is not simply involved in performance control but rather has a role in supervising control. Regarding a hierarchy in control for movement, SMA seems to range over Ng also in this respect. However, some authors think that the proof of this hierarchical concept is still lacking (cf. Wiesendanger & Wise 1992), although also these authors state that 'in the region of the medial-frontal cortex there is a gradual change from caudal to rostral with the posterior portion being more 'motor', the anterior more 'complex'. We think this is in good agreement with our thinking and would mean that our interpretation of SMA function would mainly belong to the anterior SMA (or pre-SMA of Luppino, Matelli, Camarda & Rizzolatti, 1993). In Wiesendanger's intra cortical motor stimulation experiments, the anterior SMA was 'clearly less excitable, although more sluggish responses may be obtained, dependent on whether the animal is about to move.' Also the cortico-cortical connections were more widespread in the anterior SMA ( Wiesendanger, Rouiller, Kazennikov & Perrig 1996, Hummelsheim, Bianchetti, Wiesendanger & Wiesendanger 1988). Luppino, Matelli, Camarda & Rizzolatti ( 1993) in a detailed hodological analysis also described fundamental differences in connectivity between the two SMA portions, the pre-SMA receiving rich input from areas 46, F5, anterior cingulate, and only modest input from PG, PFG and disgranular insula. However, the SMA proper is richly linked with the MI, posterior cingulate and receives modest inputs from the PM, and postrolandic areas PC and PE and the granular insula. Again it would be rather the pre-SMA that has the properties which we attribute to the SMA on the basis of our experiments. However, it appears that the cingulate motor area (CMA) also is activated prior to voluntary movement.
SMA's role in bimanual and other interlimb coordination is well documented by our experiments, and there are little objections against the view that the SMA is important for interlimb coordination. Participation of the SMA in motor learning is also well documented. The SMA is not the only area engaged in motor learning, the lateral premotor cortex is as well (electrodes F3 and F4 in EEG, MFG [middle frontal gyrus] of both sides in SPECT) as are basal ganglia and cerebellum.
Readiness for action expresses itself in the brain by the Bereitschaftspotential, whose early component BP1 appears to be largely generated by the SMA. It still remains difficult to define the function of the SMA. Neither lesion studies in man and monkey, nor DC-potentials and fields in EEG and MEG in man, nor Emission-CT studies in man nor single unit recordings in monkeys nor direct cortical recordings in epileptic patients have yielded really unequivocal results. The reason is that the function of the SMA is so extremely task-dependent. Already Foerster ( 1936) has reported on the enormous complexity of this area. On the other hand, our hypothesis that SMA - as a prime function in voluntary self-initiated action - has to do with intention (for review see Kornhuber, Deecke, Lang, Lang & Kornhuber ( 1989) has never been falsified. On the contrary, piece-by-piece, direct cortical recordings by several groups ( Neshige, Lüders & Shibasaki 1988; Ikeda, Lüders, Burgess & Shibasaki 1993, and Rektor, Feve, Buser, Bathien & Lamarche 1994) have surely confirmed our notion. This is now established. While in the beginning, pre-movement activity in the SMA had been denied on the basis of intracranial recordings, more recent studies have confirmed our hypothesis and have ascertained that a BP can, in fact, be recorded in the SMA ( Ikeda, Lüders, Burgess & Shibasaki 1993; Rektor, Feve, Buser, Bathien & Lamarche 1994).
Does SMA activity lead in time over MI activity? On the basis of our data this is clearly so: In all our BP studies since the beginning ( Kornhuber & Deecke 1964, 1965), BP latency measurements clearly have shown that it is always at the midline (vertex Cz or FCz, at times Pz)