JENNIE BROWN: RESEARCH STATEMENT

My general research interest is in social cognition, particularly attribution, attitudes and the self concept. More specifically, I am interested in affect as an underlying process for attribution and as a determinant of attitudes and behavior. Also, I am interested in how self-cognitions are organized. Additionally, I am interested in philosophical issues pertaining to science and more specifically to psychology. Some of my research examines the philosophical arguments of Aristotle and Kant regarding what people believe and includes testing these arguments empirically.

In this area of research, I have extended attribution theory to include morality, affect and evolutionary theory. The fundamental attribution error contends that observers make trait attributions (believe that an observed behavior is indicative of the actor’s character) in response to an actor’s behavior and underestimate or ignore the information about the situation and other factors. Subsequently, researchers have discovered that the type of behavior determines whether it will lead to a trait attribution – some behaviors lead to trait attributions and some do not. Additionally, for some behaviors situational influences are taken into account. This finding supports the arguments of Kant (1797/1991) and Aristotle (c. 330 B.C.[19821]) that some morality related behaviors should never be performed (Kant’s perfect duties), while others can be performed occasionally (Kant’s imperfect duties). Some examples of perfect duties are the obligations to be honest and loyal, and some examples of imperfect duties are the obligations to be kind and charitable. Although, Kant’s and Aristotle’s arguments were philosophical (not psychological), these arguments are statements about what people believe and can be tested empirically. Indeed, people do make the distinction between perfect duty violations and imperfect duty violations (Brown, Trafimow & Gregory, 2004, Brown & Trafimow, in review, Brown, Trafmow, Chadwick & Berry, in submission). Perfect duty violations lead observers to make strong trait attributions and are not thought to be mitigated by transient factors (such as the situation, actor’s mood, luck, effort, etc.). Imperfect duty violations do not lead to strong trait attributions and are believed to be influenced more transient factors.

There are still several unanswered questions in the area of attribution research. One such question is why people distinguish between perfect duty violations and imperfect duty violations as well as make distinctions between different types of ability behaviors.[1] In Brown, Trafimow and Gregory (2004), I found that affect was related to participants’ attributions and it seems likely that this affective response may be the result of evolution shaping our attribution processes. Quite possibly it was more adaptive in our ancestral environment to avoid those individuals whose behavior was the most harmful/damaging and to appreciate high ability in others. Affect may be a quick and cheap (in terms of cognitive resources) way to do this.

In one experiment, I tested six hypotheses that were developed by applying evolutionary theory and affect to the attribution process (Brown, Trafmow, Chadwick & Berry, in review). The first and second hypotheses addressed the role of affect in ability attributions and the relationship between ability and morality attributions. The first hypothesis proposed that affect is also involved in the attribution process since is an important determinant in morality attributions. The second hypothesis predicted that morality related behaviors would be more important than ability related behaviors. In our ancestral past when resources were scarce it would be very important to know whether or not an individual could be trusted, while knowing whether or not someone possessed high ability may not be so important. Additionally, if someone is dishonest or disloyal, his behavior targets other individuals, while ability behaviors do not (unless there is some moral implication). The data supported both hypotheses; affect is involved in ability attributions and participants made stronger attribution and affect ratings for morality related behavior than for ability related behavior. This was the first time affect was examined for ability attributions and the first time that ability and morality attributions were directly compared.

Hypotheses three and four tested participants’ memory for the behaviors, and reaction times for attributions. Hypothesis three predicted that if affect was a “short-cut” by which to interpret others’ behavior, then those behaviors that caused the strongest affective reactions would have shorter attribution reaction times. This was the first time reaction times were measured in an attribution experiment and the hypothesis was supported! Perfect duty violations and diagnostic ability behaviors received faster attribution reaction times than the other types of behaviors. In hypothesis four, I predicted that participants would recall diagnostic ability behaviors and perfect duty violations better than the other types of behavior. Unfortunately this hypothesis was not supported. However, in my experimental design there were three more non-diagnostic behaviors than diagnostic behaviors and participants had a total of 32 behaviors for which to make attribution and affect ratings. Thus it may be that there was just too much information. This issue of memory for diagnostic behavior is something I would like to investigate further.

The fifth and sixth hypotheses made predictions about attribution differences based on the gender of the actor (not participant) and this was the first time that gender was manipulated in an attribution paradigm. I argued that females would be judged more harshly than males for perfect duty violations. In our ancestral environment, a male depended on his female partner’s honesty and loyalty to ensure that he were not raising another male’s offspring. Thus morality would be very important in females. This hypothesis was supported – females were judged much more harshly than males for perfect duty violations. I also predicted that ability would be more important for males, as females depended on their male partners’ ability to provide resources and safety for the female and her offspring. This hypothesis was not supported. But it may be that there was no difference because ability is so much less important than morality or maybe the ability behaviors needed to be closer to the concept of resourcefulness.

