P ERSPECTIVES sure at mid-latitudes and decreased 0¡ Climate connections. (Top) Relation between pressure at high latitudes (1). variations of the southern annular mode (SAM) Increasing greenhouse gases have 0.5 and rainfall in the Southern Hemisphere, based probably contributed to the ob- on data from a long control climate model sim- 60¡W 60¡E 0.4 served Southern Hemisphere ulation (7). Similar results are obtained with 0.3 warming at mid- and lower latitudes the climate model of Gillett and Thompson (3). and to the observed circulation 0.2 (Bottom) Time series of winter rainfall in changes (strengthening of the 0.1 southwest Western Australia. The decrease in SAM) in winter. However, the mag- Ð0.1 rainfall is consistent with the observed increas- nitude of the circulation response in ing trend in the SAM. these climate models is not nearly Ð0.2 as strong as that found in the obser- Ð0.3 Untangling the separate contributions is 120¡W 120¡E vations or in the ozone-forced mod- Ð0.4 crucial for understanding recent regional el response in summer (3). Ð0.5 climate variations, such as the rainfall The recent changes in the trends in Western Australia, and for pre- Southern Hemisphere circulation at 180¡ dicting how climate is likely to change in high latitudes have clear impacts on the future. Gillett and Thompson (3) have 5 Antarctica and the Southern Ocean. In ad- Seasonal taken an important step in this direction in dition, there may be important links be- 11-year running mean showing that the recent summer circulation tween SAM variations and rainfall in south- 4 changes in the Southern Hemisphere high ern Australia, New Zealand, and South latitudes are likely to be caused by strato- America (see the figure). Increases in the spheric ozone depletion. SAM, with increasing pressure at mid-lati- 3 tudes, are associated with decreases in rain- References and Notes 1. J. T. Houghton et al.,Eds., Climate Change 2001: The fall between 35° and 50°S. There has been 2

Rainfall (mm per day) Scientific Basis,Contribution of Working Group I to a substantial reduction (15 to 20%) in win- the Third Assessment Report of the Intergov- ter rainfall in southwest Western Australia 1950 1960 1970 1980 1990 2000 ernmental Panel on Climate Change (Cambridge over the past 50 years, associated with a Time (years) Univ. Press, Cambridge, UK, 2001). 2. D. W. J. Thompson, S. Solomon, Science 296, 895 southward shift in the winter rain-bearing (2002). weather systems (5). Fyfe has noted this ozone depletion. Hence, natural decadal cli- 3. N. P. Gillett, D. W. J. Thompson, Science 302, 273 southward shift in Southern Hemisphere mate variations are likely to be an important (2003). 4. M. P. Baldwin, D. W. J. Thompson, E. F. Shuckburgh, W. extratropical cyclones in both observational factor in these rainfall decreases. A. Norton, N. P. Gillett, Science 301, 317 (2003). data and model responses to increasing Recent climate changes in the Southern 5. Indian Ocean Climate Initiative Panel, Climate greenhouse gases (6). Hemisphere are likely to result from a Variability and Change in South West Western The observed rainfall trends in southwest complex combination of natural climate Australia (Department of Environment, WA, Australia, 2002). Western Australia are much greater than ex- processes (associated with interactions be- 6. J. C. Fyfe, J. Clim. 16, 2802 (2003). pected from most climate model simulations tween the atmosphere, oceans, and sea ice) 7. W. Cai, P. H. Whetton, D. J. Karoly, J. Clim. 16, 1525 with increasing greenhouse gases. Further- and influences (including decreases (2003). 8. I thank W.-J. Cai, CSIRO Atmospheric Research, more, they occur in a season when there is in stratospheric ozone and increases in at- Australia, for his assistance with the preparation of likely to be little influence from stratospheric mospheric greenhouse gases and aerosols). the figure.

