Abstract
Some past events incite more wonder about their causes than do others. For example, negative events require explanation more than positive events. We review social psychologists’ theoretical and empirical insights on what kinds of past events “beg explanation.” We draw on attribution theory that became popular among psychologists from the 1960s onward, on research on counterfactual reasoning, and on conversational and discursive critiques of attribution theory. We argue that factors predicting what is or is not perceived as requiring explanation are culturally and historically grounded, and that accordingly, what begs explanation varies between contexts and can change over time. Yet, drawing on the distinction between content and process, we argue that there are recognizable patterns across time and space. Specifically, we propose the relationship between events and background expectations as a rather stable predictor of what begs explanation—and as a level of analysis that can unite seemingly disparate approaches.
Introduction
Consider the following research findings: (a) in a recent global survey of representations of history, the event that was listed most often as an important “historical event” was World War II (Liu et al., 2009); (b) research on so-called “flashbulb memories” indicates that extremely detailed and vivid memories of when and how one has learned of a public event are usually produced by important, surprising, and emotionally upsetting incidents such as the assassination of US president John F. Kennedy (Luminet and Spijkerman, this issue); and (c) such unexpected, tragic, and self-relevant events also seem most likely to prompt the prolific type of explanatory thinking referred to as “conspiracy theorizing” (Van Prooijen and Douglas, this issue). As these examples illustrate, certain kinds of historical events incite more wonder about their causes than do others. We examine what social psychology can teach us about what kinds of past events “lay historians” (i.e. lay people thinking about historic events, Klein, 2013) perceive as requiring explanation.
We focus on lay explanations (rather than explanations generated by academically trained historians) because social psychology has mostly been concerned with explanations created by ordinary people. Yet, we expect that our analysis will be interesting for “professional” historians. Professional and lay understandings of the past inform and influence each other (Tosh, 2008), and historians, too, pay close attention to ordinary people’s explanations (e.g. Millstone, 2014) and to popular and public understandings and uses of the past (e.g. Gillis, 1994; Hobsbawm, 1992). We will mention several examples for possible points of connection between the works of social psychologists and historians, 1 and hope that our historian readers will see further ways in which the approaches and findings we review might inform historical research.
For social psychologists, we hope to show how engaging with history—both the psychology of history and the history of psychology—can inform and enrich social psychological research. Focusing on the psychology of explanation as an example, we show how engaging with history allows us to go beyond merely summarizing the insights gained by previous work; by critically examining how key insights were reached in the past we can carve out how original insights might be gained in the future.
Throughout this article, we will be going back and forth between discussing the history of psychology, the psychology of history, and their interaction, and expect that different readers will find different aspects most appealing. We first turn to psychologists’ attempts to understand how people explain events, specifically, to theories of “causal attribution” that captured the attention of social psychologists in the United States from the 1960s onward. However, our discussion will focus less on how people explain than on what they seek to explain in the first place. To address this question, we will discuss insights gained by attribution research as well as by later developments in the field including critiques of this classic approach. We focus on what begs explanation because, as we shall see below, what is (or is not) perceived as requiring explanation can subsequently shape how explanations are constructed. The question of what begs explanation is thus the primary one.
We will argue that while the content of explanations is culturally and historically grounded and can therefore change over time, at least some of the psychological processes underlying the perceived need for explanation will be more stable. More specifically, we propose the relationship between (historic) events and implicit background expectations as one potentially stable ingredient in the process of explaining, thus providing a level of analysis that can unite seemingly disparate approaches. Following Vendler (1967), we define events quite broadly, namely as any segments of time that can be distinguished from the surrounding “flow” of time, by virtue of (being perceived as) having an identifiable beginning and ending. We suggest that a more complete understanding of how people make sense of the past can be reached by redirecting attention from the events themselves to their taken-for-granted background aspects.
The development of attribution theory in social psychology
Most psychological work on explanations concerns causal attribution, that is, the process by which people attribute causes to events. 2 This approach departs from Fritz Heider’s (1958) attempts to understand naïve psychology. Heider analogized people as “intuitive scientists” who systematically analyze each other’s behavior in order to understand the underlying motivations and intentions as well as to account for successes and failures. While Heider’s work was mainly theoretical, his analyses were influential during a time when social psychology was increasingly being imagined, funded, and actualized in the United States as an experimental science. During the 1960s, social psychologists constructed increasingly elaborate experimental scenarios, and explanations of social behavior focused more and more on cognitions in the minds of individuals (Collier et al., 1991).
