Cognitive Neuroscience in the Decade of the Brain
The Decade of the Brain has seen great progress in discovering what the mind is in terms of what the brain does. Advances in cognitive neuroscience have come through the convergence of three main sources of knowledge. First, animal experimentation has become increasingly sophisticated in its ability to illuminate the neural machinery of mental operations that have clear relevance to human behaviors. Second, experimental studies of patients with focal or degenerative lesions have provided a steady flow of new insights into the functional architecture of the human brain. These studies have become increasingly informative because magnetic resonance imaging (MRI) provides superior localization of brain lesions, and experiments have taken advantage of modern models and methods in cognitive science.
In regards to understanding normal human brain-behavior relations, these sources of knowledge have notable limitations. Animals offer only partial models of human behaviors (in fairness, the converse is also true). Lesion studies examine not what a region of the brain does, but rather what the rest of the brain can accomplish in the absence of functions mediated by the injured region. A third source of information suffers from neither of these limitations: Imaging human brain functions. The best known and most influential technique for imaging human brain functions is positron emission tomography (PET). PET studies have been prominent and influential in identifying large-scale distributed neural networks that mediate various aspects of perception, attention, language, memory, and thinking.
With very few exceptions, functional brain imaging has had little influence on our thinking about age-associated changes in human abilities and performance. This is not because such information is of little interest to those studying aging. Indeed, there are compelling reasons to understand age-associated changes in behavior in terms of age-associated changes in brain function. Many age-associated declines in performance are likely to reflect declines in brain functions (although other issues, such as social influences no doubt matter as well). Therefore, amelioration of age-associated declines in performance are likely to succeed through biological interventions, and those interventions ought to target the specific neural systems responsible for specific declines in abilities.
To the extent that age-associated changes in brain function have been considered, they have been considered by analogy to focal or degenerative lesions seen in disease. For example, age-associated declines in memory have been compared to mnemonic deficits shown by patients with frontal-lobe or medial temporal-lobe lesions (Moscovitch & Winocur, 1992), or patients with degenerative diseases such as Parkinson's or Alzheimer's disease (Gabrieli, 1991, 1994). These analyses infer that age-associated degeneration in brain areas clearly injured in focal or degenerative disease conditions accounts for milder, age-associated changes in memory performance. Such analogies are useful in conceptualizing the brain basis of age-associated changes in performance, but they have a major limitation: Healthy aging subjects do not have either focal lesions or degenerative diseases. Structural (MRI) brain imaging studies sometimes provide some degree of corroboration, but this is necessarily a weak source of evidence. It is weak because much of the brain shows age-associated changes, and linking ubiquitous structural changes to specific mental changes is difficult. Further, when seeking to understand the brain basis of age-associated changes in mental functions we do not want information about brain structure so much as we want information about brain function.
We are, however, largely ignorant about age-associated changes in human brain function. Due primarily to practical limitations in the use of PET scanning, there is a remarkable dearth of information about age-associated changes in human brain function. This is, however, about to change dramatically for aging, as it will for many areas of psychological research. The dramatic change is due to advances in MRI techniques that allow for functional magnetic resonance imaging or fMRI. In the remainder of this article, we briefly discuss three issues: (1) how PET works and why it has not been influential in aging research, with a few exceptions; (2) how fMRI works and why it will be influential in aging research; and (3) a few of the major challenges and opportunities already visible in applying fMRI to aging.
PET Imaging
In typical PET studies, a subject receives an injection of water labeled with a positron-emitting radioactive isotope into a vein in the arm. The radioactive water accumulates locally in the brain as a function of blood flow. Blood flow in the brain, in turn, is thought to reflect neural activity, i.e., the more active a brain region is, the greater the blood flow in that region. Thus, by localizing radioactive emissions in the brain, one can infer local brain activity. Most PET studies examining aging have involved resting studies where a relatively long-lasting isotope (e.g., 18F-fluorodeoxyglucose or FDG) is used while a subject rests in the scanner. Such studies have poor temporal resolution - a single image is created over a long period of about 40 minutes. Consequently, they have poor cognitive resolution - all brain activity over the 40 minutes is averaged into a single scan.
