Chemicals in the Brain that Impact Learning and Memory

SCIENCE'S COMPASS

records global ice volume, ocean tempera- ture, and tlie mixing ratio of fresh water and seawater. To separate out the freshwater in- put, Maslin and Bums removed the ice vol- ume and ocean temperature components by subtracting an independent planktonic oxy- gen isotope record south of ¿le Amazon and upstream in the NBCC, The residual was further adjusted for the effects of tempera- ture and rainfall amount on the oxygen iso- tope composition of river water.

The net result is an indirect measure of Amazon River outflow that is broadly consistent with the global methane curve from the Green- land tee Sheet Project 2 (GISP2)

Retrieving sediment and ice cores. The JOIDES Resolution drilling ship took the core analyzed by Maslin and Burns (7 7), (Inset) The site of C1SP2, one of the Greenland ice cores used to measure past methane concentrations,

ice core (see the figure) (5, 6). The best match is during the Younger Dryas (13,000 to 11,600 years a g o ) , when ice-core methane and reconstructed Amazon dis- charge both dropped to 60% below modem values (2, 11). Both records exhibit anoma- lous peaks, which occur 11,600 years ago in the methane record and 11,800 years ago in the Amazon outflow. The latter was proba- bly due to increased rainfall in the lowlands rather than meltwater from Andean glaciers.

The overall trend in Amazon outflow tracks summertime solar insolation at tO°S, which reached a minimum between 12,000 and 10,000 years ago and a maximum in the past 3000 years. These insolation differences are thought to regulate the intensity of con- vection over the Amazon Basin and the Cen- tral Andes, which in tum affects westward penetration of Atlantic moisture and southern extension of the Intertropical Convergence Zone (ITCZ), On page 2291 of this issue,

, Mayle et al. (75) also summon increasing summer insolation at 10°S to explain south- em expansion of Amazonian rainforests in eastern Bolivia during the past 3000 years,

Maslin and Bums' elegant study is proba- bly not the final word. The authors make sev- eral key but unproven assumptions to quanti-

fy Amazon discharge from the foraminiferal record. For example, the dependence of the oxygen isotope composition of rainfall on temperature and rainfall amounts over the Amazon Basin can be complicated by changes in the position of the ITCZ, which may push isotopically depleted moisture in- land (7Ó), Trade wind intensities along the northem South American coastline, which changed dramatically during déglaciation (/ 7), also could have modulated the position and width of the Amazon freshwater plume,

affecting its mixing with the NBCC (18). Furthermore, little at- tempt has been made to allow for the effects of rising sea level on the extent of Holocene wetlands. During the last ice age, when sea level was 100 m below that of to- day, the increased gradient caused the Amazon and its tributaries to incise tens of meters below their floodplains. Ten thousand years ago, sea level was still 25 m below

modern levels, and it rose only gradually throughout the Holocene, Incised valleys slowly backfilled with sediment, but tribu- taries originating in sediment-starved low- lands could not keep up with the rising water, resulting in large freshwater lakes (79), These lakes are only now being drowned in sedi- ment, implying that the maximum extent of methane-producing wetlands in the Amazon Basin may depend more on rising sea level than on increasing rainfall.

Finally, it remains unclear how orbital modulation of seasonal insolation might force tropical precipitation. During the

PERSPECTIVES: N E U R O S C I Ë N C Ë

past t million years, increases in lowland Amazon Basin precipitation have coincid- ed with ice-melting events and maximum June insolation at 65°N (20), not maximum January insolation at 10°S, Physical mech- anisms for high-latitude forcing of the tropics could involve changes in oceanic heat transport, as well as remote telecon- nections with the Asian Monsoon and Pa- cific climate (27,22).

