In “People deserve safety on college campuses, ideas don’t,” Andrew J. Perrin and Christian Lundberg make an important argument against the idea of viewpoint diversity. They write:

Emphasizing viewpoint teaches students to not bother separating ideas from the people who hold them. Viewpoint is a visual metaphor that attaches what a person believes to where they sit: Viewpoints are properties people own and express, not ideas to be evaluated. It’s a classic ad hominem fallacy that renders argument fruitless.

We all draw on experience, and our experiences are influenced by our social positions. That is why demographic diversity is intellectually valuable. If, for example, men monopolize the conversation, then issues and solutions that are more obvious to other genders will probably be overlooked, or, at best, underplayed. The fact that some individuals demonstrate exceptional insight into others’ experiences does not negate this point. (See “Dear Mrs Amartya Sen, men will never understand us.”)

However, the metaphor of a viewpoint makes people’s ideas look like automatic functions of their social positions. It overlooks the diversity and freedom of individuals in any given social group; it makes reasoning and argument look fruitless; it implies that incorporating individuals with additional viewpoints will automatically improve a group and should be the main goal; and it suggests that a critical assessment of an idea is an attack on the person who holds it. As Perrin and Lundberg conclude:

Settling for exposure to viewpoints — as if they were infections to which one might develop antibodies — places them outside the realm of argument and reason. We fail those on the political left by ignoring conservative arguments instead of engaging them. Meanwhile, conservative students learn that their ideas are something others should be exposed to rather than meaningfully engaged.

I believe that the metaphor of a viewpoint is deeply rooted, and that challenging it could be quite fruitful. Put more generally, the image of a point in space is remarkably widely used to define people and ideas.

The most familiar example is the left-to-right political spectrum, which allows a person, an opinion or position, or a party or movement to be located at one point on a straight line. People or ideas can easily be visualized as points in two-dimensional space if they are located along two axes at once. For instance, Americans have often been described as liberal versus conservative on economics and on race, as two separate dimensions. Three dimensions are harder to depict on paper or a flat screen, although a three-dimensional model can be rotated and presented meaningfully on a plane. In any case, mathematics allows adding more than three dimensions, even though we can’t picture them visually, by simply tagging a given person, idea, or party or movement with many variables at once. Prevailing statistical methods, such as factor analysis, treat people, ideas, or groups as points in many-dimensional space and envision differences as the distance between positions. Many models try to explain why a person occupies a given point based on other known information about the same individual, such as party identification or race.

If a model employs many dimensions, it can incorporate any amount of quantitative data about the people and ideas being studied. Since each person or idea has a good chance of occupying a unique position in multidimensional space, there is relatively little danger that individuals will be casually lumped together in large groups.

However, some kinds of information must be lost in a model based on points in space. First, this metaphor conceals the way that ideas may connect to each other. If respondents are asked many questions on a survey, standard statistical methods capture correlations among their answers but cannot detect logical relationships among any individual’s ideas. Does a person believe one thing because of another belief, or despite it, or as two disconnected ideas? The structure of individuals’ thinking—if there is any—is lost. In contrast, when we read an impressive political argument or speech, we are primarily interested in its structure: in why (or whether) each point implies the next, or qualifies it, or contradicts it. A metaphor of points in space makes everyone look much more simple-minded than any careful speaker or writer.

To be sure, some of us probably fail to connect our separate ideas in reasonable ways, but we cannot know how many from standard survey research. The metaphor of points in space is biased against detecting complexity of thought, if there is any (Levine 2022).

Importantly, large bodies of research based on this model find that people are not responsive to arguments, that their beliefs are either incoherent or driven by indefensible biases, that they supply reasons after the fact to rationalize what they already desire—in short, that anything remotely resembling a deliberative democracy is psychologically naïve. Paul Sniderman—who dissents in interesting ways from what he calls “the textbook view of citizens’ capacity to reason about politics”—summarizes the consensus of his fellow political scientists as follows. “Average citizens’ knowledge about politics and public affairs is threadbare; their political beliefs minimally coherent, indeed, often self-contradictory; their support for core democratic values all too likely to crumble in the face of a threat, real or imaginary” (Sniderman 2017, pp. 42, 107).