NEUROSCIENCE period, subjects were led to believe that they were only observing the game. During the experimental phase, however, they be- the of Social Loss came active participants in the game. Jaak Panksepp Within a few throws of the ball, the two other “players” (actually computerized he Greek philosopher Zeno of Citium roimaging experiment that seeks to discov- stooges) stopped throwing the ball to the (356 to 264 B.C.), the founder of er whether the metaphor for the psycholog- subjects, leading them to feel excluded (2). TStoicism, considered pain to be one ical pain of social loss is reflected in the The subjects experienced emotional dis- of nine forms of . We often speak neural circuitry of the human brain. Using tress as indicated by substantial blood- about the loss of a loved one in terms of functional magnetic resonance imaging changes in two key brain areas. One of painful , but it is still not clear to (fMRI), they show that certain human these areas, the anterior , what extent such metaphors reflect what is brain areas that “light up” during physical has been implicated in generating the aver- actually happening in the human brain? pain are also activated during emotional sive of physical pain. Enter Eisenberger and colleagues (1) on pain induced by social exclusion. Eisenberger and colleagues demonstrate page 290 of this issue with a bold neu- You might how one measures that the greater the feeling of social dis- the feeling of social exclusion while the tress, the more this brain area becomes ac- The author is at the J. P. Scott Center for subject is lying in an MRI machine. tivated. The other brain region, in the pre- Neuroscience, Mind and Behavior, Department of Eisenberger et al. circumvented this obvi- frontal cortex, showed an opposite pattern , Bowling Green State University, Bowling ous problem in a clever way. In their study, of activity, becoming more active when the Green, OH 43403, USA, and at the Falk Center for Molecular Therapeutics, Department of Biomedical the 13 participants observed a virtual ball- distress was least. In other words, the two Engineering, Northwestern University, Evanston, IL tossing video game while brain blood flow brain areas involved in the distressing feel- 60201, USA. E-mail: [email protected] was monitored by MRI. During a baseline ings of social exclusion responded in op-

www.sciencemag.org SCIENCE VOL 302 10 OCTOBER 2003 237 P ERSPECTIVES posite ways to the degree of social pain ex- ducing the activity of the (subgenual) ante- and the periaqueductal central gray area of perienced. This suggests that the anterior rior cingulate and increasing the activity of the brain stem (see the figure). The latter cingulate is more important for elaborating the posterior cingulate (15). The most pro- two areas are known to control feelings of feelings of emotional distress, whereas the found effects observed by Eisenberger et physical pain. in hu- , already implicated in al. seem to be centered in the central cin- mans, especially grief and intense loneli- emotional regulation (3), counteracts the gulate region that is known to integrate ness, may share some of the same neural painful feeling of being shunned. and cognition. This region is acti- pathways that elaborate physical pain. These results are consistent with the vated during male sexual (12) and Given the dependence of mammalian idea that aversive feelings of social exclu- during stressful cognitive tasks requiring young on their caregivers, it is not hard to sion and physical pain arise, in part, from (16). comprehend the strong survival value con- the same brain regions. They dovetail nice- The feelings induced by experimental ferred by common neural pathways that ly with what we know about separation dis- games in the laboratory, such as elaborate both social attachment and the af- tress in other animals. In our work a quar- “CyberBall” in the Eisenberger et al. study, fective qualities of physical pain. ter of a century ago, we examined the neu- are a pale shadow of the real-life feelings Throughout history poets have written rochemistry of social attachments in ani- that and other animals experience about the pain of a broken heart. It seems mals (4, 5). We found that the same neuro- in response to the sudden loss of social that such poetic insights into the human chemicals that regulate physical pain also control the psychological pain of social loss. Indeed, plant opioids (such as morphine) as well as endogenous brain opioids (espe- cially endorphins)—known to alle- viate physical pain—also alleviated separation distress (as measured by cries) in dogs, guinea pigs, chicks, rats, and primates (6). How can we further elucidate the neural mechanisms that under- lie the emotional pain induced by social exclusion? Two strategies might help. If the participants in the Eisenberger et al. study were to be given opioids, one would predict that they would feel less distress at being shunned, and that the brain areas implicated in elaborating The emotional pain of social loss. There are remarkable similarities between regions of the guinea pig brain such feelings would not be as pro- that when activated provoke separation distress and areas of the human brain that are activated during feel- foundly activated. The administra- ings of . During separation distress in guinea pigs, the most responsive brain areas are the anterior cin- tion of opioid receptor antagonists gulate (AC), the ventral septal (VS) and dorsal preoptic areas (dPOA), the bed nucleus of the stria terminalis should intensify both effects. Other (BN), the dorsomedial (DMT), and the periaqueductal central gray area of the brain stem (PAG) (18, brain chemicals, such as oxytocin 19). In humans experiencing sadness (17), it is the anterior cingulate that is most responsive, but other areas and prolactin (6) that are also pow- that are also activated include the DMT, PAG, and insula. The correspondence between the brain regions ac- erful regulators of separation dis- tivated during human sadness and those activated during animal separation distress suggests that human tress in animals, may have effects feelings may arise from the instinctual emotional action systems of ancient regions of the mammalian brain. similar to those of opioids but they OB, olfactory bulb; CC, corpus callosum; CB, . are too difficult to manipulate ex- perimentally in humans. support. It will be interesting to study more condition are now supported by neurophys- Brain imaging has yielded a plethora of intense emotional states arising from pro- iological findings. Will the opposite also neural correlates of affective states (3) in- found personal loss with fMRI, which prove to be the case—that socially support- cluding the response to breathlessness (7), should allow us to probe even deeper into ive and loving feelings reduce the sting of the craving for chocolate (8), winning the the regions of the mammalian brain that pain (20, 21)? A reasonable working hy- lottery (9), the sex-specific appeal of pret- control separation distress (6). A step in pothesis is that social feelings such as ty faces (10), the of peak musical this direction is the visualization by are constructed partly from brain neural (11), human (12), positron emission tomography of regions circuits that alleviate the feelings of social male sexual arousal (13), and even rectal of the human brain activated during sad- isolation. Will we eventually discover that distension (14). The human brain areas im- ness (17) (see the figure). These regions the feeling of a broken heart arises from the plicated by Eisenberger et al. in feelings of correspond to some of the deeper brain ar- rich autonomic circuits of the brain’s lim- social pain have many other functions. The eas activated during separation distress in bic system that control cardiac neurody- cingulate gyrus (a long ribbon of tissue at animals. Localized electrical of namics? Will we find that people we con- the brain’s midline) contains two distinct many subcortical brain sites provokes sep- sider “cold” or “warm” influence different “emotional” zones: The far anterior region aration cries in mammals (6, 18, 19). These thermoregulatory neural pathways in our elaborates negative feelings, whereas pos- sites include not only the anterior cingu- brains? terior regions elaborate positive feelings. late, but also the bed nucleus of the stria As exemplified by the Eisenberger et al. The effects of -elevat- terminalis, the ventral septal and dorsal study, such poetic insights garner some ing drugs and placebos both depend on re- preoptic areas, the dorsomedial thalamus, support from neurophysiological research,

