Highlights

• • Sensitive measures are needed to assess the effects of bilingualism on executive functions (EFs).

• • Bilingualism is made up of several complex experiences that interact with EF.

• • We used mouse-tracking to examine the relation between L2 experiences and EF.

• • L2 age of acquisition and switching frequency predicted EF on Stroop and Flanker tasks.

• • The effects of L2 experiences on EF depend on the attentional demands of the task.

1. Different bilingual experiences predict different executive functions: Evidence from mouse-tracking

Cognitive psychology relies on fundamental tools such as accuracy and reaction time (RT) to glean insights into cognitive processes, decision-making and response strategies. These variables, being both cost-effective and universally applicable across various paradigms, have been the cornerstone of cognitive research. However, their limitations become apparent when it comes to elucidating the dynamic nature of decision-making processes over time (Freeman & Ambady, Reference Freeman and Ambady2010). When canonical measures fall short and risk overinterpretation, it becomes imperative to explore newer and more nuanced metrics. Therefore, in this study, we used a dynamic mouse-tracking paradigm to dissect bilinguals’ response selection and attentional processes.

Bilingualism is one of the important experiences that increases neuroplasticity and reshapes executive functions (EFs), which are defined as higher-order cognitive functions such as working memory capacity, inhibitory control and task switching, all of which involve attentional control (Bialystok, Reference Bialystok2017; Diamond, Reference Diamond2013; Friedman & Miyake, Reference Friedman and Miyake2017). Bilinguals aptly navigate competition between two active languages in their minds (Kroll et al., Reference Kroll, Bobb and Hoshino2014, Reference Kroll, Dussias, Bice and Perrotti2015; Thierry & Wu, Reference Thierry and Wu2007; Wu & Thierry, Reference Wu and Thierry2010), fostering neural adaptations that exert a lasting impact on EF regions (reviews in Deluca et al., Reference DeLuca, Segaert, Mazaheri and Krott2020; Grant et al., Reference Grant, Dennis and Li2014; Grundy et al., Reference Grundy, Anderson and Bialystok2017a; Pliatsikas, Reference Pliatsikas2020). These neural adaptations often translate into bilinguals outperforming their monolingual counterparts on nonverbal EF tasks (e.g., Bialystok, Reference Bialystok2009; Bialystok et al., Reference Bialystok, Craik, Klein and Viswanathan2004; Bialystok et al., Reference Bialystok, Poarch, Luo and Craik2014; Costa et al., Reference Costa, Hernández, Costa-Faidella and Sebastián-Gallés2009; Prior & MacWhinney, Reference Prior and MacWhinney2010; Singh et al., Reference Singh, Fu, Rahman, Hameed, Sanmugam and Agarwal2015). Nonetheless, these effects are not universally replicated (reviews in Gunnerud et al., Reference Gunnerud, Ten Braak, Reikerås, Donolato and Melby-Lervåg2020; Grundy, Reference Grundy2020; Lehtonen et al., Reference Lehtonen, Soveri, Laine, Järvenpää, De Bruin and Antfolk2018; Ware et al., Reference Ware, Kirkovski and Lum2020).

Two potential reasons for this inconsistency arise. First, the cognitive measures may lack the necessary sensitivity to delve beyond conventional metrics, thereby failing to capture the intricate aspects of decision-making over time. Second, the wide array of individual differences in bilingual experiences is often overlooked (Yurtsever et al., Reference Yurtsever, Anderson and Grundy2023). Despite the recent emphasis on recognizing bilingualism as a complex construct interacting with sociolinguistic variables and contexts (Gullifer & Anderson, Reference Gullifer and Anderson2023; Backer & Bortfeld, Reference Backer and Bortfeld2021; Chung-Fat-Yim et al., Reference Chung-Fat-Yim, Calvo and Grundy2022; Leivada et al., Reference Leivada, Westergaard, Duñabeitia and Rothman2021; Luk & Grundy, Reference Luk and Grundy2023; Pot et al., Reference Pot, Keijzer and De Bot2018; Titone & Tiv, Reference Titone and Tiv2023; Yamasaki et al., Reference Yamasaki, Stocco and Prat2018), many studies examining the relations between bilingualism and EFs still dichotomize groups into “monolingual” and “bilingual.” This oversimplified approach can be problematic given that it overlooks crucial linguistic and contextual variations that could potentially obscure real effects (Grundy, Reference Grundy2020; Yurtsever et al., Reference Yurtsever, Anderson and Grundy2023).