Factor analysis is a statistical technique. It is often described as scientific, where “science” means a cumulative, empirical research project of testing hypotheses with data. Famous contributors to the statistical study of political opinions and behavior who have used a point-in-space model are English-speaking social and behavioral scientists like Charles Spearman, who invented factor analysis, and Philip Converse and his colleagues, who pioneered academic political survey research with the American National Election Studies.

A strangely similar metaphor is also influential in a very different tradition: Continental European political philosophy. Until the late 1800s, the words “culture” and “religion” had made sense only in the singular. People either had culture or not; they were either religious or not. But Romantic-Era thinkers began to see deep plurality. There were many cultures, religions, and nations (or peoples), understood as distinct in fundamental ways. These thinkers imagined that individuals saw the world from the perspective of their respective cultures or religions. Two people from different cultures would behold a different reality, although people who shared a culture would share a common worldview. A word for everything that can be seen from a given point is “horizon.” Perspective, viewpoint, and/or horizon were keywords in the thought of Herder, Hegel, Marx, Nietzsche, Heidegger, and many other highly abstract European philosophers.

Again, a person with a perspective occupies a point in space. This metaphor generates insights—there may be a French, or a modern, or a bourgeois perspective on certain topics—but it also obscures and creates conundrums. If people hold all their beliefs because of their fundamental perspectives or viewpoints, then a critique of any of their beliefs can be taken as an objection to the person and their right to speak. In that case, arguments about ideas can seem uncivil and even threatening.

Furthermore, if human beings are assigned to cultures on a “one-to-a-customer basis,” (Wolcott 1991, p. 247), and if each culture fundamentally shapes how all its members understand the world, then how can anyone know anything objectively, including the nature of other people’s cultures? Surely everything we think is relative to our perspective. Deep cultural relativism leads to basic skepticism or even nihilism, as Nietzsche most famously argued.

One way out is to argue that a fair institution is one that treats all cultures and religions equally and neutrally. For instance, the great American political philosopher John Rawls assumes “that a modern democratic society is characterized … by a plurality of reasonable but incompatible comprehensive doctrines” (Rawls 1993, pp. xvi, 59). Each of these doctrines determines each person’s values. Rawls concludes that a fair government must be neutral among these doctrines; indeed, he sees justice as fairness. Demands for “viewpoint diversity” on college campuses have a similar logic. However, critics have argued that neutrality is impossible (liberal institutions inevitably reflect specific values) and mere fairness among perspectives is an unsatisfactory account of justice.

Whether it is invoked in a statistical model or a work of political philosophy, a point in space from which one sees the world is a metaphor. It should not be taken too literally. We have other ways of describing the complexity of human interactions. We can model conversations as games with players and moves. We can envision ideas flowing through society on a hydraulic model, with pressure and viscosity (Allen 2015). Caroline Levine shows that literary writing often makes use of four forms—wholes, rhythms, hierarchies, and networks—to represent social phenomena (C. Levine 2015).

Indeed, we live in a period of fascination with networks: electronic, neural, social, semantic, and many other kinds. This means that we have powerful new techniques for analyzing networks, and many recent studies apply these techniques to people and ideas in ways that offer insights about politics.

This is why I have been working with colleagues to replace the metaphor of points in space with one of networks. I have introduced the technical term idiodictuon for the network of ideas that each individual holds, where the connections among ideas are reasons.

In this model, when people discuss issues, they are sharing ideas and connections that others may choose to incorporate into their respective idiodictuons. Whether we encounter another person’s ideas depends on whether we are connected to that person in some kind of social network. Human beings who discuss within a network of relationships form a phylodictuon (a shared network of ideas, including ones that conflict).