In another line of research, I tested the assumptions that attributions in response to perfect duty violations always lead to strong trait attributions and are expected to be exhibited in future situations (Brown, Trafimow & Gregory, 2004). I hypothesized that if the situation was extreme enough or created enough positive affect (or positive feelings in the observers), then it may not be believed to be indicative of the actor’s character (it will not lead to a strong trait attribution. I tested the effects of five different situations on attributions and generalizations (expectancies about future behavior). These five situations differed in extremity. Participants were informed that an actor behaved dishonestly to a friend in order to: 1. no reason (control condition), 2. in order to gain a million dollars, 3. in order to avoid losing a million dollars, 4. to save his own life or 5. to save someone else’s life. There was no significant difference between the first four conditions; participants made strong negative trait attributions and believed that the actor would engage in the same type of behavior in the future. Basically, participants believed that a person who was dishonest, even to save his own life or gain or avoid losing a great deal of money, was still dishonest! Apparently, participants have high expectations for the exhibition of moral behavior. However, participants thought about the behavior very differently when the actor engaged in a dishonest behavior to save someone else’s life. Participants made the opposite attribution when the actor was dishonest to save someone else’s life. They believed that the actor was honest. And even though they had only observed one dishonest behavior, they expected the actor to engage in honest behavior in the future. Finally, it appears that this phenomenon is highly related to the affect that the observer experiences in response to the behavior. Participants felt a great deal of positive affect in response to the actor being dishonest to save someone else’s life, but not in the other conditions. Thus, it appears that affect influenced participants’ trait attributions and generalizations. Subsequently, it appears that the previous assumptions about the attributional process need to be qualified. If a situation causes enough positive affect, it can inhibit or even reverse the negative trait attribution and generalization associated with perfect duty violations.

I have also examined the development of children’s attributional processes (Brown & Trafimow, in review). Participants ages 7-17 were presented with scenarios in which an actor engaged in a dishonest or unkind behavior and were then asked to make a trait attribution for that actor. Trait attribution ratings were not significantly different for perfect duty violations or imperfect duty violations for ages 7-13. Children made strong negative trait attributions regardless of whether the behavior was dishonest or unkind. However, at age 15 there was a change in this trend. At ages 15 and 17 participants make significantly weaker trait attributions for imperfect duty violations. This finding is intriguing because much of the attributional research has focused on the violations of perfect duties (this is where all the action is). But this data demonstrates that the distinction between different duty violations develops as a result of the perception of imperfect duty violations changing by becoming less negative. Additionally, the way in which children’s attributional process develops may be quite adaptive. When we are unable (because we are too young or inexperienced) to distinguish between behaviors that may be very damaging or harmful (i.e., dishonest) and those that may not be as damaging (i.e., unkind) and when we are dependent on others (survival depends on the kindness and honesty of others), we believe that all negative behavior is indicative of an actor’s character. However, as our social thought processes become more sophisticated, we believe that imperfect duty violations are not indicative of an actor’s character and perfect duty violations are indicative.

Another line of my research investigates attributions for the self (Brown & Trafimow, 2003) A great deal is known about how people interpret the behavior of others, but it is not clear how people interpret their own behavior. Different theoretical perspectives suggest different answers to this question. The literatures on defensiveness and self-serving attributions suggest that the answer is “no.” According to these literatures, although people may give more weight to negative information when making trait attributions about others, they give more weight to positive information when making trait attributions about the self. This is because people are motivated to have positive views of themselves, and therefore they emphasize their positive behaviors and de-emphasize their negative ones. The upshot is that positive behaviors should have a stronger effect on trait attributions than negative behaviors when people make trait attributions about themselves. However, there is literature suggesting that negative behaviors, at least in the domain of morality, are more informative about people’s underlying traits than positive behaviors are (Brown, Trafimow & Gregory, 2004; Brown & Trafimow, in review). This difference in diagnosticity (informativeness about an actor’s character) should be true regardless of whether the behaviors are performed by someone else or by oneself. Thus, negative behaviors should lead to stronger trait attributions, even if the trait attributions are about oneself. I tested these arguments in an experiment and found that participants believed that it would take fewer negative behaviors than positive for them to consider a particular behavior indicative of their character (Brown & Trafimow, 2003). So, there is a negativity bias in self attributions! This finding is consistent with the diagnosticity hypothesis, but it contradicts the prediction made from the perspective that people may avoid or repress negative self-cognitions. The present experiment provides a clear demonstration that the motivation for diagnostic information can even overcome the need to maintain a positive self image, at least in the context of the paradigm I used.

Another area of research I am working on involves attitudes and affect as predictors of behavior (Brown & Trafimow, in review; Trafimow, Brown, Grace, Thompson & Sheeran, 2002; Trafimow, Sheeran, Lombardo, Finlay, Brown & Armitage, 2004). Ajzen & Fishbein (1980) developed the Theory of Reasoned Action (TRA) in order to have a more systematic way of understanding and predicting behavior. This theory argues that behavioral intentions (an actor’s intention to engage/not engage in a behavior) are a function of attitudes toward the behavior and subjective norms (perceptions of what “important others” think an actor should do). Attitudes are a function of affect (how one feels about engaging in a behavior) and cognition (what one thinks will be the consequences of engaging in a behavior). By using these variables, the prediction of behavior is greatly improved. The TRA has been successful in predicting many different behaviors including addictive behaviors, more common behaviors (attending church, cleaning one’s room, playing video games, Trafimow, Brown, Grace, Thompson & Sheeran, 2002), athletic behaviors like exercising and mental practice and health behaviors like wearing seatbelts and using condoms.

