P ERSPECTIVES 20 to 30 minutes to reach a stable state. can change without affecting CA1 activity these questions still waiting to be answered, it What do these findings suggest about spa- suggests that the CA1 region responds to di- will be interesting to see how future studies tial processing within the hippocampus? One rect inputs from the entorhinal cortex rather gradually tease apart the components of the possibility is that CA3 encodes the animal’s than to information routed through CA3 (9, entorhinal-hippocampal memory system. location with reference to the current envi- 10). But why would CA1 echo the entorhinal This process will lead to a clearer picture of ronmental context. This would be a useful ex- representation? One possibility is that CA1 how neural networks in these regions interact tension to the coding scheme in the entorhi- melds information that was not explicitly ex- during memory storage and retrieval. Now, nal cortex, where “place cells” appear to be amined by Leutgeb et al. into the spatial rep- where did I leave my keys? relatively immune to changes in the environ- resentation. Such information could include ment (8). The context-dependent representa- nongeometric features of the environment, References 1. M. Fyhn et al., Science 305, 1258 (2004). tion in CA3 might allow spatial memory to such as odor or behavioral contingencies, or 2. S. Leutgeb et al., Science 305, 1295 (2004). differentiate between different environments internal states such as hunger, thirst, or moti- 3. J. O’Keefe, J. Dostrovsky, Brain Res. 34, 171 (1971). and so, for example, allow me to locate my vation. Alternatively, because remapping is 4. E. A. Maguire et al., Science 280, 921 (1998). 5. I. Lee, R. P. Kesner, Hippocampus 14, 301 (2004). car in the supermarket parking lot without in- known to take place in this region following 6. R. U. Muller, J. L. Kubie, J. Neurosci. 7, 1951 (1987). terference from occasions when I parked it at manipulations of the spatial relationship be- 7. K. J. Jeffery et al., Neurosci. Biobehav. Rev. 28, 201 (2004). 8. G. J. Quirk et al., J. Neurosci. 12, 1945 (1992). the mall. But what about CA1? Leutgeb et tween local and distal cues (10), CA1 may 9. V. H. Brun et al., Science 296, 2243 (2002). al.’s finding that the pattern of activity in CA3 help to represent intercue associations. With 10. I. Lee et al., Nature 430, 456 (2004).

BEHAVIOR suggested a connection between striatal ac- tivation and the satisfaction one might de- Sweet Revenge? rive from punishing a defector, they do not establish a directional relationship between Brian Knutson the two. Thus, in a clever internal analysis, the investigators observed that the degree of ou’ve been waiting in line in traffic In each interaction, subjects chose to give striatal activation during no-cost punish- for what seems like hours, when a their partners money, which was then ment predicted the extent to which subjects Yred sports car whips past on the quadrupled. Next, partners who received the chose to punish at a personal cost (that is, shoulder. Eventually, the sports car creeps money had a chance to reciprocate, or to pay under less satisfying conditions). This find- back into view—the driver has run out of back half to the subject. ing suggested to the in- shoulder and signals to be let in. Instead of If partners decided not to vestigators that striatal giving way, you stare ahead and accelerate, reciprocate, or defected, activation indexed sub- inching dangerously close to the bumper in subjects could choose to jects’ anticipation of front of you. After squeezing back the in- administer punishment. satisfaction, rather than truder, you can’t help but notice a smile At this point, their brains satisfaction per se. creep onto your face. were scanned. These findings fit a Judges worry, whereas filmmakers de- De Quervain and co- fresh piece into the rap- light, in the fact that revenge feels good. workers first asked idly expanding puzzle Evolutionary theorists argue that such an whether choosing to of reward processing as “eye-for-an-eye” strategy makes sense, pre- punish a defector would revealed by brain imag- venting future damage to one’s self or kin (1, recruit brain circuits im- ing. Ironically, punish- 2). Yet, in cases ranging from inconsiderate plicated in reward pro- “Go ahead, make my day.” Dirty Harry ment of defectors in this drivers to Nazi war criminals, even unrelated cessing. They found that succinctly informs a norm violator that study activated the same onlookers seem highly motivated to seek re- when subjects adminis- he anticipates deriving satisfaction from regions (that is, striatum venge, often in spite of personal cost. From tered a monetary pun- inflicting altruistic punishment. and MPFC) that were the standpoint of self-interest, punishing ishment to defectors, a activated when people those who violate the interests of strangers— subcortical region of the brain called the rewarded cooperators in a recent functional a form of revenge called altruistic punish- striatum increased its consumption of oxy- magnetic resonance imaging (fMRI) study ment—seems irrational. Enter de Quervain gen (that is, was “activated”). The investiga- (6). These seemingly diametrically opposite and colleagues (3) on page 1254 of this is- tors interpreted this to indicate that punish- social behaviors are united by a common sue, who offer an alternative explanation— ing a defector activates brain regions related psychological experience—both involve the instead of cold calculated reason, it is pas- to feeling good about revenge rather than anticipation of a satisfying social outcome. sion that may plant the seeds of revenge. brain regions related to feeling bad about As presaged by comparative research (7), Using an elegant laboratory task designed having been violated. Indeed, these striatal humans also show increased striatal activity to elicit acts of revenge among human volun- foci lie near brain areas that rats will work during anticipation of nonsocial rewards teers, de Quervain and colleagues appear to furiously to stimulate electrically (4). The such as monetary gains (8) and pleasant have captured this complex emotional dy- investigators then asked whether the stria- tastes (9). Together, these findings imply that namic of schadenfreude with a positron tum would be activated even when adminis- for certain parts of the striatum, it’s the feel- emission tomography (PET) camera. During tering the punishment carried a personal ing that counts. the task, subjects played games involving cost. They found that the striatum was still As with any compelling study, the find- real money with a series of different partners. activated when subjects chose to administer ings raise additional questions for future re- punishment at a personal cost, as was a re- search. Although PET measures absolute The author is in the Department of Psychology, gion in the medial prefrontal cortex (MPFC) metabolism (and can even provide neuro- Stanford University, Stanford, CA 94305, USA. E-mail: that has been implicated in balancing costs chemical information), its spatial and tem-

