Participants

In total, 135 people (49 males and 86 females), recruited by email and social networks, participated in the experiment. Thirty-three participants were excluded owing to not achieving ten consecutive correct answers in up to 320 trials of the pre-training phase of the FAST task (see procedure). Therefore, 102 participants, aged 17–65 (M = 32, SD = 10), concluded the experiment. Twenty-five of them were residents of Boa Vista, Curitiba, and São Paulo (Group A), the three Brazilian cities that welcomed the most refugees at that time, according to the Response for Venezuelans (R4V, 2022), an initiative that coordinates organizations (including UN Agencies, civil society, and NGOs) on their efforts under Venezuela’s refugee sheltering, and 77 were residents of other 40 cities (Group B) not top ranked by the R4V initiative.

Apparatus

Data were collected in a single session administered online, lasting approximately 15 minutes, in the participants’ preferred environment, using participants’ notebooks or desktop computers. An application was specifically developed for this study, using JavaScript and OpenSesame (Mathôt et al., 2012). The application was hosted for access on the MindProbe server, owned by the European Society for Cognitive Psychology, a solution based on the JATOS platform (Lange et al., 2015). Participants provided demographic data, read and accepted the Informed Consent Form (ICF), completed the Xenophobia Expression Items, a procedure evaluation assessment, and the Functional Acquisition Speed Test task through this application.

Demographic Data

After registering their agreement with the ICF, participants provided details about their email address to receive a copy of the ICF, as well as their gender, age, birth city, current living city, educational level, and racial or ethnic background.

Explicit Attitudes Measures

Explicit attitudes of xenophobia were measured using a Portuguese translation of the Xenophobia Expression Items (van der Veer et al., 2011), a hierarchical cross-national scale developed to measure fear-based xenophobia. The scale included 14 statements, for which participants indicated their level of agreement on a Likert scale from one to six. Four statements were favorable to immigration, and ten were unfavorable, presented in randomized order. The four favorable statements used a reverse scale to improve experimental control. The full wording of the items, together with the results obtained, is presented in Table 3.

Implicit Attitudes Measure

Implicit measures were collected using a FAST task with eight sets of stimuli. Table 1 presents the English version of the original Portuguese words presented to the participants. Four sets of neutral stimuli (N1, N2, N3, N4) were part of the Pre-training task block, composed of words familiar to the participants but not related to the research topic, and four sets of different stimuli (A1, A2, B1, B2), composed of words related to the topic, were part of the Training. The stimuli for the Training blocks were synonyms to Brazilians and Venezuelans, and words with positive and negative valence that are commonly associated with Brazilians and immigrants, as reported in the literature (Delfim, 2017; Holanda, 1936; Mota, 2019; Souza, 2018).

Table 1 FAST task stimuli setsProcedure Evaluation Assessment

Another assessment using the Likert scale was applied for participants’ indication of their agreement level with three statements favorable and three unfavorable to the experiment. Its full items’ wording, together with the results, is presented in Table 6.

Procedure

Participants received a link via email to access and participate in the research. On accessing it, participants were asked to register their consent with the terms of the research and enter demographic information. A screen defining specific terms used in the experiment was then displayed to minimize potential interpretations to the stimuli, that may be influenced by cultural and regional variations or individual experiences. Subsequently, participants were exposed to the FAST task or the Xenophobia Expression Items. The order of exposition to the questionnaire or task was randomized and balanced. At the end, the participants responded to the Procedure Evaluation Assessment.

Xenophobia Expression Items

The presentation of the xenophobia questionnaire began by displaying the task instructions to participants: “Next, 14 statements will be presented on this screen. To indicate your level of agreement or disagreement, choose the option that reflects whether you strongly disagree, moderately disagree, slightly disagree, slightly agree, moderately agree, or strongly agree. Press the spacebar to begin.” The statements were presented individually on the screen in randomized order. The progress from one statement to the next was automatic once a response option was selected.

Functional Acquisition Speed Test (FAST)

The FAST task was applied in two phases: Pre-training, of up to 320 trials, to introduce participants to the task, and Training, to measure participants’ performance and learning curves. The Training phase had two blocks of 48 trials: consistent and inconsistent blocks.

The task consisted of participants emitting responses by pressing the Z or M keys on their keyboards when a stimulus was displayed at the screen center. Each trial comprised a stimulus displayed for up to 1.5 seconds, followed by 0.5 seconds of feedback for response or lack of response in time, and 0.2 seconds of interval between trials. Correct responses were accompanied by 0.5 seconds of positive feedback, displayed as “CORRECT” on the screen, whereas incorrect or no responses elicited 0.5 seconds of negative feedback, indicated by “INCORRECT” in the center of the screen. FAST's 1.5-second response time, contrasting with the 3 seconds originally proposed (O'Reilly et al., 2012), is founded on Marcelino’s (2019) comparison of FAST training performances in 1.5- and 3-second settings, which showed no statistically significant difference.

The following instructions preceded the Pre-training: “INSTRUCTIONS: In a moment, some words will appear in the center of this screen. Your task is to learn which key to press when a word appears in the shortest possible time. IMPORTANT: During this phase, you must only press the Z or M keys. If you take too long to press one of them or press another key, your answer will be considered INCORRECT. Locate them on your keyboard now. This session is just a training session to familiarize you with the task. To help you learn, a message will be displayed informing you whether you pressed the CORRECT or INCORRECT key after each word. Press SPACEBAR to begin.”