In a slightly different line of research (Trafimow, Brown, Grace, Thompson & Sheeran, 2002), I examined the way in which children ages 8-17 years old choose to engage in behavior. I discovered that children are very similar to adults in the way they form behavioral intentions using attitudes and subjective norms. Younger children differed from older children in their ratings of different behaviors, but the relationships between the variables did not change as a function of age. Children as young as 8-years-old engage in a broad range of behaviors primarily based on their attitudes just like adults. Furthermore, children are primarily attitudinal in their intentions; they choose to perform behavior based primarily on how much they like or dislike performing the behavior. Thus, it is quite likely that any attitudinal intervention/manipulation that involves the TRA variables and is useful for adults may also be successful on children as young as 8-years-old.

I have also tested TRA methodology (Brown & Trafimow, in review). More specifically I have examine the effects of experience with TRA variables on the relationships among the variables. Theoretically, it is unclear whether experience would increase or decrease the relationship among the variables. In order to test this, I measured theory of planned behavior variables twice one week apart [the theory of planned behavior is basically the TRA with two additional variables – perceived behavior control (how much the behavior is under one’s control) and perceived difficulty (how easy or difficult the behavior is)]. In this study I found that experience with the variables (intentions, attitudes, subjective norms) increases the relationships among the variables and prediction of behavior.

I also worked on another line of research that involves manipulating the role of affect and cognition in determining attitudes. Some of my past research has demonstrated that attitudes are primarily determined by affect rather than cognition (Trafimow, Sheeran, Lombardo, Finlay, Brown & Armitage, 2004). The fact that people’s attitudes, intentions and behavior are affectively driven leads to people choosing to engage in many harmful behaviors. Some examples of these behaviors are not wearing seat belts, having unprotected sex, overeating and not exercising. However, if people’s attitudes were cognitively driven or if they could associate affect with their cognitions, they may be able to make better choices regarding harmful behaviors. I am attempting to classically condition participants to associate affect with cognition (Trafimow, Brown & Simon, data collection). If participants feel positive affect towards engaging in “useful” behaviors, then they may be better able to choose to engage in behaviors based on their cognitive values rather than their affective values.

I also investigated the role of social consequences on subjective norms and attitudes (Trafimow, Brown, Zarate, data collection). Researchers generally assume that social consequences (an actor’s beliefs about what others will think or do if she engaged in a particular behavior) are a normative factor, not an attitudinal factor. But a strict interpretation of the TRA means that social consequences are attitudinal because they are consequences. However, one could argue that whether social consequences are normative or attitudinal is determined by whether the participants are individualists or collectivists. I am currently collecting data from Mexican Nationals and Americans in order to clear up these issues.

Researchers (Trafimow, Triandis and Goto, 1991) have successfully argued that each individual (regardless of his/her culture) possesses a collective (cognitions about group memberships) and private (cognitions about traits, states and behaviors) self-concept. My research has discovered a third self that mediates the private and collective self-concepts, the relational self-concept (Trafimow, Brown, Cone, Johns & Madson, in review). This self-concept contains cognitions about how individuals relate to others (e.g. parent-child, husband-wife, etc.). First, I found that it takes less time for participants to transition from the relational self to the other self-concepts than the time required to transition between the private and collective self-concepts. Second, it takes less time to transition between the different relational self-cognitions than it takes to transition between different collective self-cognitions and different private self-cognitions. Third, I demonstrated that relational self-cognitions are highest in self-rated importance. Considering these findings, it seems that the relational self is the most frequently accessed self-concept (for both Mexican- and Caucasian-Americans). Finally, the theoretical implications for evolutionary theory and developmental theory are discussed.

In the future I would like to examine the effects of gender on how self cognitions are organized. There is evidence that women tend to be more “collective” (or access their collective self-concept more) than men in both individualistic and collectivistic cultures. However, it is unclear whether men (across cultures) typically access their private self-concept more than women or if they differ depending on their culture. It could be that men in collectivistic cultures access their collective self-concept as much as their female counterparts, while men in individualistic cultures access their private self-concept more than their female counterparts. Or it may be that across cultures men tend to access their private self-concept more than women. There is evidence that females (across cultures) are superior to males in reading non-verbal cues including facial expressions. This may be indicative a general trend whereby females’ think about others more, which suggests that they may be accessing their collective self-concept more than men. Thus, it may be that across cultures men access their private self-concept more than women.

[1] It is important to note that where morality is concerned negative behaviors (dishonesty and disloyalty) are considered diagnostic, but where ability is concerned positive information is given more weight.