It's no secret that cognitive functions tend to diminish in old age, but a new study from Washington University has identified at least one mental task that older adults seem to perform as well, if not better, than their younger counterparts.
Ironically, it is the older adults' diminished ability to hold important contextual clues in working memory that seems to explain their superior performance on a simple test requiring them to quickly identify a specific sequence of letters on a computer screen.
"Our new findings are startling because researchers had previously found it nearly impossible to identify a task involving attention or memory that older adults perform as well as young people," said Todd S. Braver, Ph.D., an assistant professor of psychology in Arts & Sciences. "In this experiment, older adults not only completed the task with fewer errors, but amazingly, their reaction times were as fast as the younger adults, and that's pretty much unheard of."
Perhaps even more important to science's broader effort to unravel the mechanisms of human thought: A sophisticated new "context-driven" computer model of the human brain predicted the quirk and led researchers to test for it.
The model was developed by researchers here in collaboration with Jonathan Cohen, M.D., Ph.D., professor of psychology at Princeton University.
Known as a "connectionist neural network model," the system relies on a complex, interconnected set of computer algorithms to mimic various human thought processes. The connectionist model is based on the theory that there is no single physical area of the brain that controls the thought process. Rather, cognition is carried out in an intricate web of interlocking and interacting brain regions, which together spur our thoughts and actions.
"Our model predicted specific effects of aging on brain function, and these effects were confirmed through empirical studies of actual human behavior," Braver said. Other members of the research team are Deanna M. Barch, Ph.D., an assistant professor of psychology, and psychology doctoral student Beth A. Keys.
The computer model's success, coupled with other recent breakthroughs in cognitive research, has led the team to propose a novel theory of brain function -- that our ability to retrieve, hold onto and evaluate important contextual clues is essential in a range of critical executive-level mental functions.
It's clear that a number of complex cognitive functions depend heavily on our ability to grasp context. In language, for instance, a single word, such as "pen," could have one of several distinct meanings, depending on its use in a sentence.
In this study, researchers suggested that context-processing also plays an important underlying role in cognitive processes, such as short-term memory and attention. Each process, they argue, can be understood as being supported by a chemical neurotransmitter system that continually rewards the brain for holding onto informational tidbits relevant to the task at hand. In healthy young adults, the reward system for contextually important information is so strong that sensory distractions and other nonessential information are blocked out of working memory, effectively maintaining attention and goal orientation.
"A wide range of age-related impairments in cognitive control may, in fact, be due to a single fundamental deficit in the mind's ability to properly represent and maintain task-relevant context," Barch said.
The researchers also believe, she added, that "these cognitive declines might be due to disturbances in the functional interactions between the prefrontal cortex and the dopamine neurotransmitter system, which serve as the neural mechanisms underlying context representation and maintenance."
Although science now understands a great deal about the specific functions of various brain regions and how these functions are affected by dopamine and other chemical neurotransmitters, very little is understood about how these functions interact and what central operational mechanics govern the intricate workings of human thought. In recent years, computer modeling has emerged as a viable option for exploring the underlying forces that shape cognition.
In this study, the computer model helped researchers recognize that the sheer power and agility of young people's contextual processing skills might leave them vulnerable to errors in certain context-related decision-making situations. More interesting, the model also produced the very counterintuitive suggestion that the diminished cognitive skills of older adults might leave them less susceptible to making some of those same mistakes.
To test the computer's hypothesis, the researchers analyzed data from an experiment in which two large groups of younger and older adults were instructed to observe a series of letters flashed across a computer screen and to push one button if the letter "X" appeared immediately following the letter "A" or another button for any other combination of letters.
As expected, older adults performed much worse than the younger ones in situations where the first letter presented was not an "A" and the second letter was an "X." Young people quickly realized that without an initial "A" in the sequence, the two letters presented would never meet the A-X target guidelines. Older adults made more errors in this situation, apparently taken in by the appearance of the "X" and losing track of the context -- the fact that it was not preceded by an "A".
Interestingly, as the computer model had predicted, researchers were able to lull young minds into a trap by front-end-loading the test with an abnormally high percentage of positive A-X combinations. Young minds, easily capable of keeping context information in working memory, apparently jumped to the conclusion that any "A" presentation was statistically likely to be followed by the target letter "X." Older minds, struggling to keep context in working memory, were less primed to expect an "X" as the next letter.
In situations where the next letter was, in fact, not an X, healthy older adults actually gave the youngsters a run for their money, performing with fewer errors and with reaction times that were as good, if not better, than those of their younger counterparts.
In studies such as this one, and in others using sophisticated advances in functional neuroimaging techniques, researchers have begun to identify important pieces of the puzzle of human cognition. Now, researchers such as Barch and Braver are hoping that computer brain modeling could help science to begin fitting these pieces together.
"Compared to the human brain, which has billions and billions of neurons, our models are quite simple, but they hold the potential to be very powerful in predicting and understanding the basic mechanisms of brain function," Braver said. "By systematically 'damaging' cognitive functions in computer models of the healthy brain, we hope to unlock secrets about what goes awry with normal aging and brain diseases."
