Belief Polarization Research Articles

The Epistemological Significance and Implications of Belief Polarization

A principal assumption in the epistemology of disagreement is that we, as rational subjects, assess evidence neutrally in order to justify our beliefs. However, the existence of the phenomenon of Belief Polarization threatens the validity of this basis. Since its introduction into philosophical discussion in Thomas Kelly’s paper titled, Disagreement, Dogmatism, and Belief Polarization, the phenomenon of Belief Polarization has been thoughtlessly overlooked. Given serious consideration, there seem to be widespread epistemological implications due to the existence of Belief Polarization. Specifically, Belief Polarization brings to light significant claims about the nature of justification and belief forming processes, specifically concerning evidence gathering. As this paper will argue, given awareness of Belief Polarization, rational subjects should be less confident in their justification of belief forming processes. In other words, rational subjects should not be fully confident in the objectively based truth-value of their beliefs.

Bounded Memory and Biases in Information Processing

Before choosing among two actions with state-dependent payoffs, a Bayesian decision-maker with a finite memory sees a sequence of informative signals, ending each period with fixed chance. He summarizes information observed with a finite-state automaton. I characterize the optimal protocol as an equilibrium of a dynamic game of imperfect recall; a new player runs each memory state each period. Players act as if maximizing expected payoffs in a common finite action decision problem. I characterize equilibrium play with many multinomial signals. The optimal protocol rationalizes many behavioral phenomena, like �stickiness,� salience, confirmation bias, and belief polarization.

Decision theory with resource-bounded agents.

There have been two major lines of research aimed at capturing resource-bounded players in game theory. The first, initiated by Rubinstein (), charges an agent for doing costly computation; the second, initiated by Neyman (), does not charge for computation, but limits the computation that agents can do, typically by modeling agents as finite automata. We review recent work on applying both approaches in the context of decision theory. For the first approach, we take the objects of choice in a decision problem to be Turing machines, and charge players for the "complexity" of the Turing machine chosen (e.g., its running time). This approach can be used to explain well-known phenomena like first-impression-matters biases (i.e., people tend to put more weight on evidence they hear early on) and belief polarization (two people with different prior beliefs, hearing the same evidence, can end up with diametrically opposed conclusions) as the outcomes of quite rational decisions. For the second approach, we model people as finite automata, and provide a simple algorithm that, on a problem that captures a number of settings of interest, provably performs optimally as the number of states in the automaton increases.

A social epistemology of aesthetics: belief polarization, echo chambers and aesthetic judgement

How do we form aesthetic judgements? And how should we do so? According to a very prominent tradition in aesthetics it would be wrong to form our aesthetic judgements about a particular object on the basis of anything other than first-hand acquaintance with the object itself (or some very close surrogate) and, in particular, it would be wrong to form such judgements merely on the basis of testimony. Further this tradition presupposes that our actual practice of forming aesthetic judgements typically meets, or at least approximates, this ideal. In this paper I target this descriptive claim and argue—by appeal to some empirical work concerning belief polarization and echo chambers in aesthetics—that our actual practice of forming aesthetic judgements is heavily dependent on social sources such as testimony. I then briefly consider what normative implications this descriptive claim may have.

Bounded Memory and Biases in Information Processing

I develop a model of Bayesian learning constrained by a finite memory. Before he chooses between two actions with state-dependent payoffs, a decision-maker (DM) can observe a process of informative signals, terminating each period with a fixed chance. The DM must always summarize information into one of finitely many memory states, and so must devise a transition rule specifying how to update the memory state after each new signal is observed. I characterize the optimal protocol; I show that it is an equilibrium of a dynamic game of imperfect recall, where a new player runs each memory state each period. I generally prove that all players act as if their transition choices maximize expected payoffs in a common one-shot finite action Bayesian decision problem. Specializing to multinomial signals, I prove that expected payoffs increase in the expected number of signals, but are bounded away from full information as that number explodes. I fully characterize equilibrium play with a large expected number of signals: The DM ignores all but the two most extreme signals, and sometimes mixes on the extreme signals (“sticky” behavior). The optimal protocol is consistent with many behavioral phenomena, like the role of salience, the importance of first impressions, confirmation bias, and belief polarization. My results subsume those of an older computer science literature, and offer a counterpoint to the informational herding literature.

