1
Counterfactual Thinking:
A Critical Overview
Neal J. Roese
Northwestern University
James M. Olson
University of Western Ontario
If matters had fallen out differently, she wondered, might she not have met some other man? She tried to picture to herself the things that might have been—that different life, that unknown husband. He might have been handsome, intelligent, distinguished, attractive. .. .
—Flaubert (1857/1950, p. 57)
Tales of human suffering are replete with examples of thoughts of what might have been, of what could have occurred if only a few subtle details had been different. The tormented mistress of Flaubert's novel, Madame Bovary, bored with her provincial life, dreams of a better one and a better husband. The ability to imagine alternative, or counterfactual, versions of actual occurrences appears to be a pervasive, perhaps even essential, feature of our mental lives (e.g., Hofstadter, 1979, 1985).
The term counterfactual means, literally, contrary to the facts. Some focal factual outcome typically forms the point of departure for the counterfactual supposition (e.g., Madame Bovary's angst). Then, one may alter (or mutate) some factual antecedent (e.g., her marriage) and assess the consequences of that alteration. Thus, counterfactuals are frequently conditional statements and, as such, embrace both an antecedent (e.g., "If Madame Bo vary had married a better man") and a consequent ("she would have been happier"). For present purposes, we restrict our use of the term counterfactual to alternative versions of past or present outcomes, although we are aware that others have also used the term to describe future possibilities (e.g., Hoch, 1985; M. Johnson & Sherman, 1990). Thus, in the present analysis, an essential feature of a counterfactual conditional is the falsity of its antecedent, in that it specifies a prior event that did not occur (Goodman, 1947). The consequent may or may not be false; if it is, the mutation of the antecedent undoes the factual outcome.1
Experimental social psychologists have attempted to identify empirically the precursors, underlying processes, and consequences of counterfactual thinking in research programs dating back almost 20 years. Yet, fascination with counterfactual suppositions, with "possible worlds," dates back considerably further, at least as far back as the writings of classical Greek philosophers. Our purpose in this chapter is to introduce readers to the topic of counterfactual thinking as social psychologists have conceptualized it.
We frame this discussion by first considering the rich intellectual tradition dedicated to understanding the nature of counterfactual conditionals. We then review the theoretical basis of counterfactual research, examining the mechanisms underlying counterfactual reasoning. We then discuss factors that influence the generation of such thoughts and the consequences of considering them. These determinants are conceptualized in terms of outcome-based motivational factors and antecedent-based factors that may differentially influence the mere generation and the semantic content of counterfactual thoughts. Further, we note that the consequences of counterfactual generation derive from either of two distinct mechanisms: contrast effects and causal inferences. By referring explicitly to these two mechanisms in our survey of the consequences of counterfactual thinking, we provide a more rigorous conceptual integration of previously insular findings. To conclude, we discuss some conceptual issues of concern in interpreting the extant literature and in moving toward future research.
Counterfactual Thinking: Historical Roots
Philosophical Perspectives
The consideration of what might have been and what may yet be has long fascinated philosophers. Plato and Aristotle examined in their writings the epistemological status of subjunctive suppositions and unseen yet tangible ideal forms. These ideas were embraced more vigorously by the German philosopher, Leibniz, in the seventeenth century. Leibniz argued that an alternative reality (or "world") is possible to the extent that it contradicts no formal laws of logic. Thus, there could well be an infinite number of possible worlds. Leibniz's interest in possible worlds was largely theological: In formalizing proofs for the existence of God, he argued that God must have contemplated all possible worlds before creating this one, and that this one represents the best of all possible worlds (see Rescher, 1969 for a historical review).
Discussions of possible worlds met with considerable skepticism until the 20th century, when the study of modalities (i.e., nonexistent possibilities) achieved greater respect following its more rigorous elucidation via newly developed systems of symbolic logic (see Kvart, 1986; Loux, 1979; Reichenbach, 1976). Controversially, Lewis (1973b, 1986) argued that possible worlds exist to the same extent as this one: "Our actual world is only one world among others. We call it alone actual not because it differs in kind from all the rest but because it is the world we inhabit" (1973b, p. 85).2 However, more moderate voices echoed the axioms of cognitive psychology in affirming Aristotle's dictum that actuality is prior to possibility (Rescher, 1979). That is, any conception of possible worlds must be fabricated from a cognitive substrate derived from tangible experience: "Unrealized possibility ultimately roots in the mind-correlative capabilities of the real" (Rescher, 1975, p. 217).