A radioisotope with a short half-life, however, overcomes these limitations and allows for activational studies. The isotope oxygen-15 has a half-life of about two minutes. Therefore, one can perform multiple injections in a session, with each injection providing a single picture of brain activity over about one minute (better temporal resolution). After a ten minute interval, another injection can occur because the radioactivity from the previous injection is virtually nonexistent. One can then compare brain activity across two conditions (scans), an activation and a control condition. Ideally, these two conditions are matched in all regards, except for one mental operation, so that subtraction of the images will visualize the brain region or regions that mediate that mental operation. Most oxygen-15 PET studies analyze brain activity by taking scans from a group of subjects (often about 10 subjects), translating the scans into a common, idealized brain space, and analyzing the group's data across the control and activation conditions. These studies have provided our best views of normal human brain function.
One would think that such a technique, therefore, would have provided major insights about age-related changes in human brain functions. With the notable exception of a few studies from Grady, Haxby, and their collaborators at NIH (e.g., Grady et al., 1994), however, there are virtually no activational studies of aging. Probably the most important reason for this gap in our knowledge is that production of short half-life radioisotopes requires an on-site cyclotron, a facility that is expensive to build and maintain. Thus, only a few places in the world produce the vast majority of PET activational studies, and aging has not been a priority for research at those centers. Although more medical centers are now capable of performing PET activational studies and will likely form collaborations with researchers interested in aging, there are likely to be no more than about a dozen such centers in the United States. Further, the advent of fMRI makes it unlikely that many more places will invest in creating new PET facilities given the economic climate of most medical centers.
fMRI
Advances in MRI now allow for imaging of brain function via fMRI.
fMRI, like PET, measures changes in microvasculature surrounding metabolically active brain tissue. fMRI detects changes in the magnetic state of blood, which is dependent upon the degree of oxygenation. Therefore, fMRI is noninvasive - subjects can participate without any sort of injection. There are a variety of fMRI techniques, and all have a good temporal resolution by virtue of generating a brain image in two seconds or less. fMRI also has good spatial resolution, revealing differences of less than 2 mm in the visual cortex (Engel et al., 1994). Further, one can readily take a structural MRI of an individual's brain just before or after the fMRI because the same machine produces both sorts of MRI images. This facilitates mapping the functional activity onto exact brain structures (one cannot take a structural image of a brain in a PET scanner, so localization must be inferred by an indirect registration process). The combination of good temporal resolution (many data points) and good spatial resolution allows for activational studies, and enhances the possibility of obtaining reliable activation in individual subjects. For example, in our early studies we have been able to show cognitive and memory effects in individual subjects, as well as accurately identify the language-dominant hemisphere in individuals in whom language dominance was known by Wada testing (Demb et al., in press; Desmond et al., in press). Other researchers have reported comparable fMRI findings in attention, perception, language, working memory, problem solving and motor control.
Most important, for practical purposes, is that fMRI can be performed on clinical MRI scanners in thousands of U.S. hospitals. Although most fMRI studies today come from a few centers, especially Massachusetts General Hospital, Pittsburgh, Wisconsin, and Minnesota, there is little doubt that fMRI will soon be performed in hundreds and then perhaps thousands of locations following relatively modest upgrades in hardware and software.
Implications for Research on Aging
Soon, fMRI will be widely available. It has no known risks and can yield reliable single-subject brain images. fMRI, therefore, ought to be useful for studies examining commonalities or individual differences in aging, from both cross-sectional and longitudinal perspectives. Consequently, many psychologists will have the opportunity to become involved, via collaborations with radiology or neurology departments, in the study of age-associated changes in brain function.
