G— Functional Magnetic Resonance Imaging of the Brain in Nonhuman Primates: A Prospectus for Research on Aging
Thomas D. Albright
WHAT IS FUNCTIONAL MAGNETIC RESONANCE IMAGING?
In general terms, the goal of functional brain imaging is to obtain a spatially and temporally localized measure of neuronal activity through noninvasive means. By relating such activity patterns to behavioral measures of sensory, perceptual, cognitive, or motor events, it becomes possible to identify the neuronal structures and events that underlie these processes. Notwithstanding the fact that the brain lies within a closed cavity, the last 50 years have seen the development of a variety of noninvasive techniques for assessing brain activity (e.g., electroencephalography, magnetoencephalography, positron emission tomography). All have severe limitations, owing to the indirectness of the measurements.
An ingenious new technique, known as functional magnetic resonance imaging (fMRI), has emerged as the best and most promising tool for this purpose. This technique exploits variations in magnetic susceptibility that arise from molecular binding of oxygen to hemoglobin, which can be used to detect blood flow changes associated with neuronal activity. These neuronal activity-elated signals can be isolated with a spatial resolution (1–2 mm) approaching columnar structure in the neocortex, and temporal resolution compatible with most perceptual and cognitive operations. In other words, fMRI is an extraordinary new window through which one can probe the neural machinery of cognition.
To date, the three principal research areas of fMRI application have been sensory (primarily visual) processing, memory, and language, all of which
depend on brain systems that are at risk for varying rates of change with aging. In the field of sensory processing, a major development has been the identification of a hierarchy of functionally specific visual processing areas in the human brain (Sereno et al., 1995), which may be homologues of areas that have been studied extensively at the cellular level in nonhuman primates. Not only does this knowledge provide a solid foundation for the interpretation of a large body of existing neuropsychological data on human brain function (with concomitant clinical relevance), but it sets the stage for the first direct and in-depth comparisons between human and nonhuman primate data. In the field of memory, several recent studies have identified brain regions that are active while volunteers hold specific types of information in short-term memory (e.g., Smith and Jonides, 1998) Other studies have demonstrated changes in the nature of the fMRI response—expansion of the zone of activation, for example—associated with the learning of new sensorimotor tasks (e.g., Karni et al., 1998). These observations provide direct evidence for selective plasticity of neuronal structures as a substrate for learning. Finally, language has become a special focus of fMRI investigation, as it addresses issues that are difficult if not impossible to approach using animal models. Several studies have explored the functional modularity of language processing areas by these means (e.g., Posner and Pavese, 1998).
WHAT IS THE RELEVANCE OF FMRI TO UNDERSTANDING THE AGING BRAIN?
Inasmuch as an understanding of the relationship between the aging process and the brain must be built on a general understanding of brain organization and function, all of the functional imaging applications cited above are of relevance to aging research. In addition, there are specific areas of research in which functional imaging can make a unique contribution. Perhaps the most obvious of these is the use of imaging technology to compare brain organization and function across different age groups, as a complement to psychological and neurological characterization of normal and pathological states. Indeed, this is an area of tremendous potential owing to the unprecedented ability offered by fMRI to identify specific brain abnormalities associated with age-related sensory, motor, memory, and language dysfunction.
LIMITATIONS OF FMRI FOR STUDIES OF THE HUMAN BRAIN
The reasons for the rapid development of fMRI techniques and their application in studies of human brain function are obvious. In spite of the many realized and potential gains, however, and the fact that fMRI stands to significantly advance research on aging of the brain, there are some notable limitations to this approach when it is applied strictly to humans. These
limitations exist because practical and ethical considerations preclude using fMRI in conjunction with other powerful techniques that have been developed for use in studies of nonhuman species. Two of the most salient examples illustrate the point.
First, manipulation of the genome (i.e., creation of transgenic animals) and subsequent controlled expression of specific genes are becoming extremely valuable tools for assessing the functional contributions of specific cells, subcellular components and signaling pathways (see Picciotto, 1999, for a review). Methods of this type are now applied routinely using germ-line transgenic manipulations in mice, and their enormous relevance to research on the aging brain is only hinted at by recent highly publicized demonstrations that gene-induced alterations of neurotransmitter receptor concentrations has a marked influence on learning and memory (e.g., Mayford et al., 1996; Tang et al., 1999). To completely understand the underlying mechanisms and implications of such findings, brains that have been functionally altered will be probed using many standard anatomical and physiological techniques. To that list of techniques we can now add fMRI, which can be applied to rodents with very high spatial resolution, and which offers a view of global changes in brain organization and function that result from genetic manipulations.