In light of these considerations, the present study employed a sensitive and dynamic mouse-tracking paradigm (Freeman & Ambady, Reference Freeman and Ambady2010) to examine how individual bilingual experiences are related to EFs. This approach is a cost-effective alternative to functional neuroimaging. Neuroimaging techniques have long been useful in examination of cognitive efficiency between monolinguals and bilinguals. We will first discuss neuroimaging evidence and then compare how mouse-tracking can complement and extend the current dialogue.

Neuroimaging techniques such as electroencephalography (EEG) and functional magnetic resonance imaging provide one path to examine cognitive processes that are not observable at the behavioral level alone and can measure the efficiency of performance on EF tasks (Luck, Reference Luck2014; van Heuven & Coderre, Reference van Heuven and Coderre2015). For example, Abutalebi et al. (Abutalebi et al., Reference Abutalebi, Della Rosa, Green, Hernandez, Scifo, Keim, Cappa and Costa2011) compared German–Italian bilinguals and Italian monolinguals on a Flanker task and found that bilinguals used the anterior cingulate cortex (ACC) less than monolinguals in dealing with the conflicting stimuli, despite similar behavioral performance. The ACC is highly involved with monitoring and moderating conflict in the brain (Botvinick et al., Reference Botvinick, Braver, Barch, Carter and Cohen2001; Carter et al., Reference Carter, Braver, Barch, Botvinick, Noll and Cohen1998; Shenhav et al., Reference Shenhav, Botvinick and Cohen2013; Sheth et al., Reference Sheth, Mian, Patel, Asaad, Williams, Dougherty and Eskandar2012); thus, Abutelabi et al.’s findings demonstrate that their bilinguals were more efficient at handling conflict resolution in the ACC than monolinguals. The finding that bilinguals use the ACC more efficiently than monolinguals has been replicated several times with different designs, populations and tasks (e.g., Anderson et al., Reference Anderson, Grundy, Grady, Craik and Bialystok2021; Rodríguez-Pujadas et al., Reference Rodríguez-Pujadas, Sanjuán, Fuentes, Ventura-Campos, Barrós-Loscertales and Ávila2014).

EEG studies provide complementary findings for more efficient cognitive processing in bilinguals than monolinguals. Zunini et al. (Reference Zunini, Morrison, Kousaie and Taler2019) used a task-switching paradigm to demonstrate that French–English bilinguals had smaller switching and mixing costs than monolinguals, with mostly similar electrophysiological patterns in the brain, suggesting that bilinguals need the same number of resources to perform the task better than monolinguals. However, bilinguals did have larger N2 and smaller P3 amplitudes than monolinguals. These patterns suggest that bilinguals prioritize early attention to facilitate faster task-set reconfiguration at later stages in processing, consistent with more efficient processing for bilinguals than monolinguals. Recent reviews of the literature suggest that these are common findings: bilinguals demonstrate enhanced processing at early and more automatic stages to alleviate the burden from later stages of processing, which is a sign of efficiency (Antoniou, Reference Antoniou2023; Grundy et al., Reference Grundy, Anderson and Bialystok2017a). Together, neuroimaging techniques provide a useful avenue by which to examine the efficiency of processing on EF tasks, but the cost and accessibility associated with such technology limits its use. Fortunately, alternative methods, such as MouseTracker (Freeman & Ambady, Reference Freeman and Ambady2010), provide a behavioral solution to examine multiple cognitive processes and response strategies before response execution, without the cost (Spivey et al., Reference Spivey, Grosjean and Knoblich2005).

