Neuropsychologia 40 (2002) 1300–1312 Interhemispheric neural summation in the split brain with symmetrical and asymmetrical displays Matthew Roser ∗ , Michael C. Corballis Department of Psychology, Research Centre for Cognitive Neuroscience, University of Auckland, Private Bag 92019, Auckland, New Zealand Received 20 March 2001; received in revised form 6 November 2001; accepted 6 November 2001 Abstract The present study, investigates interhemispheric integration in the split brain. Four split-brained, two acallosal and 14 normal subjects carried out a simple reaction time task in which they responded to stimuli presented either singly in the left visual field, singly in the right visual field, or simultaneously in both visual fields. Stimuli were white against a black background and bilateral stimuli were either symmetrical or asymmetrical around the central vertical meridian. For unilateral stimuli, the difference in response time (RT) between crossed and uncrossed hand-field combinations (crossed–uncrossed difference, or CUD) measured interhemispheric transfer time. RTs to bilateral and unilateral stimulus displays were compared to provide a measure of redundancy gain (RG). Normal subjects exhibited small CUDs and RGs. Split-brained and acallosal subjects were found to have much longer CUDs, and to show enhanced RGs which could not be explained by a probability (race) model, implying subcortical neural summation. This summation did not depend on the preservation of symmetry, suggesting that it may not occur at the retinotopically organized superior colliculus, but at another site such as the pons or reticular formation. © 2002 Elsevier Science Ltd. All rights reserved. Keywords: Redundancy gain; Commissurotomy; Callosotomy; Callosal agenesis 1. Introduction A well established finding in studies of simple reaction time is that response times (RTs) to stimuli such as lumi- nance onsets are faster if there is more than one stimulus presented. This is known as the redundancy gain (RG) and can often be explained as the result of probability summa- tion [25,35]. If the two stimuli trigger independent response preparation processes and a response is made as soon as the first process is completed, then on average, responses will be made faster than when there is only one process in oper- ation, as long as the two RT distributions overlap. As RTs to redundant signals are determined by the first process to be completed, this is known as the “race model”. If however there is an enhancement of RT greater than that which can be explained by probability summation, then coactivation or neural summation is implied [7,36]. Coactivation mod- els involve activation from redundant processes combining towards reaching a single criterion for response initiation [25,30]. Thus, activity on otherwise independent channels is summed together to produce a speeded response [36]. ∗ Corresponding author. Tel.: +64-9-373-7599; fax: +64-9-373-7450. E-mail address: m.roser@auckland.ac.nz (M. Roser). The race model seems sufficient to explain RG data in most normal subjects [29,36], although there are excep- tions [28]. However split-brained and acallosal subjects often show RGs which exceed the predictions of the race model by a far greater extent than the small violations occasion- ally observed in normal subjects [7,18,33,36]. This is despite the fact that redundant stimuli, displayed one in each visual field, are presented separately to hemispheres which have been disconnected at the cortical level by commissurotomy or callosotomy, or by congenital failure of callosal fibers to develop, as in callosal agenesis. Indeed bilateral presentation seems to be a necessary condition for race-model violations in at least one split-brained subject, since there was no viola- tion when redundant stimuli were presented to a single hemi- sphere and the split-brained subject showed similar mean RT effects from redundant targets as normal subjects [36]. In ad- dition to this, crossed responses, or those made by the hand contralateral to the visual field in which the stimulus was dis- played, are usually slowed relative to uncrossed responses. In normals this crossed–uncrossed difference (CUD) is typ- ically in the range of 2–6 ms [2,32] and has been used as an index of callosal transmission times. In split-brained sub- jects it is much longer, commonly around 70 ms, although there is considerable variation [7,14,36]. The CUD is also prolonged in acallosal subjects for whom CUDs of around 0028-3932/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII:S0028-3932(01)00219-6