Belief polarization is not always irrational.

Belief polarization occurs when 2 people with opposing prior beliefs both strengthen their beliefs after observing the same data. Many authors have cited belief polarization as evidence of irrational behavior. We show, however, that some instances of polarization are consistent with a normative account of belief revision. Our analysis uses Bayesian networks to characterize different kinds of relationships between hypotheses and data, and distinguishes between cases in which normative reasoners with opposing beliefs should both strengthen their beliefs, cases in which both should weaken their beliefs, and cases in which one should strengthen and the other should weaken his or her belief. We apply our analysis to several previous studies of belief polarization and present a new experiment that suggests that people tend to update their beliefs in the directions predicted by our normative account.

Infostorms

AbstractIt has become a truism that we live in so‐called information societies where new information technologies have made information abundant. At the same time, information science has made us aware of many phenomena tied to the way we process information. This article explores a series of socio‐epistemic information phenomena resulting from processes that track truth imperfectly: pluralistic ignorance, informational cascades, and belief polarization. It then couples these phenomena with the hypothesis that modern information technologies may lead to their amplification so as to give rise to what are called “infostorms.” This points to the need for studying further the exact relations between information technologies and such infostorms, as well as the ways we may design technologies to avoid being misled away from what we have good reasons to believe.

Rumor clustering, consensus, and polarization: Dynamic social impact and self-organization of hearsay

The “bottom-up” self-organization of shared sense-making and group decision-making through rumor (unverified information statements in circulation) was investigated in two computer-mediated laboratory experiments on the effects of network clustering (i.e., structural “cliquishness”). Participants in 27 (Study 1) and 33 (Study 2) 16-person laboratory-created networks at three institutions discussed ambiguous situations (e.g., “a professor was found dead”) and then chose one of four possible rumors in a judgment task (e.g., “he was killed by an angry student”) to explain each situation. Static lattice, “ribbon” (street-like), “family” (connected clusters), random, and dynamic-random configurations were employed. Network clustering led to rumor clustering (emergence of homogenous pockets of rumor choices). There was also evidence for increased consensus, rumor persistence, and belief polarization. Belief polarization was amplified by rumor clustering and consensus. In addition, the extent to which “neighbors” were unified in their disagreement (versus agreement) with the participant tempered confidence increases and strongly affected the selection of rumors that “made the most sense.” Results explain rumor persistence and variation, document the role of patterns of connectivity and dynamic social influence processes in unverified collective beliefs, and suggest modification of Dynamic Social Impact Theory to include belief polarization mediated by emergent “echo chambers.”

Polarization and Belief Dynamics in the Black and White Communities: An Agent-Based Network Model from the Data.

Public health care interventions-regarding vaccination, obesity, and HIV, for example-standardly take the form of information dissemination across a community. But information networks can vary importantly between different ethnic communities, as can levels of trust in information from different sources. We use data from the Greater Pittsburgh Random Household Health Survey to construct models of information networks for White and Black communities--models which reflect the degree of information contact between individuals, with degrees of trust in information from various sources correlated with positions in that social network. With simple assumptions regarding belief change and social reinforcement, we use those modeled networks to build dynamic agent-based models of how information can be expected to flow and how beliefs can be expected to change across each community. With contrasting information from governmental and religious sources, the results show importantly different dynamic patterns of belief polarization within the two communities.