Sometimes it seems like there is a "revolutionary" technique every few years. The skeptical reader may think that fMRI will join other techniques such as PET, MEG, or SPECT, as being of more intellectual than practical interest to those who study aging. As fMRI researchers we may be biased, but we think that this is the technique for the "rest of us" because it finally has all the properties needed to become a widely available tool for imaging the functioning human brain.
In addition to opportunities, fMRI provides considerable challenges. First, it is a new technique that is still undergoing development. For example, there is still considerable debate about optimal fMRI image analysis, especially in regarding artifacts from large, draining veins that may be at a considerable distance from the location of neural activity. Second, there are a variety of age-associated brain changes that will need to be considered in imaging studies, including aging effects on microvasculature, synaptic density, oxygen uptake, and brain volume. Would reduced activation that appears proportional to reduced brain volume suggest preservation or reduction of the mental operation associated with activity in that brain region? Third, age-associated changes that influence performance, such as the rate of stimulus presentation, may have to be considered carefully. Fourth, there will be considerable challenges in making images so comparable across testing sessions that longitudinal analyses will be accurate. Fifth, cross-sectional analyses of fMRI signal has just begun to be performed, and will entail a variety of unknown problems.
As psychologists, we may worry whether brain imaging will become such a powerful and influential technique that purely behavioral studies will appear obsolete to the public or funding agencies. This concern may be correct in one sense (although as researchers engaged in brain imaging, we may be more than a little biased in this regard). Those studying aging, whether involved in imaging research or not, who ignore the approaching burst of brain imaging studies may find their theories to be less compelling than other theories that successfully integrate findings from both behavioral and brain-imaging studies.
Such a concern would be misplaced, however, in that psychological analysis of age-associated changes in abilities and performance will actually become more important due to the power of brain images. The meaning of an activational study rests entirely on the interpretation of the behavioral conditions that are being compared, i.e., on the psychological analysis of the mental operation whose brain location is imaged by fMRI. Moreover, psychological theory must guide interpretation of brain activity and guard against naive theorizing. For example, many researchers suspect that aging has a disproportionate impact on frontal-lobe function. By a naive view, one may expect less frontal activation in older than younger subjects. Grady et al., (1994), however, found that for a certain task older subjects showed greater frontal-lobe activation than younger subjects. Thus, age-related declines in performance may be associated with either increases or decreases in regional activity. Only a psychologically informed framework will help explain what reduced, increased, or equivalent activation in older versus younger brains actually means. Therefore, the anticipated widespread use of fMRI to examine age-associated changes in brain function ought actually to enhance the importance of psychological experimentation and interpretation.
We are half way through the Decade of the Brain, but we are just at the beginning of what ought to be a revolution in our understanding of how the aging brain makes the aging mind.
References
Demb, J. B., Desmond, J. E., Wagner, A. D., Vaidya, C. J., Glover, G. H., & Gabrieli, J. D. E. (1995). Semantic encoding and retrieval in the left inferior prefrontal cortex: A functional MRI study of task difficulty and process specificity. Journal of Neuroscience (in press).
Desmond, J. E., Sum, J. M., Wagner, A. D., Demb, J. B., Shear, P. K., Illes, J., Glover, G. H., Gabrieli, J. D. E., & Morrell, M. J. (1995). Language lateralization in Wada-tested patients using functional MRI. Brain (in press).
Engel, S. A., Rumelhart, D. E., Wandell, B. A., Lee, A. T., Shadlen, M. N., Chichilnisky, E. J., & Glover, G. H. (1994). Measuring human primary visual cortex topography using functional magnetic resonance imaging. Nature, 369, 525.
Gabrieli, J. D. E. (1991) Differential effects of aging and age-related neurological diseases on memory subsystems of the brain. In F. Boller & J. Grafman (Eds.), Handbook of Neuropsychology (pp. 149-166). Amsterdam: Els