In one classic study, participants read essays for or against Fidel Castro and were asked to guess the alleged author’s true opinion (Jones and Harris, 1967). An enduring contribution of these studies was the discovery that people attribute behavior disproportionally to causes that lie within the person (such as the author’s personal opinion) and do not fully take situational pressures into account (such as the information that the author had been instructed to take a certain position). This tendency was later dubbed the “fundamental attribution error” (Ross, 1977). However, the influence of history on this work was not limited to the choice of debate topics (i.e. Fidel Castro). As social psychology became experimental, it relied increasingly on inferential statistics to interpret results. In turn, statistical tools became a heuristic for theory development (Gigerenzer, 1991), most obviously in a model of attribution that Kelley (1967) presented in analogy to the statistical procedure analysis of variance. According to that model, people systematically examine how much an observed behavior varies between individuals, contexts, and over time to decide whether this behavior is caused by the person or the situation. While not explicitly focused on attribution “biases,” this model still contributed to an enduring trend in social psychology to interpret ordinary people’s attributions for events as biased against a background normative ideal of a statistically oriented social science.
More recent work on causal attribution has moved away from this tendency to interpret lay explanations as irrational and biased; it has also moved away from the dichotomy between explanations that focus on characteristics of individuals versus the surrounding circumstances. In partial return to Heider’s original work, recent models suggest that lay explanations are more sophisticated. For example, people distinguish between intentional and unintentional behaviors, and between different types of reasons and conditions for intentional behaviors (Böhm and Pfister, 2015; Malle et al., 2000). These developments in attribution research may be useful for historians, as they, too, have examined the role of (perceived) rationality and irrationality in public discourse and representations of “the public” (e.g. Knights, 2005).
Together, research on causal attribution has led to clear insights into how people explain certain types of events when asked directly why they occurred. Some of these insights have been fruitfully applied to historical cognition (e.g. Bilewicz et al., this issue). Yet, most classic research on causal attribution is only indirectly relevant to our question of what begs explanation in history. Experiments on causal attribution have mostly addressed the question of which explanations people prefer when given a choice between different options. The question, what people perceive as requiring an explanation in the first place (whether and when they spontaneously make attributions), has received considerably less attention in attribution research. We turn to this question next.
Psychological research on what events beg explanation
Drawing on the different psychological approaches that we discuss in this article, we have synthesized an overview of characteristics of events that beg explanation (Table 1). Research on spontaneous attributions has suggested three factors that robustly predict the active construction of explanations. People spontaneously explain events that are unexpected or appear surprising; they explain negative events more than positive ones; and they tend to explain events that are self-relevant (see Weiner, 1985). Although attribution research has mostly been concerned with explanations of the behavior of individuals in everyday human interactions, these findings go some way toward understanding which historical events will prompt an explanation. Both novelty/surprise and self-relevance/importance are also important predictors of flashbulb memories (Luminet and Spijkerman, this issue), and they also align with findings of global surveys about world history (Liu et al., 2009). Wars are most often named as events of historic importance, and wars are both negative and surprising when set against the desire of almost all people to live in peace. Moreover, participants in these surveys most often mentioned recent events and events in their own countries, suggesting an influence of self-relevance.
What social psychology can tell us about when people do (or do not) feel inclined to ask why an event occurred.
Please note that these determinants are not mutually exclusive and often overlap. For example, changes, human actions, and later observations are all perceived as more “mutable” than their counterparts.
