Processing Conditional Relations As Biconditionals: Some Poor Consequences and Rich Opportunities

Thomas K. Srull, University of Illinois at Urbana-Champaign
ABSTRACT - When A implies B, it does not follow that B implies A; yet people usually assume that it must. One thesis of the present paper is that processing conditional relations as biconditionals is logically fallacious, but psychologically irresistible. It is argued that this tendency can account for both the counterfactual reasoning results reported by Folkes and Lassar and the construction-of-reality results reported by Shrum, O'Guinn, Semenik, and Faber. More generally, the case is made that processing conditional relations as b conditionals is at least partially responsible for the following heuristics: availability, representativeness, time estimation, mental simulation, beliefs in a just work, similarity and attraction, and salience and perceptions of influence.
[ to cite ]:
Thomas K. Srull (1991) ,"Processing Conditional Relations As Biconditionals: Some Poor Consequences and Rich Opportunities", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 764-767.

Advances in Consumer Research Volume 18, 1991      Pages 764-767


Thomas K. Srull, University of Illinois at Urbana-Champaign


When A implies B, it does not follow that B implies A; yet people usually assume that it must. One thesis of the present paper is that processing conditional relations as biconditionals is logically fallacious, but psychologically irresistible. It is argued that this tendency can account for both the counterfactual reasoning results reported by Folkes and Lassar and the construction-of-reality results reported by Shrum, O'Guinn, Semenik, and Faber. More generally, the case is made that processing conditional relations as b conditionals is at least partially responsible for the following heuristics: availability, representativeness, time estimation, mental simulation, beliefs in a just work, similarity and attraction, and salience and perceptions of influence.

My fascination with cognitive psychology just never seems to go away. It is an insightful and quickly advancing discipline, and I believe it has much to offer to consumer researchers of all disciplinary backgrounds and theoretical perspectives.

Contemporary cognitive psychology continues to be dominated by the information processing model. People are said to represent the present in perception, the past in memory, and the future in goals, plans, and intentions. To put it another way, people are said to represent the subjective reality of the present in perception, the subjective reality of the past in memory, and the subjective reality of the future in goals, plans, and intentions.

Among other things, the preceding papers illuminate three points that cognitive psychologists occasionally acknowledge, but rarely address explicitly. First, representations of reality are often constructed, sometimes haphazardly, and rarely stored as coherent entities that are simply retrieved and activated into consciousness. The system is flexible and adaptable, and the software is just as important as the hardware.

The second point is that a representation of reality is constructed for a purpose. In other words, there are functional determinants of information processing. In the present context, representations of reality are constructed so that one can reason about them. It is clear that such reasoning can occur both forward and backward in time. In the "forward" case people ask themselves such questions as: what would it be like to get married and have children, what will my colleague say when I tell her that I lost the manuscript, or would I enjoy working on the client side? In the "backward" case, people ask such questions as: would I have been happier if I hadn't passed up that fellowship, would my colleague be friendlier if I already finished that last draft, or what could I have bought with the money if I hadn't lost it at the race track? As one can see, however, the boundary between backward and forward processing sometimes becomes blurry.

The third point flows naturally from the fact that people reason both forward and backward in time. Specifically, once a representation of reality is constructed to reason-with, it is often compared to imagined alternatives. These alternative constructions of reality may be placed in the past (what if he hadn't said that), the future (what will I do if he says this), or the present (what does it mean if he is mad rather than just busy).

To see the value of an information processing perspective, let us begin by considering two basic effects that have been reported in the preceding papers. Then I will attempt to take a little closer look at each of them.

Folkes and Lassar have reported some very intriguing effects counterfactual reasoning. Their basic effect is that consumers experience greater anger when a product breaks down shortly after a warranty expires. Their interpretation is that such an event makes it very easy to engage in counterfactual reasoning. Specifically, because the product failure occurred so close in time to the expiration date of the warranty, it is (relatively) easy to imagine oneself using the product in such a way that the failure would have occurred while it was still covered by the warranty. They have provided an important and ingenious way to examine counterfactual reasoning within the domain of consumer behavior and experience. Counterfactual reasoning is a special case of Kahneman and Tversky's (1982) simulation heuristic in which reactions to an event are determined by the way in which people construct different ways in which the event might have played out.

