Relative Shift in Activity from Medial to Lateral FrontalCortex During Internally Versus Externally GuidedWord GenerationBruce Crosson, Joseph R. Sadek, Leeza Maron, Didem Go¨kc¸ay,Cecile M. Mohr, Edward J. Auerbach, Alan J. Freeman,Christiana M. Leonard, and Richard W. BriggsAbstract& Gol dberg (1985) hypothesized that as language outputchanges from internally to externally guided pr oduction,activity shifts from supplementary motor area (SMA) to lateralpremotor areas, including Broca’s a rea. To test this hypothesis,15 right-handed native English speakers performed three wordgeneration tasks varying in the amount of internal guidanceand a repetition task during functional magnetic resonanceimaging (fMRI). Volumes of significant activity for each taskversus a resting state were derived using voxel-by-voxelrepeated-measures t tests (p < .001) across subjects. Cha ngesin the size of activity volumes for left medial frontal regions(SMA and pre-SMA/BA 32) versus left lateral frontal regions(Broca’s area, inferior frontal sulcus) were assessed as internalguidance of word generation decreased and external guidanceincreased. Co mparing SMA to Broca’s area, Goldberg’shypothesis was not verified. However, pre-SMA/BA 32 activityvolumes decreased significantly and inferior frontal sulcusactivity volumes increased significantly as word generationtasks moved from internally to external ly guided. &How does the human brain initiate the expression ofthought i n language? A comprehensive understandingof the brain’s language systems necessitates an answerto this question. An appreciation of language initiationmechanisms could facilitate treatment development forlanguage initiation deficits in aphasia. For some time,cognitive neuroscientists have known t hat the medialfrontal cortex plays a role in language initia tion. How-ever, the nature of this involvement has yet to bedetermined.Medial frontal lesions cause akinetic mutism, inwhich speech is initiated only with significant externalprompting (Barris & Schuman, 1953; Nielson & Jacobs,1951). Regarding the absence of spontaneous lan-guage, Luria’s (1966) report of a medial frontal lesioncase indicated that thoughts were not present toexpress. This phenomenon suggests that the medialfrontal cortex plays a role in initiating the cognitiveaspects of spontaneous language. After evaluating em-pirical evidence on this subje ct, Picard and Strick(1996), Passingham (1993), and Goldberg (1985) con-cluded that the degree of involvem ent of medialfrontal structures, and which medial structures partici-pate, depends upon the nature of the language that isinitiated.Goldberg (1985) suggested that involvement of med-ial frontal cortex depends upon whether language istriggered by internal or external contingencies. He fo-cused on the divergent roles of the supplementarymotor area (SMA) and lateral premotor cortex. Goldbergspeculated that Broca’s area was prominent in the lateralpremotor cortex of humans (p. 578), especially whenconsidering language functions . Traditionally, Broca’sarea has been design ated as the posterio r portion ofthe inferior frontal gyrus, i.e., pars opercularis (Brod-mann’s area [BA] 44) and pars triangu laris (BA 45).Although recent literature indicates important functionaldivisions within medial BA 6 that might have influencedhis conclusions (see below), SMA was considered toconsist of the entire medial BA 6 at th e time Goldbergwrote his review. He surmised that SMA was prim arilyinvolved in internally generated language and actions,whereas lateral premotor cortex was involved in lan-guage or actions that are externally referenced.Passingham’s (1993) subsequ ent review on medialversus lateral premotor cortex arrived at a similar con-clusion. In reference to selection of movement, Passing-ham believed that lateral premotor cortex made greatercontributions when movements were driven by externalcues, and medial premotor cortex played a greater rolewhen no external cues were available (i.e., when move-ment was driven from internal model s). However, Pas-singham (1993) also concluded that neither internallyUniversity of Florida© 2001 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 13:2, pp. 272–283cued nor externally cued movements were the exclusivedomains of SMA and lat eral premotor cortex, respec-tively. Rather, it was the balance of activity within thetwo cortices that was important.Yet, the lit erature has not unambiguously supportedthis conclusion. In compari ng right-han d motor tasks,Deiber et al. (1991) found less activity in the left SMA forexternally than inter nally cued tasks. However, lessactivity in left prefrontal cortex (BAs 9 and 46) alsowas observed for externally than for internally guidedmovements. In addition, lateral premotor activity wasgreater for internally cued movement than for a fixed-movement control task, whereas this region did notdemonstrate significant activity changes for externallycued movement versus the sa me co ntrol task. Withrespect to language, Frith, Friston, Liddle, and Frack-owiak (1991) compared generation of words beginningwith the letter F (internally driven word production ) torepetition of words (externally driven word production).Word generation produced more activation of both themedial frontal (centered in BA 32) and lateral frontal(centered in BA 46) cortex. Similar but less extensivechanges occurred for an internally as opposed to anexternally guided finger movement task. Unfortunately,neither Deiber et al. (1991) nor F rith et al. (1991)explored whether changes for medial and lateral frontalcortex were similar in m agnitude. Although both areasshow decreases from internally to ext ernally drivenactivity, a difference in the relative proportions ofchange could indicate a shift in the balance of medialversus lateral frontal activity. Further, these studies werenot consistent in what elements of medial frontal (BA 32or medial BA 6) or lateral frontal (BAs 9 and 46 or lateralBA 6) cortex were involved in the activity changes.Subsequent to Goldberg’s review, Matsuzaka, Aizawa,and Tanji (1992) and others (e.g., Luppino, M atelli,Camarda, & Rizzolatti, 1993) i ndicated that medial BA6 can be divided into a posterior region mainly con-nected to lateral motor and premotor systems (SMAproper), and an anterior region primari ly connected tolateral frontal cortex (pre-SMA). In monkeys, areas with-in the cingulate sulcus (cingulate moto r