It is generally good for a phylodictuon to encompass diverse ideas and ideas from diverse people (which are different matters), yet the job of a wise community is to improve its collection of ideas and how they are organized, not merely to ensure that all available ideas are included. As Perrin and Lundberg write, “Confronting serious ideas means that while every person deserves safety on campus, no idea does; all ideas deserve the respect that a real stress test brings.”

See also: individuals in cultures: the concept of an idiodictuon; Mapping Ideologies as Networks of Ideas; a mistaken view of culture; Teaching Honest History: a conversation with Randi Weingarten and Marcia Chatelain; etc.

Sources: Perrin, Andrew J. and Christian Lundberg, “People deserve safety on college campuses. Ideas don’t,” The Boston Globe, March 29; Paul M. Sniderman, The Democratic Faith: Essays on Democratic Citizenship (Yale University Press); Peter Levine, “Mapping ideologies as networks of ideas,” Journal of Political Ideologies, 2022, DOI: 10.1080/13569317.2022.2138293; Harry F. Wolcott, “Propriospect and the acquisition of culture., Anthropology & Education Quarterly 22, no. 3 (1991): 251-273; John Rawls (Political Liberalism. New York: Columbia University Press, 1993); Danielle Allen, “Reconceiving Public Spheres: The Flow Dynamics Model,” in Allen and Jennifer S. Light, From Voice to Influence: Understanding Citizenship in a Digital Age, University of Chicago Press, 2015, pp. 178-207; Caroline Levine, Caroline, Forms: Whole, Rhythm, Hierarchy, Network (Princeton University Press, 2015).

Presumably, each person should hold a structured set of political opinions. For instance, if you want more government spending without any new taxes, you should be OK with deficits (unless you dispute that deficits will result). If you want a specific right for yourself, you should support the same right for other people, because fairness demands equal protection (unless another worthy principle overrides that conclusion). If you think individual liberty is a high priority, you should oppose censorship (unless you think restricting speech is necessary for a different reason). Each of your beliefs should predict several others, forming a tight network.

In the early 1960s, Philip Converse argued that most Americans’ beliefs were hardly structured at all (Converse 1964). Knowing what a person believed about x would not help you predict what that person believed about y.

Converse’s article has been cited more than 12,000 times and has generated a large literature. Some studies have confirmed his basic finding (e.g., Kinder & Kalmoe 2017). Some use different methods or datasets to challenge his conclusion by finding structure (e.g., Boutyline & Vaisey 2017 or Levine 2022). Some have contested Converse’s interpretation. For instance, maybe people are ambivalent about issues, holding views on both sides. A multiple-choice survey misses their ambivalence and gives a misleading impression that people are inconsistent when they really feel conflicted (Zaller & Feldman 1992).

There is also a line of research that finds that most Americans (Achen & Bartels 2016, p. 268; Sniderman 2017) and Europeans (Galina 2023) hold structured political beliefs, but their structures come from the leaders of their political parties.

It may not be self-evident what a conservative or a progressive party should think about each new topic, from COVID vaccination to Russia’s invasion of Ukraine to bailing out a bank in Silicon Valley. But parties do form views, and their voters generally follow suit. Most people’s lists of issue positions mirror those of their party’s leaders. A given person’s ideas may or may not cohere, but they probably correlate closely with the positions of that person’s party.

I do not think that this research settles the empirical issues. The reality is complicated, with many dynamics at work. There are methodological challenges, such as the limitations of surveys that I mentioned earlier. And it’s not completely clear what causes what. (Maybe party elites are affected by their grassroots members or by some third force, such as celebrities.) Nevertheless, I believe there is at least some important truth to the theory that parties organize people’s thinking for them–or, I should say, for us. As Paul Sniderman puts it, “parties organize the choice environment” for voters “and define what goes with what” (Sniderman 2017 p. 71).

How should we assess this situation? Is it good or bad, and does it require some kind of remedy?