In the near future it will also be possible to extend this genetic approach to other species, including nonhuman primates, by exploiting the capacity of genetically engineered viruses to introduce novel genes into the brains of adult animals. Moreover, this pairing of techniques from molecular biology and functional imaging will provide a powerful means to investigate both the contributions of specific cells or cellular components to age-related changes in brain function and to evaluate genetic intervention as a means to influence the course of such changes. None of this research can be performed using humans as subjects.
A second realm in which great potential exists for using fMRI in conjunction with other methods is single-and multiunit electrophysiology in nonhuman primates. The technology for this type of physiology, which is often profitably coupled with behavioral analysis, has evolved over the past 40 years to become a preeminent experimental approach in the field of cognitive neuroscience (e.g., Hubel, 1988). Its applications in the field of aging research are few to date, but when combined with fMRI in nonhuman primates, this approach promises to tie functional imaging data on age-related changes in the brain to a firm foundation of cellular physiology. The next section focuses specifically on this potential, considering the likely gains from it and how it can be realized.
FMRI IN NONHUMAN PRIMATES: WHAT DO WE HAVE TO GAIN?
Ironically, it is the evident value of fMRI for human studies that occasionally prompts concerns about its utility in nonhuman primates. Monkeys, after all, have long been subject to fine-grained analyses of structure and function using other techniques that are unsuitable for work with humans. It would thus appear that monkey fMRI both fails to exploit the considerable advantages associated with the use of human subjects and lacks the power of existing techniques appropriate for monkeys. These concerns overlook, however, the enormous potential associated with combined application of fMRI and more traditional techniques in the same animal, which can be accomplished only with nonhuman primates. This next section begins by outlining the benefits to neuroscience generally of combining these approaches. It is followed by specific illustrations of what can be gained by studying the aging brain.
The focus of this discussion is on rhesus monkeys (Macaca mulatta) and other macaque species as animal models for aging studies, largely because (1) they have been the most well-studied nonhuman primates in the field of neuroscience, (2) there is extensively documented evidence of their similarity to humans with respect to a wide variety of sensory, perceptual, and cognitive functions, and (3) they are particularly amenable to the sorts of experimental manipulations commonly used for studies of the brain bases of cognitive decline with aging. Nonetheless, there may be specific advantages associated with the use of other nonhuman primates, including some great apes and prosimians. For example, the rhesus monkey life span (in captivity) is 25–30 years, which makes life-span studies prohibitive. The mouse lemur (Microcebus murinus), in contrast, is a potentially attractive subject for study because of its small size, rapid aging, and ease of breeding. Such benefits must, of course, be weighed against the fact that there are far fewer details known of brain anatomy and physiology in these species in comparison with macaques, but a case can sometimes be made for their use in studies of the aging brain and, hence, for fMRI studies of aging in nonhuman primates.
General Benefits for Neuroscience
As indicated above, fMRI has already begun to demonstrate its worth for understanding the neuronal substrates of sensory, motor, and cognitive function. Such an understanding is of obvious relevance to aging research, even if age-related changes are not the primary subject matter. Likewise, much of what we expect to learn from the merger of fMRI with other neurobiological techniques in nonhuman primates has general relevance to neuroscience,
which is addressed here. Specific application of this general knowledge to aging research, and a consideration of new paradigms for aging research afforded by these research tools, follows in a later section.
Evaluation of Cellular Events That Underlie the Vascular Signals of fMRI
The fMRI technique enables quantification of local changes in cerebral blood flow. The use of these measures as an index of neuronal function is predicated on the assumption that there is a specific relationship between neuronal activity and the magnitude of the hemodynamic response. While that assumption is a reasonable first-order approximation, the broad nature and potential import of claims resulting from human fMRI studies demand that the relationship between fMRI signal change and neuronal response be investigated thoroughly. Studies using fMRI in nonhuman primates will make this possible.
