race and regional identity (Converse 1964, 14). Second, non-elite voters' views regarding policies or problems that bear on their group interests tend to hang together in a fairly coherent and integrated way (Converse 1964, 38–43).¹² Finally, the more strongly and explicitly a policy relates to someone's group identity, the more stable that person's commitments relating to that policy are (Converse 1964, 45–6). Seen in this light, Converse's evidence suggests that people do have integrated and stable systems of political commitments. These commitments are simply structured along different axes – group identities – than the partisan axes elites often employ.

One might perhaps worry that social group identity is an inadequate basis for structuring one's political commitments – and this may be why Converse underplays the significance of this mass group-based structuring. However, it is unclear why this should be. To be a member of a particular social group is, in part, to be subjected to a wide range of group-specific constraints and opportunities. Rural and urban communities, black and white communities, and middle-class and working-class communities are affected by different laws, social norms, economic realities and geographic constraints. The descriptive knowledge of these group-specific constraints, and the normative concerns stemming from this knowledge, are deeply relevant to assessing a wide range of political problems, policies, politicians and parties. Therefore, prima facie, it seems epistemically reasonable to structure one's political commitments based on one's social group memberships.¹³

This does not mean that any structuring based on social group identities is reasonable. Take, for instance, a white-supremacist world view. Such a world view arguably constitutes a political model – specifically, a model that interprets and structures political information through the lens of false beliefs about the racial superiority of white people. Yet this model is profoundly misguided. The central property that it represents as salient – namely, the inherent superiority of white people relative to other races – is not instantiated in the real world. Thus, the white-supremacist model fails to yield political understanding.

However, this is compatible with my present argument. I am not suggesting that all voters possess models capable of yielding political understanding. Rather, my aim is more limited: I am responding to an objection which claims that voters do not commonly, let alone generally, possess such political models. In this context, the important point remains that voters do commonly appear to possess integrated systems of political commitments and that the axes along which these are structured (social group identities) seem, in principle, epistemically reasonable. This, to reiterate, is precisely what Harris-Lacewell's exploration of black political ideologies suggests. Her investigation indicates that these ideologies are widely instantiated among black Americans (Harris-Lacewell 2004, ch.3) and that they organize information in a broadly useful way – not least by shedding light on the nature, causes and potential remedies of social inequality in the United States (Harris-Lacewell 2004, 17–22).

Thus, the objection at hand falls short. The examinations of mass belief systems pioneered by Converse do not establish that people generally lack political models (which would make them incapable of political understanding). Nevertheless, they do yield important methodological insights. In particular, they highlight a promising way to measure political understanding on a large scale. To begin, examining the stability of, and coherence between, voters' different commitments helps to establish the extent to which voters possess a potential vehicle of political understanding: a relatively integrated system of politically relevant commitments. Moreover, asking people open-ended questions where they can explain their views (rather than multiple-choice questions) allows us to ascertain to what extent voters grasp the interdependencies between these different commitments.

Yet this is not sufficient. A set of cognitive commitments can hang together, or cohere, in different ways. It can be unified by liberal concerns, conservative concerns, rural concerns, racial concerns and so on. The problem with Converse's analysis, to reiterate, is that it considers an insufficiently broad range of potential unifying themes and so prematurely concludes that people lack coherent models of political reality.

To avoid this pitfall, we need wide-ranging hypotheses regarding what might unify people's commitments. This is one area where political ethnography has an important role to play. As discussed earlier, listening to voters at length can help identify unanticipated ways in which their views hang together. This is precisely what Cramer does. She argues that rural Wisconsinites' seemingly paradoxical constellation of views (for example, their combined support for limited government and increased healthcare) can make sense once we appreciate the structuring role of the urban–rural divide in their thinking.

Thus, ethnographic research can complement the quantitative methods outlined earlier by helping to generate hypotheses regarding how different cognitive commitments might hang together within a political model. Only by putting these two kinds of methods together can we adequately assess the large-scale incidence of political understanding.

The Epistemic Costs of Political Modelling

I have argued that just as falsehoods can promote understanding within scientific models, so can they promote understanding within political models. However, one might worry that there is a crucial disanalogy: although political falsehoods can have some benefits for political understanding, they also invariably engender serious epistemic costs.

