The foreign language effect and the gambler’s fallacy: evidence from neurophysiology

Universidade do Minho Escola de Psicologia Márcia Filipa Rodrigues Costa The Foreign Language Effect and the Gambler’s Fallacy: Evidence from Neurophysiology outubro de 2022 The Foreign Language Effect and the Gambler’s Fallacy: Evidence from Neurophysiology Título Título Título Título Título Título Título Márcia Costa UMinho | 2022 Universidade do Minho Escola de Psicologia outubro de 2022 Márcia Filipa Rodrigues Costa The Foreign Language Effect and the Gambler’s Fallacy: Evidence from Neurophysiology Dissertação de Mestrado Mestrado Integrado em Psicologia Trabalho efetuado sob a orientação do(a) Professor Doutor Pedro Moreira Professora Doutora A n a P a u l a S o a r e s ii DIREITOS DE AUTOR E CONDIÇÕES DE UTILIZAÇÃO DO TRABALHO POR TERCEIROS Este é um trabalho académico que pode ser utilizado por terceiros desde que respeitadas as regras e boas práticas internacionalmente aceites, no que concerne aos direitos de autor e direitos conexos. Assim, o presente trabalho pode ser utilizado nos termos previstos na licença abaixo indicada. Caso o utilizador necessite de permissão para poder fazer um uso do trabalho em condições não previstas no licenciamento indicado, deverá contactar o autor, através do RepositóriUM da Universidade do Minho. Licença concedida aos utilizadores deste trabalho Atribuição-NãoComercial-SemDerivações CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/ Agradecimentos Ao meu orientador, Professor Doutor Pedro, agradeço as inúmeras aprendizagens, a disponibilidade e o sentido de humor. À Professora Doutora Ana Paula, a partilha de conhecimento e ajuda valiosa. Obrigada por acreditarem no meu potencial. Agradeço aos meus pais, por estarem sempre por perto, por acreditarem sempre em mim e me apoiarem de todas as formas possíveis. Ao meu querido irmão, Pedro, pelo carinho e preocupação. À minha família, pelas mensagens de apoio constantes e pela motivação. Aos meus amigos, um eterno obrigada por compreenderem a minha ausência em determinados momentos ao longo deste percurso e por me apoiarem incondicionalmente. O Efeito da Língua Estrangeira e a Falácia do Jogador: Evidência da Neurofisiologia Resumo Estudos recentes mostraram que, quando usam uma segunda língua (L2), pessoas bilíngues tendem a ser menos enviesadas, um efeito experimental designado por Efeito de Língua Estrangeira ( Foreign Language Effect , FLE). Alguns destes estudos investigaram o FLE em contexto de tomada de decisão de risco, onde a Falácia da Mão Quente ( Hot Hand Fallacy , HHF) e a Falácia do Jogador ( Gambler’s Fallacy , GF) surgem frequentemente, sugerindo que apenas a primeira é influenciada por este efeito. Neste trabalho, procurámos clarificar os resultados obtidos por estes estudos, empregando uma tarefa que envolve um jogo de cartas, manipulando valência (positiva vs. negativa) e língua (Português vs. Inglês). Medidas comportamentais e psicofisiológicas de 41 participantes foram recolhidas. Os resultados mostram uma proporção semelhante de GF em ambas as línguas. A valência do feedback revelou-se significativa para a amplitude de SCR (maior amplitude para feedback negativo), mas nenhum efeito significativo de língua emergiu. As implicações destes resultados são discutidas. Palavras-Chave: Efeito de Língua Estrangeira, Bilinguismo, Tomada de Decisão, Falácia do Jogador, Psicofisiologia The Foreign Language Effect and the Gambler’s Fallacy: Evidence from Neurophysiology Abstract Recent studies have shown that using a second language makes individuals less prone to bias, an experimental effect known as the Foreign Language Effect (FLE). Some of these studies have tested the FLE in risky decision-making context, where the Hot Hand Fallacy (HHF) and the Gambler’s Fallacy (GF) often emerge, suggesting that only the first is affected by this effect. Here, we aimed to further clarify the results obtained by these studies, using a card gambling task, manipulating valence (positive vs. negative) and language of feedback (Portuguese vs. English). Behavioural and psychophysiological measures of 41 Portuguese-English bilinguals were collected. Results revealed a similar proportion of GF in both language contexts. We found a significant effect of valence for SCR amplitude (higher amplitude for negative feedback), but no significant effect of language emerged. The implications of these results are discussed. Keywords: Foreign Language Effect, Bilingualism, Decision-Making, Gambler’s Fallacy, Psychophysiology Index Background ........................................................................................................................................ 8 Methods ........................................................................................................................................... 12 Participants .................................................................................................................................. 12 Instruments .................................................................................................................................. 12 Second Language Characterization Questionnaires and Tests .................................................... 