A primary goal of such studies will be to characterize directly the relationship between the magnitude and temporal dynamics of the magnetic resonance (MR) response (e.g., to a sensory, perceptual, cognitive, or motor event) and the corresponding properties of a neuronal response (i.e., a change in the frequency of action potentials). In practice, this can be accomplished by recording both MR and neuronal responses under conditions that are identical and computing the correlation between the observed responses. If the correlation is significant and largely invariant over a range of conditions in nonhuman primates, it should be possible to draw quantitative conclusions about the nature of the neuronal response elicited in humans when only the MR signal is available.
The possibility also exists that the relationship between the vascular MR signal and neuronal activity varies in a context-specific manner—over time (in a circadian and/or age-related manner), cognitive demands, and/or attentional, motivational, or motor state. To the extent that this is true, correlational measures of the MR/neuronal relationship are valid predictors only if they have been assessed in the same context in which a prediction is to be made. In principle, knowledge of context-specific correlations will enable one to compute a normalized index of neuronal function across different contexts.
Linking Human Neuropsychology to Fine-Grained Measures of Neuronal Function
Historically, studies of brain function in humans have focused on the behavioral or cognitive effects of focal brain lesions in clinical populations, while animal studies using electrophysiological techniques have enabled fine-
grained investigation of the cellular substrates of sensory, perceptual, cognitive, or motor events. fMRI will allow an unprecedented comparison of functional activity patterns in monkeys and humans engaged in identical tasks, which will, in turn, permit identification of functional homologies and the ability to directly relate the vast literature on cellular response properties in monkeys to the field of human neuropsychology.
Use of fMRI As an Adjunct to Traditional Methods Used in Nonhuman Primates
There remains much to be gained from the application in nonhuman primates of traditional experimental techniques, such as single-cell electro-physiology, anatomical tract tracing, and behavioral analyses of the effects of focal brain lesions. fMRI offers a valuable means to guide such experiments. For example, cortical regions that are active under specific perceptual conditions can be rapidly identified, and that information can then be used to guide the positioning of single-cell microelectrodes for more in-depth analyses. Similarly, fMRI can be used to identify functionally specific zones for injection of anatomical tracers or placement of focal lesions.
Specific Benefits for Research on the Aging Brain
General knowledge of brain organization and function is obviously needed to understand the aging brain, and fMRI is a valuable method to obtain that knowledge. Moreover, as we have seen, the joint application in nonhuman primates of fMRI and traditional research methods offers a means to accurately interpret human fMRI data on the aging brain, and to link a vast human neuropsychological literature on aging to cellular structures and events. In addition, there are questions specifically directed at aging phenomena that can be profitably addressed only by using fMRI in nonhuman primates. Generally speaking, the richness of this approach for aging research is found in the fact that it offers a means to interpret the cellular substrates of fMRI signals seen at different points in an animal's lifetime and, furthermore, to manipulate substrates identified by these means and evaluate the consequences of doing so for cognitive function.
Do fMRI Signals Change As a Function of Age?
fMRI signals reflect blood flow. Any attempt to assess neuronal function across different age groups using MR signals as a dependent measure is likely to be confounded by the fact that brain tissue undergoes significant age-related changes in vascularization (both vascular patterning and responsiveness), tissue volume, shifts in receptor subpopulations and signaling, among
many other phenomena of tissue remodeling. In other words, because of such changes, the correlation between MR signal and neuronal activity may change as a function of age. A potential consequence of this confound is the possibility that an MR signal seen in a 20-year-old human may have an altogether different neuronal origin than the same signal in a 60-year-old, which—if unrecognized—could have profound implications for research and clinical diagnosis. fMRI in nonhuman primates offers a means to eliminate this confound. Specifically, by evaluating the correlation between the MR signal and neuronal response (outlined above) as a function of age, it should be possible to establish ''normalization factors" that can be used to appropriately compare MR signals across different age groups.
In addition to their use for data normalization, any observed age-related changes in the MR-neuronal response correlation may be of interest in their own right. Such changes may be indicative of a decline in the metabolic requirements of specific cell groups and could perhaps be diagnostic of early age-related loss of function. Data informative of the correlation between MR signal and the activity of specific cell populations can be obtained only in nonhuman species.
