The Machinery of Thought

adly on special tests he has devisedthat require subjects to remember theirearlier responses during the test. Healso cites PET studies of healthy humansthat find heightened activity inthe same region when subjects are performingthe tasks he uses. The findingis the same whether the tasks involvespatial cues or not. “The material doesnot seem to matter—the process is crucial,”Petrides says.Other researchers have found evidenceto support the notion that thehigher parts of the prefrontal cortex arekey for self-monitoring. In an experimentby Mark D’Esposito and his associatesat the University of Pennsylvania,volunteers performed either one orboth of two tasks that, separately, didnot require working memory. One taskrequired subjects to say which words ina list read aloud were the names of vegetables,whereas the other asked themto match a feature of a geometric figureseen in different orientations. FunctionalMRI showed that the dorsoventralprefrontal cortex became active onlywhen subjects attempted both tasks simultaneously.And in April at a meetingof the Cognitive Neuroscience Society,D’Esposito presented a meta-analysisof 25 different neuroimaging studies.The analysis supported Petrides’s generalnotion that tasks involving morecomputation involve higher regions ofthe prefrontal cortex. “It was amazingthat this came out,” D’Esposito says.D’Esposito’s analysis also confirmedearlier indications that humans, far morethan monkeys, represent different typesof information in different halves of thebrain. The meta-analysis did not, however,detect the upper/lower distinctionbetween spatial and object workingmemory that Goldman-Rakic espouses.Asymmetry of the human hemispheresis becoming apparent to other researchersas well. John D. E. Gabrieli and hiscolleagues at Stanford University haveused functional MRI to study the brainsof volunteers who were solving pictorialpuzzles such as those often found onintelligence tests. The puzzles were ofthree types. One group was trivial, requiringthe subject simply to select asymbol identical to a sample. A secondgroup was a little harder: people had toselect a figure with a combination offeatures that was absent from an arrayof sample figures. The third group containedmore taxing problems that requiredanalytical reasoning.Gabrieli’s study sheds some light onBRAIN REGIONS showmore activity as subjectshave to remember moreletters in a memory task.One such region (closest tothe viewer) is in the rightprefrontal cortex.the debate over the organizationof the prefrontalcortex. When volunteerspondered the intermediateclass of tasks, whichmost resembled the tasksother investigators haveused when studying workingmemory, the right sideof the higher part of theprefrontal cortex was prominently active.Moreover, the activity was in areasthat other researchers have found to beused when cues about spatial locationare stored. This result fits Goldman-Rakic’sidea that working memory for spatiallocation is stored in the higher regionsof the prefrontal cortex, becausethese intermediate tasks all demandedthat subjects visualize features in differentlocations.When the volunteers in Gabrieli’s experimentworked on the hard problems,however, the prefrontal cortices of thesubjects became even more active, onthe left as well as the right side. The addedcomplexity produced a pattern ofactivation like that Petrides has foundduring his tests of self-monitoring.Gabrieli’s data thus provide somesupport for Petrides’s theory of a higherexecutive level in the prefrontal cortex,as well as for Goldman-Rakic’s viewthat domain-specific regions exist there.“There are definitely domain-specificplaces,” Gabrieli says. “And there areJONATHAN D. COHENFurther Readingothers that rise above that.” In otherwords, both sides in the debate over domain-specificorganization of the prefrontalcortex may have a point. Yet inJune, Matthew F. S. Rushworth of theUniversity of Oxford and his colleaguesreported in the Journal of Neurosciencethat monkeys with large lesions in theirlower prefrontal cortex could still performwell on delayed-choice tests. Thefinding casts new doubt on the theorythat object working memory residesthere and seems to support Petrides.It may take years before the outstandingquestions about the prefrontal cortexare settled and the operation of thebrain’s executive functions are pinneddown to everyone’s satisfaction. “If youput a theory out, people will attack it,”Goldman-Rakic muses. “Everyone iscontributing.” And the modus operandiof the brain’s decision-making apparatusis slowly becoming visible. “Weare getting,” Goldman-Rakic observes,“to the point where we can understandthe cellular basis of cognition.”SACognitive Neuroscience. Special section in Science, Vol. 275, pages 1580–1610; March15, 1997.Down-Regulation of the D 1 and D 5 Dopamine Receptors in the Primate PrefrontalCortex by Chronic Treatment with Antipsychotic Drugs. Michael S. Lidow,John D. Elsworth and Patricia S. Goldman-Rakic in Journal of Pharmacology andExperimental Therapeutics, Vol. 281, No. 1, pages 597–603; April 1997.Temporal Dynamics of Brain Activation during a Working Memory Task. J. D.Cohen, W. M. Perlstein, T. S. Braver, L. E. Nystrom, D. C. Noll, J. Jonides and E. E. Smithin Nature, Vol. 386, pages 604–608; April 10, 1997.Transient and Sustained Activity in a Distributed Neural System for HumanWorking Memory. S. M. Courtney, L. G. Ungerleider, K. Keil and James V. Haxby, ibid,pages 608–611.Integration of What and Where in the Primate Prefrontal Cortex. S. ChenchalRao, Gregor Rainer and Earl K. Miller in Science, Vol. 276, pages 821–824; May 2, 1997.The Machinery of Thought Copyright 1997 Scientific American, Inc. Scientific American August 1997 83