New Approaches to Functional Neuroenergetics

Meeting Report

Susan M. Fitzpatrick Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/11/4/467/1758540/089892999563454.pdf by guest on 18 May 2021 James S. McDonnell Foundation

Douglas Rothman Yale University

glucose metabolism. Depending on the form of the la- INTRODUCTION beled tracer used, PET either directly measures the regional rate of glucose metabolism (CMRglc) or the The coupling between brain energy metabolism and coupled parameters of the cerebral metabolic rate for neuronal activity has, for more than a century, allowed oxygen (CMRO2), and the cerebral blood ºow (CBF) rate. researchers to monitor brain function (Roy & Sher- The fMRI blood oxygenation level dependent (BOLD) rington, 1890; Siesjo, 1973; Sokoloff, 1981). A break- signal is sensitive to both CMRO2 and CBF (Ogawa et al., through in this effort, and one that made human studies 1998). routine, was the development 20 years ago of positron In a functional imaging study, a subject performs ex- emission tomography, or PET (see Raichle, 1998). In perimental tasks while the signal is acquired. Most cog- combination with experimental paradigms and models nitive neuroscience imaging studies rely on the PET CBF developed in cognitive psychology, PET allowed the ªrst or the BOLD fMRI measurement. The acquired signal is high-resolution metabolic maps of functionally special- analyzed to provide images of the spatial distribution and ized regions of the human brain. A drawback of the PET temporal dynamics of CBF or BOLD contrast. The func- technology was its reliance on cyclotron-generated tional image is an image of the increment in signal short-lived radioisotopes. The subsequent development intensity during a task relative to a baseline state in of functional magnetic resonance imaging (fMRI) made which the subject rests in the scanner. Cognitive proc- functional brain mapping widely available to scientists esses are localized by functional imaging using experi- (Kwong et al., 1992; Ogawa, Menon, Kim, & Ugurbil, mental paradigms and analyses based upon theories of 1992). Almost weekly, new brain imaging results are cognitive neuroscience. highlighted in scientiªc journals and the popular media As an illustrative example, consider a study designed as providing new insights into the biological basis of to assess whether the frontal lobe is involved in the human brain function and neurological and psychiatric general cognitive skill of verbal working memory. The disorders. subject would perform tasks requiring this cognitive The application of PET and fMRI to localize cognitive skill, such as remembering lists of words, while being processes is based on the assumption that functional scanned. In one strategy the degree of involvement of neuronal activity increases when a region is involved in verbal working memory in each task would be varied, performing a cognitive task (Posner & Raichle, 1994). but the requirements for other cognitive skills would be These functional neuronal activities are involved in the held constant. The relative intensity of the functional communication of information between neurons and imaging signal in the frontal lobe during each task would include neurotransmitter release and action potentials. be statistically correlated with the verbal working mem- The energy required for these and other brain processes ory component. A positive correlation would support is provided almost exclusively by oxidative glucose me- the involvement of frontal lobe neuronal activity in this tabolism (Siesjo, 1978). Functional imaging measures skill (Posner & Raichle, 1994, provides an excellent sum- either glucose metabolism or neurophysiological pa- mary of how imaging studies are designed). rameters coupled to glucose metabolism (Sokoloff, Given the enthusiasm with which functional imaging 1981). Regions of increased functional neuroenergetic is used to answer sophisticated questions about the demand are identiªed by the corresponding increase in functional architecture of the brain, it is easy to forget

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892999563454 by guest on 29 September 2021 that there are long-standing controversies concerning of metabolism with development) deªnitive answers the neuroenergetic basis of the signal. To deªne some of would be obtained for these fundamental, long-standing the outstanding questions that remain in the ªeld, and questions. to encourage research to better understand the meaning of these measurements, the James S. McDonnell Founda- What are the molecular mechanisms that couple tion organized a workshop, Cerebral Metabolism and neuroenergetic requirements to CMRglc and CMRO2? Human Cognition: A Workshop Discussing New Ap- Although most functional neuroimaging studies assume proaches to Functional Neuroenergetics.1 The workshop that signal changes reºect neuronal activity, Pierre Magi- took place over two days with focused talks and ani- stretti (University of Lausanne) presented results identi-