Modern philosophical treatment of the modal nature of reality assumes some consideration of all possible worlds. Within this field, the study of counterfactuals occupies a more tightly circumscribed domain. Loux (1979) explained, "When I say that if the Blue Jays were to win the pennant, Toronto would go wild, I am not saying that in every possible world where the Blue Jays win the pennant, Toronto goes wild; for there obviously are possible worlds where the Blue Jays win the pennant and few, if any, of the citizens of Toronto find the event very interesting" (p. 33). Rather, a counterfactual typically posits one possible world that is imaginally very close to the real world, containing only a very few (or just one) features that differentiate it from this world.
Specific discourse aimed at counterfactuals per se has generated fascinating controversies, frequently focusing on the truth value of counterfactual statements (e.g., Chisholm, 1946; Goodman, 1947, 1983; Nute, 1980). As noted in our introduction, all counterfactuals are necessarily false, insofar as their antecedents refer to some state of affairs that was not so. Evaluation of the veracity of counterfactual propositions cannot, therefore, proceed simply via rules of logic; additional, inherently psychological principles of reasoning must be posited. For example, Rescher (1964) suggested that counterfactual propositions will be accepted as true to the extent that they are predicated on alterations to episodic specifics as opposed to negations of general laws. Thus, for example, counterfactuals involving deletions of someone's behavior last Tuesday will be viewed as more "truthful" than deletions of the earth's gravity. In general, the question of what constitutes true, or natural, counterfactuals (e.g., "If I had a million bucks, I could buy a new house") as opposed to more absurd statements (e.g., "If I were a million bucks, I'd be green") continues to generate controversy (e.g., Lewis, 1973a; Nute, 1980), as too does the problem of inferring the causal contingency between a counterfactual antecedent and consequent (e.g., To what extent does having a million bucks lead to buying houses?). As we discuss later, these same problems have stimulated and informed psychological theories of counterfactuals (e.g., Kahneman & Miller, 1986; Wells & Gavanski, 1989).
Psychological Perspectives
Interest among psychologists in counterfactuals is, of course, a much more recent development. Within the field of cognitive psychology, researchers have examined counterfactuals from the perspective of explicating basic processes of learning and memory. For example, Fillenbaum (1974) discovered that memory for counterfactual statements (e.g., "If he had caught the plane, he would have arrived on time") was slightly more accurate than for semantically similar causal statements (e.g., "He did not arrive on time because he did not catch the plane"), suggesting at the very least that people are adept at cognitive manipulations of counterfactual information. Although causal information is easily abstracted from counterfactual conditionals, people are well able to encode counterfactual information in memory as semantically complex conditional propositions, rather than first reducing it to simpler, basic propositions (Carpenter, 1973). Regarding the content of counterfactual thoughts, Revlis and colleagues provided experimental confirmation for Rescher's (1964) conjecture that people prefer belief-contravening propositions that involve alterations to particulars rather than universals (Revlis & Hayes, 1972; Revlis, Lipkin, & Hayes, 1971). In this way, counterfactuals maintain consistency with general laws of the world, with changes occurring mainly to narrowly episodic elements (see also Braine & O'Brian, 1991).