REFLECTIONS ON BIASED ASSIMILATION AND BELIEF POLARIZATION

Where Taber and Lodge view belief polarization to indicate a “partisan motivation,” Lord et al. (1979) believed it to be consistent with a desire for accuracy: A “weak” study articulating an opposing viewpoint might simply sharpen participants' initial belief of the wisdom of their prior beliefs. This polarization, Taber and Lodge show, correlates with political sophistication: The more partisan a participant, the more time spent reading the opinions of the other side—in order to critically refute them. Taber and Lodge attribute such findings to people's personal identification with their beliefs. However, participants may have spent so much time reading opinions of the other side because the arguments are unfamiliar. Further, Lord et al. (1984) were able to induce people to act with “skepticism” toward their own views by simply suggesting that they “consider the other side.”

Belief-Oriented Segregation in Policy Networks

Many policy networks are characterized by belief-oriented segregation, where actors with shared belief systems are clustered together and few opportunities exist for communication across coalitions of like-minded stakeholders. This inhibits the ability of network actors to effectively learn about, and successfully manage, complex policy problems. Despite the importance of understanding why these structures emerge, the processes that generate belief polarization in networks are not well-studied. This paper derives a general agent-based model of network formation and belief change from the Advocacy Coalition Framework (ACF), a prominent theory of the policy process that has been widely applied to the study of belief conflict in contentious policy systems. Simulation results suggest that the ACF assumption of biased information processing plays a critical role in the emergence of belief-oriented segregation in networks. This model provides a starting point for re-thinking the role of cognitive bias in social and policy learning, as well as the relationships between belief change and the evolving structure of policy networks.

Disagreement, Dogmatism, and Belief Polarization

Disagreement, Dogmatism, and Belief Polarization

MISPERCEPTIONS ABOUT PERCEPTUAL BIAS

▪ Abstract Do people assimilate new information in an efficient and unbiased manner—that is, do they update prior beliefs in accordance with Bayes' rule? Or are they selective in the way that they gather and absorb new information? Although many classic studies in political science and psychology contend that people resist discordant information, more recent research has tended to call the selective perception hypothesis into question. We synthesize the literatures on biased assimilation and belief polarization using a formal model that encompasses both Bayesian and biased learning. The analysis reveals (a) the conditions under which these phenomena may be consistent with Bayesian learning, (b) the methodological inadequacy of certain research designs that fail to control for preferences or prior information, and (c) the limited support that exists for the more extreme variants of the selective perception hypothesis.

Self‐categorization theory and belief polarization among Christian believers and atheists

Intergroup comparison may strengthen the beliefs of group members. The resultant belief polarization, according to self‐categorization theory (Turner, 1987), is mediated by the perceptual extremitization of the in‐group norm. To test the theory, 51 Christian believers and 52 atheists completed a belief questionnaire twice. Firstly, they estimated the typical beliefs of the in‐group (in‐group comparison condition) or out‐group (out‐group comparison condition), or indicated their own beliefs (self condition). The effects of comparison were tested by later asking all subjects to express their own beliefs on the same questionnaire. Only the in‐group condition produced polarization. As predicted, the polarization was mediated by the extremitization of the in‐group norm among atheists. Among believers, contrary to the prediction, the polarization was associated with the moderation of the in‐group norm, suggesting that the believers were responding as individuals concerned with their personal identities rather than as group members concerned with their social identities.

Anticipatory Belief Polarization as a Function of the Expertise of a Discussion Partner

This study examined change that occurs in persons' beliefs when they anticipate being observed discussing an issue with an individual whose position on the issue opposes their own. Subjects expected to discuss an issue with a partner who held an opposing view and was much more, more, equally, or less expert on the issue. While expecting a discussion with the equally expert partner, subjects produced more arguments countering their partner's position and supporting their own position than did control subjects, who did not expect to discuss the issue. As a consequence of this preparation for the discussion, subjects expecting the equally expert partner adopted a more extreme position opposing the partner (belief polarization) than control subjects and maintained their changed beliefs after the anticipated discussion was canceled. Subjects expecting partners less expert, more expert, or much more expert than themselves did not change their beliefs. The findings of the study were assumed to be the outcome of a social comparison process motivated by subjects' self-presentational concerns.