Let us consider again the false belief that urban areas receive more tax dollars per capita than rural areas. As suggested earlier, this falsehood may highlight a genuine distributive disadvantage. However, it obviously also obscures the fact that rural areas actually receive as many tax dollars per capita as urban areas. So too with the political ideologies Harris-Lacewell explores. The fact that these ideologies screen off certain features of the social situation allows them to highlight more strongly the pivotal role of race in explaining social inequality. Yet, as Harris-Lacewell (2004, 181–91) acknowledges, the flip side is that many of these ideologies de-emphasize other important social factors, such as gender or class. This is epistemically problematic both because it conceals sources of social inequality other than race, and because it risks obscuring how racial inequality interacts with these other sources of inequality (Crenshaw 1989).

Thus, one might worry that even when political falsehoods are pro tanto beneficial to understanding, they remain epistemically detrimental overall. What is worse, their epistemic costs may be difficult to eliminate. As we saw earlier, the integration of cognitive commitments within a broader model or account makes them more stable than if they were held in isolation. While this stability makes it harder for voters to abandon true cognitive commitments, it also makes it harder for them to abandon false cognitive commitments. Therefore, not only does the presence of falsehoods within voters' political models give rise to important epistemic costs, but it may also be difficult to revise voters' political models so as to eliminate these costs.¹⁴

The first thing to notice in response is that this phenomenon is actually not specific to the political case. No model, scientific or otherwise, can produce epistemic benefits without simultaneously incurring epistemic costs. Models advance understanding by exemplifying properties of their target – that is, by instantiating those properties in a way that renders them salient. Salience, however, is a comparative matter: nothing is salient if everything is. Hence, Elgin (2017, 270) notes, a model can exemplify a property of its target only ‘by overshadowing or ignoring others’.

And in non-political cases, too, the features obscured by a model's ‘felicitous falsehoods’ may be important ones. A subway map helps people understand the structure of a city's public transit system. But it does so by distorting the physical distances between stations and abstracting from streets and buildings. Likewise, the Hardy-Weinberg model, though central to population genetics, yields understanding by severely distorting reality: it represents the population as infinite and unaffected by mutation, migration or genetic drift (Elgin 2017, 263).

Nevertheless, the fact that scientific models obscure (potentially important) features of reality does not mean that we should not use them. Nor does it mean that scientific models do not play an indispensable role in achieving understanding. What it shows, Elgin (2017, 270) suggests, is that to adequately understand a scientific phenomenon, we need multiple scientific models:

different models of the same target may make different features manifest.… if what one model highlights is that in some significant respects the [atomic] nucleus behaves like a liquid drop, and another model highlights that in some other significant respects it behaves as though it has a shell structure, there is in principle no problem.… The models afford different perspectives on the same reality.

We should think of political models along similar lines. Put differently, we should not take the fact that ‘felicitous’ political falsehoods generate epistemic costs as a decisive problem. All models, political and non-political, necessarily involve such costs. However, these costs do mean that to achieve an adequate measure of political understanding, we must make use of different political models, which highlight – and obscure – different aspects of political reality.¹⁵

In this view, the epistemic costs of rural consciousness do not entail that this model cannot contribute meaningfully to political understanding. Rather, they suggest that adequate political understanding requires complementing rural consciousness with a contrasting political model – say, an ‘urban’ consciousness, which tends to foreground the challenges urban residents face (and, notably, what they actually contribute and receive from taxes). Likewise, the fact that some political models highlight race at the expense of de-emphasizing or obscuring class and gender does not nullify their contribution to political understanding. Rather, it calls for complementing them with alternative models that foreground the political significance of class and/or gender.

However, one might think that this is where the analogy between science and politics really breaks down. Scientists can move back and forth between different models of scientific reality. But one might doubt that ordinary voters would do this. This, in turn, is because of a broader disanalogy between science and politics: scientists often recognize that the scientific models they use contain falsehoods; by contrast, ordinary voters are typically committed to the truth of their political models. The following section turns to this important concern.

The Commitment to Political Models

Scientists often know that the falsehoods in their models are false. Although many scientists find it useful to represent energy as a fluid, they overwhelmingly know that energy is not a fluid. Voters, however, frequently lack this knowledge. The rural Wisconsinites Cramer (2016) interviewed, for instance, really believe that they receive fewer tax dollars. Thus, unlike scientists, ordinary voters are generally committed to their political models: they take these models, and their various components, to be true.