Use of fMRI Signals to Guide and Evaluate Manipulation/Intervention Studies
Because fMRI reflects neuronal activity with good spatial and temporal precision, it promises to be a valuable tool for the identification and characterization of age-related loss of function and specific brain pathologies. Once age-related neuronal changes are detected, experiments can be designed to address hypotheses about their origins or to evaluate intervention intended to promote recovery—in each case using the characterized MR signal as a dependent measure. For example, one might use the MR signal as a means to evaluate the effects of estrogen administration on the activity of specific cell groups as a function of age. Perhaps the most exciting and promising prospect in this domain is the use of genetic manipulations (i.e., creation of transgenic animals and selective expression of novel genes in targeted cell populations) to alter cell signaling, metabolism, or firing rate in a precisely targeted manner, with the goal of promoting recovery from selective age-related loss of function. As indicated above, genetic manipulations of this type will soon be possible in nonhuman primates using genetically engineered viruses for introduction of novel genes. As a dependent measure, the MR signal can provide a repeated and highly localized assessment of the effects of such manipulations on brain function. Perhaps one day successful interventions can be employed in humans. At the present time, nonhuman primates provide an ideal model for studies of this type.
FMRI IN NONHUMAN PRIMATES: OVERCOMING METHODOLOGICAL CHALLENGES
What Are the Challenges and How Can They Be Approached?
There are a number of significant technical challenges involved in carrying out fMRI experiments in monkeys. While apparatus and procedures for behavioral control, and for immobilization of animals in a confined space, have been well developed for single-cell electrophysiological experiments and can be adapted for use in fMRI, there are some novel components that must be introduced to satisfy the constraints of the magnet environment. Solutions to many of these problems have been attained and are described thoroughly in published reports of initial studies (Stefanacci et al., 1998; Logothetis et al., 1999). These technical problems and solutions are summarized briefly here.
To begin with, most conventional fMRI facilities use a horizontal bore magnet. The methods that have been refined over the years for single-cell electrophysiology in alert monkeys involve restraining the animal in a natural "sitting" position, in which the vertebral column is vertically oriented. Monkeys are particularly receptive to behavioral conditioning under these conditions, and the vertical body orientation (compared to a horizontal, or "crouching" posture) frees the hands for use in making behavioral responses. A vertical bore magnet is thus preferred for use with nonhuman primates, particularly if the same animals are to be trained for use in single-cell electrophysiological experiments. Vertical bore magnets are now commercially available, but they are costly and highly specialized. A less satisfactory solution involves restraining (through a combination of behavioral conditioning and apparatus) the alert animal in a horizontal position to accommodate the bore of the magnet. Stefanacci et al. (1998) have demonstrated that this is a feasible, though suboptimal, alternative.
A second challenge arises from the fact that the magnet itself is a significant source of audible noise. Such noise poses two problems. First, to exploit the potential of fMRI in nonhuman primates, the experimenter must extract a measure of the subject's sensory, perceptual, or cognitive state, which can only be accomplished via some sort of behavioral response. Audible magnet noise is distracting and/or annoying to such a degree that it can interfere with expected cognitive events and the performance of animals on complex behavioral tasks. Second, unattenuated audible noise may preclude any clean investigation of auditory processing by the brain. These problems may be overcome using a combination of habituation to the magnet environment and noise attenuation devices. Recent studies have demonstrated the potential effectiveness of these solutions (Stefanacci et al., 1998; Logothetis et al., 1999).
Third, the magnet environment demands that the apparatus for restraining the subject must be made from nonferrous materials that do not interfere
with fMRI signal acquisition. To meet this demand, Stefanacci et al. (1998) designed and constructed a specialized restraint device constructed entirely of plastic parts (e.g., nylon screws), which successfully avoided distortion of the magnetic field.
Finally, owing to both space limitations and magnetic interference, specialized devices must be designed for sensory stimulation and measurement of behavioral responses. Problems of this variety are similar to problems faced in human fMRI, but there are a number of unique constraints associated with the size and behavioral repertoire of monkeys. Initial studies have begun to offer workable solutions to these problems (Stefanacci et al., 1998; Logothetis et al., 1999).
Initial Evidence of Feasibility and Validation
fMRI research with nonhuman primates (rhesus monkeys) has thus far focused solely on development and validation of the technology. The initial target of investigation has been the visual cortex. Research progress is described thoroughly in published reports of initial studies (Stefanacci et al., 1998; Logothetis et al., 1999). Highlights of that progress are briefly summarized here.