mated discussions on a variety of issues related to fying the astrocyte as the primary site of cerebral Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/11/4/467/1758540/089892999563454.pdf by guest on 18 May 2021 neuroenergetics and functional imaging. Some of the glucose consumption. Magistretti proposed that astro- major questions identiªed and the results of these dis- cytic glucose uptake is mechanistically coupled to cussions are described in this report. Selected references neuronal energy requirements by the use of nonoxida- are provided for key points raised by the participants. tive glycolysis to provide energy for removing glutamate, released by the nerve terminals, from the synaptic cleft. What is the relationship between the imaging signal The lactate produced as a result of the incomplete oxi- being acquired and the cellular localization of neuro- dation of glucose is then transported from the astrocyte nal activity? to the neuron as a substrate for oxidative energy meta- Louis Sokoloff (NIMH) reviewed studies on the cellular bolism (Pellerin et al., 1996). Robert Shulman’s in vivo location of glucose metabolism in peripheral neurons. results from rat and human brain supported and ex- These studies showed that the uptake of radio-labeled tended the proposal by Magistretti. Shulman described deoxyglucose during electrical stimulation of peripheral 13C NMR spectroscopic studies showing that the rate of neurons occurs mainly in the synaptic regions. The rate astrocytic glutamate uptake increased with glucose me- of glucose metabolism was close to linear with the tabolism with close to a 1:1 stoichiometry (Sibson et al., stimulation frequency, which he interpreted as support- 1998). He proposed that 80% of cortical glucose oxida- ing a direct relationship between increases in CMRglc tion is coupled to neuronal glutamate release and uptake measured by PET and the total electrical activity of a by the astrocyte. Rolf Gruetter (University of Minnesota) region (Yarowsky, Kadekaro, & Sooloff, 1983). Gordon discussed his own 13C NMR spectroscopy data of human Shephard (Yale University) addressed the issue of the brain with enhanced sensitivity that is consistent with variations in neuroenergetic requirements as a function Shulman’s ªndings (Gruetter, Seaquist, Kim, & Ugurbil, of cell type and brain region. He pointed out that the 1998). A lively discussion initiated by Albert Gjedde (Uni- relationships measured in peripheral neurons do not versity of Denmark) questioned the localization of glu- hold for all classes of neurons. He presented high-reso- cose metabolism to the astrocyte. Among the objections lution deoxyglucose autoradiography studies of the ol- were the presence of a high glucose transporter and factory bulb that showed that unmyelinated axons have glycolytic enzyme capacity in neurons as well as the lack a high glucose metabolic rate (Greer, Stewart, Kauer, & of direct evidence in vivo. Although no consensus was Shepherd, 1981). Because unmyelinated axons are com- reached, it was agreed that the results presented pro- mon in the cerebral cortex, they may contribute sig- vided important new insights into the potential coupling niªcantly to the neuroenergetic requirements measured of the functional imaging signal to neurotransmission. by functional imaging. Shepherd, collaborating with Yale colleague Robert Shulman, is correlating metabolic and Are functional neuroenergetic requirements tempo- electrical maps of neuronal activity in the olfactory bulb rally dependent? with ultra-high-resolution fMRI (Yang et al., 1998). Wil- Marcus Raichle (Washington University) and Louis Sok- liam Greenough (University of Illinois) observed that oloff each presented data suggesting, in the human and functional imaging of the developing or learning brain, rat brain, respectively, that the fractional increase in in which the density of synapses and other neuronal CMRglc is greater than CMRO2 during sensory stimula- structures changes with time, may help in the assign- tion. Mark Mintun (Washington University) discussed his ment of the neuroenergetics measured by functional recent PET results, showing that at longer stimulation imaging to speciªc cellular structures. Along these lines times CMRO2 increases, consistent with NMR and mi- Kevin Behar (Yale University) discussed NMR measure- crodialysis data showing a transient increase in brain ments in the postnatal rat cortex that showed that lactate during sensory stimulation (Prichard et al., 1991). CMRO2 increased with age and correlated with synaptic Several discussants pointed out that even during the development (Behar, Ariyan, Mason, Haddad, & Novotny, transient period the majority of the increase in cerebral 1997). It was concluded that by further development of energy requirements is supplied by glucose oxidation. these experimental strategies and methods (e.g., higher- Albert Gjedde described studies suggesting that the resolution autoradiography; more detailed correlation mismatch is highly stimulation and region dependent