Pre-training consisted of a single block of up to 320 trials, where only unrelated sets of familiar words were presented (stimuli N1, N2, N3, and N4). Selecting the same key (Z or M) to stimuli N1 and N2 and the other key to stimuli N3 and N4 was considered a correct response. The block ended when participants achieved 10 consecutive correct answers or reached 320 trials. Next, the participants were exposed to the Training phase.

The instructions before the first block of Training stated: “NOW IT’S FOR REAL! From now on, your task is to learn which key to press when a word appears, minimize errors, and respond as quickly as possible, remembering that you should only press the Z or M keys. If you take too long to press them or press any other key, your answer will be considered INCORRECT. Locate them on your keyboard now. To help you learn, you will receive a message informing you whether you pressed the CORRECT or INCORRECT key. Press SPACEBAR to begin.” The second block of Training began with the instruction: “VERY GOOD! Now, words and keys have been shuffled. Your task is to relearn which key to press when a word appears, making as few mistakes as possible, and as quickly as you can. REMEMBER: You can only press the Z or M keys. If you take too long to press them or press another key, your answer will be considered INCORRECT. Locate them on your keyboard now. To help your learning, you will receive messages informing you whether you pressed the CORRECT or INCORRECT key again. Press SPACEBAR to begin.”

During the Training phase, four sets of words related to the research topic (stimuli A1, B1, A2, B2) were combined in pairs of sets associated with the Z or M keys to create the Consistent Block and Inconsistent Block conditions. The Consistent Block deemed correct responses based on stereotypes: press Z or M (randomly determined), when encountering stimuli from the Venezuelan-related word set (A1) and negative valence words (B1), and the other key when encountering stimuli from the Brazilian-related word sets (A2) and positive valence words (B2). The Inconsistent Block presented stimuli that deviated from stereotypes, requiring participants to press one key in response to Venezuelan-related stimuli (A1) and positive valence words (B2), and the other key in response to Brazilian-related stimuli (A2) and negative valence words (B1).

If participants commanded <back> or <refresh> on their browser at any time, the application was interrupted, recorded partial results, and restarted as a new data collection under the same user code. This way, it was possible to identify participants who tried more than once and evaluate whether a particular performance had sufficient data and complete phases to be either included in the data analysis or excluded. Participants who had completed the study multiple times would only have their first completion considered, and those who had interrupted and restarted the study would only have their final completion considered if the interruption occurred until the Pre-training phase and the total number of trials in that phase across all participations did not exceed 320.

For balancing purposes and preventing order effects, by random choice, half of the participants were exposed to the Consistent Block at first, and half to the Inconsistent. The keys associated with each pair of stimulus sets were also randomized between participants, aiming to eliminate possible influences on performances arising from the position of the keys.

Procedure Evaluation Assessment

After completing the FAST task and the Xenophobia Expression Items, participants received the following instruction: “We are finishing... Now, we would like to know your opinion about participating in this experiment. Next, six statements will be presented on this screen. You must record your opinion, indicating whether you: strongly disagree, moderately disagree, slightly disagree, slightly agree, moderately agree, or strongly agree. Press the SPACEBAR to start.” Participants then viewed the assessment items, one per screen, with the response to the presented item being a condition for advancing to the next one. At the end of participation, the application redirected the participant to a thank-you message page while recording performance data.

Data Analysis

Data analysis was conducted using MS Excel and SPSS software, based on data generated by the application and available through text files on the MindProbe server. Initially, each participant’s learning curve was calculated by applying the MS Excel “slope” function to the data of reinforced trials and response time in these trials. The curves were calculated separately for Consistent and Inconsistent Block performances, and then subtracted to obtain the difference between the curves. Positive difference values represented faster learning curves in Consistent Blocks, and negative values represented faster learning in Inconsistent Blocks.

Although statistical tests pointed out the normality of the total participants’ sample data and stratified samples, settling for a more rigorous statistical analysis, samples with fewer than 30 elements were treated as nonparametric, and samples larger than this were treated as parametric, in accordance with the rule of thumb proposed by van Belle (2002). Normality of the data was assessed in SPSS via the Kolmogorov–Smirnov parametric test, which indicated normality for the total sample of participants’ values across all variables examined (number of correct answers, response time, average response time, total response time, accuracy, task elapsed time, and curves’ difference). Normality tests for values stratified by city groups and gender produced identical positive outcomes. The group A participants’ sample was tested using the nonparametric Shapiro–Wilk test, as it was a small stratum, and for the others, the Kolmogorov–Smirnov test was repeated.

The significance data analysis was conducted by applying the Student’s t-test for samples including all participants or comparing different results of the same large sample, such as participants from group B. The nonparametric Wilcoxon and Mann–Whitney tests were applied when comparisons involved at least one smaller strata result, such as group A, with Wilcoxon applied to the tests in paired samples and Mann–Whitney on independent samples. Correlations between performance variables were also calculated by treating total or larger samples as parametric and smaller ones as nonparametric, with the Pearson correlation coefficient applied to the former and the Spearman coefficient to the latter. The effect size was calculated using Cohen’s d for samples considered parametric and Pearson’s r for samples treated as nonparametric.