This is a review of the lecture by Dr. Ellen Bialystok[1], distinguished Research Professor of Psychology at York University, who examines studies of cognitive development and decline across the lifespan with a focus on how these processes are modified by the selective attention that the bilingual experience engenders.

The talk addresses the title question by posing whether the concept of neuroplasticity, as a proxy for measurable changes in brain mechanisms and functions, can be also led by bilingualism. As language seems to be the most widely used brain activity, it promotes constant activation. For bilinguals, this means simultaneous and constant need for selection. She contends that a primary consequence is a reorganization in attentional networks. In so doing, bilingualism would lead to lifelong adaption of how attention is engaged.

Research developed with subjects from infancy to old age shows differences in their response to the effects of bilingualism. She draws attention to recent discussions on the discrepancies in the research. A vast majority of the controversy, though, relates to accuracy and response time (RT) by young adults.

A widely replicated article (BIALYSTOK; MCBRIDE-CHANG; LUK, 2005[2]) relates how results for a Simon task (congruent and incongruent trials) show identical superior patterns for bilinguals (BL) compared to monolinguals (ML) across all age brackets. However, another study (PAAP; GREENBERG, 2013[3]) with identical results gave an oppositional interpretation. And that started a series of disputes on the research basis so far. The contradictory effects found claim for an explanation, though.

With a track record of twenty years, cognitive differences found in tasks of executive functions (EF) abilities have been performed under the unity and diversity model (MIYAKI et al., 2000[4]). This structural model consists of three elements: updating, shifting and inhibition. Performance on the tasks reflects the engagement of each component separately. Within each component, several tasks are aligned to show its function. The integrity in the component reflects its independence, and the theoretical framing dictates the hypothesis generation and reporting of results for empirical predictions.

The model served as a starting point for bilingualism research and has shaped predictions issued by research since then. But problems emerged: the predicted correlations reflecting inhibitions were not accurately found; and across a wide range, research kept showing a higher RT for BLs (BIALYSTOK et al., 2004[5]; MARTIN-RHEE; BIALYSTOK, 2008[6]; EMMOREY et al., 2008[7]; CHUNG-FAT-YIM; HIMEL; BIALYSTOK, 2019[8]). This led to the conclusion that the issue may not relate to inhibition.

Additionally, this issue pointed to a conflation of inhibition. Research with developmental and fMRI data showed that there are two kinds of inhibition: the one used to execute a response (response inhibition) and another that requires interference suppression (BUNGE et al., 2002[9]; MARTIN-RHEE; BIALYSTOK, 2008[6]). In bilingual research, response inhibition shows no group differences. However, for interference suppression, BLs are faster than MLs. Taken together, research establishes that there is no general inhibition process.

Choices to resolve the issue fall between two possibilities: either that predictions are wrong and there is no cognitive effect of bilingualism, or that the model is wrong. In revisiting the model, relations between components are indeed questionable and problematic if predictions be taken seriously. Alternatively, a look at a common denominator – attention - is of use as where it is placed answers for performance on the tasks. Thus, the effect of bilingualism may be better found in the processes that modulate attention.

A revision of studies for all age brackets was then performed to separate information about RT from attention mechanisms (COMISHEN; BIALYSTOK; ADLER, 2019[10]; GRUNDY; ANDERSON; BIALYSTOK, 2017[11]). Taken together, results point towards a higher RT for MLs (a sequential congruent effect), i.e., MLs take longer to switch between tasks, meaning that BLs surpass monolinguals in allocating their attention.

Statistically meticulous research performed by Zhou and Krott (2018[12]) investigated RT distribution and attention with exponential gaussian analysis. They examined performance of MLs and BLs in the three big tasks for EF function (Stroop, Flanker and Simon). Results show that, regarding RT, MLs performed the same as BLs. However, for attention, MLs took longer than BLs. This gives the modulation of attention hypothesis a robust basis.

To examine language processing, data from a recent study by Bialystok et al. (2020[13]) show that BLs fare worse than MLs in linguistic tasks (vocabulary scores). However, when combining linguistic benefits for MLs and executive function benefits for BLs, data show that even linguistic tasks requiring selection and attention yield results similar to nonverbal EF tasks.

The controversy then lies in determining why, among so many positive bilingual effects, a consistent pattern (within the component model) does not emerge. To choose between a rock and hard place, some points had to be analyzed, such as confounding factors (like socioeconomic status or SES), publication bias and sample sizes in relation to positive results rejections.

To confront SES and bilingualism confounding effects, a study by Hartanto and colleagues (2018[14]) shows that, besides SES being the most important predictor of EFs, there is an interaction effect, i.e., bilingualism is especially helpful for those in the lower SES strata. Another study by Brito and Noble (2018[15]) examined cortical brain volume. Results show that bilingualism boosts brain development in the most disadvantaged groups. Regarding publication bias, as it does not challenge the validity of the results, it is not a validity check. And for sample size, method emerges as more important than size. As for null results, a definition of bilingualism needs more scrutiny.

Bilingualism is taken as an independent variable and contended as being a continuum that encompasses a full range of experiences. These in turn could depict analyses of degrees in relation to outcomes with more granularity. Many studies show complex non-linear relations (brain/behavior) that reflect how different bilingual experiences are important for outcomes (ANDERSON et al., 2018[16]; BIALYSTOK; CRAIK; LUK, 2008[17]; CHUNG-FAT-YIM et al., 2020[18]; DELUCA et al., 2019[19]; GUERRERO; SMITH; LUK, 2016[20]). In examining reasons to reject null results, the type of tasks used are demystified as demands on processing matter more than their classification per se. And finally, relative to the context of language use, adaptive control models showing how languages are used in context could provide better answers (such as the one by GREEN; ABUTALEBI, 2013[21]).

In sum, EFs remains central to cognitive processing and development, academic achievement, and long-term wellbeing. Additionally, they are associated with mental resilience. But the biggest payoff, argued by Bialystok, lies in old age; in the face of normal cognitive decline for which there is no effective intervention (as in dementia), the single resource is cognitive reserve. This means maintaining high levels of cognitive function through sustained use, and it implies a dissociation between what is happening in the brain (decline) as compared to what happens in the mind (cognitive function preservation). The brain can show some atrophy, but that may not hold for cognitive function.

A clear indication of this is the age of diagnosis for dementia and, across studies, MLs show signs earlier than BLs (KOWOLL et al., 2016[22]; PERANI et al., 2017[23]). A proxy for dementia advancement is in the levels of metabolic glucose uptake, which get lower as dementia increases. By this measure, BLs take longer in showing dementia. When examining bilingual countries in relation to dementia rates, another study by Klein et al. (2016[24]) which held life-expectancy as the independent variable, shows a lower incidence of dementia for more BL countries.

The last argument posited by Bialystok loops back to the mechanism in the unity and diversity model to examine transfer versus adaptation. She argued that a bilingual experience modifies selective attention. That, in turn, would impact language and non-verbal selection in different ways independently and in differing degrees, i.e., neuroplasticity in action. Considering that there is no transfer for BLs in infancy nor in old age, a single conclusion remains: research must look elsewhere. And this outlook should consider bilingualism both as reconfiguring attentional networks and as a source of neuroplasticity leading to adaptations both in brain and behavior.