468 Journal of Cognitive Neuroscience Volume 11, Number 4

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892999563454 by guest on 29 September 2021 (Marret et al., 1993). Robert Shulman proposed an expla- measurement for not appropriately taking into account nation for the transient mismatch based upon the in- the physical constraints on blood volume. Ugurbil and creased use of glycogen metabolism during the others vigorously defended the BOLD measurement. De- transition period between the resting and stimulated spite the controversies the discussants agreed that many state. Other discussants noted that there are large vari- of these issues could be resolved through appropriate ations in the reported degree of mismatch between functional imaging studies in animal models to allow CMRglc and CMRO2. (Marret et al., 1993; Seitz & Roland, correlation with invasive methods that allow measure- 1992). Results from BOLD fMRI studies ªnd temporal ments of blood ºow and neuroanatomy at extremely variations between CBF and CMRO2 only for the ªrst few high spatial resolution. Because BOLD fMRI provides the

seconds of stimulation (Ogawa et al., 1998). It was con- highest spatial and temporal resolution of the metabolic Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/11/4/467/1758540/089892999563454.pdf by guest on 18 May 2021 cluded that with improvements in both fMRI and PET, as functional imaging methods, the resolution of these is- well as the more detailed chemical picture provided by sues is critical for continuing progress in functional im- MRS, a better understanding will emerge of the neuroen- aging. ergetic basis and functional importance of this phe- nomenon. What are the molecular mechanisms that regulate CBF? Does blood ºow spatially and temporally correlate Data supporting several candidates for the molecular with functional neuroenergetic requirements? regulation of CBF during functional activation were pre- Workshop participants are using several different tech- sented, including adenosine, nitric oxide, and adrenaline. niques and experimental models in an effort to address A novel proposal for CBF regulation was put forth by this question. Marcus Raichle’s PET studies suggest that Joseph Williams (Washington University). Working with during sensory stimulation, the observed fractional in- Thomas Woolsey on the rat whisker barrel, he showed crease in CBF is similar to that seen for CMRglc, with that the plasma lactate/pyruvate ratio, as well as other both exceeding the increase in CMRO2. Kamil Ugurbil factors that alter the brain redox potential, effect CBF. (University of Minnesota), acquiring BOLD functional Although the results from several speakers were prom- images with a 4 Tesla magnet system from the activated ising, no single factor explained the full range of CBF human visual cortex, reported that a rise in CBF is not autoregulation. The consensus among discussants was localized to the active optical columns (Hu, Le, & Ugur- that these mechanisms warrant further investigation. bil, 1997). The adjacent nonactive columns also experi- ence an increase in blood ºow. Superior localization of How will understanding of functional neuroenergetics the active columns was achieved using the BOLD signal aid in localizing cognitive processes? immediately after the start of the stimulation. This early If one goal of brain imaging studies is to map cognitive BOLD signal is believed, based on optical data (Malonek operations onto their neurobiological correlates, it is et al., 1996), to reºect primarily oxygen consumption. important to have a sense of how an understanding of Robert Turner (Wellcome Institute of Neurology) and neuroenergetics contributes to this goal. Randy McIntyre others supported Ugurbil’s proposal of a spatial uncou- (University of Toronto) is tackling this issue by studying pling of CBF and local neuroenergetic requirements the nature of the electrophysiological activity underlying based on optical data (Malonek et al., 1996). This conclu- cognitive function in humans and primates. McIntyre’s sion was debated as participants discussed the well- data suggests that both an increase in the total number documented autoradiographic data from rat studies and frequency of neurons ªring and alterations in pat- showing a tight spatial coupling of the changes in blood terns of ªring in activated tissue accompany increases in ºow and oxidative metabolism, electrical measurements, functional activity. Several discussants pointed out that it and fMRI (Greenberg, Hand, Sylvestro, & Reivich, 1979; is not known whether functional imaging is more sensi- Ueki, Linn, & Hossmann, 1988). Jeff Dunn (Dartmouth tive to changes in neuronal ªring patterns or total elec- College) supported a tight spatial coupling of CMRO 2 trical activity. The results presented by Shulman and and CBF with his own data, showing that the oxygen Magistretti supporting a relationship of the functional tension in the brain remained low during activation. imaging signal with neurotransmitter release (Pellerin et Thomas Woolsey (Washington University) suggested that al., 1996; Sibson et al., 1998) were interpreted by some CBF responds to local neuronal activity prior to CMRglc to support a tighter relationship with total electrical and CMRO2, a view supported by his elegant studies of activity. Seiji Ogawa (Lucent Technologies) described rat whisker barrels (Moskalenko et al., 1996). He also studies in which BOLD fMRI correlated well with suggested an alternate explanation for the primate opti- evoked potential recordings in sensory stimulation para- cal measurements based upon imprecision in the spatial digms in animals (Ogawa et al., 1998). These studies were correspondence of the vessels on the cortical surface also interpreted as supporting a correlation with total with the columns that they serve. Louis Sokoloff criti- electrical activity. It was generally agreed that more stud- cized the standard interpretation of the BOLD fMRI ies deªning the electrophysiological basis of the electro-