Why might this be problematic for political understanding? One might worry that if someone does not know that the falsehoods in their model are false, those falsehoods cannot contribute to understanding at all. However, this initial worry is too strong. As discussed earlier, falsehoods can facilitate understanding by helping to exemplify important properties of a target – that is, by helping to represent those properties in a way that makes them more salient. This process of exemplification does not depend fundamentally on whether falsehoods are recognized as false. Rather, it depends on whether, by screening off or distorting certain features of the target, the falsehoods actually make some of its properties more visible.

To see this, imagine someone who uses a subway map without knowing that it distorts the physical distance between stations. Even in this case, the map's distortions arguably help to highlight the structure of the public transit system. Indeed, they simplify the map, such that the user in question can easily see how different stations are ordered, how many stops separate two stations, where it is possible to change lines and so on. What this case illustrates is that falsehoods can help exemplify properties even when those falsehoods are not known to be false. Therefore, even if voters do not know that the falsehoods in their political models are false, it need not follow that these falsehoods cannot contribute at all to political understanding.

However, there is another worry. Even if a political model's falsehoods contribute somewhat to political understanding, they come at an epistemic cost. As argued earlier, the standard way to overcome these costs is to use multiple models. Yet the practice of multiple modelling might seem to depend on knowing that models contain falsehoods. A scientist may decide to use different models on different occasions precisely because they know that each model contains falsehoods (and so epistemic costs). By contrast, if a voter sees their political model as the correct model – such that they do not realize it contains falsehoods – that commitment might seem to preclude using multiple models.

The first step to addressing this concern is to note that, in fact, scientists do not always view their models in a purely instrumental fashion. On the contrary, according to Imre Lakatos, scientific communities have historically been organized into different research groups that are deeply committed to particular models and that compete to establish their superiority (Lakatos 1968; Zollman 2010). Importantly, this commitment does not mean that scientists cannot represent scientific reality using alternative models: they can and do consider their rivals' models, partly to consider what might be wrong with them. For Lakatos (1968), this interaction and competition between scientists who are committed to different models stimulates the growth of scientific understanding.

What this suggests is that even where people are committed to specific models – such that they do not realize that these models contain falsehoods – it is possible to bring multiple models together in a fruitful way. However, doing so requires multiple modelling to be a social process. Instead of a single person moving readily between different models that they consider to be useful but incorrect, it requires different groups of people coming together to share the models they respectively embrace – and perhaps to argue over which model is most accurate.

In the political context, participatory democracy is the response to this call for a socially distributed process of multiple modelling. Participatory ideals of democracy, such as the ideals of ‘deliberative’ or ‘communicative’ democracy, promote the active participation in public discourse of people from different social groups and different walks of life (see, for example, Anderson 2006; Chambers 2018; Landemore 2013; Young 2000). In particular, participatory democracy invites diverse groups to interact and exchange narratives expressing their perspectives on matters of public interest. Thus, it encourages the expression, within the same public arena, of different models of political reality.

In doing so, participatory democracy creates greater opportunities, for voters and their representatives, to make use of the multiple political models that are distributed across democratic societies. Even if voters have a default political model, they or their representatives can still imaginatively adopt other perspectives. For example, hearing the testimony of rural inhabitants can help me imaginatively reconstruct their rural consciousness. Although this may not be my own perspective on political reality, I can imaginatively adopt it.

Sharing multiple political models via this participatory democratic process can contribute meaningfully to political understanding. Indeed, imaginatively adopting someone else's political model can help draw one's attention to important properties of the political environment, even – as we have seen – if that model omits, distorts or otherwise falsifies important aspects of political reality. Even if the rural narrative I listen to falsely claims that rural areas receive fewer tax dollars per capita than urban areas, imaginatively inhabiting it can help draw my attention to the genuine rural disadvantage that my own default political model had obscured or overlooked.¹⁶

Still, one might worry that this social process of multiple modelling, whereby voters imaginatively adopt the political models of others in their community, remains importantly disanalogous from the way scientists engage in multiple modelling. In the former case, one might suggest, voters believe that their default political model is accurate and take other, conflicting models to be false – even though they are willing to hypothetically consider or entertain them. By contrast, scientists realize that none of the models they are switching between is completely accurate. This difference seems important: it is partly because of this realization that scientists are able to use different models for different purposes, based on where they seem more appropriate.