Signal Acquisition and Identification of a Specific Locus of fMRI Response
This first step in validating the fMRI technique with nonhuman primates was recently accomplished independently by two groups (Stefanacci et al., 1998; Dubowitz et al., 1998). The MR signal was measured in the visual cortex in response to a stimulus known to elicit robust responses from individual cortical neurons. The MR measurements were low resolution and intentionally nonspecific, but cortical MR signals were robust and confined to regions known to exhibit significant neuronal responses to the stimuli presented. The observed correlation, though not precisely quantified, appeared quite high.
Spatial Precision of the fMRI Response
The second step in the validation process, which was recently undertaken by two additional research groups (Vanduffel et al., 1998; Logothetis et al., 1999), involved establishing the correlation between specific spatial patterns of MR and neuronal signal activation in the visual cortex. This step exploited the known topographic mapping of the visual field onto the cerebral cortex:
the spatial topographic pattern of neuronal activity known to be elicited by a given visual stimulus was directly correlated with the topographic pattern of MR activation elicited by the same stimulus. Initial results suggest that the degree of correlation is quite high.
WHERE DO WE GO FROM HERE?
The future of this approach to understanding brain function rests on: (1) continued validation of the methodology, (2) identification of the limits of the approach, and (3) application of the methodology to important and feasible scientific goals, including understanding the aging process.
Continued Validation of the Methodology
Having established that the correlation between the fMRI signal and neuronal activity is both modality and spatially selective, the third step in the validation process will involve measurements of the magnitude and time course of both MR and neuronal signals as a function of some stimulus dimension, such as contrast. (A similar approach has been used successfully to identify the relationship between neuronal and intrinsic optical signals recorded from visual cortex—Everson et al., 1998.) The observed correlation in nonhuman primates will provide a means to predict the magnitude and time course of the neuronal response in humans, given only the MR signal.
What Are the Limitations of fMRI for Use in Nonhuman Primates?
The maximal spatial resolution of fMRI obtainable with magnetic field strengths suitable for use with humans and nonhuman primates has improved by a factor of ~5 over the last decade, from around 1 cm in 1990 to around 2 mm today. Foreseeable changes in technology will improve this to approximately 0.5 mm over the next decade (Chen and Ugurbil, 1999), which is on the scale of functional columns in the sensory cortex. Despite these welcome advances, fMRI will never replace cellular electrophysiology as a means to investigate the neural foundations of cognitive function—there are physical upper bounds on the spatial and temporal resolution that can be achieved. However, as summarized above, fMRI is an extremely valuable complement to more conventional cellular studies of neuronal activation and patterns of connectivity. Indeed, it is primarily because fMRI in nonhuman primates lends itself (unlike human fMRI) to such combined use of techniques that it is an approach with enormous potential.
The preceding sections have reviewed specific types of scientific applications. The next section speculates briefly about what these applications are likely to reveal.
General Benefits for Neuroscience
The promise of monkey fMRI is considerable, but research efforts to date have achieved nothing more than validation of specific aspects of the technology. The next few years should see use of this technology—in conjunction with traditional electrophysiological, anatomical, and behavioral analyses—to topics ranging from the nature of sensory representations in the cerebral cortex and the neuronal mechanisms for plasticity of those representations, to mechanisms of memory storage and retrieval, to the cellular substrates of decision making, motor control, and even consciousness. Because this methodology provides a "global" view of brain events associated with cognition, it is likely to lead to many new discoveries regarding the interactions between major brain systems, such as those responsible for vision and memory, and the manner in which different sources of information (sensory, mnemonic, etc.) are integrated to yield coherent perceptual experience and behavior. This knowledge is a foundation on which a better understanding of age-related cognitive changes can be built.
Specific Benefits for Research on the Aging Brain
As indicated above, one can identify two general types of scientific applications of fMRI in nonhuman primates that are of relevance for an understanding of the aging brain and cognition: (1) collection of data that will serve as a basis for normalization of the MR signal across different age groups and (2) use of fMRI as a dependent measure for evaluation of the effects of specific manipulations of brain neurochemistry, organization, or function. The likely outcome of the first of these applications is straightforward (although the data may be complex in detail), and it will be invaluable for MR signal interpretation as an adjunct to cognitive studies of aging.