13 Impulsive Behavior Scale (UPPS) ............................................................................................... 13 Psychophysiological Measures .................................................................................................. 14 The Gambler’s Fallacy Task ...................................................................................................... 14 Stimuli ..................................................................................................................................... 16 Procedure ..................................................................................................................................... 17 Data Analysis ................................................................................................................................ 18 L2 Characterization Questionnaires and Tests and UPPS-P Data Analysis ................................... 18 Behavioral Data Analysis ........................................................................................................... 18 Psychophysiological Data Analysis ............................................................................................. 18 Results ............................................................................................................................................. 19 L2 Characterization Questionnaires and Tests and UPPS-P Results ................................................ 19 Behavioral Results ........................................................................................................................ 20 Psychophysiological Results .......................................................................................................... 24 Discussion ........................................................................................................................................ 25 References ....................................................................................................................................... 28 14 thrilling experiences), and positive urgency (PU; i.e., tendency to act rashly under extreme positive emotions; Whiteside & Lynam, 2001). Psychophysiological Measures Skin Conductance Responses (SCRs) is a measure of the electrodermal system that records the momentary increase in electrical conductance of the skin due to additional sweat in the eccrine ducts, reported as numbers of responses over a period of time. We used James One portable sensors that are connected to the computer via Bluetooth (see Moreira et al., 2019 for a quality comparison with Biopac MP36). Participants’ baseline signal was registered for one minute, while they were instructed to remain calm and looking at a fixation cross at the center of the screen. Heart Rate (HR) is a measure of the cardiovascular system that consists of the number of beats per unit of time, usually one minute. HR was also recorded using JamesOne sensors, which record the blood volume pulse waveform. SCRs and HR were collected simultaneously to behavioral data, throughout the experiment. The Gambler’s Fallacy Task The experimental task was built using PsychoPy’s software (v.2021.2.3). We applied a gambling task consisting of a card guessing game (adapted from Xue, et al., 2012b). In this task, participants are asked to decide what they think the computer's choice of cards is (either red or black card). Each trial starts with the simultaneous presentation of these two cards, on the left and right sides of the screen. First, the computer (C1) chooses one card in 1000 ms. Then, the participant (P2) guesses which card was chosen by the computer by pressing the corresponding button within 2000 ms. In our study, besides the choices of both the computer and the participant, verbal feedback in either English or Portuguese (depending on the block) was also displayed during 6000 ms to allow for a EDA measure, similar to what Agren et al. (2018) did. The total of points detained was shown at the end of the entire game. A correct decision rewarded the player with 1 point, while an incorrect decision or an absent response resulted in a loss of the same amount. It was stated in the instructions that the computer chooses the cards randomly (i.e., the probability of choosing each card was 50%) and that it was a game of chance. 15 A win followed by a change of response represents a use of the GF. Similarly, a loss followed by a maintenance of the response demonstrates the use of the GF. The participant expects that, as the streak (i.e., same card selected by the computer over a period of time) gets longer, there is a correction of the probabilities. This means that, in a win situation, the other option becomes the winning one, and that, in a loss situation, the current choice has an increasing probability of becoming the winner. This win-shiftlose-stay strategy consists in deviating from the computer's last choice. The opposite strategy is the winstay-lose-shift (WSLS), used in the remaining trials (Xue et al., 2012). To guarantee that the probability that a streak will continue or break is always 50%, the computer’s choices followed a sequence