Tracing the bilingual advantage in cognitive control: The role of flexibility in temporal preparation and category switching

Abstract

The present study addressed the question whether bilinguals are characterised by increased cognitive flexibility. Mechanisms of cognitive flexibility were compared between a group of Hungarian-Polish bilinguals and a group of Hungarian monolinguals. The first task explored the effects of temporal orienting (ability to voluntarily orient attention to a certain point in time when a relevant event is expected) and the efficiency of switching between preparatory time intervals of different duration (sequential effects). The second task – the social category switching task – tapped into the mechanisms of switching between 2 types of categories (age and gender) and employed socially relevant stimuli (faces). The results of the first task revealed similar temporal orienting effects for both groups; however, the pattern of sequential effects differed between the groups, showing that bilinguals were less affected by the duration of the preceding preparatory interval. In the social category switching task, bilinguals showed reduced switch costs in the RT measure when categorising gender, and greater accuracy in the specific switch and no-switch conditions. We suggest that bilinguals are characterised by an enhanced mechanism of cognitive flexibility, which is applied to a temporal domain (efficient switching between preparatory intervals of different duration), and extends to the cognitive control processes in social categorisation tasks.

Keywords:

• Bilingualism
• Cognitive flexibility
• Temporal preparation
• Switching
• Sequential effects
• Social categorisation

This research was supported by the European Society for Cognitive Psychology Early Career Stimulus awarded to AM and by the Foundation for Polish Science (FNP) subsidy awarded to ZW (FOCUS programme). AM, MB, and ZW were also supported by the same FNP subsidy and JL was supported by a research grant PSI2011–22416 from the Spanish Ministry of Innovation. We thank Samyogita Hardikar for proofreading previous versions of the manuscript and to Borysław Paulewicz for his help with data analysis. We also wish to thank to the Józef Bem Polish Cultural Society of Budapest and the Polish Institute in Budapest for allowing us to conduct the study in their facilities.

Notes

¹ An analogous version of the task that employed emotional and gender stimuli was developed by López-Benítez, Carretero-Dios, Acosta, and Lupiáñez (2012).

² The pattern of data did not change when data from 1 participant (whose mean RT was above 3 standard deviations from the overall mean) were excluded and data were normalised (by means of log-transformation).

³ RT data in the SCST task were normally distributed, as indicated by Kolmogorov–Smirnov tests for RTs in each condition of the task (p>.05). We performed an additional analysis to answer the question whether the gender of participants interacts with the switch costs for female and male gender categorisation. We calculated switch costs separately for female and male pictures, and conducted ANOVA with gender of stimuli (male, female) as a within-subject factor and gender of participant (male, female) as a between-subjects factor. The main effect of gender of stimuli was not significant (F<1), neither was the main effect of group of participants (F<1). However, we observed a significant interaction between gender of stimuli and gender of participants, F(1, 41)=4.25, p=.046, : for female participants, the magnitude of switch costs were larger when they were categorising male pictures as compared to when categorising female pictures (156 ms vs. 134 ms), whereas for male participants the pattern was reversed, the magnitude of switch costs was larger when the categorisation of a female picture was required as compared to the categorisation of a male picture (141 ms vs. 112 ms).

⁴ In the SCST task RT and ERR variables were – in general – positively and significantly correlated (except for 2 conditions, namely the complete repetition, no-switch, gender task condition and the complete repetition, no-switch, age task condition). RT and ERR in the SCST task therefore seem to be related measures of cognitive flexibility. Taking into consideration that the effects observed on the error proportion scales are usually nonlinear, we additionally performed mixed effect logistic regression on the ERR data. The mixed effect logistic regression showed a pattern of results very similar to that obtained with ANOVA. The results clearly showed that bilinguals were significantly more accurate than monolinguals when switching into another task under condition of complete repetition of stimuli, and under condition of partial repetition of stimuli (but in this case only when switching into the gender task). Bilinguals also showed a significantly higher accuracy on no-switch trials: under the partial repetition condition (both for the gender and age task) and the complete alternation condition (only for the gender task).

⁵ This might be due to the fact that age classification requires a distinction between less well-delimited classes (young/old), whereas gender judgements are based on more qualitative distinctions between 2 well-defined categories (men/women) (see Mouchetant-Rostaing, & Giard, 2003). It has been shown that the visual clues on which gender judgements are based are more isolated (eye and eyebrow regions) and more prominent than those (skin texture and colour) required for age judgements (Brown & Perrett, 1993; Roberts & Bruce, 1988).