Shrum, O'Guinn, Semenik, and Faber have also examined the construction of reality. But they have used a different domain and a very different set of methodologies. Their basic effect is that estimates of frequency or the probability of various key events is influenced by the amount of television to which one is exposed. More specifically, events that are often depicted on television receive from heavy viewers higher estimates of frequency for the general population. Their interpretation is that this is based on an availability heuristic. Because heavy viewers of television find it very easy to retrieve specific instances of key events (such as having a maid), they give higher estimates of frequency than do light viewers of television.

Although these are quite distinct streams of research, and both the phenomena and interpretations vary considerably, I believe there is a set of very basic psychological mechanisms that plays a role in each of them. Moreover, these mechanisms are so basic and so ubiquitous that they play a role in many other areas of consumer research as well. To the degree my claims are true, they allow (and, I would argue, demand) the construction of more general theories of consumer behavior, ones that will bring coherence to what have historically been considered quite disparate areas of research. In the following section, I will try to make the case that we are much closer to this state of affairs than generally recognized.


Many judgements and inferences, especially those made in the context of consumer behavior, are made in terms of conditionals. When one states, for example, that Japanese cars are better than American cars, this is just a shorthand way of saying that the probability that a car is high in quality is greater if it was built in Japan than if it was built in the United States. The same is true whenever we infer that an object has a particular feature (e.g., low price) because it belongs to a particular category (e.g., country-of-origin).

The likelihood (or conditional probability) that an object belongs to a category if it has a particular feature is not logically the same as the likelihood (or conditional probability) that the object has the feature if it belongs to a particular category. Generating a few extreme examples that have very small base rates is the easiest way to see the fallacy in such logic. For example, the probability that one is rich given that one drives a Maserati is quite high. At the same time, however, the probability that one drives a Maserati given that one is rich is quite low. Similarly, the probability that one is a man given that one smokes cigars is high; the probability that one smokes cigars given that one is a man is low.

Although logically such inferences are not justified, psychologically they appear to be compelling. I believe that conditional relations are typically processed as biconditionals, even though they should not be from a rational perspective. In other words, people believe that if A implies B, then B must imply A as well. This very basic tendency can, in my view, account for several of the effects reported in the preceding papers, as well as for most examples of heuristic processing that have been portrayed in the literature. Heuristics, of course, are generalized rules that people use to simplify very difficult judgments. Such rules do not always produce correct judgments, but they require far less extensive processing than would be needed to arrive at a judgment through more algorithmic means. A few examples of how many of the heuristics that have been reported result from processing conditional relations as biconditionals will help to illustrate just how important these low-level cognitive mechanisms are in everyday life.


The results reported by Shrum and his colleagues ultimately rest upon the use of an availability heuristic. Most people believe that if a particular event occurs frequently, it will be easy to remember. Based on this belief, they assume that if something is easy to remember, it is likely to have occurred frequently. Consequently, they infer the frequency with which events have occurred from the ease with which instances of them come to mind. This rule, denoted by Tversky and Kahneman (1973) as the availability heuristic, has been demonstrated in numerous areas of research (for a review, see Sherman & Corty, 1984). For example, people infer that more English words begin with the letter k than have k as the third letter, although the reverse is actually true. This occurs because it is easier to generate words that begin with k. To use an example that is closer to the effects of the mass media, 80% of subjects infer that death is more likely to occur from an accident than from a stroke. In reality, however, strokes cause far more deaths than do accidents (Lichtenstein, Slovic, Fischoff, Layman, & Combs, 1978). Consistent with the work of Shrum et al., however, accidents are often written up in very vivid detail in newspapers and are portrayed more often on television. Thus, instances 0 of them come to mind more easily.


People often believe that members of a given category have particular attributes. Consequently they often infer that objects that have the attribute must belong to the category. In one study, for example, subjects were randomly assigned to play the roles of quizmaster and contestant in a game. The quizmaster was asked to generate questions that would stump the contestant and, because each person has available a huge storehouse of rather esoteric information, this was easily done. Afterwards, both observers and the participants themselves rated the quizmaster as smarter. Because people believe that smart people are able to stump others, they also inferred that one who stumps others must be smart. Similarly, because dull people are believed to be incapable of answering many questions, people infer that one who cannot answer questions is likely to be dull (see Kahneman & Tversky, 1971 for other examples).