generated by a Bernoulli process ensuring: (1) equal number of black and red cards, (2) switch of card choice on half of the trials, and (3) streak length in an exponential distribution (Xue et al., 2012). The minimum streak length was, obviously, of one, and the maximum of six. This process is important because the optimal strategy for individuals in this type of game would be to choose the red or black card randomly. To verify if the task was understood, participants completed five practice trials. To guarantee the same language activation induced by the presentation of the practice trials, 5 practice trials were also displayed before the beginning of the second block. The procedure consisted of two 126-trial blocks, one with instructions and feedback in Portuguese and the other one in English. Block order was counterbalanced across participants. To minimize possible participants’ exhaustion, the blocks were separated by a pause. Similar to what Xue et al. (2012) did, the computer's last five choices were displayed on the left side of the screen to reduce the working memory load. An example of a task trial can be seen in figure 1. Figure 1 Example of an Experimental Task Trial 16 Note. The total amount of verbal feedback was 6000ms. Inter stimulus intervale (ISI, mean = 700ms) was varied to reduce expectancy effects. A fixation cross placed at the center of the screen was displayed during this time. Stimuli Positive feedback words were shown for wins and negative feedback words were shown for losses (Table 1). For each feedback valence condition, three English words and three Portuguese words were used. These words were non-cognate words (words only similar in meaning) once cognates words (words that are similar in form and meaning; Kroll & Groot, 2009), such as the Portuguese word “excelente”, would strongly activate the homonymous word in English (“excellent”; see Costa et al., 2005). Furthermore, these words were balanced for various linguistic measures known to affect processing, such as lexical frequency and length, and affective variables, such as arousal and valence, so that these parameters were equivalent between and within languages both for the positive and negative valenced words. The English words were selected from the BRM database (Warriner et al., 2013). Portuguese words were selected from ANEW-PT database and a previous pilot study (N=206 females, M=22.70 years, SD=1.53), aiming to increase the number of words in ANEW-PT database (Soares et al., 2012). Lexical frequency values (i.e., Lg10WF) were retrieved from SUBTLEX-US (Brysbaert & New, 2009) and SUBTLEX-PT (Soares et al., 2015). 17 Table 1 Feedback Words Used in the Gambling Task Procedure After the researcher explained the task, participants gave written informed consent. Excessive sweating was cleaned from participants’ hands to enhance signal quality. Then, two electrodes with solid gel were then placed on the participants’ left-hand palm (to record EDA) and another one to the index finger of same hand (to record HR). Participants could now start the experimental task by reading the instructions. Following the completion of the experimental task, the researcher presented the questionnaires, explaining what was demanded and addressing any doubts. Participants first evaluated all feedback words regarding their arousal and valence, using the self-assessment manikin scale (Bradley & Lang, 1994). They also reported the subjective frequency of those words (i.e., perceived frequency of contact with each word) using a scale from 1 (never) to 7 (several times per day). Participants whose evaluation significantly deviated from the values retrieved from the words’ database were excluded from further analysis. This was meant to serve as a manipulation check procedure. Since no major differences were found between participants ratings of the feedback words and the metrics from the databases, no data was excluded. Participants then responded to the LHQ, LexTALE, CGET, and, finally, UPPS. The questionnaires assessing the language characteristics were presented after the experimental task to prevent any possible effect on data originating from this previous exposure to English content. Arousal Valence Language Words Mean SD Mean SD N letters Frequency Portuguese Bestial 6.10 2.1 7.50 1.2 7 3 Deslumbrante 4.60 2.5 7.10 1.3 12 2.45 Espantoso 6.40 2.0 7.20 1.4 9 3.16 Errado 5.5 2.0 3.2 1.4 6.0 4.19 Engano 5.10 1.83 3.33 1.49 6 3.32 Fracasso 6.4 2.34 1.7 0.99 8 2.93 English Amazing 6.29 2.58 8 1.05 7 3.62 Awesome 5.85 3.41 8.5 0.76 7 3.20 Beautiful 5.41 2.92 7.43 1.83 9 4.15 Mistake 4.78 2.95 3.07 0.58 7 3.72 Wrong 5.43 2.34 3.5 1.22 5 4.43 Failure 4.79 2.25 1.89 1.18 7 3.01 18 Data Analysis L2 Characterization Questionnaires and Tests and UPPS-P Data Analysis Descriptive data for the LexTale, CGET and self-reported proficiency (i.e., average score of the 4 selfreported measures already mentioned) was calculated, as well as for the 5 dimensions of UPPS-P. Unidimensional reliability for each of the dimensions was computed. Behavioral Data Analysis All data from PsychoPy was firstly pre-analyzed in Excel, where we calculated the proportion of GF and the average response times, for each