Neuropsvlhologia. Vol. 28, No. 7, pp. 665 671, 1990. 0028 3932/90$3.00+ 0.00 Printed in Great Britain. ~ 1990PergamonPressplc CROSS-MODAL MATCHING BY AMNESIC SUBJECTS C. SHAW, R. W. KENTRIDGEand J. P. AGGLETON* Department of Psychology, University of Durham, South Road, Durham, DH1 3LE, U.K. (Received 8 June 1989: accepted 20 December 1989) Abstract The present study examined the performance of two groups of amnesic subjects on a cross- modal identification task. It was found that subjects with Korsakoff's disease did not differ from alcoholic controls on their ability to match the tactile feel of an arc with the visual appearance of the full circle from which the arc was taken. The postencephalitic subjects were, however, impaired on this same task. All groups performed normally on two intramodal control tasks. The postencephalitic group, like the Korsakoff subjects, were also poor at identifying common objects from tactile cues. The results are consistent with the notion that limbic regions in the temporal lobe are important for cross-modal associations. INTRODUCTION INTERESTin the ability of amnesic patients to perform cross-modal associations arises from evidence that the amygdala may be necessary for this function. It has been shown that extensive lesions of the amygdala can produce a very severe tactile-to-visual cross-modal recognition impairment in monkeys [17]. This severe deficit contrasted with the normal or near-normal performance of the same animals on intramodal versions of the task, i.e. visual or tactile delayed nonmatching-to-sample [! 7]. No evidence was found that hippocampec- tomy affected the cross-modal version of the task [17]. These findings provide important evidence that the primate amygdala might be necessary for integrating the different sensory components of a complex, multimodal stimulus; a role consistent with its extensive cortical connections [I, 4, 5, 23]. In an attempt to replicate this result in humans a recent study compared cross-modal recognition before and after bilateral amygdalotomy [14]. This study, which was deliberately modelled on the procedure used with monkeys, failed to find any evidence of an impairment in either tactile-to-visual or visual-to-tactile associations [14]. One problem with this study was, however, the failure to eliminate verbal cues. Although the stimuli were nonsense paperboard forms differing in size and shape, it is unclear whether this was sufficient to exclude verbal mediation by the subject. In addition, the presence of ceiling effects may have obscured any mild deficit following the surgery. For these reasons it was decided to reinvestigate cross-model matching using a slightly different procedure. The present study examined the performance of two groups of amnesic subjects, a postencephalitic group and an alcoholic Korsakoff group, on the "Arc-Circle test" devised by NEBES[18, 19]. In this task the subject blindly explores with his or her fingers an arc taken from one of three sizes of complete circle lying in free view before the subject. The subject then *To whom correspondence should be addressed. 665