Importantly, unexpectedness is not inherent in events themselves but describes the relations between events and expectations. War is surprising if we expect people to want to live in peace, but not when we consider that violent intergroup conflicts are frequent. Accordingly, a full psychological explanation of surprisingness must account for the often silent expectations or background “norms” of how events will unfold. Kahneman and Miller (1986) developed a systematic account of the cognitive process by which such mental event norms are constructed. They argued that an event (e.g. a war) will activate memories of similar events (other wars) which will then form the normative expectancy against which the current event is compared. Differences between the current event and that background event norm, both beg explanation and seem to begin to explain how the event might have turned out differently—an example of how what begs explanation influences how explanations are constructed. For example, the Hundred Years’ War implicitly brings other wars to mind. Based on these, one forms an implicit expectation of wars’ characteristics, such as their duration. This makes the Hundred Years’ War appear abnormally long, thus prompting such questions as Why did this war last so long? Could it have ended earlier? How so?
Kahneman and Miller’s (1986) norm theory inspired considerable experimental research on counterfactual thinking (see Roese, 1997)—the tendency to think counter to the facts after an event has occurred. Rather than as “intuitive scientists” systematically analyzing behavior, these approaches saw people as mental “reconstructors” (Alicke et al., 2015) who made sense of events by imagining how things might have turned out differently. This research concluded that people are particularly likely to ponder what if (not)? and to spontaneously construct explanations when it is easy to imagine an alternative course of events (Kahneman and Tversky, 1982). For example, because it is easier to imagine that individuals could have acted differently than to imagine different laws of nature, events (perceived as) caused by human action, such as an assassination or a suicide, will beg explanation more so than comparable events (perceived as) caused by “natural causes,” such as dying of old age.
One insight of particular relevance to historical cognition is the importance of temporal order. Causes precede effects and it seems easier to imagine that a situation turned out differently (e.g. Franz Ferdinand of Austria could have survived the assassin’s shots in Sarajevo in 1914) than to imagine a different background situation (e.g. Franz Ferdinand could have stayed at home). However, even when two events are not causally related, lay people spontaneously imagine alternatives to later events more so than to earlier events. Teigen (2004) demonstrated this by presenting participants with statistical records collected in different years, for example, the number of Norwegian immigrants to the United States in 1870 and 1880. When asked what might explain the difference, participants focused their explanations overwhelmingly on the later statistics. They took the earlier statistics as background norms and asked the counterfactual question why things changed rather than stayed the same; rarely, did they ask the equally valid question why the first statistic was different from the later one(s).
Individual historical events, however, are compared against the background “flow” of time, making the counterfactual question why the event occurred at all the central one. Both psychologists (Zacks et al., 2001) and historians (White, 1973) have noted that time does not come pre-packaged as “events” but needs to be actively constructed as such. Events are what appears to stand out against a background of a more normal (less spectacular) state of affairs. Accordingly, (sudden) changes, differences, and actions suggest the presence of an event and provide occasions for explanation more than continuity, sameness, and inaction do. These observations are important because events and transitions between events can be constructed in a number of different ways. For instance, the same transition can be narrated as an ending or a beginning, as the familiar cry of succession The king is dead, long live the king exemplifies. However, we argue that beginnings are more essential for understanding why an event occurred. Accordingly, we have recently shown that when thinking about a particular event, people are more interested in learning about its beginning than its ending (Teigen et al., 2017, see also Zerubavel’s work on the significance of origin myths, e.g. Zerubavel, 1993).
Understanding event norms as cognitions that are constantly constructed and re-constructed also means that events can change those background norms. Indeed, the most surprising historical events are those—often “firsts”—that irrevocably change our assumptions about what we can take for granted. For example, women’s right to vote may have seemed unthinkable at the beginning of Western democracies and its introduction meant a remarkable change for contemporaries. Yet, it is now so taken for granted that people rarely ask how women’s right to vote came about—and consequently often fail to recognize the role of feminist movements in achieving these rights (Vernet and Butera, 2005). Accordingly, one central goal of gender history has been to counteract such forgetting by illuminating women’s agency in the past (see J. W. Scott, 1999). Such changes in background norms may also help to explain the hindsight bias, that is, the inclination, after an event has occurred, to perceive its outcomes as more predictable than they could have been beforehand (see Klein et al., this issue).