streak length and language. Using Jasp software (0.16.2), we calculated the descriptive statistics (e.g., mean, SD) of the same variables. Afterwards, we partially replicated the analysis of the authors that originally built the Gambler’s Fallacy Task (Xue et al., 2012). We computed a lagged logistic regression to examine the influence of streak length (1 to 6), outcome feedback (gain vs. loss), and the interaction of these two factors on the participant’s subsequent strategy (GF or WSLS). For each participant two models were computed, one for each language block. Our main interest here was the overall accuracy of these models, which indicates how predictable is the subject's behavior once we know the introduced variables. Using t-test paired samples, we compared the accuracy of the two language blocks (Portuguese vs. English) for each participant. Like Xue et al. (2012), we divided the streak length into two categories: short (1 to 3) and long (4 to 6). The reasoning for this resides in the expectation that, as mentioned earlier, for longer streak lengths, the probability of using GF strategy is higher. Thus, we conducted repeated measures ANOVA using streak length (short vs. long) and language (Portuguese vs. English) as within-subjects factors to compare the differences in the proportion of GF as well as in RT, and, thus, verify if language moderates the relationship between streak length and GF, as well as with RT. We also computed Pearsons’s correlation coefficients (or the equivalent non-parametric Spearman’s correlations when the normality assumption was not met) to examine possible correlations between the proportion of GF, RTs, accuracy of the two logistic regression models, and the results of each questionnaire. Psychophysiological Data Analysis Using EDA recorded data files and PsychoPy’s output file, we compared the onset times for feedback presentation, since studying participants reactivity to this event is our main interest using these measures. Thus, for each participant, a file of events, containing baseline and feedback onset times, feedback duration (6000ms) and conditions (language and valence) was built. Afterwards, we analysed participants 19 data individually. The data from each subject was analyzed with NeuroKit2, a Python toolbox for neurophysiological data processing, such as ElectroDermal Activity (EDA). NeuroKit2 allows users to manage data, extract events and epochs from signals, process signal, among other operations (Makowski et al., 2021). After cleaning the data, EDA was divided into its tonic (slow changes in EDA signal) and phasic (rapid and marked changes in EDA signal) components (Posada-Quintero & Chon, 2020). SCR amplitude was also computed. We extracted the events and epochs (i.e., the interval from 1s after feedback onset to 1s after its offset). Finally, means for SCR amplitude, EDA Tonic (adjusted to participants baseline response) and EDA Phasic were calculated for each condition. Unfortunately, due to technical errors, we lost a high proportion of our HR data, which made impossible its valid analysis. We also lost some of our EDA data, resulting in a total of 25 participants. Results L2 Characterization Questionnaires and Tests and UPPS-P Results The descriptive statistics for UPPS-P and L2 proficiency assessment tests are presented on table 2. The internal consistency obtained for UPPS-P achieved adequate levels for all the dimensions, ranging from 0.863 to 0.962. For self-reported proficiency questions, the internal consistency was of 0.937, also indicating an adequate level. Table 2 Descriptive statistics, normality checks, and internal consistency for L2 proficiency questionnaires and tests and UPPS-P dimensions Mean SD ShapiroWilk P-value of Shapiro-Wilk Internal Consistency (Cronbach's alpha) AoA English Speaking 7.9 2.426 0.923 .01* --- AoA English Reading 8.275 2.025 0.907 .003** --- AoA English Writing 8.45 2.195 0.907 .003** --- Estimated number of years learning English 11.6 2.753 0.977 0.576 --- 20 Percentage of time per day using Portuguese 79.644 14.393 0.902 .002** --- Percentage of time per day using English 20.356 14.393 0.902 .002** --- Self-Reported English Reading Proficiency 5.675 1.309 0.791 < .001** --- Self-Reported English Writing Proficiency 4.8 1.572 0.927 .012* --- Self-Reported English Speaking Proficiency 5.15 1.369 0.895 .001** --- Self-Reported English Listening Proficiency 5.675 1.269 0.859 < .001** --- Self-Reported Proficiency 5.325 1.276 0.923 .010* 0.937 Estimated capacity to learn new languages 4.756 1.179 0.866 < .001** --- LexTale 70.235 12.284 0.976 .535 --- CGET 17.951 4.806 0.928 .013* --- UPPS-PU 33.878 3.422 0.97 .343 0.962 UPPS-SS 30.756 3.787 0.987 .92 0.863 UPPS-NU 27.78 3.328 0.96 .158 0.91 UPPS-PM 19.073 6.976 0.922 .008** 0.938 UPPS-PS 21.707 4.445 0.953 .092 0.925 Behavioral Results Table 3 shows descriptive statistics for the main measures for both Portuguese (PT) and English (EN), and the corresponding results of normality assumption checks. More detailed descriptive data can be consulted in appendix A. Table 3 Descriptive statistics for the main measures 21 Mean SD Shapiro-Wilk P-value of Shapiro-Wilk GF (all) 0.471 0.111 0.932 0.017 GF (Portuguese) 0.477 0.131 0.978 0.604 GF (English) 0.464 0.126 0.933 0.018 GF (short streak) 0.433 0.098 0.872 < .001 GF (long streak) 0.509 0.160 0.975 0.493 GF (Portuguese, short streak) 0.439 0.103 0.923 0.009 GF (Portuguese, long streak) 0.515 0.208 0.965 0.232 GF (English, short streak) 0.426 0.121 0.926 0.011 GF (English, long streak) 0.502 0.183 0.988 0.937 RT (all) 0.539 0.096 0.976 0.524 RT (Portuguese) 0.534 0.105 0.985 0.841 RT (English) 0.544 0.117 0.954 0.097 RT (short streak) 0.544 0.103 0.968 0.286 RT (long streak) 0.534 0.100 0.962 0.179 RT (Portuguese, short streak) 0.542 0.114 0.981 0.699 RT (Portuguese, long streak) 0.525 0.123 0.977 0.562 RT (English, short streak) 0.546 0.117 0.954 0.097 RT (English, long streak) 0.542 0.128 0.953 0.092 Accuracy LRa Portuguese 65.589 8.729 0.915 0.005 Accuracy LRa Englishb 65.558 7.575 0.958 0.145 a) Logistic Regression. b) This measure presents missing data for one participant. Consistent with our expectations and previous findings (e.g., Xue et al., 2012; 2018), there was a significant effect of streak length on GF [ F (1,40) =10.934, p = .002, ƞ2= .099], such that longer streak lengths have higher GF (M=0.509,SD=0.160) than shorter streak lengths (M=0.433 SD=0.098), indicating that participants were more likely to deviate from the computer’s choice after long streaks (Figures 2 and 3). However, no statistically significant effects were found for language [ F (1,40) = 0.393, p = .534, ƞ2=.003], suggesting that language of feedback did not influence the strategy used for gambling. 22 Finally, no significant effect was found for the interaction between language and streak length on the GF [ F (1,40)= 5.158e-4, p = .982, ƞ2 =3.175e-6; Figure 2]. For RT no significant effects of streak length [ F (1,40)=0.984, p =.327, ƞ2 =.005], language [ F (1,40)=0.345, p =.560, ƞ2 =.005], nor the interaction of both factors [ F (1,40)=0.419, p =.521., ƞ2 =.002] were found. Figure 2 Differences in GF for the different streak lengths and for each language condition Note. a) shows the differences in proportion of GF for all streak lengths, and b) the differences for short streak length (<4) vs. long streaks lengths (≥4). It is clear that, for short streaks, WSLS strategy is preferred, whereas for long streaks, a shift to GF occurs, a result also obtained by Xue et al. (2018). 0 0.1 0.2 0.3 0.4 0.5 0.6 123456 Gambler's Fallacy Streak Length English Portuguese GF a) b) 23 Figure 3 Histogram of individual differences in the use of the GF strategy for short (<4) and long streaks (≥4) for each language condition The accuracy of the logistic regression models for Portuguese (M=65.589, SD=8.729) and English (M=65.558, SD=7.575) conditions revealed no statistically significant differences [ t (39)= 0.551, p = .585, d =0.087], indicating that streak length, outcome feedback and their interaction allowed a similar accuracy in the prediction of the participant’s next strategy (deviate vs follow computer’s last choice) in both Portuguese and English contexts. These accuracy values are similar to the 63.9 value obtained by Xue et al. (2018), who used an identical model. A significant positive correlation emerged between RT for Portuguese and UPPS-PS dimension ( r =0.311, p =.048). There’s also a significant negative correlation between RT in English condition and Lextale performance ( s =-0.388, p =.012) and, also, with the Selfreported proficiency ( s =-0.314, p =.048). RT in Portuguese for short streak also presents a significant correlation with UPPS-PS ( r =0.314, p =.046). The accuracy of the individual logistic regression models for English was also significantly associated with the UPPS–SS [ r =-0.342, p =.031; all correlations are presented in appendix B]. 30 Geipel, J., Hadjichristidis, C., & Surian, L. (2015). How foreign language shapes moral judgment. Journal of Experimental Social Psychology , 59 , 8-17. https://doi.org/10.1016/j.jesp.2015.02.001 Grosjean, F. (2010). Bilingual: Life and reality. Harvard University Press. https://doi.org/10.4159/9780674056459 Hadjichristidis, C., Geipel, J., & Savadori, L. (2015). The effect of foreign language in judgments of risk and benefit: The role of affect. Journal of Experimental Psychology: Applied , 21 (2), 117-129. https://doi.org/10.1037/xap0000044 Harris, C. L., Aycicegi, A., & Gleason, J. B. (2003). Taboo words and reprimands elicit greater autonomic reactivity in a first language than in a second language. Applied Psycholinguistics , 24 (4), 561. https://doi.org/10.1017/S0142716403000286 Hayakawa, S., Tannenbaum, D., Costa, A., Corey, J. D., & Keysar, B. (2017). Thinking More or Feeling Less? Explaining the Foreign Language Effect on Moral Judgment. Psychological Science, 28 (10), 1387-1397. https://doi.org/10.1177/0956797617720944 Hayakawa, S., & Keysar, B. (2018). Using a foreign language reduces mental imagery. Cognition , 173 , 8-15. https://doi.org/10.1016/j.cognition.2017.12.010 Hayakawa, S., Lau, B. K. Y., Holtzmann, S., Costa, A., & Keysar, B. (2019). On the reliability of the foreign language effect on risk-taking. Quarterly Journal of Experimental Psychology , 72 (1), 29-40. https://doi.org/10.1177/1747021817742242 Karataş, M. (2020). Making decisions