The idea that learning to speak two languages is good for your brain has come to be widely accept as fact, particularly in popular media. Studies have shown that bilingual speakers of all ages outperform monolinguals on certain cognitive performance measures. Other studies show delays in the onset of dementia and some even claim enhanced intelligence.

But a handful of attempts to replicate some of these seminal findings have failed to confirm this “bilingual advantage.” The number of studies that have not found a tie between bilingualism and better cognition has risen dramatically over the past few years.

A heated debate over this issue now rages in the research community and has gained prominent attention recently with a series of articles in the journal Cortex. A paper by Kenneth Paap of San Francisco State University and colleagues kicked off the fireworks in August, arguing that the evidence now suggests either no bilingual advantage exists, or it only occurs under certain as yet undetermined circumstances. Twenty-one commentaries and a summary by Paap and colleagues followed in October.

The authors were reacting to the intense optimism about the benefits of bilingualism generated by a large number of studies published over the last decade or so. Ellen Bialystok, a psychologist at York University, Toronto, and her colleagues, produced much of the important early work that helped to dismiss the outdated idea that bilingualism could be detrimental to children's intellectual development. Later studies went further, finding that bilingual children actually perform better than monolinguals in tests of “executive functions” –processes that control thought and behavior and enable complex cognitive tasks like problem solving .

A study published in 2004 made a particularly big splash. Bialystok and colleagues compared the cognitive control performance of elderly bilinguals and monolinguals. Not only was this the first study to find advantages in older bilingual adults, the results implied that being bilingual might mitigate age-related cognitive decline. Subsequent studies in patient populations have shown that being bilingual can delay the onset of dementia by about four to five years.

The measure Bialystok's group used, called the “Simon” task, involves asking participants to respond on their left or right, depending on the color of a box that appears on a screen. Researchers typically calculate how much longer participants take to react when the color they must respond to on their right appears on the left, or vice-versa, compared to when the location doesn't conflict. This test is designed to measure inhibitory control, since participants must suppress a reflex to respond with the hand on the same side as the box.

Executive function consists of three components: inhibition, switching between tasks, and updating working memory (the capacity to hold information in mind in the face of distraction). A common explanation of the bilingual advantage suggests that the constant practice of suppressing the language they are not using exercises inhibition. Another possibility is that constantly switching between languages exerts executive function when switching among different tasks. These hypotheses inspired a massive research effort investigating the extent, mechanisms, and causes of the bilingual advantage, involving over 200 studies and a huge amount of data.

Paap and colleagues identified several problems with this body of evidence. When researchers study groups in natural settings outside the laboratory, they can't control factors that may differ between groups, such as socioeconomics, immigrant status, and cultural differences. Attempts to match these factors among groups or account for them statistically are inevitably imperfect, leaving the possibility that differences in performance are due something other than language skills. An even thornier problem has to do with causality. Does being bilingual influence cognition, or does a person's cognitive ability affect the probability of acquiring multiple languages?

The researchers also collected the results of all tests comparing executive functions between bilinguals and monolinguals published since 2011, finding that 83 percent of them found no difference between the two groups. There was also a tendency for studies with positive results to have used smaller samples, whereas those using larger study populations were more likely to find no effect. Smaller sample sizes have a greater probability of producing a spurious result by chance.

Here is where the debate heats up. Some “skeptics” accuse “believers” of dismissing dissenting views and ignoring calls for higher standards. Believers accuse skeptics of selectively citing studies that support their argument, while not taking account the large body of research that supports the bilingual advantage. Bialystok claims the majority of negative findings come from studies of young adults, in which benefits of bilingualism are not often found because most young adults are at peak performance regardless of language skills.

The impact of speaking more than one language, Bialystok says, is more often found in children and in older adults. “There is a huge amount of research the skeptics never cite showing that if you give executive control tasks to bilingual children, they perform better.” She also points to brain imaging studies that find differences even in the absence of performance differences—a claim that the skeptics take issue with. Bilingualism, the naysayers contend, could conceivably reorganize the brain without leading to any benefits.