238 10 OCTOBER 2003 VOL 302 SCIENCE www.sciencemag.org P ERSPECTIVES adding a humanistic touch to the mind- References and Notes 14. M. K. Kern et al., Am. J Physiol. 281, G1512 (2001). brain sciences. Feelings of love and loneli- 1. N. I. Eisenberger, M. D. Lieberman, K. D. Williams, 15. H. S. Mayberg et al., Am. J. 159, 728 (2002). ness, and the thoughts they provoke, are Science 302, 290 (2003). 16. C. S. Carter et al., Proc. Natl. Acad. Sci. U.S.A. 97, 1944 2. To visualize the game, see www.psy.mq.edu.au/ (2000). constructed in part from neural pathways in staff/kip/cyberball.htm. 17. A. R. Damasio et al., Nature Neurosci. 3, 1049 (2000). the brain that regulate core emotional re- 3. K. L. Phan et al., NeuroImage 16, 331 (2002). 18. B. H. Herman, J. Panksepp, Science 211, 1060 (1981). sponses, such as playfulness, sexuality, and 4. J. Panksepp et al., Neurosci. Biobehav. Rev. 4, 473 19. J. Panksepp et al., in The Physiological Control of Mammalian Vocalizations,J.D.Newman, Ed. (Plenum, friendship, as well as separation distress in (1980). 5. J. Panksepp, in Progress in Theory in Psycho- New York, 1988), pp. 263–299. 20. J. Panksepp. Psychopharmacology 72, 111 (1980). our fellow creatures (6). pharmacology,S.J.Cooper, Ed. (Academic Press, 21. When I lost my daughter 12 years ago in a horren- Of course, scientists should continue to London, 1981), pp. 149–175. dous traffic accident, among her papers I found a po- 6. J. Panksepp, (Oxford Univ. be skeptical about such hypotheses until em that is now carved on her tombstone. The last they are supported by solid research such Press, New York, 1998). stanza is particularly pertinent to the question of as that carried out by Eisenberger and co- 7. M. Liotti et al., Proc. Natl. Acad. Sci. U.S.A. 98, 2035 whether love can reduce the emotional pain of loss. (2001). workers. Although there are many species When your days are full of pain, 8. D. M. Small et al., Brain 124, 1720 (2001). And you don’t know what to do, differences in the emotional systems that 9. B. Knutson et al., J. Neurosci. 21, RC159 (2001). Recall these words I tell you now we share as ancestral gifts with other ani- 10. I. Aharon et al., Neuron 32, 537 (2001). —I will always care for you mals (6, 22, 23), the field of neuroscience 11. A. J. Blood et al., Proc. Natl. Acad. Sci. U.S.A. 98, Full poem is published in A. Miller, A Road Beyond will be more productive if it remains open 11818 (2001). Loss (Memorial Foundation for Lost Children, Bowling 12. J. Decety, T. Chaminade, Neuropsychologia 41, 127 Green, OH, 1995). to the similar nature of human and animal (2003). 22. J. Panksepp, Psychol. Rev. 110, 376 (2003). affective experiences. 13. J. Redoute et al., Hum. Brain Mapp. 11, 162 (2000). 23. J. Panksepp, & Emotion 1, 17 (2000).