The prototypical mouse-tracking paradigm involves a binary forced-choice decision task between two options represented as buttons on the computer screen (Schoemann et al., Reference Schoemann, O’Hora, Dale and Scherbaum2021). Mouse-tracking involves using a computer mouse to measure x- and y-axis coordinates as participants initiate and move toward a target over time. Motor responses, such as the movement of one’s arm to select an option, are frequently updated through our perceptual and cognitive processing. These measurements of ongoing and continuous movement metrics, as measured through mouse-tracking software, give insights into the dynamics of cognition and decision-making (Freeman & Ambady, Reference Freeman and Ambady2010). This allows for a dynamic set of measures to be calculated, including the time it takes to initiate a movement after the stimulus appears, and the deviation from the most direct path to the correct response (see Figure 1). These mouse-tracking metrics have been shown to be more reliable than overall RTs (Grundy, Reference Grundy2022).

Figure 1.

The mouse-tracking paradigm that allows for the examination of several behaviors that occur before response completion. Reused with permission from Grundy (Reference Grundy2022).

Spivey et al. (Reference Spivey, Grosjean and Knoblich2005) were the first to implement a mouse-tracking methodology to capture continuous cognitive dynamics during a spoken language task. Participants were presented with two objects on the screen, one target and one competing, displayed in the upper right or left corners. Participants were given auditory instructions on which stimulus to select with a mouse click, 500 ms after the images were presented. Competing stimuli were chosen to assess the impact of competition on lexical activation via motor output. The trials consisted of either phonologically dissimilar (control items; e.g., fountain-chair) or similar words (distractor items; e.g., sink-rink). Results showed that the distractor items yielded longer total response times than the control items. In addition, distractor items yielded greater deviations from the target compared to the control condition, indicating larger effects of spatial attraction toward the competing object in the distractor conditions than in the control conditions.

Bartolotti and Marian (Reference Bartolotti and Marian2012) investigated language competition between monolinguals and bilinguals using eye-tracking and mouse-tracking methodologies. Participants learned words in an artificial language (e.g., shundo → acorn) and then completed a word recognition task. In this task, they clicked on pictures corresponding to the translation of the learned words (e.g., a picture of an acorn as the target, or a shovel as the phonological competitor). Their findings revealed no group differences in accuracy and response-time metrics. However, the competitors affected the mouse movements of monolinguals and bilinguals differently. Monolinguals exhibited greater deviation along the x-coordinates toward the competitors than targets. In contrast, competitors did not affect bilinguals’ x-coordinate motion but did affect their y-coordinates. This pattern suggested that bilinguals were equally drawn to both the target and competitor. In other words, they experienced language competition between responses, but they resolved it earlier and more successfully than monolinguals. Moreover, bilinguals used executive control not to eliminate interference completely, but to maintain goals.

Incera and McLennan (Reference Incera and McLennan2016) used mouse-tracking to examine Stroop task performance between monolinguals and bilinguals. The authors demonstrated that bilinguals took longer to initiate their responses, but once movement toward the target began, they were more efficient in their path to the correct response. Incera and McLennan concluded that bilinguals behaved more like experts than monolinguals, in that they took longer to process initial demands to facilitate performance at later stages of response execution. Incera and McLennan (Reference Incera and McLennan2018) followed this study using a diverse sample of bilinguals from different language backgrounds and examining bilingualism on a continuum. The authors found that higher bilingualism scores, as measured by the percentage of time exposed to a second language, were associated with a reduced Stroop effect (incongruent minus congruent trials), as measured by a smaller deviation toward the incorrect response (i.e., more direct path to correct response) on incongruent than congruent trials. This finding was only borne out of the Stroop task, not the Flanker task, and the authors suggested that this could be because bilingualism leads to better prepotent response inhibition (EF component assessed on the Stroop task) rather than resistance to distractor interference (EF component assessed on the Flanker task). Alternatively, it could be that bilingualism is more likely to influence EF on more demanding tasks (Stroop) than less demanding tasks (Flanker). Kousaie and Phillips (Reference Kousaie and Phillips2017) further highlighted the importance of examining both tasks, given that bilinguals were faster and more accurate on the Stroop task than monolinguals. However, bilinguals only outperformed monolinguals on accuracy of the Flanker task, not RTs. They also demonstrated using ERPs that distinct cognitive processing between monolinguals and bilinguals emerged depending on whether the Flanker or Stroop task was used. These findings highlight the importance of examining both the Stroop and Flanker tasks within the same experiment.