References and Notes 1, T, P, Cuilderson, R, C, Fairbanks, J, L Rubensonte, Sci-

ence 263,663 (1994), 2, L C, Thompson et ai. Sc/ence 269,46 (1995), 3, M, Stute et ai. Science 269,379 (1995), 4, P, Colinvaux et ai. Science 274,85 (1996), 5, J, P, Severinghaus, E, J, Brook, Science 286,930 (1999), 6, D, Raynaud et aL. Quat. Sei Rev. 19,9 (2000), 7, I, Farrera et ai. Clim. Dyn. 15,823 (1999), 8, C, O, Seltzer, D, Rodbell, S, J, Burns, Geology 28, 35

(2000), 9, L, C,Thompson, Quat Sei Rev. 19,19 (2000),

10, J,LBetancourt etal..Sc/ence289,1542 (2000), 11, M, A, Maslin, S, J, Bums, Science 290,2285 (2000), 12, R, H, Meade, Quat. int. 21,29 (1994), 13, The site is located at 6°N, 49°W, 3346-m water depth.

It is one of 17 sites drilled on the Amazon Fan during O D P L e g l 5 5 ,

14, M,A, Maslin et ai.J. Quat Sd. 15,419 (2000), 15, F, E, Mayle, R, Burbridge, T, J, Killeen, Science 290,2291

(2000). 16, K, RozanskI, L Araguás-Araguás, R, Confiantini, in Cli-

mate Change in Continental Isotopic Records. P, K, Swart et ai. Eds,, vol, 78 of Geophysical Monograph Se- ries (American Geophysical Union, Washington, DC, 1993), pp, 1-36,

17, K, A, Hughen, J, T, Overpeck, L C, Peterson, S, Trumbore, Nature 380,51 (1996),

18, P, W, Jewell, R, F, Stallard, C, L, Mellor, / Sediment. Petroi 63, 734 (1993),

19, C, Keim et ai. paper presented at Manaus '99—Hydro- logical and Geochemical Processes in iarge-Scale River Basins. Manaus, Brazil, 15 to 19 November 1999,

20, S, E, Harris, A, C, Mix, Quat. Res. 51,14 (1999), 2 1 , Z, Liu, J, Kutzbach, L Wu, Geophys. Res. iett. 27, 2265

(2000), 22, A, C, Clement, R, Seager, M. A, Cane, Pateoceanography

14,441(1999),

Boosting Working Memory Trevor W. Robbins, Mitul A. Mehta, Barbara J. Sahakian

M any parts of the brain are involved in the formation and storage of long- and s h o r t - t e r m memory.

Working memory—a form of short-term memory that depends on different popula- tions of" brain neurons, in particular those in the prefrontal cortex—serves to main- tain temporary, active representations of in- formation that can be rapidly recalled (7), Neurons in the prefrontal cortex and asso-

T, W, Robbins is in the Department of Experimental Psychology, University of Cambridge, Cambridge CB2 3EB, UK, E-mail: twr2@cus,cam,ac,uk M, A, Mehta is in the MRC Cyclotron Unit, Imperial College School of Medicine, Hammersmith Hospital, London W l 2 ONN, UK, E-mail: mitul,mehta2(B>csc,mrc,ac,uk B, J, Sahakian is in the Department of Psychiatry, Univer- sity of Cambridge, Addenbrooke's Hospital, Cam- bridge CB2 2QQ, UK,

ciated areas receive input from cholinergic pathways comprising neurons that release the neurotransmitter acetylcholine, which originate in the reticular core of the brain- stem and basal forebrain (see the figure). This anatomical organization leads to an obvious strategy for improving working memory: increasing the amount of acetyl- choline in synapses. That this strategy works is demonstrated by Furey et al. (2) on page 2315 of this issue. Using function- al magnetic resonance imaging (fMRI), these authors show that enhancing cholin- ergic activity with the drug physostigmine (which blocks the breakdown of acetyl- choline) improves the efficiency of work- ing memory in humans.