People typically believe that events are easier to remember in detail if they have occurred more recently. Consequently, they often infer that an event has occurred more recently if they can recall a lot about it. The use of this heuristic was demonstrated by Brown, Rips, and Shevell (1985) who found that subjects' judgments of the temporal order of events depicted by the mass media (e.g., the death of John Lennon vs. the eruption of Mt. St. Helens) were predictable from the relative amounts of knowledge they could retrieve about the events in question.


The results reported by Folkes and Lassar appear to be due to some type of mental simulation process. People generally believe that events are easier to imagine if they are likely to occur than if they are improbable. Consequently, they infer that events are more likely to occur if they are easy to imagine. Many demonstrations of the use of this simulation heuristic have been reported by Kahneman and Tversky (1982). It was found in one study, for example, that people believe a disease with easy-to-imagine symptoms is more likely to affect them than an equally prevalent disease with difficult-to-imagine symptoms (Sherman, Cialdini, Schwartzman, & Reynolds, 1982).

The effects of generating explanations of an event on predictions of its occurrence provide an additional example. In several studies, for instance, subjects were asked to explain a hypothetical event involving either themselves or another, and then to predict the actual likelihood of its occurrence. The process of generating an explanation for the hypothetical event presumably made the occurrence of the event easier for subjects to imagine, and this increased their predictions of its likelihood (Ross, Lepper, Strack, & Steinmetz, 1977; Sherman, Skov, Hervitz, & Stock, 1981).

Folkes and Lassar have extended this principle by further assuming that people will be more upset by the occurrence of an undesirable event if they can easily imagine how it might have been avoided (making it seem more likely that it could have been avoided) than if they cannot. This also appears to be a general principle. In another study, for example, subjects were asked to consider two persons who are caught in a traffic jam and arrive at the airport thirty minutes late for their scheduled departure. One person finds that his Right left on time, while the other learns that his flight was delayed by twenty-five minutes, and only left five minutes beforehand. Which man is more upset at having missed his flight? It is easier to imagine how five minutes could have been saved (by not stopping to buy a paper, by not having a second cup of coffee, etc.) than to imagine how thirty minutes could have been saved. Consequently, the second passenger is more likely to believe that he could have made his Sight, and so he is more disappointed by failing to do so (Kahneman & Tversky, 1982). In an analysis that has much to do with issues such as product usage, Gleicher, Kost, Baker, Strathman, Richman, and Sherman (1990) have recently examined the separate effects of imagining alternative actions (e.g., performing a maintenance procedure) as opposed to inactions (e.g., not performing a maintenance procedure). Once again, however, the use of the simulation heuristic ultimately depends on processing conditional relations as biconditionals.


As a result of personal experience, parental upbringing, reading Horatio Alger, or religious training, people often believe that persons will be punished or experience adversity if they are bad, but that they will experience joy and success if they are virtuous. As a result, there is a tendency to infer that people are bad if something terrible happens to them but are virtuous if they experience joy and success. This is the root of blaming the victim and the principle was first proposed by Lerner and Simmons (1966) as the just world" hypothesis. As an example, subjects who witness someone experience pain (for reasons that are ostensibly beyond the person's control) evaluate the person more negatively if they believe that the pain will continue than if they believe the pain will be terminated or offset by a positive experience (Lerner & Simmons, 1966). More generally, subjects' disparagement of a sufferer is directly proportional to the amount of pain they believe he or she must endure.


People usually learn from experience that individuals like one another if they have similar attitudes or interests. Treating this conditional relation as a biconditional, they infer that people have similar attitudes or interests if they like one another. The reverse is also true. Because people believe that individuals dislike one another if they have dissimilar attitudes, they infer that people must have dissimilar attitudes if they dislike each other (see Insko, 1984; Rosenbaum, 1986; Wyer, 1974).


Finally, people discover from experience that they pay more attention to participants in a social interaction, and have better memory for what they say and look like, if the participants are influential, or if they otherwise dominate the interaction, than if they do not. As a result, people infer that a participant has been more dominant of influential if they recall having paid more attention to that participant, or if they have a more vivid memory of what the person looked like. Experiments using quite different paradigms have supported this conclusion.- For example, in a study by Taylor, Fiske, Close, Anderson, and Ruderman (1974), subjects observed a group discussion in which one person was rendered unique by virtue of his or her gender, race, or clothing. The unique participant was later judged to be more influential in the group discussion than persons who were similar to one another in these respects. Other studies show that subjects attribute more responsibility in conversations to people who are in their real or imagined line of vision as they observe the F conversation (Regan % Totten, 1975; Storms, 1973). These studies converge on the conclusion that salient aspects of a person, which may have y nothing to do with the person's actual responsibility L for what occurred in a situation, are likely to capture S subjects' attention and, therefore, to be contained in the representation of the situation that is stored in memory. Consequently, the person is likely to be attributed influence and responsibility for what went on in the situation when this representation is retrieved and used as a basis for judgments.