in foreign languages: Weaker senses of ownership attenuate the endowment effect. Journal of Consumer Psychology , 30 (2), 296-303. https://doi.org/10.1002/jcpy.1138 Keysar, B., Hayakawa, S. L., & An, S. G. (2012). The Foreign-Language Effect: Thinking in a Foreign Tongue Reduces Decision Biases. Psychological Science , 23 (6), 661-668. https://doi.org/10.1177/0956797611432178 Kirova, A., & Camacho, J. (2021). Can You Make Better Decisions If You Are Bilingual?. Languages, 6 (1), 43. https://doi.org/10.3390/languages6010043 31 Lemhöfer, K., & Broersma, M. (2012). Introducing LexTALE: A quick and valid lexical test for advanced learners of English. Behavior research methods , 44 (2), 325-343. https://doi.org/10.3758/s13428-011-0146-0 Li, P., Sepanski, S., & Zhao, X. (2006). Language history questionnaire: A web-based interface for bilingual research. Behavior research methods , 38 (2), 202-210. https://doi.org/10.3758/BF03192770 Liu, L., Margoni, F., He, Y., & Liu, H. (2021). Neural Substrates of the Interplay between Cognitive Loading and Emotional Involvement in Bilingual Decision Making. Neuropsychologia , 151 , 1-62. https://doi.org/10.1016/j.neuropsychologia.2020.107721 Liu, S. (2021). The Behavior of Judges under the Gambler’s Fallacy [Master's thesis, University of Chicago]. Knowledge@UChicago. https://knowledge.uchicago.edu/record/2855 Luk, G., & Bialystok, E. (2013). Bilingualism is not a categorical variable: Interaction between language proficiency and usage. Journal of cognitive Psychology , 25 (5), 605-621. https://doi.org/10.1080/20445911.2013.795574 Siegel, E. H., Sands, M. K., Van den Noortgate, W., Condon, P., Chang, Y., Dy, J., Quigley, K. S., & Barrett, L. F. (2018). Emotion fingerprints or emotion populations? A meta-analytic investigation of autonomic features of emotion categories. Psychological bulletin , 144 (4), 343. http://dx.doi.org/10.1037/bul0000128 Soares, A. P., Comesaña, M., Pinheiro, A. P., Simões A. & Frade, S. (2012). The adaptation of the Affective Norms for English Words (ANEW) for European Portuguese . Behavior Research Methods, 44 (1), 256-269. https://doi.org/doi:10.3758/s13428-011-0131-7 Soares, A. P., Machado, J., Costa, A., Comesaña, M., & Perea, M. (2015). On the advantages of frequency measures extracted from subtitles: The case of Portuguese. Quarterly Journal of Experimental Psychology , 68 (4), 680-696. https://doi.org/10.1080/17470218.2014.964271 Warriner, A. B., Kuperman, V., & Brysbaert, M. (2013). Norms of valence, arousal, and dominance for 13,915 English lemmas. Behavior Research Methods, 45 , 1191-1207. https://doi.org/10.3758/s13428-012-0314-x Woumans, E., Van der Cruyssen, I., & Duyck, W. (2020). Crime and punishment: Morality judgment in a foreign language . Journal of Experimental Psychology: General, 149 (8), 1597–1602. https://doi.org/10.1037/xge0000736 32 Yang, W., Rauwolf, P., Frances, C., Molina-Nieto, O., Duñabeitia, J. A., & Thierry, G. (2022). ChineseEnglish Bilinguals Prefer Being Truthful in the Native Language. SSRN . https://dx.doi.org/10.2139/ssrn.4074355 Niszczota, P., Pawlak, M., & Białek, M. (2021). Bilinguals are less susceptible to the bias blind spot in their second language. International Journal of Bilingualism. https://doi.org/10.31234/osf.io/9vu76 Stankovic, M., Biedermann, B., & Hamamura, T. (2022). Not all bilinguals are the same: A meta-analysis of the moral foreign language effect. Brain and Language, 227, 1– 13. https://doi.org/10.1016/j.bandl.2022.105082 Miozzo, M., Navarrete, E., Ongis, M., Mello, E., Girotto, V., & Peressotti, F. (2020). Foreign language effect in decision-making: How foreign is it? Cognition, 199 , Article 104245. https://doi.org/10.1016/j.cognition.2020.104245 Vives, M. L., Aparici, M., & Costa, A. (2018). The limits of the foreign language effect on decision-making: The case of the outcome bias and the representativeness heuristic. PloS one , 13 (9), Article e0203528. https://doi.org/10.1371/journal.pone.0203528. He, Y., Margoni, F., Wu, Y., & Liu, H. (2021). fMRI evidence reveals emotional biases in bilingual decision making. Brain Structure and Function , 226 (5), 1405-1421. https://doi.org/10.1007/s00429021-02246-3 Hu, J., Li, X., Li, J., Zhang, W., Lan, Y., Gao, Z., & Gao, S. (2022). Valence-differential mechanisms of the foreign language effect in decision-making under risk. Journal of Multilingual and Multicultural Development , 1-14. https://doi.org/10.1080/01434632.2022.2043333 Puntoni, S., De Langhe, B., & Van Osselaer, S. M. (2009). Bilingualism and the emotional intensity of advertising language. Journal of Consumer Research , 35 (6), 1012-1025. https://doi.org/10.1086/595022 33 Iacozza, S., Costa, A., & Duñabeitia, J. A. (2017). What do your eyes reveal about your foreign language? Reading emotional sentences in a native and foreign language. PloS one , 12 (10), e0186027. https://doi.org/10.1371/journal.pone.0186027 Jończyk, R., Boutonnet, B., Musiał, K., Hoemann, K., & Thierry, G. (2016). The bilingual brain turns a blind eye to negative statements in the second language. Cognitive, Affective, & Behavioral Neuroscience, 16 (3), 527-540. https://doi.org/10.3758/s13415-016-0411-x Pavlenko, A. (2012). Affective processing