Fitzpatrick and Rothman 469

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892999563454 by guest on 29 September 2021 encephalogram, EEG (and MEG) signal and the neuroen- how to extend present methodologies to achieve higher ergetic basis of the functional imaging signal are needed speciªcity. to better understand the relationships between these mapping methodologies. How will neuroenergetically “correct” functional imag- Several discussants gave examples of how a better ing help us understand how the brain works? understanding of the functional imaging signal would aid Although most of the conference focused on issues of in their research. Kia Nobre (Oxford University) dis- understanding the neuroenergetic basis of the functional cussed how knowledge of the electrophysiological cor- imaging signal, there remains the question of what are relates of the functional imaging signal would assist in the possibilities and limits of functional imaging in study-

her studies using functional imaging to localize signal ing brain function. Several discussants offered sugges- Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/11/4/467/1758540/089892999563454.pdf by guest on 18 May 2021 sources and EEG mapping to provide high temporal tions of how understanding the neuroenergetic basis of resolution information. Several possibilities for combin- functional imaging may lead to novel understanding of ing electrophysiological and metabolic imaging studies cognition. Stephen Hanson (Rutgers University) pro- were discussed. posed that there has been a lack of new insight into An animated discussion arose over new PET ªndings cognition provided by functional imaging due to unnec- presented by Marcus Raichle that demonstrated in some essary assumptions of the present statistical paradigms. regions decreases in CBF, compared with the “at rest” or These assumptions result in regions being identiªed as control condition, during cognitive tasks (Raichle, 1998). active only if the time course of the imaging signal These ªndings were considered paradoxical because it agrees with the expectations of the experimenter. He is generally assumed in functional imaging studies that suggested that less restrictive assumptions, made possi- functional energy demand is increased during tasks. Vari- ble by improvements in functional imaging methods, will ous explanations were offered by workshop participants reveal novel insights into cognitive function. Several par- for this ªnding, including increased inhibition of total ticipants commented that as our understanding of the neuronal activity of the region by projecting dopaminer- neuroenergetic meaning of the imaging signal is im- gic and serotinergic neurons, an increase in inhibitory proved, it is essential to develop a parallel understanding GABAergic neuronal activity, and physiological con- of the “neuronal code,” in which electrical and synaptic straints on CBF that lead to an uncoupling between CBF activity leads to information processing. Once this under- and glucose metabolism. Robert Shulman proposed, standing is developed, neuroenergetically speciªc func- based on his ªndings of high neuroenergetic activity in tional imaging, particularly ultra-high-resolution fMRI the resting cortex, that the negative CBF signal did not obtained with the new high magnetic ªeld systems, may require a novel neurobiological explanation. He pro- provide a window on information processing in humans posed that in these regions internal mental processes at a neuronal network level. during the resting baseline condition simply have a greater neuroenergetic demand than the processes in- CONCLUSIONS duced by the externally directed task (Shulman & Roth- man, 1998). Although consensus was not achieved, the Although there were many vigorous scientiªc disagree- participants felt that the discussion was valuable in pro- ments, the participants could agree that the conference viding a speciªc example of how understanding the played an important role in identifying key areas where underlying neuroenergetics would distinguish interpre- our understanding of neuroenergetics and functional tations of the role of a region in supporting cognitive imaging needs to be strengthened. The ªnal session of processes. the workshop concentrated on the future studies Several discussants supported the importance of de- needed to address the questions raised during the meet- veloping functional imaging methods of speciªc ing. Of particular concern is the critical need for re- neuroenergetic processes. Douglas Rothman (Yale Uni- search spanning the many levels of complexity versity) suggested that a lack of proportionality between separating neuroenergetic processes at a synaptic level the functional imaging signal and speciªc neuroener- from complex human behavior deriving from the coor- getic processes could lead to false negatives if propor- dinated activity of vast networks of neurons. Given the tionality is assumed in the statistical analysis. These experimental uncertainties at all levels of study, the abil- pitfalls may be particularly severe for studying patient ity to test for consistency across levels will greatly groups with altered coupling between cerebral blood strengthen our understanding of cognitive processes. ºow and glucose metabolism. Steven Foote (NIMH) It was generally agreed that the gaps in our knowledge stated that improved neuroenergetic speciªcity in func- remaining after decades of research on cerebral glucose tional imaging would be valuable for the studies his metabolism should send a cautionary note to those institute is funding to apply functional imaging to study newly interested in brain imaging. The common usage of psychiatric disease. Along these lines there were several the functional image as synonymous with a neuronal stimulating discussions among the participants about activity map should be avoided or, at minimum, be care-

