Rosalyn Moran images and models the human brain to understand how it changes with age.

Rosalyn Moran is studying how our brains react to their environment by modeling the brain's 'software' using machine learning techniques.

"We use computational theories to expand our ideas of what the brain is by looking at the algorithms and software running in our brains. We're trying to look in the black box that is our brain and see what's really going on," says Moran, who has a joint appointment as an assistant professor in electrical and computer engineering and the Virginia Tech Carillon Research Institute.

Images of the brain's different responses for younger and older subjects.

Moran's main focus is on understanding how "the aging brain is a function of its environment and the length of time it has been in its environment." She is modeling this via computer algorithms, using deep learning techniques. Deep learning uses hierarchies in artificial systems that can learn deeper and more abstract concepts. Moran is looking into whether the aging brain, like these models, has a better grasp of high-level concepts. "As you age, you get deeper and deeper into the hierarchy, and you can grasp more and more abstractions," according to Moran. "This can have good and bad effects."

To study deeper abstractions, Moran and her collaborators reviewed a neuroadjusted automaton game that is one of the tasks set up for the Roanoke Region Brain Study, which is scanning hundreds of brains of all ages. In this game, proposers must split a pool of money between himself or herself and a receiver. If the receiver accepts, they both get the money. If the receiver rejects the offer, neither gets the money. According to Moran, this is a basic study of game theory and concept of fairness.

In the study, the results of this game differed by age. As the proposers became more and more greedy (offering smaller and smaller sums to the receivers), younger players started to refuse the offers more often. Older players, however, were more likely to accept even low offers. The question that arose, says Moran, is whether this proves more abstract thinking or whether people simply become more forgiving as they age.

The graph shows how the brain reacts differently to stimulation as it ages. Moran has discovered that learning signals go deeper in older brains.

"What we found is that you get learning signals that go deeper and deeper in the older participants," says Moran. "They can see different groups of people, and notice the difference between a more or less generous group." Younger participants, however, "can't distinguish these different environments and do not adapt their behavior."

"The interesting effect that you see is that you're more economically rational, and it seems to be an effect of being able to learn deeper and deeper interactions." Presumably, she continues, this is because older people have simply had more social interactions.

"Now we're looking at the brain regions responsible for this behavior." Moran is building a model of the brain and will compare it to an MRI model. "We might expect activity in regions of interest such as the prefrontal cortex in the older individuals, and we also expect to see interactions with the mid brain regions," like those that control dopamine and serotonin.

The prefrontal cortex is implicated in decision making, planning, and moderating social behavior. Dopamine and serotonin are implicated in risk/reward and happiness.

Using expectation maximization, machine vision, and pattern recognition, her team hopes to discover how the brain learns to adjust to these situations. This might, Moran notes, be similar to how an artificial brain learns.