Reaction to “On the Confirmability and Disconfirmability of Trait Concepts”

[First published on Nathen’s Miraculous Escape, November 7, 2008.]

Some time ago, I was present for a conversation between two of my friends in which it came out that one had lied to the other. It was immediately clear that there was nothing the one who had lied could do to make the other believe anything else he said. It could easily take years of acting with perfect integrity, I thought, to build that trust back, if it could be built back. This is a major point of the article, “On the Confirmability and Disconfirmability of Trait Concepts,” by Rothbart and Park: Where favorable trait ascriptions, like “honest” and “truthful,” are easy to lose, unfavorable trait ascriptions, like “deceitful” and “deceptive” are difficult to lose. This is a problem not only for individuals like my friend who lied, but for whole groups of people laboring under the unjustly and inaccurately received negative trait ascriptions called stereotypes.

Rothbart and Park had participants rate 150 trait adjectives on how easy they were to acquire and lose in three dimensions: how often the environment provides opportunities to show or disprove each trait, to what extent each trait had clear, corresponding behaviors to indicate it, and how many times a behavior would need to be seen before a corresponding trait could be inferred or disproved. What they found in part was that two of the dimensions, the first and third were related to how favorable a trait was rated. That is, when it comes to how often the environment provides opportunities to show or disprove each trait, favorable traits are easy to get and easy to lose, whereas unfavorable traits are hard to get and hard to lose. On the dimension of how many times a behavior would need to be seen before a corresponding trait could be inferred or disproved, favorable traits were hard to get and easy to lose, where unfavorable traits were easy to get and hard to lose. Notice that in both of these relationships, unfavorable traits were difficult to lose; there are few occasions provided to disprove them, and disconfirming behaviors seem more ambiguous than confirming behaviors.

This pattern may apply to the traits ascribed to groups as well as those ascribed to individuals, as trait ascriptions are part of our schemas about individuals as well as groups. Schemas are cognitive information-sorting systems that function in part as filters on our perception, causing schema-consistent behaviors to be memorable and schema-inconsistent behaviors to go unnoticed more easily and be forgotten more easily, if noticed (Aronson, Wilson, & Akert, 2007). This may be a problem-compounding effect for those who are victims of a negative cultural stereotype.1 Rothbart and Park’s (1986) participants were not asked to consider race, and reported that it was difficult to imagine someone disconfirming a negative trait. It seems that it would be that much more difficult for a person of color to disconfirm a negative trait that coincided with the stereotype of their group. At least three factors would converge against it: The environment doesn’t provide many opportunities to disconfirm unfavorable traits, disconfirming behaviors are much more ambiguous than confirming behaviors, and stereotypers2 are less likely to notice disconfirming behaviors and more likely to forget them.

There is another troubling point. Rothbart and Park (1986) predicted that, once acquired, a negative stereotype will persist in a stereotyper’s mind unless they have contact with the stereotyped group, for otherwise there is no opportunity to observe disconfirming behavior. That reasoning is congruent with the evidence they collected, but is somewhat contrary to, or at least complicated by, the evidence collected by Correll, Park, Judd, and Wittenbrink (2002) investigating “shooter bias,” the increased tendency for experiment participants to mistakenly shoot African American target images more than white target images. The authors of this study found that the bias was predicted in part by the level of contact with African Americans. In keeping with Rothbart and Park’s prediction, increased contact with African Americans did correlate with decreased personal endorsement of stereotypes of African Americans, but also with this increased, non-conscious acting on a stereotype, that an African American is more likely to have a gun.3

There are caveats to consider in this line of thinking. First, “gun-possessing” was not among the traits rated by Rothbart and Park’s participants, and so it is a stretch to compare shooter bias with even the related traits that were included, like “violent,” “mercenary,” and “hostile.” Second, trait ascription may be more closely related to the personal endorsement of a stereotype than the unconscious application of a stereotype, like shooter bias. Third, Rothbart and Park used imaginability as their operationalization of their three major factors. In other words, to determine how many occasions the environment provides for confirming or disconfirming a trait, for example, they asked participants how easy it was to imagine such occasions for each trait. Though Rothbart and Park found that their operationalization provided reliable statistics in that they were stable from participant to participant, imaginability may or may not accurately predict how difficult a trait will be to gain or lose. Fourth, Rothbart and Park predicted that contact with a stereotyped group would be necessary to disconfirm a trait, which is probably true, but it does not follow that contact would be sufficient to disconfirm a trait.4

Still, that stretch is worth entertaining because of how closely all three elements, trait ascription, belief in stereotypes, and acting on stereotypes, are related conceptually. The possibility that contact with a stereotyped group could simultaneously be involved in decreasing belief in the stereotype while increasing the amount of unconsciously acting on the stereotype begs for more investigation. One possible place to start is to check if the effects replicate in other stereotypes, perhaps having participants rate job applications attributed to either whites or African Americans, to see how bias in the ratings might correlate with participants’ personal endorsements of African American stereotypes and contact with African Americans.5 There is also the possibility that “unconsciously acting on a stereotype” breaks down so that on tasks involving time-forced errors, like the shooter-bias video game, bias correlates positively with level of contact, while on tasks that leave time for conscious deliberation, like the rating of job applications, bias would correlate negatively with level of contact.6 This could be checked by decreasing the amount of time each participant had with each application to see if time-forced biases resulted.

It does seem clear that, at least in people’s imaginations, unfavorable trait ascriptions are difficult to shed, and that this probably applies both to individuals and to groups. The complexities of that process remain to be investigated, or have, perhaps, been investigated in work by Rothbart and Park (and others) since their 1986 paper. It also remains how Rothbart and Park’s findings relate to others’ such at Correll and colleagues’ (2002): Just how deep does changing one’s mind about a stereotype go? Under what circumstances will changing one’s mind about the truth of a stereotype produce a change in behavior? Are some behaviors immune to influence by such a mind change? If so, what processes will affect these behaviors?



Aronson, E., Wilson, T. D., & Akert, R. M. (2007). Social psychology. Upper Saddle River, NJ: Pearson Education.

Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2002). The Police Officer’s Dilemma: Using ethnicity to disambiguate potentially threatening individuals. Journal of Personality and Social Psychology, 83, 1314-1329.

Rothbart, M., & Park, B. (1986). On the confirmability and disconfirmability of trait concepts.Journal of Personality and Social Psychology, 50, 131-142.


Very good, tight paper, with lots of interesting questions and thoughts that might be pursued. This is just about the epitome of what I’m looking for in a reaction paper – strong evidence of critical thinking, application of constructs across domains, integration of disparate material, and demonstration of knowledge that you probably would not have had before taking this course. 50/50


Btw, may I have permission to post this paper as an example of a great reaction paper for future classes? J


1 Awk., rephrase.

2 Strange use of this word; I might rephrase.

3 This last sentence is hard to parse…I’d rephrase. Also, did Correll et al. show that Ps thought AAs were more likely to have a gun, or were they just more likely to shoot even when a gun wasn’t present?

4 Good point.

5 This would address the first part: whether contact reduces personal bias and/or belief in the cultural ST – but how would this relate to further automatic application of the same ST?.

6 An interesting idea, but in a non-salient domain, do you think you would see effects? I guess the question is: how much does knowledge of cultural ST or endorsement of ST relate across domains – such as “blacks are poor workers” with “blacks are more likely to have a gun.”