香京julia种子在线播放

Implicit biases include attitudes that form through experiences and operate outside an individual’s awareness. Due to implicit cognitive biases, people may often attribute certain qualities or characteristics to all members of a particular group, a phenomenon known as stereotyping or prejudice. The effects of implicit bias on behavior have sparked a lot of attention, particularly when it comes to discrimination in important areas of life including healthcare, law enforcement, housing, criminal justice, and education (FitzGerald et al., 2019). Given this, society is interested in figuring out how to minimize implicit bias in the general population, as well as among practitioners who work in these fields.

To this date, the most widely accepted indicator of implicit biases is the Implicit Association Test (IAT; Greenwald et al., 1998). In the IAT task, participants are asked to categorize words as referring to negative or positive valence together with either images or words. For example, concepts such as “black” or “white” people could be associated with “good” or “bad”. Response latency to this categorization decision is then calculated and used as a measure that quantifies the strength of implicit associations between categories for a given individual. The IAT has undergone extensive testing and has been shown to be insensitive to procedural variation (Nosek et al., 2005) and less susceptible to individual response biases than questionnaires (Steffens, 2004). The IAT has also been shown to be reliable across test-retest experimental designs (Bosson et al., 2000; Dasgupta et al., 2000; Greenwald and Farnham, 2000; Greenwald and Nosek, 2001).

Implicit measures were initially assumed to be hardly susceptible to change (Fazio et al., 1995). However, with multiple studies showing that implicit associations are vulnerable to laboratory interventions, this viewpoint has changed to one of implicit malleability (for reviews, see Blair, 2002; Gawronski and Bodenhausen, 2006; Lai et al., 2013). Intriguingly, a recent study that evaluated several interventional approaches showed that those that were more self-relevant, emotional, and vivid were also more effective in inducing a change in implicit cognitive biases (Lai et al., 2014). This is in line with previous research indicating that messages that are related to the one’s “self” are more likely to be processed and become personally relevant (e.g., Petty and Cacioppo, 1990; Blankenship and Wegener, 2008).

Motivated by the need to find ways to decrease levels of implicit biases among the general population, we here aimed to investigate whether the involvement of non-professional artists in a training program on the performing arts influenced implicit measures. Acting requires an actor to understand a character’s mental world and the experience of that character’s feelings. The willingness to take another person’s point of view is a critical element in social functioning that promotes empathy for others (Davis, 1983) and changes representations of an out-group to be more self-like (Galinsky and Moskowitz, 2000). Indeed, it has been shown that imagining an event from another person’s point of view increases the perception of sharing their own personality attributes (Davis et al., 1996).

Here, we asked non-professional artists to enroll in a training program of 14 consecutive full days, whereby they were requested to devise, prepare and implement a real show that expressed the notion of “cancer and neurodegenerative diseases.” In fact, adverse medical conditions are an important source of implicit cognitive biases in the population. Studies in patients under negative health conditions showed they feel stigmatized and reported feelings of exclusion, rejection or devaluation (Weiss et al., 2006). Stigma-induced discrimination in turn, has been associated with different negative outcomes such as more psychological distress (Berjot and Gillet, 2011), more depressive symptoms (Roberts et al., 2020) less self-esteem (Liu et al., 2020) and the reduction of perceived quality of life (Chambers et al., 2012, 2015).

In the current study, artists were immersed into a preparation stage that involved meeting groups of scientists that researched different aspects of the illness, from molecular biology to patient care. During the 14-days program, artists were exposed to information that allowed them to better understand the illness from a biological point of view, as well as to help them invoke realistic behavior from a patients’ perspective, thereby fostering self-relevancy, emotionality, and vividness in their training program. We hypothesized that this particular interventional approach would influence implicit cognitive bias to illness, a stigma that is extended among the general population and clinical practitioners (Link et al., 2004; Brohan et al., 2010). To quantify the degree to which the training program influenced implicit cognitive bias on an “illness,” artists (i.e., the experimental group) were requested to complete an IAT before and after the training program that included images of sick/healthy people and words related to good/bad attributes. Two matched control groups of healthy participants were also included in the study. The first control group (no intervention) completed the IAT sessions separated by the same interval of days. The second control group underwent an information-based intervention and completed the IAT sessions separated by the same interval of days. The data from the control studies was used to assess for the specificity of possible changes between the two IAT sessions in the experimental group.