This is an important concern. But we can soften its force by means of two observations. The first observation is that we should not over-idealize the case of multiple modelling in science. As explained earlier, historians of science, such as Lakatos (1968), have argued that scientists are often strongly committed to the correctness of a particular scientific model. When such committed scientists consider the rival models advocated by other research programmes, they may well believe that these rival models are false and may consider them largely with an eye to demonstrating why they are inferior or inadequate. To the extent that this picture of scientific communities is accurate, the suggested disanalogy between voters and scientists may be less strong than it initially appears.

The second observation takes a different approach. It concedes that scientists – or, at least, many scientists – realize that none of the models they switch between is completely accurate. However, it suggests that, over time, the social process of multiple modelling enacted by participatory democracy can give rise to forms of multiple modelling that approximate this. Even if one initially assumes that other voters' political models are false, imaginatively adopting those other models can gradually lead one to realize that they capture important features of political reality that one's default model does not. For example, through exposure to rural narratives, an urban voter may come to realize that rural areas face patterns of socio-economic disadvantage that are scarcely visible to those living in urban areas. As a result, this urban voter may come to accept that neither urban consciousness nor rural consciousness is completely accurate. Based on this realization, they may then decide to adopt the perspective of rural consciousness (perhaps by soliciting testimony from rural residents) when considering questions that require a sensitive understanding of the disadvantages faced by rural areas, while sticking to urban consciousness when considering problems that predominantly affect urban areas.

Let us summarize. We started with the concern that voters often seem committed to particular political models. However, as I have argued, this is not a decisive problem for political understanding – in part, because the multiple modelling required for adequate political understanding can take a socially distributed form. Moreover, as we have seen, participatory democracy is crucial to realizing this social process of multiple modelling.

This last point bears emphasizing. As discussed in the Introduction, the evidence that voters accept many political falsehoods has motivated a theoretical move away from participatory democracy, and towards ideals that de-emphasize citizen engagement (see, for example, Bell 2015; Brennan 2016; Green 2010; López-Guerra 2014; Parvin 2015; Somin 2016). We can now see that this move is doubly problematic. Not only does it overlook the fact that accepting political falsehoods does not necessarily undermine political understanding, but it also overlooks the fact that participatory democracy seems positively needed for adequate political understanding. This, to reiterate, is because any single political model, with its felicitous falsehoods, can only yield partial political understanding. Participatory democracy defines a social process that helps to overcome this partiality.

This is not to say that pooling multiple models through inclusive democratic participation is easy. Voter apathy, intergroup dislike, disparities in political power and the fragmentation of different social groups across different spatial and occupational arenas all make it extraordinarily difficult to get different social groups to publicly share their accounts of political reality.¹⁷

However, this should not detract from the main point. The point is not that participatory democracy is easy to realize; rather, it is that it constitutes the political ideal that responds most explicitly to the need for diverse political models in decision making. That, the analogy with science suggests, is what is needed for policy making to be guided by comprehensive political understanding.

Conclusion

When examining voters' competence, we should focus on their level of political understanding. This matters because understanding has a complex relationship to falsehood: the presence of falsehoods within a model of reality can improve as well as impair our understanding of that reality. This complex relationship, I have argued, has crucial implications for democracy and political ignorance.

Methodologically, it means that from the observation that voters often accept falsehoods about politics, we cannot directly infer that they lack political understanding. To determine when falsehoods destroy political understanding and when they instead enhance it, measurements of political ignorance should consider how those falsehoods function in voters' broader political models.

Empirically, this analysis tentatively alleviates the puzzle with which we began. Although many real-world political falsehoods impair political understanding, some falsehoods arguably facilitate it. To that extent, the observation that voters accept many political falsehoods may be reconciled with ethnographic findings suggesting that voters understand political reality in meaningful ways.

Finally, this analysis has key normative upshots for democracy. The first is that widespread calls for abandoning participatory ideals of democracy because of voters' political ignorance are premature. In light of the foregoing methodological and empirical insights, the evidence of political ignorance is less conclusive than it is often thought. However, we can go further. Given what political understanding is and how individual models of political reality contribute to it, participatory democracy seems positively desirable. Political understanding calls for multiple models of political reality. By bringing together diverse narratives that express complementary models of political reality, participatory democracy is the political response to that call.