470 Journal of Cognitive Neuroscience Volume 11, Number 4

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892999563454 by guest on 29 September 2021 fully qualiªed. However, there was optimism generated ing of the brain. Annual Review of Biophysics and Bio- by the progress being made in closing these gaps. The molecular Structure, 27, 447–474. Ogawa, S., Tank, D. W., Menon, R., Ellermann, J. M., Kim, S. G., achievement of a detailed understanding of the neuro- Merkle, H., & Ugurbil, K. (1992). Intrinsic signal changes ac- energetic basis of the functional imaging signal was companying sensory stimulation: Functional brain map- agreed to be critical for understanding the neurobiologi- ping with magnetic resonance imaging. Proceedings of cal basis of human cognitive behavior. the National Academy of Sciences, USA, 89, 5951–5955. Pellerin, L., & Magistretti, P. J. (1994). Glutamate uptake into astrocytes stimulates aerobic glycolysis: Mechanism cou- Note pling neuronal activity to glucose utilization. Proceedings of the National Academy of Sciences, USA, 91, 10625– 1. A listing of the participants can be found on the Meetings 10629. Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/11/4/467/1758540/089892999563454.pdf by guest on 18 May 2021 Section of the JSMF website www.jsmf.org. Posner, M. I., & Raichle, M. E. (1994). Images of mind. New York: Freeman. Prichard, J. W., Rothman, D. L., Novotny, E. J., Petroff, REFERENCES O. A. C., Kuwabara, T., Avison, M., Howseman, A., Hanstock, C., & Shulman, R. G. (1991). Lactate rise detected by 1H Behar, K. L., Ariyan, C., Mason, G. F., Haddad, G., & Novotny, NMR in human visual cortex during physiological stimula- E. J. (1997). Determination of the rate of glucose utilization. Proceedings of the National Academy of Sciences, tion and the tricarboxylic acid cycle from (1–13C)glucose USA, 88, 5829–5831. in the developing rat brain. Symposium on Perinatal Cere- Raichle, M. E. (1998). Behind the scenes of functional brain bral Blood Flow Metabolism, Satellite Meeting of XVII In- imaging: A historical and physiological perspective. Pro- ternational CBFM, Hershey, PA, June 12–15. ceedings of the National Academy of Sciences, USA, 95, Fox, P. T., Raichle, M. E., Mintun, M. A., & Dence, C. (1988). 576–772. Nonoxidative glucose consumption during focal physi- Roy, C. S., & Sherrington, C. S. (1890). On the regulation of ologic neural activity. Science, 241, 462–464. the blood supply of the rat brain. Journal of Physiology Greenberg, J., Hand, P., Sylvestro, A., & Reivich, M. (1979). Lo- (London), 11, 85–108. calized metabolic-ºow couple during functional activity. Seitz R. J., & Roland, P. E. (1992). Vibratory stimulation in- Acta Neurological Scandinavia, 60, supplement 72, 12– creases and decreases the regional cerebral blood ºow 13. and oxidative metabolism: A positron emission tomogra- Greer, C. A., Stewart, W. B., Kauer, J. S., & Shepherd, G. M. phy (PET) study. Acta Neurological Scandinavia, 86, 60– (1981). Topographical and laminar localization of 2-deoxy- 67. glucose uptake in the rat olfactory bulb induced by electri- Sibson, N. R., Dhankhar A, Mason, G. F., Rothman, D. L., Be- cal stimulation of olfactory nerves. Brain Research, 217, har, K. L., & Shulman, R. G. (1998). Stoichiometric cou- 279–293. pling of brain glucose metabolism and glutamatergic Gruetter, R., Seaquist, E. R., Kim, S., & Ugurbil, K. (1998). Lo- neuronal