Caryn Lerman, a University of Pennsylvania psychologist who studies risky health behavior, hoped that computerized brain training could strengthen activity in parts of the brain used in making good decisions or resisting bad ones. That could prove useful in fighting obesity or cigarette addiction.

Alas, Lumosity, the commercial brain-training games program, failed the test.  It was no better than playing some regular games developed in the Drexel University Replay Lab.  Lest you conclude that computer games are good for you, people who took just the cognitive-function test three times improved as much as those who played Lumosity and the Drexel games.

"We're all disappointed because everyone is looking for a fun and easy way to improve cognitive performance, and this does not seem to be it," Lerman said.

Although studies on the impact of brain training have had mixed results, this one adds to mounting evidence that many who use the programs get better at the tests themselves, but not much else.

The study, which was co-led by Joseph Kable, a Penn neuroscientist, was published Monday in the Journal of Neuroscience.

Lerman and her team will now study whether adding electrical or magnetic brain stimulation to the mix leads to better results. "That's where the attention is now going in the field," she said.  "We're not pursuing studies of cognitive training alone anymore." Our brains change throughout life as we learn new things, and Lerman thinks there is still potential to enhance performance.

The Penn team's study involved healthy young volunteers. Younger brains are likely to be more modifiable than those of older people. Still, the new results do not necessarily apply to older people who use online programs in an effort to stave off cognitive decline.

Lerman said she picked Lumosity because previous research, funded by the program's maker, found it outperformed crossword puzzles in improving several types of brain functioning.  (The makers of the program last year agreed to pay a $2 million settlement to the Federal Trade Commission in response to charges that the company had made unfounded claims about the games' ability to help users perform better at work or school or combat age-related cognitive changes.)

The Penn study, a randomized controlled trial, included brain imaging during decision-making. It tested how much people were willing to delay rewards and how they evaluated risks after exposure to exercises meant to improve executive function, used to make decisions. The 128 study participants, who were on average in their mid-20s, practiced either with Lumosity or video games for 10 weeks.

The decisions involved were meant to evaluate impulsiveness or a tendency toward risky behavior. Participants could choose between a smaller immediate reward and a larger one they could collect in the future. They also chose between smaller rewards they were likely to receive versus larger rewards that were less likely.  Being willing to delay gratification and being risk averse make people less likely to develop addiction, obesity, or financial or health problems, Lerman said.  Some studies have also found  those qualities also affect the likelihood of relapse after abstaining from an addiction.

Her previous work has found that engagement of parts of the brain involved in self-control predicts whether smokers can refrain from smoking.

The study found "no evidence" that brain training altered neural activity during decision-making or that it affected choices involving risk and delayed reward.  People using Lumosity did get better at its specific tasks, but they were no better on standardized cognitive tests than those who played computer games.

Dawn Mechanic-Hamilton, a neuropsychologist who directs the Penn Memory Center's cognitive fitness programs, said that in the past, she included the BrainHQ training program in classes for people with mild cognitive impairment, often a precursor to dementia. She thinks it has better scientific support than Lumosity.  She now devotes her class time to focusing on either emotional issues or strategies that help people compensate for memory problems.

The Lerman study, she said, leaves open the possibility that brain-training games could still be helpful for people with specific deficits. Much of the practice in Lerman's study, she said, was outside of the cognitive domains that have been found to be most responsive to training: processing speed and working memory.  There's little evidence, though, that those improvements in testing transfer to everyday life, "in part because we don't have very good measures of what people do every day."

She is both skeptical of brain training and intrigued that there are signs it could be helpful if researchers could figure out whom to target.

In the meantime, people ask her all the time whether they should try brain training.  Her response: "You should do something that is cognitively challenging to you.  Whatever that is, it should be engaging to you, and it should get harder."  That could be learning a new language or joining a book club, she said.  If you can pair learning with social activity, that's extra good.  If you're spending a lot of time sitting passively in front of a television, BrainHQ would probably be an improvement.

As for what is most likely to reduce the odds of dementia, she had a quick answer: physical exercise.