Intracranial ERPs in humans during a lateralized visual oddball task: I. Occipital and peri-Rolandic recordings

Abstract

Objectives: This study investigated the relative participation of each cerebral hemisphere during a lateralized task that can be performed by a single hemisphere. This first of two articles focuses on recordings from visual and motor cortices.

Methods: Intracranial event-related potentials (ERPs) were recorded from occipital and/or peri-Rolandic sites in 8 patients with intractable epilepsy while they performed a lateralized visual oddball task.

Results: As expected, lateralized visual (N150, P200, N250) and motor (N/P400, N/P550) ERP effects were found for occipital and peri-Rolandic recordings, respectively. These reflect an advantage for direct over indirect sensory/motor pathways. More surprisingly, some occipital recordings were paradoxically larger in amplitude for indirect than for direct visual stimulus lateralization, and other occipital sites were sensitive to motor response factors. Likewise, one peri-Rolandic site exhibited a slow wave component that was sensitive to visual sensory factors. There was also pervasive bilaterally-symmetric ERP activity as reflected by P3-like and slow wave-like components.

Conclusions: These findings argue against a hemispheric independence model of information processing. With the exception of initial stimulus input and final response output pathway effects, the processing in this simple task engages both hemispheres in a roughly symmetrical fashion, even though a single hemisphere may be adequate for task performance.

Introduction

More than 3 decades of behavioral laterality research has conclusively demonstrated cerebral hemispheric asymmetries in cognition, anatomy, and physiology. While the existence of hemispheric asymmetries is undisputed, there is an ongoing debate concerning the extent to which the left cerebral hemisphere (LH) and the right hemisphere (RH) function independently of each other. Independent hemispheric processing has been most clearly demonstrated in patients with complete forebrain commissurotomy (i.e. split-brain patients), in which the corpus callosum, and possibly other forebrain commissures, is surgically severed for the treatment of epilepsy. Yet, even in these patients there is a surprising amount of interhemispheric interaction via subcortical interhemispheric pathways (Trevarthen and Sperry, 1973, Sergent, 1983, Sergent, 1986, Zaidel, 1995), or by elaborate cross-cueing strategies (Gazzaniga and Hillyard, 1971, Corballis and Trudel, 1993), in which one hemisphere provides a signal that can be detected by the other hemisphere (e.g. by movements of the tongue, which is under bihemispheric control). In neurologically-normal individuals, it is even more difficult to assess the extent to which the two cerebral hemispheres operate independently, versus operating in some form of coupled interaction. In behavioral laterality studies, it is a common practice to equate the performance of the left or right visual field (VF), ear, or hand condition with the contralateral hemisphere. While this approach commonly produces findings that are consistent with clinical and neuroimaging evidence for a particular task-specific hemispheric specialization, there is often little consideration for the influence of interhemispheric interactions. For example, in divided VF studies a stimulus presented to the left visual field (LVF) is initially received by the RH. The question becomes, what happens to the information from that point on? A strong version of a hemispheric independence model proposes that the stimulus information is always processed by the directly-receiving hemisphere (Dimond, 1972). In this case, the corpus callosum may actively maintain hemispheric independence via interhemispheric shielding or inhibition (Liederman, 1986, Zaidel et al., 1990). In this view, there may be conditions during which the normal brain behaves more or less like a split brain (Zaidel et al., 1990). This does not presuppose that interhemispheric inhibition and interhemispheric transfer of information need be mutually exclusive. For example, lateralized visual event-related potential (ERP) findings reveal that visual information received by one hemisphere is rapidly transferred to the contralateral hemisphere, albeit with a corresponding temporal delay and signal (i.e. amplitude) diminution (Rugg et al., 1986, Halliday, 1993). Nevertheless, precedence may be given to the directly stimulated hemisphere, such as through initial allocation of attentional resources (Schweinberger et al., 1994), and processing in the indirectly stimulated hemisphere may be inhibited at later stages of information processing. Thus, ERP measures of ongoing brain activity compliment behavioral laterality approaches in providing a richer understanding of interhemispheric interactions that follow lateralized stimulus presentations.

