1
The Philosophy of Polarization Phenomena
1.1 The Aim of the Book
Group polarization is a tendency of deliberating groups to incline toward more extreme positions than initially held, on average, by most of their individual members prior to deliberation. When a group polarizes, the average post-group response is more extreme in the same direction than the average of the pre-group response (Myers & Lamm 1976). The phenomenon is very well-established in social psychology, where many empirical studies have detected polarization effects in groups discussing a wide variety of topics. Outside laboratory conditions, group polarization is believed to feature centrally (and cross-culturally) in a high number of socially relevant phenomena such as jury decisions, political debates, financial decision making, extremism, terrorism, and online interaction, hence its practical and theoretical significance. However, although widely investigated in empirical psychology, this important tendency of groups has not attracted much attention among philosophers. With this book, we aim to fill this gap. Whereas social psychologists have tried to pin down the causes of group polarization, our aim is to investigate its metaphysics and epistemology in a novel and programmatic way.
In this chapter, we will state more precisely what our project is, draw some conceptual distinctions concerning polarization phenomena, and vindicate the place of philosophy when it comes to theorizing about them. We distinguish, in Ā§1.2, polarization from extremism; in Ā§1.3, two general senses of polarization; in Ā§1.4, polarization from disagreement. In Ā§1.5, we compare group polarization to belief polarization and, in Ā§1.6, to political polarization. In Ā§1.7, we state the two guiding questions of our investigation, and our methodological approach and vindicate the place of philosophical theorizing about polarization phenomena.
1.2 Polarization and Extremism
Moving to extremes is not the same as polarizing. Moving to extremes is compatible with shifting to any of the two poles of a given scale, whereas polarization only involves moving toward the nearāor already preferredāpole (Fraser et al. 1971). Thus, when one polarizes, one also moves to an extreme, but if one moves to an extreme, one does not necessarily polarize, as one can shift to the opposite pole. Polarizing is just a way of moving to extremes.
Neither moving to extremes nor polarizing necessarily implies that one has become an āextremistā. Extremism, in general, consists in adopting an extreme attitude on a given scale, and one might move slightly to an extreme (e.g., polarize) without thereby adopting what constitutes the extreme attitude on the scale. For example, if one uses the leftāright political spectrum to classify someoneās political views, that personās views count as extremist if they are located at any of the two poles of the scale: far-left or far-right. However, that person might only move to extremes or polarize from, say, a center-right position to a right view without yet becoming a far-right extremist. That is, one does not become an extremist until one adopts a view that is located at a pole of a given scale.
Polarizing and shifting to the opposite pole are the two ways of moving to extremes and hence of becoming an extremist. An individual whose political attitudes are located on the center-right of the political spectrum at t1 and shift to far-right at t2 becomes a far-right extremist by polarizing: that personās attitudes move toward the near or already preferred pole. By contrast, an individual whose political attitudes are located on the far-left of the political spectrum at t1 and shift to far-right at t2 becomes a far-right extremist by shifting from one extreme to another, i.e., without polarizing.
1.3 Bidirectional and Unidirectional Polarization
In the different empirical literaturesābut also in several normative disciplines as well as in the mediaāthe term āpolarizationā is used in two different general senses that we can characterize as follows:
- Bidirectional polarization: for any individual or collective agents X and Y, if at t1 X and Y respectively hold conflicting attitudes A1 and A2, then X and Y bidirectionally polarize if A1 and A2 become more extreme toward their near or already preferred poles at t2.
- Unidirectional polarization: for any individual or collective agent X, if at t1 X holds attitude A, then X unidirectionally polarizes if A becomes more extreme toward the near or already preferred pole at t2.
Bidirectional and unidirectional polarization apply both to individuals and groups, but the minimal number of agents involved differs. Unidirectional polarization involves at least one individual or group whose attitude becomes more extreme. By contrast, bidirectional polarization involves at least two agents, individuals or groups, whose attitudes move further away in opposite directions.
The crucial difference between them is, precisely, that in bidirectional polarization, the agents involved polarize in opposite directions, whereas in unidirectional polarization, they polarize in the same direction. For example, when media say that US society is polarized, what is meant is that US society is polarized bidirectionally. By contrast, when we say of an individual that they have become radicalized, what is meant, among other things, is that that individual has unidirectionally polarized toward an extreme worldview. We should keep this general difference in mind, because the same term, āpolarizationā, is often used in these two senses to refer to importantly different phenomena.
Although it is certainly not the same to say that a group has āpolarizedā, meaning that it has bidirectionally polarized internally, as saying that it has āpolarizedā, meaning that it has adopted a more extreme view (i.e., unidirectionally polarized), the two senses of polarization are conceptually related in that bidirectional polarization necessarily involves unidirectional polarization but not the other way around. Saying that the conflicting attitudes, A1 and A2, of two agents bidirectionally polarize is conceptually the same as saying that A1 and A2 unidirectionally polarize in opposite directions. However, an agentās unidirectional polarization is not an instance of bidirectional polarization unless there is another agent whose attitude unidirectionally polarizes in the opposite direction. Unidirectional polarization is accordingly more fundamental, at least conceptually.