One way to think about those questions is to choose a model for understanding political parties. Here are three, amongst others:

1. Activists and leaders of political parties are highly interested in issues. Most people defer detailed consideration of issues to the leadership of their preferred party. That makes good sense, in the same way that it’s often wise to delegate a decision to a committee of passionate volunteers. Supporters of a party can assess its general direction and use “voice” (becoming involved in the party’s decisions) or “exit” (leaving the party or just voting for a different one), if they are dissatisfied (Hirshman 1970). Perhaps a given political system needs more voice–more participatory opportunities within each party–or more viable parties, so that voters can exit more easily. But we should not be worried by the general finding that people take their cues from party leaders. People are wisely delegating the nitty-gritty work of political analysis to those who enjoy it most. Voters are learning from the more extensive thinking of party leaders.
2. Party leaders are politicians, defined as people who want and pursue political offices for themselves. They will choose positions on issues to improve their chances of winning. Their self-interest is not shared by ordinary people, who want good outcomes. Insofar as people take their cues from party leaders, they are being used as means to the politicians’ personal ends. In a better democracy, more of us would exhibit individual structures of ideas, and political leaders would have to cater to our views, not the reverse.
3. Parties basically reflect social interests. In a given system, there may be a party for the farmers, for the urban middle classes, for the observant Catholics, and for a linguistic minority. In the USA, the electoral system forces the concatenation of interests into two umbrella parties, but they are basically coalitions of such interests. Therefore, voters will primarily seek a party that protects the interests that they consider most important (not necessarily material ones). However, a party must also take positions on many other issues. Leaders choose positions that maximize their party’s political appeal and leverage so that they can protect the voters’ core interests. Voters assess parties as tools for protecting their interests, and as long as they are basically satisfied with a given party, they will mimic its specific issue stances.

I think the truth is some mix of these ideas, depending on the political system. After all, countries differ in respect to how many parties they have, whether and to what degree their major parties are ideologically or demographically distinguished, whether coalitions are built inside parties or among them, whether parties exercise discipline over politicians, to what extent intellectual work is conducted inside the parties compared to other sectors of the society, and which kinds of people constitute the party “elites.” (In the USA, official members of the party committees are less influential than nominally nonpartisan pundits and celebrities.)

As usual, empirical evidence is relevant to our political judgments, but it is insufficient. What should we do if “parties organize the choice environment”? Sniderman clarifies the empirical literature and offers some important normative guidance. But his argument makes me want to think harder about how specific parties in specific political systems play their structuring roles and whether their approaches to choosing and combining positions are acceptable.

See also: Mapping Ideologies as Networks of Ideas; are Americans ‘innocent of ideology’?; what if people’s political opinions are very heterogeneous?; two theories of American political parties; social class inversion in the 2022 US elections, etc.

References: Achen, C. H., Bartels, L. M. (2016), Democracy for Realists: Why Elections Do Not Produce Responsive Government; Princeton University Press; Boutyline, A., & Vaisey, S. (2017), Belief network analysis: a relational approach to understanding the structure of attitudes, American journal of sociology, 122(5), 1371-1447; Converse, P.E. (2006) The nature of belief systems in mass publics, Critical review 18.1-3 (2006): 1-74; Kinder D.R. & Kalmoe, N.P. (2007), Neither liberal nor conservative: Ideological innocence in the American public, University of Chicago Press; Gallina, M (2023), Solving the (false) dilemma: an ecological approach to the study of opinion constraint,” Political studies; Hirschman, A. O. (1970), Exit, voice and loyalty: responses to decline in firms, organizations, and states (Cambridge, MA: Harvard University Press); Levine, P (2022), Mapping ideologies as networks of ideas, Journal of Political Ideologies: 1-28. Sniderman P.M. (2017), The Democratic Faith: Essays on Democratic Citizenship (New Haven, CT: Yale University Press), Zaller, J. & Feldman, S. (1992), A simple theory of the survey response: answering questions versus revealing preferences, American Journal of Political Science, 36:3: 579-616