Brains of human subjects performing a visual recognition task were imaged first dur-

www,sciencemag,org SCIENCE VOL 290 22 DECEMBER 2000 2275

SCIENCE'S COMPASS

Frontal Cortex

ACh

ing infusion of physostigmine, and then, on a subsequent day, during infusion of a saline placebo (2). The visual recognition task com- prised three stages—3 seconds to visualize a human face (encoding), a 9-second pause during which the face is "held" in working memory (memory), and then presentation of the original face and a new face, requiring that one face be recognized (recognition). In the new work (2), and in two previous studies using positron emission tomography (PET) (5), physostigmine accelerated the subjects' ability to recognize visual stimuli (human faces). In the PET studies, this improve- ment correlated with a decrease in , / brain activity in the dorsolateral prefrontal cortex—a region of the brain considered crucial for accurate working memory— and an increase in brain activi- ty in regions of the visual cor- tex. Because of the poorer tem- poral resolution of PET com- pared with fMRI, the PET work did not provide information on the parts of the brain that were activated at each stage of the vi- sual recognition task.

With fMRI, Furey et al. (2) now show that the increased ac- tivity in the visual cortex after physostigmine treatment oc- curred during the encoding of faces. T h e r e f o r e , improved working memory performance may be due, in part, to enhance- ment of the earliest stages of vi- sual processing in the cortex, possibly through an increase in the signal-to-noise ratio of neu- ronal information processing {4). Increased visual processing in response to physostigmine is consistent with results from other work in which animals were infused intracerebrally with selective c h o l i n e r g i c agents (5), but it is unclear how the Furey results relate to other findings in experimental ani- mals. For example, Furey and colleagues suggest that, for certain types of memory, boosting the input of visual infor- mation leads to reduced activity in the pre- frontal cortex. However, injection into the rat prefrontal cortex of muscarinic or nico- tinic receptor antagonists—which prevent acetylcholine from binding to its recep- tors—^produces different profiles of impair- ment on two working memory tasks; only the more demanding task was impaired by the nicotinic receptor antagonist (6). This raises the possibility that acetylcholine might have different effects depending on whether it binds to muscarinic or nicotinic

receptors. Consistent with this notion, in both normal volunteers and patients with Alzheimer's disease, nicotine improves per- formance on working memory tasks that demand heightened attention (4).

The drug-induced changes seen by Furey and co-workers in the prefrontal cortex during face recognition, unlike those' in the posterior regions of the brain, were not preferentially as- sociated with any particular stage of the task. The authors choose to explain this finding in terms of the Petrides model of working memo-

BS-ACh

DA ACh From NBM From LC

Multitasking in the brain. The main ascending cholinergic and nnonoaminergic pathways in the brain and their possible contri- butions to working memory (2, 3, 11 12!). Different neurotrans-

mitter pathways—acetylcholine (ACh), dopamine (DA), nore- pinephrine (NE)—modulate working memory through sepa- rate mechanisms. It remains unclear whether the serotonin pathway (not shown) is involved in working memory ( 12). For clarity, the back-projections from the frontal cortex and the projections between the neurotransmitter groups have been omitted. DLPFC, dorsolateral prefrontal cortex; VLPFC, ventro- lateral prefrontal cortex; VTA/SN, ventral tegmental area/sub- stantia nigra pars compacta; LC, locus coeruleus; NBM, nucleus basalis of Meynert; BS-ACh, brainstem cholinergic neurons. The projections depicted reflect possible modulatory influ- ences on working memory. Anatomically, the NBM and LC project to most of the cortical mantle and the VTA/SN has fewer projections in more posterior regions ( 13).

ry (7). This model assigns the more passive ("on-line") short-term maintenance of infor- mation (8) and the more active ("executive") processing of information held on-line to the ventral and dorsal regions of the prefrontal cortex, respectively. The investigators postulate that decreased dorsal prefrontal cortex activity reflects reduced requirements for "executive" operations after increased posterior cortical ac- tivity. But not all activity in the prefrontal cor- tex was reduced during the face recognition task after physostigmine inftasion; increased activity was stiU observed in the inferior pre- frontal cortex. As the authors point out, it is

unclear whether the activity of this area sub- sumes the venti-olateral prefrontal cortex. If this area is close to the venti-olateral prefrontal cortex (area BA47), then this might reflect en- hancement of the entire network of "on-line" working memory (7). The precise relationship between working memory and different re- gions within the prefrontal cortex is currently the subject of intense debate (8). The Furey et al. study can now be extended with different working memory tasks that vary in their de- gree of "executive" and perceptual require- ments. Thus, the finaj interpretation of drug- related changes in the prefrontal cortex will ul- timately depend on exactly which parts of the prefrontal cortex carry out each stage of work- ing memory and on the exact brain regions where drug-induced changes in activity occur.