In sum, research on counterfactual thinking seems more directly relevant to the question of what begs explanation in history than does classic attribution research. In fact, counterfactual thinking is a common, albeit controversial, approach to understanding history (Tetlock and Lebow, 2001). It allows both lay historians and experts to approach the causal question for events that are unique and not amenable to experimentation. Historians have been critical of the conflation of historical counterfactual thinking with causal thinking (Fischer, 1970). Psychologists have also shown that the question of why an event happened and the question of how an event might be undone are more distinct than Kahneman and Miller (1986) originally suggested (Mandel and Lehman, 1996). Most importantly though, norm theory highlights that the answer to the question “what begs explanation” has many features, that this answer lies in the relationship between the background norm and the event itself, and that different background norms can be constructed by the same event. For example, women’s suffrage may inspire the question of why it came about so late or the opposing question of how it was possible at all, depending on the context in which the question is asked and on what background expectations are most available in that context. We now turn to later developments in the field of attribution research that were informed by this approach.
Conversational and discursive approaches to attribution
In the late 1980s, critics of classic attribution models (in the United States and elsewhere) suggested that explanations are best understood not by analyzing cognitive processes in the minds of individuals, but also by understanding explanations as taking place in, and taking the form of, conversation (see Hilton, 1990). Explanations are thus embedded in interpersonal relationships and guided by rules of discourse. Conversation partners typically rely on a “common ground” of shared knowledge (Clark, 1996), that is, information that all parties are assumed to know or see as self-evident. Explanations will typically not evoke aspects presumably known by all parties involved in the conversation, but rather, it will focus on “new” information that goes beyond this common ground. For example, explanations of why the Cuban Missile Crisis did not result in a war will not focus on the observation that a war between countries with nuclear weapons is undesirable—not because that fact is considered irrelevant, but because it is considered common ground and thus goes without saying.
Understanding explanations as conversations also highlights that explanations have consequences. Explaining something or allowing it to “go without saying” not only engages the common ground; it also reinforces the respective shared ideas or builds new common ground (Kashima et al., 2007). That is, explaining things can give them a reality that they did not previously possess. Anderson et al. (1980) demonstrated this by presenting their participants with bogus information about either a positive or a negative association between risk-seeking and effectiveness as a firefighter that participants were then asked to explain. Later, some participants were informed that the data were fake and that they had been randomly assigned to read about either a positive or a negative association. Yet, when these participants were subsequently asked what they thought the true relationship was, they still endorsed the relationship that they had explained. Thus, the mere act of explaining (vs not explaining) created a subjective reality that was subsequently difficult to change—a process that occurred regardless of the specific content of the explanation. In a second experiment, merely reading about a positive or negative association (without being asked to explain it) had a similar, but weaker belief-creating effect. A similar effect has been observed by historians of the early modern period in Protestant England. Once political crises were explained as something triggered by Catholics, governments and the press continued to characterize subsequent crises as plotted by the same minority (Lake, 1989).
An even more radical critique of classic attribution theories than this conversational approach was issued by discursive psychologists. They argued that the main reason why people engage in explaining is not to (individually or conversationally) make sense of events; instead, explanation is a social action, and is oriented toward doing such things as blaming, shifting responsibility, apologizing, and claiming ownership (Edwards and Potter, 1993). For example, explaining the Nazi Party’s success in Germany in the 1930s with the conditions of the Versailles Treaty can be an attempt at understanding the Nazis’ rise to power, an excuse for the allure of Nazi rhetoric for contemporary Germans, an attribution of (partial) blame to the Allied Powers, or all of the above, depending on context. More generally, in any engagement with the past in which individual or group interests are at stake and whenever (collective and/or individual) emotions are involved, explanations of events are “not a neutral business of making the best sense of experience” (Edwards and Potter, 1993: 24–25), an aspect of historical explanation especially conspicuous in modern histories of former European colonies (Woolf, 2011). Discursive psychologists argue that explanations are inextricably linked to social dynamics and relations between groups; they are actions with consequences, located in social contexts. Hilton et al. (1996) provide an example of such consequences of different lay explanations of past events for intergroup relations in the present. Their participants in the UK and France were more supportive of European unification the more they attributed Hitler’s rise to power to the conditions of the Versailles Treaty—and less supportive to the extent that they explained it via more “internal” attributes of Germany such as a lack of a democratic tradition before World War I. Similarly, in Germany, the question of how to explain the Nazis’ rise to power and/or the Holocaust—asked by historians as well as by lay people—has repeatedly caused fierce public and political disputes, most prominently in the so-called Historikerstreit (the “historians’ dispute”) in the 1980s (Berger, 1997).