in bilingual speakers: Disembodied cognition? International Journal of Psychology, 47 (6), 405-428. https://doi.org/10.1080/00207594.2012.743665 Harris C. L., Gleason J. B., & Ayçiçeǧi, A. (2006). When is a First Language More Emotional? Psychophysiological Evidence from Bilingual Speakers. Bilingual Minds: Emotional Experience, Expression, and Representation. (pp.257-283). Del Maschio, N., Del Mauro, G., Bellini, C., Abutalebi, J., & Sulpizio, S. (2022). Foreign to whom? Constraining the moral foreign language effect on bilinguals’ language experience. Language and Cognition , 1-23. https://doi.org/10.1017/langcog.2022.14 Fedeli, D., Del Maschio, N., Sulpizio, S., Rothman, J., & Abutalebi, J. (2021). The bilingual structural connectome: Dual-language experiential factors modulate distinct cerebral networks. Brain and Language, 220, Article 104978. https://doi.org/10.1016/j.bandl.2021.104978 Gullifer, J. W., Chai, X. J., Whitford, V., Pivneva, I., Baum, S., Klein, D., & Titone, D. (2018). Bilingual experience and resting-state brain connectivity: Impacts of L2 age of acquisition and social diversity of language use on control networks. Neuropsychologia , 117 , 123-134. https://doi.org/10.1016/j.neuropsychologia.2018.04.037 Sulpizio, S., Del Maschio, N., Del Mauro, G., Fedeli, D., & Abutalebi, J. (2020). Bilingualism as a gradient measure modulates functional connectivity of language and control networks. NeuroImage , 205 , 116306. https://doi.org/10.1016/j.neuroimage.2019.116306 34 Liu, C., Wang, H., Timmer, K., & Jiao, L. (2022). The foreign language effect on altruistic decision making: Insights from the framing effect. Bilingualism: Language and Cognition , 1-9. https://doi.org/10.1017/S1366728922000128 Soares, A. P., Comesaña, M., Pinheiro, A. P., Simões, A., & Frade, C. S. (2012). The adaptation of the Affective Norms for English words (ANEW) for European Portuguese. Behavior Research Methods , 44 (1), 256-269. https://doi.org/10.3758/s13428-011-0131-7 Huang, X., Zhang, H., Chen, C., Xue, G., & He, Q. (2019). The neuroanatomical basis of the Gambler's fallacy: A univariate and multivariate morphometric study. Human Brain Mapping, 40 (3), 967– 975. https://doi.org/10.1002/hbm.24425 Xue, G., Juan, C. H., Chang, C. F., Lu, Z. L., & Dong, Q. (2012b). Lateral prefrontal cortex contributes to maladaptive decisions. Proceedings of the National Academy of Sciences , 109 (12), 44014406. https://doi.org/10.1073/pnas.1111927109 Xue, G., He, Q., Lei, X., Chen, C., Liu, Y., Chen, C., Lu, Z., Dong, Q., & Bechara, A. (2012a). The gambler’s fallacy is associated with weak affective decision making but strong cognitive ability. PLoS ONE, 7 (10), Article e47019. https://doi.org/10.1371/journal.pone.0047019 Kroll, J. F., & De Groot, A. M. (2009). Handbook of bilingualism: Psycholinguistic approaches . Oxford University Press. Makowski, D., Pham, T., Lau, Z. J., Brammer, J. C., Lespinasse, F., Pham, H., Schölzel, C., & Chen, S. H. (2021). NeuroKit2: A Python toolbox for neurophysiological signal processing. Behavior Research Methods , 53 (4), 1689-1696. https://doi.org/10.3758/s13428-020-01516-y Silva Moreira, P., Chaves, P., Dias, R., Dias, N., & Almeida, P. R. (2019). Validation of wireless sensors for psychophysiological studies. Sensors, 19 (22), 4824. https://doi.org/10.3390/s19224824 Posada-Quintero, H. F., & Chon, K. H. (2020). Innovations in electrodermal activity data collection and signal processing: A systematic review. Sensors , 20 (2), 479. https://doi.org/10.3390/s20020479 35 Tversky, A., & Kahneman, D. (1971). Belief in the law of small numbers. Psychological bulletin , 76 (2), 105. https://doi.org/10.1037/h0031322 Whiteside, S. P., & Lynam, D. R. (2001). The Five Factor Model and impulsivity: Using a structural model of personality to understand impulsivity. Personality and Individual Differences , 30 (4), 669-689. https://doi.org/10.1016/S0191-8869(00)00064-7 Xue, G., He, Q., Lei, X., Chen, C., Liu, Y., Chen, C., Liu, Y., Chen, C., Lu, Z., Dong, Q. & Bechara, A. (2012). The gambler’s fallacy is associated with weak affective decision making but strong cognitive ability. PloS one , 7 (10), e47019. https://doi.org/10.1371/journal.pone.0047019 Xue, G., Juan, C. H., Chang, C. F., Lu, Z. L., & Dong, Q. (2012). Lateral prefrontal cortex contributes to maladaptive decisions. Proceedings of the National Academy of Sciences of the United States of America , 109 (12), 4401-4406. https://doi.org/10.1073/pnas.1111927109 Zhang, Y., De Zoysa, A., & Cortese, C. L. (2022). Foreign language effect in accounting uncertainty expressions: Interpretation and probabilistic estimation. Journal of International Accounting, Auditing and Taxation. Zheng, L., Mobbs, D., & Yu, R. (2020). The behavioral and neural basis of foreign language effect on risktaking. Neuropsychologia , 136 , 1-8. https://doi.org/10.1016/j.neuropsychologia.2019.107290 36 Appendix A Table A1 Proportion of trials using the GF strategy following different streak lengths in each language Table A2 Mean Reaction Times in milliseconds following different streak lengths in each language condition Streak Length 1 2 3 4 5 6 Portuguese Mean 0.552 0.544 0.532 0.544 0.523 0.509 Standard Deviation 0.120 0.141 0.113 0.119 0.191 0.220 English Mean 0.546 0.547 0.545 0.565 0.576 0.485 Standard Deviation 0.124 0.124 0.127 0.183 0.182 0.200 Appendix B Streak Length 1 2 3 4 5 6 Portuguese Mean 0.401 0.428 0.489 0.427 0.545 0.573 Standard Deviation 0.103 0.126 0.192 0.205 0.310 0.363 English Mean 0.379 0.410 0.489 0.404 0.530 0.573 Standard Deviation 0.114 0.149 0.183 0.218 0.280 0.363 37 UPPS-P LexTale CGET Self-Reported Proficiency PU SS NU PM PS GF (All) 0.026 (p=.874)b 0.006 (p=.971) b -0.023 (p=.886)b -0.084 (p=.600) b -0.110 (p=.494) b 0.116 (p=.469) b 0.207 (p=.195) b 0.161 (p=.319) b GF (Portuguese) -0.108 (p=.503) -0.082 (p=.611) 0.049 (p=.761) 0.005 (p=.978) -0.175 (p=.272) --- 0.290 (p.