I have tried to show in this paper that the processing of conditional relations as biconditionals is a ubiquitous phenomenon. There is a difference between logic and psychologic and, as a result of such processing, many erroneous judgments and inferences are made. I also believe, however, that the information processing model offers the most useful perspective for understanding the causes, correlates, and consequences of such processing. In all of these areas, work has just begun.

Consumer behavior is obviously a domain in which conditional relations play a central role. Consumers actively infer one attribute from another, and understanding how and when they do so is critical for a more complete understanding of both prepurchase and postpurchase activities. Very little is understood about these matters however. I have tried in the present paper to place this issue in the larger context of information processing, and I hope that my comments will further stimulate researchers to examine the cognitive psychology of consumer behavior and consumer experience.


Brown, N.R., Rips, L.J., & Shevell, S.K. (1985). The subjective dates of natural events in very-long-term memory. Cognitive Psychology, 17, 139- 177.

Gleicher, F., Kost, K.S., Baker, S.M., Strathman, A.J., Richman, S.A., & Sherman, S.J. (1990). The role of counterfactual thinking in judgments of affect. Personality and Social Psychology Bulletin, 16, 284-295.

Insko, C.A. (1984). Balance theory, the Jordan paradigm, and the Wiest tetrahedron. In L Berkowitz (Ed.), Advances in experimental social psychology (Vol. 18). New York: Academic Press.

Kahneman, D., & Tversky, A. (1971). Subjective probability: A judgment of representativeness. Cognitive Psychology, 3, 430-454.

Kahneman, D., & Tversky, A. (1982). The simulation heuristic. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgments under uncertainty: Heuristics and biases. New York: Cambridge University Press.

Lerner, M.J., & Simmons, C.H. (1966). Observers' reaction to the "innocent victim": Compassion or rejection? Journal of Personality and Social Psychology, 4, 203-210.

Lichtenstein, S., Slovic, P., Fischoff, B., Layman, M., & Combs, B. (1978). Judged frequency of lethal events. Journal of Experimental Psychology: Human Learning and Memory, 4, 551 -578.

Regan, D., & Totten, J. (1975). Empathy and attribution: Turning observers into actors. Journal of Personality and Social Psychology, 32, 850-856.

Rosenbaum, M.E. (1986). The repulsion hypotheses: On the nondevelopment of relationships. Journal of Personality and Social Psychology, 51, 1156- 1166.

Ross, L., Lepper, M.R., Strack, P., & Steinmetz, J. (1977). Social explanation and social expectation: Effects of real and hypothetical explanations on subjective likelihood. Journal of Personality and Social Psychology,35, 817-829.

Sherman, S.J., Cialdini, R.B., Schwartzman, D.F., & Reynolds, K.D. (1982). Imagining can heighten or lower the perceived likelihood of contracting a disease: The mediating effect of ease of imagery. Unpublished manuscript, Arizona State University, Tempe, AZ.

Sherman, S.J., & Corty, E. (1984). Cognitive heuristics. In R.S. Wyer & T.K. Srull (Eds.), Handbook of social cognition (Vol. 1), (pp. 189286). Hillsdale, N.J.: Erlbaum.

Sherman, S.J., Skov, R.B., Hervitz, E.F., & Stock, C.B. (1981). The effects of explaining hypothetical future events: Prom possibility to probability to actuality and beyond. Journal of Experimental Social Psychology, 17, 142-158.

Storms, M. (1973). Videotape and the attribution process: Reversing-actors' and observers' point of View, Journal of Personality and Social Psychology, 27, 165-175.

Taylor, S.E., Fiske,>A.T., Close, M.M., Anderson, C.E., & Ruderman, A. (1974). Solo status as a psychological variable. Unpublished manuscript, University of California, Los Angeles, CA.

Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and availability. Cognitive Psychology, 5, 207-323.

Wyer, R.S. (1974). Cognitive organization and change: An information processing approach. Potomac, Md.: Erlbaum.