Applying a discursive framework to the question of what begs explanation in history, explaining should happen whenever individual or group interests are at stake. This may sound very similar to classic attribution approaches’ insight that self-relevance triggers attribution. Yet, a discursive perspective not only suggests a more contextual answer to this question, but foregrounds that explanations are not just reflective of cognition, but are also performative. As speech acts, explanations can be made strategically by historians, politicians, or anybody else having a stake in present-day representations of the past, as they can serve political agendas in the present. This might almost sound like a truism to historians who (together with colleagues in other social sciences and humanities) have long embraced the idea that language is performative (see Woolf, 2011). Psychologists have, however, been a bit slower to pick up this idea, illustrating differences in “common ground” between disciplines.
One example of the performative aspect of what is explained and what is considered common ground are explanations of group differences and related power dynamics. Foucault’s (1995 [1975]) description of the emergence of disciplinary power in Western Europe illustrates a form of power generated by academic disciplines such as psychiatry and criminology that disempowers people such as mentally ill or criminal individuals by making them particularly visible and by simultaneously making those exercising this power invisible. In line with this argument, social psychological research on the explanation of group differences consistently shows that people—lay persons as well as psychologists—tend to focus explanations of group differences (e.g. gender differences) on the behavior of historically lower-status groups (e.g. women), not on the behavior of historically higher-status groups (e.g. men). These habits of explaining reinforce the behavior of higher-status groups as the (invisible) standard that goes without saying. Accordingly, research shows that such explanations have consequences, for example, for the status and power ascribed to the respective groups, for stereotypes, and for self-perceptions of members of stigmatized groups (Bruckmüller, 2013; Hegarty and Bruckmüller, 2013). These findings are important to psychologists because they make us uncomfortably aware of political implications of our seemingly neutral explanations of human behavior. They are important to historians because they warn that explanations can themselves influence history. Historians are aware that limiting accounts of history to the role of those in power could reinforce existing power relationships (J.C. Scott, 1990). Yet, our findings on the explanation of group differences warn that particularizing historical analyses that focus on subordinate groups as special (e.g. “women’s history,” “indigenous history”) while deeming the history of powerful groups simply “history,” will not be enough to challenge existing power relationships (see also J. W. Scott, 1999, preface). It might even subtly perpetuate them.
What social psychologists can learn by engaging with history
We set out to examine what social psychology can teach us about what, in history, seems to beg explanation. Along the way, we have narrated a brief history of attribution research. Some contemporary psychologists speculated that future historians of psychology would deem the 1970s the decade of attribution theory. Yet, present-day social psychologists also remember the 1970s as a decade of crisis (Faye, 2012). Social psychology’s social relevance was called into question, as its experiments seemed increasingly removed from real-world concerns—including movements for peace, civil rights, and women’s rights. Critics emphasized that the field failed to conceptualize how its answers to questions were embedded within historically particular frames of reference (Gergen, 1973), that those answers applied only to North American cultures (Israel and Tajfel, 1972), and that experiments were often unethical and coercive (see Collier et al., 1991). With the benefit of hindsight, we might speculate why this particular version of attribution theory emerged from social psychology’s crisis. Unlike the previous generation of social psychologists who aimed at historical levels of explanation—most notoriously, Milgram (1963) who attempted to understand the Holocaust through experiments on obedience—attribution researchers rarely attended to historical frames of reference; paper and pencil experiments raise few ethical risks and can be conducted in less well-funded times and places. As such, attribution research allowed social psychologists to go forward with a less controversial experimental science that still engaged how people made sense of events.
This work led to some clear answers to the question of what is most likely to beg explanation (Table 1). However, our table could also be read as a recipe for innovative explanations of events. As intimated above, the factors predicting what is and is not seen as requiring explanation are culturally and historically grounded, and accordingly, what begs explanation varies between contexts and can change over time. For example, while contemporaries of the disastrous 1755 earthquake in Portugal spent considerable time and effort debating its causes (specifically, whether it should be explained by natural causes or the wrath of God, Udidas, 2009), we might now simply accept that such natural disasters happen and have less interest in explaining the particulars of how this earthquake came about. Referring to the determinants summarized in Table 1, we might explain this change by differences in the extent to which people then and now perceive(d) earthquakes as caused by (super)human agency and/or by the high stakes that only contemporaries had in the implications of different explanations.