The concrete potential of the second type of scientific application is obviously dependent on specific hypotheses about the origins of age-related cognitive decline and the prospects for intervention. Nonetheless, the point can be illustrated briefly by a fantasy experiment that is not too far removed from reality. Consider, for example, the hypothesis that one aspect of age-related cognitive decline is linked to down-regulation of a specific subtype of postsynaptic neurotransmitter receptor in a specific brain region. fMRI data obtained from aged nonhuman primates confirms that the activity level in that
brain region, in association with a specific cognitive task, is lower than what is observed in younger animals. An intervention study is planned, whereby a genetically engineered virus is packaged with recombinant DNA that codes for the expression of the down-regulated receptor, along with genes that will enable temporal control of expression and spatial restriction of expression to the specific cells in question. The virus is injected into the relevant brain region, whereupon it infects cells and results in the transgenes being inserted into the host DNA. Gene expression is subsequently activated to produce the receptor protein, which is thus up-regulated to levels present in younger animals. At this point, one needs an assay of neuronal activity to determine whether the manipulation has had the desired functional effect. fMRI provides such a dependent measure, which can be compared with the preintervention state. Moreover, when obtained in conjunction with a behavioral index of cognitive function, the MR signal can be used as a basis for identifying the neuronal substrates of any cognitive gain. An intervention approach of this sort will soon be possible using new molecular genetic tools, and will clearly have broad applicability in conjunction with fMRI.
In sum, the great richness of fMRI in nonhuman primates—for aging research and for neuroscience generally—lies in the fact that the method can be applied in conjunction with many other techniques that cannot be used for research on human subjects. The "whole" that we stand to gain from such conjunctions is surely far greater than what can be learned from each technique on its own.
Chen, W., and K. Ugurbil 1999 High spatial resolution functional magnetic resonance imaging at very-high magnetic field. Topics in Magnetic Resonance Imaging 10:63–78.
Dubowitz, D.J., D.Y. Chen, D.J. Atkinson, K.L. Grieve, B. Gillikin, W.G. Bradley, Jr., and R.A. Andersen 1998 Functional magnetic resonance imaging in macaque cortex. Neuroreport 9:2213–2218.
Everson, R.M., A.K. Prashanth, M. Gabbay, B.W. Knight, L. Sirovich, and E. Kaplan 1998 Representation of spatial frequency and orientation in the visual cortex. Proceedings of the National Academy of Sciences of the United States of America 95(14):8334–8338.
Hubel, D.H. 1988 Eye, Brain and Vision. New York: Scientific American Library, W.H. Freeman.
Karni, A., G. Meyer, C. Rey-Hipolito, P. Jezzard, P., M.M. Adams, R. Turner, and L.G. Ungerleider 1998 The acquisition of skilled motor performance: fast and slow experience-driven changes in primary motor cortex. Proceedings of the National Academy of Sciences of the United States of America 95:861–868.
Logothetis, N.K., H. Guggenberger, S. Peled, and J. Pauls 1999 Functional imaging of the monkey brain. Nature Neuroscience 2(6):555–562.
Mayford, M., M.E. Bach, Y.Y. Huang, L. Wang, R.D. Hawkins, and E.R. Kandel 1996 Control of memory formation through regulated expression of a CaMKII transgenes . Science 6:1678–1683.
Picciotto, M.R. 1999 Knock-out mouse models used to study neurobiological systems. Critical Reviews in Neurobiology 13(2):103–149.
Posner, M.I., and A. Pavese 1998 Anatomy of word and sentence meaning. Proceedings of the National Academy of Sciences 95:899–905.
Sereno, M.I., A.M. Dale, J.B. Reppas, K.K. Kwong, J.W. Belliveau, T.J. Brady, B.R. Rosen, and R.B. Tootell 1995 Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. Science 268:889–893.
Smith, E.E., and J. Jonides 1998 Neuroimaging analyses of human working memory. Proceedings of the National Academy of Sciences 95:12061–12068.
Stefanacci, L., P. Reber, J. Costanza, E. Wong, R. Buxton, S. Zola, L. Squire, and T. Albright 1998 fMRI of monkey visual cortex. Neuron 20:1051–1057.
Tang, Y.P., E. Shimizu, G.R. Dube, C. Rampon, G.A. Kerchner, M. Zhuo, G. Liu, and J.Z. Tsien 1999 Genetic enhancement of learning and memory in mice. Nature 401:63–69.
Vanduffel, W., E. Beatse, S. Sunaert, P. Van Hecke, R.B.H. Tootell, and G.A. Orban 1998 Functional magnetic resonance imaging in an awake rhesus monkey. Society for Neuroscience Abstracts 24:11.