The Stroop findings from these experiments complement Incera and McLennan’s previous work by demonstrating that bilingualism enhances efficiency of performance on an EF task that cannot be measured with RTs alone. More recently, Damian et al. (Reference Damian, Ye, Oh and Yang2019) showed that Chinese–English bilinguals had smaller maximum deviations and area under the curve (AUC) for the most direct path to the correct response than English monolinguals, consistent with the previous work. Additionally, bilinguals had significantly later initiation times for both the Simon and spatial Stroop tasks than monolinguals, again consistent with the previous work. However, contrary to the previous work, bilinguals were earlier to initiate responses than monolinguals for the Flanker task. One of the reasons for this discrepancy in initiation times for the Flanker task compared to the Stroop task observed in Damian et al. (Reference Damian, Ye, Oh and Yang2019) could be that the Flanker task is less challenging, and more demanding tasks are required to observe an effect of bilingualism on EF. Alternatively, individual bilingual experiences could influence outcomes in different directions. Previous work has shown that age of L2 acquisition (L2 AoA; Deluca et al., Reference DeLuca, Rothman, Bialystok and Pliatsikas2019), L2 proficiency (Singh & Mishra, Reference Singh and Mishra2013; Tse & Altarriba, Reference Tse and Altarriba2012), L2 use (Qi et al., Reference Qi, Qin, Zhang and Yun2023), L1/L2 balance (Yow & Li, Reference Yow and Li2015), and L1/L2 switching frequency (Liu et al., Reference Liu, Yang, Jiao, Schwieter, Sun and Wang2019; Soveri et al., Reference Soveri, Laine, Hämäläinen and Hugdahl2011a, Reference Soveri, Rodriguez-Fornells and Laine2011b) all influence EFs in the brain. Thus, it is possible that variations in the L2 experiences of the sample populations across the studies drove the outcomes observed.

To resolve discrepancies in the literature, the present study used a dynamic mouse-tracking paradigm to examine the individual influences of different second-language experiences on initiation times and movement dynamics on conflict tasks. We tested young adult bilinguals with a Flanker and a Stroop task and examined conflict effects and sequential congruency effects (SCEs; Gratton et al., Reference Gratton, Coles and Donchin1992). SCEs measure the extent to which individuals’ responses are influenced by the preceding trial, shedding light on their ability to swiftly disengage attention—a valuable proxy for executive functioning. Specifically, when the previous trial is incongruent, the conflict effects that follow (incongruent–congruent) are smaller than when the previous trial is congruent. Smaller SCEs are typically interpreted as indicative of more effective conflict management (Gratton et al., Reference Gratton, Coles and Donchin1992), and SCEs become smaller with practice (Egner et al., Reference Egner, Ely and Grinband2010). Conflict and SCEs are two EF effects that have been shown to be modified by bilingualism by some (e.g., Costa et al., Reference Costa, Hernández, Costa-Faidella and Sebastián-Gallés2009; Grundy et al., Reference Grundy, Chung-Fat-Yim, Friesen, Mak and Bialystok2017b), but not others (e.g., Morton & Harper, Reference Morton and Harper2007; von Bastian et al., Reference von Bastian, Souza and Gade2016).