=066) 0.099 (p=.543) GF (English) 0.177 (p=.268) b 0.034 (p=.831) b -0.093 (p=.563) b -0.123 (p=.445) b 0.056 (p=.727) b 0.079 (p=.625) b 0.136 (p=.397) b 0.147 (p=.366) b GF (short streak) 0.031 (p=.847) b 0.146 (p=.364) b -0.306 (p=.052)*b -0.280 (p=.076) b -0.139 (p=.387) b -0.052 (p=.746) b 0.012 (p=.941) b 0.014 (p=.933) b GF (long streak) 0.062 (p=.702) -0.023 (p=.888) 0.120 (p=.453) 0.020 (p=.899) b -0.045 (p=.780) 0.206 (p=.197) 0.317 (p=.044)*b 0.196 (p=.226) GF (Portuguese, short streak) -0.074 (p=.644) b 0.173 (p=.279) b -0.194 (p=.225) b -0.232 (p=.145) b -0.293 (p=.063) b 0.130 (p=.419) b 0.117 (p=.465) b -9.417e-4 (p=.995) b GF (Portuguese, long streak) -0.114 (p=.478) -0.112 (p=.485) 0.071 (p=.658) 0.153 (p=.339) -0.126 (p=.434) 0.090 (p=.578) 0.214 (p=.178) 0.104 (p=.524) GF (English, short streak) 0.133 (p=.405) b 0.150 (p=.348) b -0.297 (p=.060) b -0.321 (p=.041)*b -0.031 (p=.848) b -0.145 (p=.366) b -0.051 (p=.753) b 0.075 (p=.645) b GF (English, long streak) 0.238 (p=.134) 0.088 (p=.584) 0.130 (p=.417) -0.110 (p=.495) 0.064 (p=.689) 0.259 (p=.102) 0.357 (p=.022)*b 0.226 (p=.160) RT (All) -0.134 (p=.402) -0.161 (p=.315) 0.074 (p=.644) 0.277 (p=.079) b 0.305 (p=.053)* -0.348 (p=.026)* -0.167 (p=.296) -0.250 (p=.120) b RT (Portuguese) 0.023 (p=.887) -0.145 (p=.364) 0.127 (p=.427) 0.230 (p=.148) b 0.311 (p=.048)* -0.226 (p=.156) -0.077 (p=.634) -0.171 (p=.290) b RT (English) -0.241 (p=.128) -0.134 (p=.402) 0.008 (p=.959) 0.167 (p=.295) 0.222 (p=.163) -0.388 (p=.012)*b -0.164 (p=.306) -0.314 (p=.048)* b RT (short streak) -0.144 (p=.369) -0.181 (p=.257) 0.120 (p=.456) 0.179 (p=.262) 0.284 (p=.071) -0.369 (p=.018)* -0.118 (p=.464) -0.223 (p=.166) b RT (long streak) -0.110 (p=0.495) -0.122 (p=.446) 0.019 (p=.904) 0.164 (p=.304) 0.292 (p=.064) -0.269 (p=.089) b -0.200 (p=.210) -0.289 (p=.070) b RT (Portuguese, short streak) -0.098 (p=.540) -0.181 (p=.258) 0.198 (p=.214) 0.296 (p=.060) b 0.314 (p=.046)* -0.305 (p=.053)* -0.085 (p=.595) b -0.054 (p=.739) b RT (Portuguese, long streak) 0.131 (p=.415) -0.080 (p=.619) 0.033 (p=.837) 0.057 (p=.724) 0.238 (p=.133) -0.102 (p=.526) -0.031 (p=.845) b -0.245 (p=.128) 38 a) Logistic Regression. b) Spearman’s coefficient. UPPS-P LexTale CGET Self-Reported Proficiency PU SS NU PM PS RT (English, short streak) -0.158 (p=.324) -0.143 (p=.371) 0.018 (p=.911) 0.113 (p=.480) 0.196 (p=.220) -0.359 (p=.021)* b -0.060 (p=.710) -0.329 (p=.038)* b RT (English, long streak) -0.297 (p=.060) -0.114 (p=.477) -0.002 (p=.992) 0.202 (p=.205) 0.227 (p=.154) -0.345 (p=.027)* b -0.242 (p=.128) b -0.307 (p=.054)* b Accuracy LRa Portuguese -0.051 (p=.751) b 0.052 (p=.745) b -0.026 (p=.869) b -0.179 (p=.262) b 0.047 (p=.769) b -0.221 (p=.165) b -0.220 (p=.167) b -0.006 (p=.972) b Accuracy LRa English 0.043 (p=.793) -0.342 (p=.031)* 0.143 (p=.377) 0.131 (p=.422) 0.120 (p=.461) -0.068 (p=.676) 0.002 (p=.989) -0.075 (p=.649) Anexo: Formulário de identificação e caracterização do projeto Conselho de Ética Comissão de Ética para a Investigação em Ciências Sociais e Humanas Identificação do documento: CEICSH 120/2021 Relatores: Emanuel Pedro Viana Barbas Albuquerque e Marlene Alexandra Veloso Matos Título do projeto: The Foreign Language Effect and the Gambler’s Fallacy: Evidence from Neurophysiology Equipa de Investigação: Pedro Miguel Silva Moreira, Centro de Investigação em Psicologia (CIPsi), Escola de Psicologia, Universidade do Minho; Ana Paula Carvalho Soares, Centro de Investigação em Psicologia (CIPsi), Escola de Psicologia, Universidade do Minho; Márcia Costa, Mestrado Integrado em Psicologia, Escola de Psicologia, Universidade do Minho PARECER A Comissão de Ética para a Investigação em Ciências Sociais e Humanas (CEICSH) analisou o processo relativo ao projeto de investigação acima identificado, intitulado The Foreign Language Effect and the Gambler’s Fallacy: Evidence from Neurophysiology. Os documentos apresentados revelam que o projeto obedece aos requisitos exigidos para as boas práticas na investigação com humanos, em conformidade com as normas nacionais e internacionais que regulam a investigação em Ciências Sociais e Humanas. Face ao exposto, a Comissão de Ética para a Investigação em Ciências Sociais e Humanas (CEICSH) nada tem a opor à realização do projeto nos termos apresentados no Formulário de Identificação e Caracterização do Projeto, que se anexa, emitindo o seu parecer favorável, que foi aprovado por unanimidade pelos seus membros. Braga, 11 de janeiro de 2022. O Presidente da CEICSH (Acílio Estanqueiro Rocha)