To assess for the existence of changes in cognitive biases specifically in the experimental group between IAT sessions, a mixed ANOVA was performed on mean reaction time (RT) to the associative blocks (Figure 1). The ANOVA included congruency (2 levels: congruent vs. incongruent) and session (2 levels: pre and post) as within subject factors and group (3 levels: experimental group, no-intervention control group and information intervention control group) as between subject factor. This analysis revealed a statistically significant main effect of session [F(1, 43) = 12.43, p = 0.001, η² = 0.22] and congruency [F(1, 30) = 28.43, p < 0.001, η² = 0.39], corroborating that participants were slower in categorizing words/images when social biased categories shared the location on the screen (i.e., sick/good and healthy/bad). Importantly, we found a triple congruency, session and group trend toward significant effect [F(2, 43) = 2.73, p = 0.07, η² = 0.11] but not a session × group [F(2, 43) = 0.96, p = 0.39, η² = 0.04] nor congruency × group interaction effect [F(2, 43) = 0.98, p = 0.38, η² = 0.04], suggesting that the congruency effects between conditions and sessions varied differently as a function of the study group. To identify the source of this effect, we ran a repeated measures ANOVA including congruency and session as within subject factors separately for each group. This analysis showed a significant main effect of congruency in the experimental group [F(1, 15) = 16.92, p = 0.001, η² = 0.53], in the no-intervention control group [F(1, 15) = 16.26, p = 0.001, η² = 0.52] but not in the information control group [F(1, 13) = 2.46, p = 0.14, η² = 0.16]. We also found a main effect of session in the experimental group [F(1, 15) = 5.54, p = 0.03, η² = 0.27] and a trend toward statistical significance in the no-intervention control group [F(1, 15) = 3.39, p = 0.08, η² = 0.18] and in the information intervention control group [F(1, 13) = 4.19, p = 0.06, η² = 0.24]. Importantly, however, this analysis revealed a significant session x congruency interaction effect in the experimental group [F(1, 15) = 10.32, p = 0.006, η² = 0.41] but not in the no-intervention control group [F(1, 15) = 0.31, p = 0.58, η² = 0.02] nor in the information intervention control group [F(1, 13) = 0.30, p = 0.11, η² = 0.19]. A student paired t-test analysis confirmed that the session x congruency interaction in the experimental group was driven by differences between sessions for the incongruent [t(15) = 3.01, p = 0.009, Cohen’s d = 0.75] rather than for the congruent [t(15) = 0.23, p = 0.82; Cohen’s d = 0.06] trials. To further assess for the specificity of the congruency effects found in the critical experimental condition, we ran a mixed ANOVA with the outcome variable mean reaction time (RT) to the practice blocks. The ANOVA included Block (2 levels: 1st vs. 2nd presentation block) and session (2 levels: pre and post) as within subject factors and group (3 levels: experimental group, no-intervention control group and information intervention control group) as between subject factors. This analysis revealed a significant main effect of Block [F(1, 43) = 9.80, p < 0.01, η² = 0.19] and session [F(1, 43) = 27.79, p < 0.01, η² = 0.39], but no interaction with group (all ps > 0.2, all η² < 0.1).

Finally, we calculated the IAT D-score, which instead of comparing within-participant differences in RTs, standardizes them at the participant level, dividing the within-participant difference by a “pooled” standard deviation (Greenwald et al., 2003). D-scores in the experimental group were M = 0.73 (SD = 0.86) and M = 0.46 (SD = 0.71) in the pre and post observation, respectively, M = 0.41 (SD = 0.56) and M = 0.55 (SD = 0.44) in the pre and post measures in the no-intervention control group and M = 0.36 (SD = 0.67) and M = 0.14 (SD = 0.69) in the pre and post measures in the information intervention control group. A mixed ANOVA, including session (2 levels: pre and post) as a within-subject factor and group (3 levels: experimental group, no-information control group and information intervention control group) as a between-subject factor, confirmed previous findings by revealing a trend toward statistically significant session x group interaction [F(2, 43) = 2.74, p = 0.08, η² = 0.12]. A paired student t-test showed statistically significant D-score difference between pre and post in the experimental group [t(15) = 2.55, p = 0.02, Cohen’s d = 0.64] but not in the no-information control group [t(15) = −1.04, p = 0.32, Cohen’s d = −0.26] nor in the information intervention control group [t(13) = 1.27, p = 0.23, Cohen’s d = 0.34].

The aim of the current study was to assess whether the involvement in a performing arts training program of 14 full days could lead to a shift in implicit measures. To this end, we conducted an experiment in which healthy participants completed an IAT to assess biased attitude to illness in two separate sessions, before and after the training program. A similar two-session IAT task was completed by two different matched control groups of healthy participants. Results showed that participants who were involved in the training program reduced their implicit bias toward illness measured through IAT in the second session. This reduction in IAT measures was not observed in the control samples, despite the two IAT measures being matched in temporal delay with the experimental group. These findings suggest that an interventional program based on the performing arts could be effective in reducing levels of implicit biases among the general population.