A notable controversy within the cognitive literature focused on counterfactual generation as a test of the Sapir-Whorf hypothesis, that is, of whether language critically constrains cognitive perceptions and representations of the world. Bloom (1981) provided evidence that expression and comprehension of counterfactual statements was substantially weaker among Chinese speakers than among American English speakers. Bloom pointed to the absence in Chinese of any syntactical or lexical equivalent of the English-language subjunctive ("if-then") form as the linguistic cause of this cultural difference. Bloom (1981, 1984) argued, therefore, that it was the availability of subjunctive linguistic structures that permitted the cognitive manipulation of might-have-been scenarios. Au (1983, 1984), however, noted that Bloom's materials may have been idiomatically biased and that Chinese speakers could reason counterfactually with little difficulty if this bias were removed (see also Liu, 1985). The current consensus seems to be that language and culture influence the content of counterfactuals and also their application in specific situations but that "most people with an adequate level of education possess the skills of hypothetical reasoning and the ability to think in a counterfactual fashion" (Markus & Kitayama, 1991, p. 234; see also Lardiere, 1992; Vorster & Schuring, 1989).
Relatedly, the work of Sternberg and Gastel suggested that the ability to reason counterfactually and to deal, more generally, with imaginal possibilities, is an important component of intelligence (Sternberg & Gastel, 1989a, 1989b). Developmental researchers have examined the age of onset of logically valid counterfactual reasoning, linking it to Piaget's fourth and final stage of cognitive development, the formal operations stage (Markovits & Vachon, 1989; Wing & Scholnick, 1986). Thus, most people are able to reach logically valid conclusions through counterfactual reasoning by the time they reach their teen years, although the ability to fabricate fictional scenarios in the context of imaginative play occurs much earlier, at around the age of 4 or 5 years.
Social Psychological Perspectives
Social psychologists seek generally to examine the global cognitive functioning of the individual within a broader, social context. The three characteristics of social psychology identified by Ross and Nisbett (1991) effectively illuminate what distinguishes sociopsychological approaches to counterfactual thinking from the approaches discussed previously. First, social psychology is predicated on situationism, such that external antecedents in social situations are viewed as prime determinants of the generation and content of counterfactual thoughts. Thus, qualitative descriptions of focal outcomes were among the first causal factors to be linked empirically by social psychologists to counterfactual thinking. Second, a focus on construal emphasizes the unique importance of individual interpretation of social situations. Thus, unlike their philosophical counterparts, social psychologists tend to ignore the objective truth value of counterfactual propositions in favor of examining their perceived plausibility and meaningfulness to the individual. Third, the goals and motivational states of individuals as they operate within a larger social system are also taken by social psychologists to be crucial determinants of counterfactual thinking. Thus, for example, aversive outcomes and thwarted plans have been shown to affect counterfactual thinking. With the social psychological approach, a high premium is also placed on the global psychological consequences of counterfactual thinking. Attributions of blame and responsibility, self-inference, emotions, coping, and adjustment are examples certifying the breadth of counterfactual effects identified by social psychologists.
In summary, the social psychological approach is perhaps the broadest and most general of the psychological approaches to counterfactual thinking, embracing the totality of an individual's feelings, perceptions, motivations, and goals within the larger context of social interaction. In the sections that follow, we examine the theories of counterfactual thinking and their supporting research explicated by social psychologists. First, we examine basic mechanisms integral to the formation of counterfactual thoughts. Next, we discuss the factors influencing counterfactual generation in terms of the determinants of their availability and semantic content. Then, we examine some of the psychological and behavioral consequences of engaging in counterfactual thought. Last, we consider directions for future research.
Basic Processes
Two theoretical questions form the heart of theory aimed at elucidating the processes underlying counterfactual thinking: What triggers counterfactual generation and, once triggered, what constrains the content of such thoughts? Norm theory has been the guiding theoretical formulation for the bulk of social psychological research on counterfactual thinking. In the next section, we first examine the process of counterfactual generation according to norm theory. We then examine the motivational underpinnings of counterfactual generation, noting how a motivational model may complement the perspective offered by norm theory. Finally, we turn to the relation between counterfactuals and causal inference.
Norm Theory
The aim of norm theory is the description of the psychology of surprise, that is, of the judgmental processes that occur in response to specific experiences and that result in post hoc inferences of surprisingness. Like other social judgment theories, norm theory focuses on the pairwise comparison between an experiential outcome and some cognitive anchor. The direction and magnitude of the discrepancy between the outcome and the anchor determine various cognitive and affective reactions. For example, a student's receiving a D on an essay when her or his standard is a B evokes negative affect (e.g., disappointme...