Researchers say they have the first scientific evidence that brain training can help ward off dementia, after a 10-year study showed a particular type of cognitive exercise was associated with significantly reduced risk of developing the condition.

Scientists claim a short course of 'speed of processing' training – designed to boost how quickly participants recognise objects – could produce cognitive benefits in older people even 10 years later. If so, it's the first intervention identified to lower dementia risk - but the research has some limitations.

"Speed of processing training resulted in decreased risk of dementia across the 10-year period of, on average, 29 percent as compared to the control," says psychiatrist Jerri Edwards from the University of South Florida.

"When we examined the dose-response, we found that those who trained more received more protective benefit."

Edwards' team analysed data from the ACTIVE (Advanced Cognitive Training in Vital Elderly) study, which monitored 2,802 healthy older adults for a period of 10 years, as they aged from 74 to 84 on average.

As part of the study, participants were randomly assigned to groups performing one of three different kinds of cognitive training, focussing on either memory training, reasoning training, or speed of processing training.

A fourth group acted as controls and didn't engage in any brain training exercises.

Participants received 10 1-hour sessions of training over a period of weeks, with a smaller group receiving a limited amount of follow-up sessions about a year after (and three years after) the initial training.

Due to death and other factors, only 1,220 of the original 2,802 were able to complete the whole 10-year study, which assessed the participants' cognitive and functional ability after the first six weeks, and at 1, 2, 3, 5 and 10 years.

Of these 1,220 participants, 260 had developed dementia by the conclusion of the study – but the researchers say the risk of developing the condition was 29 percent lower for those who had done speed of processing training, when compared against the control group.

It's the first time a cognitive training exercise has been demonstrated to have this kind of association, the researchers say, noting that the memory and reasoning exercises failed to produce any comparable significant lowering of risk.

"We need to further delineate what makes some computerised cognitive training effective, while other types are not," says Edwards.

"We also need to investigate what is the appropriate amount of training to get the best results. The timing of intervention is also important."

But as promising as the results seem, other scientists are urging considerable caution in how we interpret the team's findings.

First off, the finding that speed of processing training reduced dementia risk only just scrapes by in terms of statistical standards. Scientific convention holds that a p-value of 0.05 is the threshold for statistical relevance – any higher and it's possible the same result could occur by chance.

Here, the reduced risk p-value was 0.049, meaning the result would almost be considered statistically irrelevant – something that weakens the strength of the findings, some say.

Second, participants in the study self-reported their dementia, meaning they weren't clinically diagnosed as having the condition – a significant limitation in a study that's claiming to lower risk of developing the disease.

"It's positive that this study compared several types of brain training and was both long term and large scale in nature," says the director of research at the Alzheimer's Society in the UK, Doug Brown, who wasn't involved in the research.

"However, as it relied on self-reporting of dementia in many cases rather than a robust clinical diagnosis, the results should be interpreted with caution."

Concerns are also being raised about how such a small amount of cognitive training could produce lasting effects even a decade later.

"The results reported here, of apparent reduction in risk of dementia after 10 years following only a few hours of cognitive training, are therefore rather surprising and should be treated with caution," says old age psychiatrist Rob Howard from University College London.

"I find it implausible that such a brief intervention could have this effect and it is worth bearing in mind that the results could have occurred by chance or as a consequence of uncontrolled confounding factors.

It's not the first time we've seen the promises of brain training apps criticised, but while there are definitely limitations with the study we need to be aware of, it's also important that other researchers continue to examine this area.

Because if these results can be replicated in a separate study without the same caveats described here – that is, providing stronger statistical results in research grounded in clinical diagnoses of dementia – we could really be onto something amazing here.

For their part, the researchers are convinced their speed training hypothesis correlates with the broader conversation about how mental activity is good for your brain as you get older.

"It's completely consistent with a large literature that talks about the beneficial effects of [mental] engagement," says one of the team, psychiatrist Frederick W. Unverzagt from Indiana University.

"All of that epidemiological research has found support for the idea that those things are helpful to brain health, and in terms of risk for later development of dementia, Alzheimer's disease, engagement with those things is associated with a lower risk. It's quite consistent with that literature."

The findings are reported in Alzheimer's & Dementia: Translational Research & Clinical Interventions.