To the best of our knowledge, this is the first study showing that a training program based on acting can have an influence on implicit cognitive bias, thereby raising the possibility of extending previously established lab-based interventional techniques that can reduce implicit bias (e.g., Lai et al., 2014, 2016) to include more realistic approaches. Current findings therefore may represent a first attempt to show the promising possibilities that the performing arts may have to help treat prejudice and stereotypes. At the same time, the present study lacks a clear explanation of the mechanisms that underlie this effectiveness, as it was not designed with such purpose. Nevertheless, we can offer some speculations. Previous research revealed that most effective lab-based techniques to reduce implicit bias invoke self-relevancy, emotion and vividness (Lai et al., 2014). While designing strategies that can effectively elicit these properties in individuals may be challenging in the laboratory, inducing them by requesting people to perform as actors seems more practicable. Thus, one plausible explanation of the changes in IAT in our study was driven by the self-involvement of the participants during the training program, which in turn, induced a more thorough processing of the scenario. An alternative explanation would be that participants that engaged the performing arts training program increased their general cognitive control skills, which has been shown to influence implicit bias tasks (e.g., Payne, 2001, 2005; Amodio et al., 2008; Klauer et al., 2010). Future studies may include separate measures of executive functioning (e.g., Stroop task) in conjunction with the implicit bias tasks to be able to assess bias- and control-related ability changes due to an interventional program.

Several important questions, however, remained unanswered in the current study. The first one concerns the specificity of the effects. More concretely, it would be important to determine in the future the extent to which the reduction in implicit bias was specific to the stereotyped concept that surrounds the training program. For example, one would expect the reduction in implicit bias to be specific to illness but not to other stereotypes such as racial or gender bias. Our findings showing that only the experimental group, but not the control groups, had a reduction in implicit measures speaks about the effectiveness of the interventional program but not about its specificity at the implicit bias level. Future research may include different IAT versions at the within-subject design to scrutinize this possibility.

A second concerns refers to the small sample size of the experimental groups in our study. Though we believe current findings are well supported by effect sizes, the within-subject design and the inclusion of two control groups, future studies aiming to assess the impact of the performing arts in implicit biases should, if possible, include larger sample sizes. This would increase the statistical power of the study, thereby ensuring the effects can be extrapolated to the overall population, especially if the expected effects cannot be assessed at the within-subject level.

Another important issue is to assess the extent to which the changes in implicit bias observed in the current design persist over time. Previous research revealed that stability of changes in implicit preferences are difficult to observe in available interventional techniques (Lai et al., 2016; see, however, Peck et al., 2013). While the current research does not look at long-term effectiveness, our proposed framework engaged participants in a 14-days training program unlike previous interventional programs that involved a single session at the lab. At the speculative level, we hypothesized that engaging the participants in a consecutive day-to-day program may promote memory consolidation of learned non-biased association between concepts. A consequence of such a consolidation process would be participants’ reduction in implicit bias being accompanied by an explicit awareness of the bias. Indeed, previous research on memory consolidation showed that night sleep promotes knowledge awareness (e.g., Wagner et al., 2004). If that were the case, it would suggest that an interventional program has an impact in transforming implicit to explicit memory representations. A possible advantage of such transformation is that it may thus favor generalization (Ranganath and Nosek, 2008), thereby promoting implicit bias reduction in other conceptual domains. Future studies should examine this possibility by combining IAT measures with questionnaires that help assess attitude awareness of cognitive bias.

Society is interested in reducing the implicit biases in the general population. So far, most of the interventional techniques described in the literature that may help reduce implicit bias have been developed to be implemented in the lab. Though lab-based techniques offer experimental rigor and control, they may limit participants’ motivation and engagement with a task, thereby complicating the possibility of making reliable changes in implicit bias at an individual level. An alternative strategy is to take advantage of realistic activities that invoke solid individual self-engagement with a task, with the hope that this fosters strong changes at an implicit level. Current research underscores the powerful impact that performing theater can have in reducing implicit bias. Our findings show that implicit bias toward physical illness is reduced in a group of non-professional artists after enrolling in a 14-days training program based on theater acting. The extent to which such an interventional program has a causal effect and if so, whether such effect is specific to a specific conceptual bias (i.e., physical illness) remains to be examined in the future.

The datasets generated for this study are available on request to the corresponding author.

The studies involving human participants were reviewed and approved by the University of Barcelona’s Bioethics Commission. The patients/participants provided their written informed consent to participate in this study.

JG-A, PC, and LF contributed to the conception and design of the study. JG-A and PA programmed the computer task. PA and CV-G acquired the data. JG-A and LF performed the statistical analysis and wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.