[email protected] and benefits (5). Although these findings poral resolution are limited (in this case, to COLLECTION KOBAL CREDIT: WARNER/MALPASO/THE

1246 27 AUGUST 2004 VOL 305 SCIENCE www.sciencemag.org Published by AAAS P ERSPECTIVES 15 mm3/min). Thus, although they were able imaging studies of social behavior (12). The Back in traffic, brake lights flare ahead. to visualize activation at the head of the cau- new results also suggest that, depending on You realize that your smile was short-sight- date, the investigators may not have been able social learning, some of the same emotions ed. Your car skids to a halt. Fortunately, the to track activity in the smaller ventral part of that bring us together can also break us apart. smile didn’t cost a pile-up. This time. the striatum—the part most directly implicat- The findings of de Quervain et al. also ed in motivation (10). Fortunately, event-relat- chip yet another sliver from the rational References ed f MRI can resolve activity in smaller re- model of economic man. In fact, their sub- 1. R. H. Frank, Passions Within Reason: The Strategic gions (~4 mm3) on a second-to-second basis jects illustrated at least two types of irra- Role of the Emotions (Norton, New York, 1988). 2. R. Trivers, Q. Rev. Biol. 46, 35 (1971). (11). Techniques like this may enable future tionality: reacting on the basis of emotion- 3. D. J.-F. de Quervain et al., Science 305, 1254 (2004). investigators to make even more specific ob- al considerations and spending costly per- 4. J. Olds, P. Milner, J. Comp. Physiol. Psychol. 47, 419 servations regarding when and where activa- sonal resources to ensure that defectors got (1954). tion occurs during altruistic punishment. their due. Beyond providing a compelling 5. A. Bechara, H. Damasio, A. R. Damasio, Cereb. Cortex 10, 295 (2000). Second, while the present PET study of de- justification for adding social justice con- 6. J. K. Rilling et al., Neuron 35, 395 (2002). fectors included male subjects, the aforemen- cerns to existing economic models, the 7. S. Ikemoto, J. Panksepp, Brain Res. Rev. 31, 6 (1999). tioned f MRI study of cooperators included findings serve as a harbinger of future 8. B. Knutson, C. M. Adams, G. W. Fong, D. Hommer, J. females. Future research will undoubtedly “neuroeconomic” studies that strive to de- Neurosci. 21, RC159 (2001). 9. J. P. O’Doherty, R. Deichmann, H. D. Critchley, R. J. need to explore which social interactions most scriptively reconstruct these models using Dolan Neuron 33, 815 (2002). powerfully motivate men compared with neurobehavioral data. One can imagine the 10. D. Martinez et al., J. Cereb. Blood Flow Metab. 23, 285 women (as well as members of different so- new models accommodating both “pas- (2003). cial groups). Regardless, the findings do pow- sionate” and “rational” forces, as well as 11. B. Knutson, G. W. Fong, C. M. Adams, J. L. Varner, D. Hommer, NeuroReport 12, 3683 (2001). erfully illustrate the importance of consider- specifying when and how they come to- 12. P. D. MacLean, The Triune Brain in Evolution (Plenum ing proximal emotional mechanisms in brain gether to influence individual choice. Press, New York, 1990).