activity. Proceedings of the National Academy calized in vivo 13C-NMR of glutamate metabolism in the of Sciences, USA, 95, 316–321. human brain: Initial results at 4 Tesla. Developmental Neu- Siesjo, B. K. (1978). Brain energy metabolism. New York: roscience, 20, 380–388. Wiley. Hu, X. Le, T. H., & Ugurbil, K. (1997). Evaluation of the early Shulman, R. G., & Rothman, D. L. (1998). Interpreting func- response in fMRI in individual subjects using short stimu- tional imaging studies in terms of neurotransmitter cy- lus duration. Magnetic Resonance Medicine, 37, 877–884. cling. Proceedings of the National Academy of Sciences, Kwong, K. K., Belliveau, J. W., Chesler, D. A., Goldberg, I. E., USA, 95, 11993–11998. Weisskopf, R. M., Poncelet, B. P., Kennedy, D. N., Hoppel, Sokoloff, L. (1981). Relationships among local functional activ- B. E, Cohen, M. S., Tur ner, R., Cheng, H. M., Brady, T. J., & ity, energy metabolism, and blood ºow in the central nerv- Rosen, B. R. (1992). Dynamic magnetic resonance imaging ous system. Federation Proceedings, 40, 2311–2316. of human brain activity during primary sensory stimula- Sokoloff, L., Reivich, M., Kennedy, C., des Rosiers, M. H., Pat- tion. Proceedings of the National Academy of Sciences, lak, C. S., Pettigrew, K. D., Sakurada, O., & Shinohara, M. USA, 89, 5675–5679. (1977). The (14C)deoxyglucose method for the measure- Malonek, D., & Grinvald, A. (1996). Interactions between elec- ment of local cerebral glucose utilization: Theory, proce- trical activity and cortical microcirculation revealed by im- dure, and normal values in the conscious and anesthetized aging spectroscopy: Implications for functional brain albino rat. Journal of Neurochemistry, 28, 897–916. imaging. Science, 272, 551–554. Ueki, M., Linn, F., & Hossmann, K. A. (1988). Functional activa- Marret, S., Fujita, H., Meyer, E., Ribeiro, L., Evans, A., tion of cerebral blood flow and metabolism and after Kuwabara, H., & Gjedde, A. (1993). Stimulus speciªc in- global ischemia of rat brain. Journal of Cerebral Blood crease of oxidative metabolism in human visual cortex. In Flow Metabolism, 8, 486–494. K. Uemura, N. A. Lassen, T. Jones, & I. Kanno (Eds.), Quan- Yang, X., Renken, R., Hyder, F., Siddeek, M., Greer, C., Shep- tiªcation of brain function: Tracer kinetics and image herd, G. M., & Shulman, R. G. (1998). Dynamic mapping at analysis in brain PET (pp. 217–224). Amsterdam: Elsevier. the laminar level of odor-elicited responses in rat olfac- Moskalenko, Y. E., Dowling, J. L., Liu, D., Roainen, C. M., tory bulb by functional MRI. Proceedings of the National Sermernia, V. N., & Woolsey, T. A. (1996). LCBF changes in Academy of Sciences, USA, 95, 7715–7720. rat somatosensory cortex during whisker stimulation moni- Yarowsky, P., Kadekaro, M., & Sooloff, L. (1983). Frequency- tored by dynamic H2 clearance. International Journal of dependent activation of glucose utilization in the superior Psychophysiology, 21, 45–59. cervical ganglion by electrical stimulation of cervical sym- Ogawa, S., Menon, R. S., Kim, S. G., & Ugurbil, K. (1998). On pathetic trunk. Proceedings of the National Academy of the characteristics of functional magnetic resonance imag- Sciences, USA, 80, 4179–4183.

Fitzpatrick and Rothman 471

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892999563454 by guest on 29 September 2021