This example again illustrates the importance of distinguishing between content and process. The content of explanations—both which particular events are perceived as requiring explanation and the actual content of these explanations—may change over time. Yet, we would expect at least some of the processes that prompt explanations, such as surprise, emotional involvement, or power dynamics, to stay relatively stable over time (see Table 1, third column). We further propose that such stable processes include the psychological relationship between events and underlying implicit assumptions of what the normal state of affairs is. We further suggest a figure-background metaphor as a useful approach to understanding how people make sense of the past. A more complete and counter-intuitive understanding of any event may be reached by historians, social psychologists, or anybody else by producing figure-ground reversals and by asking questions about the non-eventful background aspects of a situation (e.g. incidents of nations living in peace that are rarely addressed by historians, see Mahatma Gandhi as cited in Woolf, 2011: 497).
In support of this claim, we note that historically, attribution theory has often advanced most obviously—at least in retrospect—when foundational assumptions that were part of the taken-for-granted background have been challenged, that is, when they became focal “figures” in critical analysis. The “fundamental attribution error” became part of the common ground early on (Ross, 1977). However, the rise of new (probabilistic rather than frequentist) statistical procedures inspired a new look at classic data leading to the conclusion that the extent and generality of this attribution error had been overstated (Morris and Larrick, 1995). Cross-cultural research demonstrated that this attributional style was culturally particular, and social and cultural psychologists have since debated the extent to which attributional patterns differ by culture and beg culturally particular or more universal explanations (e.g. Norenzayan and Nisbett, 2000). Conversational and discursive approaches have led to a re-thinking of the common ground of social psychological theory by focusing on how experimenters and participants actively construct meaning in experimental contexts. Another example of psychologists challenging the common ground of psychological science are attempts at decolonizing psychology, that is, at critically examining the role a Euro-American-centered psychological science may play in maintaining the power of a privileged minority of people in affluent societies over a disadvantaged global majority (Adams et al., 2015). One way to do so is to make visible and thereby challenge implicit Euro-American-centric assumptions of normality in psychology and to preferentially side with global majority perspectives in formulating theory (see e.g. Adams et al., 2012).
However, while experimental attribution theory’s search for universal laws can only take us so far in accounting for what begs explanation, contextual discursive studies of explanation run the risk of losing sight of such consistent habits of explanation as those summarized in Table 1. Discourse analysts have been particularly hostile to cognitive explanations in psychology, preferring to re-locate traditional topics of social psychology, such as attitudes, stereotypes, and the self, in discourse that is culturally and historically specific (Potter and Wetherell, 1987). Yet, at an ontological level, discursive and conversational approaches both require an understanding of how people hold in mind (or not) what the common ground of a conversation is, and what their stake in an argument might be. We suggest that a level of analysis akin to that posited by Kahneman and Miller (1986) that lies between event perception and norms constructed from memory traces might unify understanding of what begs explanation, where, why, and for whom. Because such norms are collectively organized, and can change in irreversible ways, they are in part historical entities. Developing this level of analysis will hence require the continued collaboration of social psychologists and historians.
Historians are of course not simply interested in what begs explanation, but in how explanations are stitched into particular narratives that organize and give meaning to experience. White (1973) proposed that history is best thought of not as an objective collection of facts but as a narrative that is selectively employed by different agents for different agendas. Psychologists have already started to make connections with these insights (e.g. Hilton and Liu, 2017; Liu and Hilton, 2005). White further urged the writing of ironic histories in which different narratives clash uncomfortably keeping alive our questions of what the past has been. We hope to have followed this advice and to ironically have shown that if we want to find out what begs explanation in history, we need to realize how history itself shapes what begs explanation.
Footnotes
Acknowledgements
This work was conducted within the framework of COST Action IS1205 “Social psychological dynamics of historical representations in the enlarged European Union.”