CHEMISTRY characteristic time window. The diffraction of x-rays and neutrons falls into this group, The Motions of as do various spectroscopies, including nu- clear or electronic magnetic resonance (4, 5), Mössbauer absorption (6), infrared, and an Enzyme Soloist Raman spectroscopy. NMR is particularly powerful because it can distinguish each Michel Orrit amino acid residue in the sequence, but it does not cover all fluctuation time scales. n page 262 of this issue, Yang et al. tlety and complexity is still in its infancy. Second, an ensemble of molecules may report a fluorescence study of the Numerical simulations (3) can handle ever be synchronously brought out of equilibri- Oconformational fluctuations of a more complexity, but the longest simula- um, and its subsequent moni- single enzyme molecule (1). As we can tions do not exceed a few nanoseconds. tored as a function of time. The initial dis- observe every time we boil an egg, pro- Moreover, computations must eventually turbance in such kinetic measurements can teins are not static. Conformational flexi- be compared with experiments. be a temperature jump, a sudden concen- bility at various levels of their hierarchical Current experiments to explore protein tration change, or—on much shorter time structure enables proteins to fulfill a wide dynamics are of two kinds. First, fluctua- scales—the breaking of a bond by a laser range of complex biological functions. tion amplitudes may be measured with a pulse (7). Under such strong perturbations, Yang et al. now follow these fluctuations range of techniques, each of which has a however, proteins may no longer be close over a range of time scales at equilibrium, to equilibrium. Furthermore, and not in response to some initial large R synchronization is short-lived. disturbance. 1 Because different individual Most proteins are highly complex R2 molecules follow different path- macromolecules (2). Driven by Brownian ways in the energy landscape, motion, they wander in an intricate multi- dimensional energy landscape, featuring Monitoring dynamics via fluores- multitudes of interconnected wells and cence. Variations of the distance R dells, valleys and passes. Hopping from between an emitting fluorophore well to well can be as fast as tens of fem- (flavin, sketched as a light bulb) and toseconds for fast backbone vibrations, or an electron acceptor and quencher as slow as hours and even days for fold- (tyrosine, sketched as a black ball) ing and maturation in large proteins—a V(R,t') cause variations of the fluorescence range of time scales spanning 18 orders intensity and lifetime. The wiggly curve schematically represents a cut of magnitude. of the potential energy of the en- To make matters worse, protein func- zyme along R. Because the multidi- tions such as catalytic reactions are deter- mensional potential depends on mined by short-range atom-atom interac- many other time-dependent atomic tions and critically depend on tiny atomic coordinates, the one-dimensional

SCIENCE displacements. Understanding of such sub- potential is time dependent and is V(R,t) shown here at two different times. The coexistence of low and high The author is with Molecular NanoOptics and Spins, Leiden Institute of Physics, Leiden University, Post Office barriers in the transient potential Box 9504, Netherlands. E-mail: orrit@molphys. R1 R2 generates stretched kinetics with a

CREDIT: HUEY/ PRESTON leidenuniv.nl R broad range of relaxation times.

www.sciencemag.org SCIENCE VOL 302 10 OCTOBER 2003 239