There is also disagreement about results in older adults. Some studies show a delay of four to five years in the onset of Alzheimer's disease, but others have not found this difference. Psychologist Angela de Bruin, of the University of Edinburgh, says it could depend how you measure disease onset. Retrospective studies ask people with dementia to report when their symptoms began, but there have also been some research now that tracks people over time. “Studies tracking people as they grow older have often not found a difference,” de Bruin says.

All this bickering takes place against the backdrop of a “reproducibility crisis” across all of psychological science (and life sciences in general). Commentators point to a system that rewards researchers for publishing positive and novel results, rather than maintaining scientific rigor and resisting pressures that lead to selective reporting of results that inflate the number of spurious findings.

A key problem relates to publication bias—the tendency for more positive results to be published than negative ones because of a combination of researchers' decisions about what to submit and editors' decisions about what to publish. De Bruin and colleagues looked for evidence of such bias in bilingualism studies by examining 104 conference abstracts, from 1999 to 2012. They classified the abstracts as supportive of the benefits of bilingualism, challenging of the idea, or mixed results. They then looked for which studies were ultimately published.

The researchers found that 68 percent of positive studies were published, whereas only 29 percent of the negative ones were. “There's a massive gap, and the only way for us to progress is to have access to all the results,” de Bruin says. They also combined the data from published studies (using a technique called meta-analysis) to assess what the sum of evidence says. This produced a significant but small positive result, but one likely to be an over-estimate. “Even though the meta-analysis showed a small effect, we don't know how reliable it is, because it's based on a selective sample,” de Bruin says. (Specifically, the data yielded an effect size of 0.3—effect sizes are standardized measures normalized according to the variability of the data, with 0.3 indicating a “small” effect.)

Paap and colleagues argue the true effect could be zero, but stress this does not imply bilingualism has any negative consequences. There is no evidence of any disadvantage in speaking additional languages for executive function or intelligence, whereas benefits unrelated to cognition are numerous and profound. “We don't question whether it's better to speak more than one language," Paap says. "It makes travel easier and more rewarding, can open career doors and other opportunities, and enables one to speak to more people, develop a deeper understanding of another culture, and understand the world from another vantage point.”

Paap also acknowledges the possibility that the effect of bilingualism on cognition is real, but only occurs in specific circumstances. If only certain types of language experience produces an advantage, divergent results could result from studying different groups that varied in the age they learned a second language, amount of use and relative proficiency. But psychologist Claudia von Bastian, at University of Colorado, Boulder, and colleagues at the University of Zurich, recently published a study quantifying bilingualism along all these dimensions, using nine executive function measures, and found no effect of bilingualism on any aspect of executive function.

Evy Woumans of Ghent University, Belgium, thinks the advantages of a second language may relate to how often a person switches between languages. She and her colleagues recently published a study comparing balanced bilinguals who switch many times a day, with interpreters who switch infrequently. They found that better language switching proficiency was associated with better executive functioning, but only for balanced bilinguals, not interpreters. “This suggests control of languages [an ability to readily switch] leads to better executive functioning, not the other way around,” Woumans says. She also advocates moving beyond a yes or no debate, to investigate what determines whether advantages occur. “It's important for both camps to agree it's not straightforward—when you have that [acknowledgement] you can collaborate to try to find out why some people find something and others don't,” Woumans says. “Then we'll be a lot closer to finding answers.”

Wouman’s comment echoes calls from many skeptics for collaborations between proponents and critics of the bilingual advantage. These efforts would ideally involve ‘pre-registration” of studies, specifying the study design, measures used to assess study outcomes, and other analysis methods. Defining these study characteristics in advance would prevent the possibility of changing methodology mid-course, a practice that can inflate spurious findings. Cortex, in fact, announced an option for researchers to pre-register studies in 2013. Unfortunately, the fever pitch of this debate suggests the two sides are not likely to take advantage of this offer any time soon.