1.4 Polarization and Disagreement
Disagreement can be classified in many ways. One way is in terms of whether it involves collectives or individuals. For example, interpersonal disagreement occurs when two or more individuals disagree amongst themselves. Those individuals can be members of different groups but, in interpersonal disagreement, individuals do not disagree in their capacity as group members but as individuals. Groups can disagree too: intergroup disagreement occurs when two or more groups disagree between them. Intergroup disagreement can involve disagreement between individuals, too, but, in such a case, the individuals who disagree do it in their capacity as group members and, specifically, as spokespersons for their groups. By contrast, intragroup disagreement is the kind of disagreement that occurs within a group between individual group members or subgroups of that group. Finally, disagreement can also take place between an individual and a group (one-versus-many disagreement).1
As we can see, whatever shape a disagreement takes (whether interpersonal, intergroup, intragroup, or one-versus-many), it always involves at least two agents (individuals or collectives) whose attitudes conflict. How does polarization relate to disagreement? First, disagreement can occur without polarization, as it does in the case of agents whose conflicting views do not change over time. For example, someone with center-right views might disagree with someone with center-left views on, say, economic policy without their views shifting a bit in any direction. Such individuals disagree but do not become more polarized. In fact, disagreement can occur with de polarization. Someone with far-right views might hold a disagreement with someone with far-left views on economic policy; upon discovering their disagreement, they might conciliate and shift, respectively, to center-right and center-left positions while still being in disagreement.
Even if disagreement and polarization are phenomena that can occur independently, disagreement between individuals, groups, or within groups that intensifies over time involves polarization. In particular, individuals, groups, or group members who disagree about some issue see their disagreement intensified when they polarize bidirectionally about that issue, i.e., when they unidirectionally polarize in opposite directions.
With all these distinctions in place, let us compare group polarization with two other polarization phenomena studied, respectively, in psychology and political science: belief polarization and political polarization.
1.5 Group Polarization and Belief Polarization
When psychologists talk about āgroup polarizationā, they mean a tendency of some groups to move to extremes following deliberation. This phenomenon ought not to be confused with a different polarization phenomenon that psychologists call attitudeāor belief polarization. Belief polarization occurs when individuals who disagree about a given proposition see their own beliefs reinforced after individually assessing a body of mixed evidence that both supports and contradicts their beliefs, absent any deliberation.
By way of illustration, in an often-cited experimental study, Lord and colleagues (1979) selected some undergraduates who firmly believed that capital punishment has a deterrent effect and others who firmly believed the opposite. Subjects were asked to individually assess two fictional but realistic pro-deterrence and anti-deterrence empirical studies, as well as prominent criticisms to such studies in the alleged literature and the replies offered by the alleged authors. The result was that subjects who initially favored the position that capital punishment has a deterrent effect (i) more strongly believed it after being exposed to the pro-deterrence study and (ii) regarded it as significantly more convincing than the anti-deterrence study; the same effect, but in the opposite direction, was found in anti-capital punishment subjects. This result confirmed that all participants were biased in favor of the study that supported their initial beliefs and, more generally, the hypothesis that belief polarization is due to biased assimilation of the evidence.2
Let us compare group and belief polarization.
First, whereas belief polarization arises in the absence of deliberationānamely when subjects privately assess and assimilate available evidence in a biased wayāgroup polarization arises following group discussion.
Second, both group and belief polarization involve individuals and groups, but whereas belief polarization is fundamentally an individual phenomenon, group polarization is fundamentally a collective phenomenon. The empirical studies on belief polarization involve mixed groups of people with similar and opposing beliefs about a given topic. However, communication is absent among them, which means that belief polarization is tested individuallyānamely, by asking subjects to read and process the relevant evidence silently, as in Lord et al. (1979). Group polarization affects groups of like-minded individuals who communicate with each other about their like-minded views, but it is tested collectively. In particular, group polarization is measured by seeing whether or not, and to what extent, the average of individual post-discussion responses is more extreme than the average of individual pre-discussion responses. The focus on average responses gives rise to the possibility that a group of individuals polarizes on average when no individual in the group polarizes (i.e., when no group member adopts a more extreme position). Fraser et al. (1971) capture this possibility as follows:
a three person group using a seven-point scale (from 1 to 7 with a neutral point at 4) could have initial scores of 7, 5 and 2 and after discussion without consensus all might put down 5. At the group level, a slight move to extremity (which in this case is also polarization) has occurred in the shift from a mean of 4.67 to one of 5.00. But individually, after discussion the new positions are in two cases less extreme and not one of the subjects has adopted a more polarized position.
(Fraser et al. 1971: 17; emphasis added)
Fraser et al.ās example is an excellent illustration of how collective unidirectional polarization, such as group polarization, can diverge from individual unidirectional polarization. In groups that polarize, individual members can unidirectionally polarize, too. For example, if all members of a group adopt a more extreme view about some issue after discussing it than the view they individually held before discussion, they all individually polarize in the same direction. In that case, the group will polarize too, as the average post-group response will be more extreme in the same direction than the average of the pre-group responses. However, this does not imply that if a group unidirectionally polarizes and adopts a more extreme view than it had, on average, before deliberating, its members will individually, unidirectionally polarize as well. This is what the example shows.
1.6 Group Polarization and Political Polarization
Another polarization phenomenon that is distinct but related to group polarization is political polarization. Political polarization refers to the kind of disagreements and divisions toward ideological poles that exist in the political life of democratic societies. As theorized in the political science literature, political polarizationāsometimes misleadingly referred to as group polarization (e.g., Iyengar & Westwood 2015)ācomes in two guises, depending on who is the subject of polarization: elite popularizationāi.e., polarization among political elites (e.g., among elected party members and relevant non-elected members)3āand mass or popular polarization, i.e., the kind of political polarization that divides large parts of society over controversial issues, especially in the context of elections.4
The distinction between elite and mass polarization far from exhausts the array of polarization phenomena that falls under the umbrella term āpolitical polarizationā. For example, Bramson and colleagues (2017) usefully give a (non-exhaustive) list of nine polarization phenomena dubbed āpolitical polarizationā in the social sciences, with a particular focus on formal models of opinion polarization. Relatedly, as an illustration of the ambiguity of the term in the political science literature, there is no agreement on how to define āmass polarizationā. As Lelkes (2016) helpfully points out, one way of conceptualizing mass p...