GEOSCIENCE ruary 2004, causing nearly 600 deaths. One big question remains. Is the subduc- What Caused the tion system still active, and does it pose a seismic risk? New evidence supports con- tinued activity. Numerous active mud volca- Great Lisbon Earthquake? noes have been identified and sampled in the Gulf of Cadiz (10). These features indi- Marc-André Gutscher cate ongoing dewatering processes, which are widespread in accretionary wedges n 1 November 1755, as worship- deep marine basin in the West Alboran Sea (compressed sediment piles formed at sub- pers in Portugal and southwestern (western Mediterranean), while shortening duction zones, like piles of dirt in front of a OSpain were gathered for mass on and thrusting continued in the horseshoe- bulldozer). Marine seismic data indicate ac- All Saint’s Day, a tremendous earthquake shaped Betic and Rif mountain belts (see tive folding and thrusting of the youngest struck, toppling many churches and killing the figure, left panel) (4, 5). A popular mod- sediments (which are a few thousand years about 60,000 people (1, 2). Many church- el concluded that this region was a prime ex- old) at the outermost edge of this accre- goers were killed, sparking a lively debate ample of “delamination” (breaking off of a tionary wedge (11). Marine heat flow data among philosophers about divine justice. deep mantle root following continental col- are also indicative of active subduction (12). Recent studies have shed light on what lision) (4). However, new data increasingly An active subduction zone off southern caused the earthquake and what the seis- support eastward subduction beneath the Iberia poses a long-term seismic risk and is mic future of the region may be. Straits of Gibraltar (see the figure, right a likely candidate for having produced the The Great Lisbon earthquake had an es- panel) (5, 6). Tomographic cross sections of Great Lisbon earthquake in 1755. How- timated magnitude (M) 8.7. It triggered a 5- the Earth show cold, dense material—a slab ever, no instrumentally recorded subduc- to 10-m-high tsunami and caused many ca- of oceanic lithosphere—descending from tion interface earthquakes have been sualties in Europe and northwestern the surface to depths of nearly 700 km (6). recorded in the Gulf of Cadiz. Hence, sub- Morocco (2). In this region, the African The chemistry of 15- to 5-million-year-old duction has either ceased, is active and plate pushes toward the northwest against volcanoes in the Alboran Sea shows that aseismic, or is active but the seismogenic southern Iberia at a rate of 4 mm/year (see they were formed in an arc setting (like that fault zone is locked. The latter interpreta- the figure, left panel). But the plate bound- of arcuate island chains in the West Pacific tion seems most likely. In this case, the ary off southern Iberia is not well defined landward of the subduction zone) (7). Gibraltar subduction would resemble the (3), and the source of the Great Lisbon Overall, the movement of tectonic Nankai and Cascadia subduction zones, earthquake has remained elusive (2). blocks in the southern Iberia region is best which are characterized by a large locked Indeed, it has been difficult to find a simple explained by a model of slab retreat (roll- zone and have a recurrence time of 100 to plate-tectonic model that explains all geo- back) during subduction, causing exten- 1000 years for great earthquakes. logical observations in the region (4, 5). sion in the region behind the subduction Additional clues can help to pinpoint During the past 15 million years, crustal zone (5, 6, 8). The southeastern limit of the likely location of the Great Lisbon thinning and extension have produced a deformation in this back-arc region ap- earthquake. The generation of a strong pears to be a major north-east trending tsunami implies a source region mostly at strike-slip fault across the West Alboran sea. The tsunami wave can be modeled and The author is at the Institut Universitaire Européen Sea (9). This fault emerges on land in compared to historical observations (2). A de la Mer (IUEM), CNRS Unité Mixte de Recherche 6538, Domaines Oceaniques, F-29280 Plouzané, northeast Morocco, right where the Al Hoce- record of past great earthquakes exists in France. E-mail: [email protected] ima earthquake (M = 6.3) struck on 24 Feb- the abyssal plains off southwest Iberia in

www.sciencemag.org SCIENCE VOL 305 27 AUGUST 2004 1247 Published by AAAS