Modtilating the activity of monoaminer- gic neuronal pathways (that release m o n o a m i n e n e u r o t r a n s m i t t e r s such as dopamine and norepinephrine) controls dy- namic neural networks in the neocortex (9, 10). For example, methylphenidate (an indi- rect enhancer of dopamine and nore- pinephrine) decreases the activity of a work- ing memory "circuit" that includes the dorso- lateral prefrontal cortex and the posterior parietal cortex, while improving overall per- formance on a memory task (70). Together with the new Furey et al. findings, these stud- ies raise the exciting possibility that aspects of working memory may be improved by drugs with selective actions on difièrent neu- rotransmitter systems, resulting in possible therapeutic benefits for patients with cogni- tive disorders such as Aljjieimer's disease.

References 1. A. D. Baddeley, Woridng Memory (Clarendon, Oxford,

1986); P. Coldman-Rakic, in iHandbooic of Physioiogy, F. Plum, V. Mountcastle, Eds. (American Physiological Society, Washington, DC, 1987), pp. 373-417.

2. M. L Furey, P. Pietrini, J. V. Haxby, Sdenee 290, 2315 (2000).

3. M. L. Furey et ai., Proc. Nati. Acad. Sci. U.S.A. 94, 6512 (1997); M. L Furey etai. Brain Res. Buii 5 1 , 213 (2000).

4. T. W. Robbins et ai, Psychopharmacoiogy 134, 95 (1997); B. J. Sahakian et ai., Br. J. Psychiatr. 154, 797 (1989); B. J. Sahakian et ai, Psychopiiarmacoiogy 110,395(1993).

5. T. C. Aigner, M. Mishkin, Behav. Neural Bioi 45, 81 (1986); P. C. Murphy, A. M. Sillito, Neurosdence 40, 13(1991).

6. S. Cranon et ai, Psychopharmacoiogy 119, 139 (1995); B. J. Everitt, T.W. Robbins, Annu. Rev. Psyehoi 48,649(1997).

7. M. Petrides, Phiios. Trans. R. Soc iondon Ser. B 351, 1455 (1996).

8. P. Coldman-Rakic, Neuroimage 11, 451 (2000); M. F. S. Rushworth, A. M. Owen, Trends Cognit Sd 2, 46 (1998).

9. P. M. Crasby et ai, Eur. J. Neurosd. 4,1203 (1992); J. T. Coull et ai., Eur.}. Neurosd. 9,589 (1997).

10. M.A. Mehta etai,}. Neurosd. 20, RC65 (2000). 11. A. F.T.Arnsten,7. Psychopharmacoi 11,151 (1997);T.

Sawaguchi, Parkinsonism Rei Dis. 7,9 (2001). 12. M. Luciana, P. F. Collins, R. A. Depue, Cerefc. Cortex 8,

218 (1998). 13. M.-M. Mesulam, In Psychopharmacoiogy: The Fourth

Ceneration of Progress, F. E. Bloom, D. J. Kupfer, Eds. (Raven, New York, 1995), pp. 135-146; D. A. Lewis, in Stimuiant Drugs and ADHD: Basic and Ciinicai Neuro- sdence, M.V. Solanto, A. F.r. Arnsten, F. X. Castellanos, Eds. (Oxford Univ. Press, New York, 2000), pp. 77-103.

2276 22 DECEMBER 2000 VOL 290 SCIENCE www.sciencemag.org

Copyright of Science is the property of American Association for the Advancement of Science and its content may not be copied without the publisher's express written permission except for the print or download capabilities of the retrieval software used for access. This content is intended solely for the use of the individual user. Copyright of Science is the property of American Association for the Advancement of Science and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.