A variety of brain imaging techniques (e.g. EEG, ERP, SPECT, PET and fMRI) have shown that both cerebral hemispheres become active during most cognitive tasks (Lassen et al., 1978, Gevins et al., 1981, Haxby et al., 1991, Roland, 1993). Relative hemispheric differences in cerebral activity have often been found and these frequently correspond to previously known clinical and/or behavioral evidence for hemispheric specialization. PET and fMRI have been particularly useful in providing precise localization of such activity, although these techniques have limited temporal resolution. Scalp ERP techniques, in contrast, have limited spatial resolution, but have the best temporal resolution of all of these functional imaging techniques, and can distinguish electrophysiological correlates of different stages of information processing with millisecond accuracy. In the present study, the ERP approach was combined with intracranial recordings from different brain regions of epileptic patients in order to assess hemispheric independence at different stages of information processing. In this manner, the activity of specific brain structures was monitored at precise points in time in response to a particular stimulus.

A simple lateralized visual discrimination task was used (distinguish a ‘+’ from an ‘o’), which is not associated with a particular hemispheric specialization. Consequently, both hemispheres should be independently capable of performing this type of discrimination. If both stimulus input and response output are lateralized to a single hemisphere (e.g. the RH for LVF input and left hand response), then a strong version of the hemispheric independence model predicts that only that hemisphere will be engaged in the full sequence of information processing required of the task (Fig. 1). It then follows that only the directly participating hemisphere should exhibit all of the corresponding ERP components. However, if the entire sequence of ERPs is present over both hemispheres during unilateral input and output, then this would argue against a strong version of the hemispheric independence model. However, bilaterally-symmetric activity at a given stage of processing is not necessarily evidence that both hemispheres contribute to the processing. Activation in one hemisphere may be accompanied by interhemispheric inhibition of the other hemisphere, and yet there may be corresponding ERP components evoked by both hemispheres since inhibition is itself a physiologically active process. However, ERPs corresponding to the subsequent stages of processing should differ between the hemispheres, being present in the active hemisphere, and absent or significantly reduced in amplitude in the inhibited hemisphere. Furthermore, ERPs generated by IPSPs should in most cases have the opposite polarity from those generated by EPSPs, so interhemispheric inhibition or shielding should also be detectable in the component on which it acts. Thus, the ERP approach together with lateralized behavioral paradigms can be used to assess the relative engagement of the two hemispheres throughout the information processing sequence.

Findings from scalp recordings during unilateral stimulus presentations generally reveal bilateral ERP components, although these are not always symmetric over the two hemispheres. For early sensory ERP components, hemispheric ERP amplitude tend to be associated with sensory pathway effects (i.e. direct versus indirect projections). As noted previously, amplitudes for visual ERP components tend to be larger for direct than for indirect VF-hemisphere combinations, as there are costs for the indirect condition that are associated with interhemispheric transfer of information. This interpretation is supported by the absence of early sensory ERP components over the indirect hemisphere in split brain patients and in individuals born without a corpus callosum (i.e. callosal agenesis) (Rugg et al., 1985, Mangun et al., 1991). Attentional factors appear to modulate this anatomical pathway effect, since the advantage for the direct condition is reduced when participants are less engaged in the task (Rugg et al., 1986). Likewise, motor components tend to have larger amplitudes over the hemisphere that is contralateral to the responding hand. As these studies are based on scalp recordings, their ability to localize activity to specific brain structures or regions is limited, particularly for deep generators. Intracranial (i.e. depth) recordings from individuals with intact corpus callosums should further our understanding of the relative contributions of each hemisphere to the generation of ERP components.

In this first of two articles, we examined visual and motor ERP components that are evoked by a lateralized visual oddball task. Our interest is in characterizing ERP activity associated with the initial input and final output stages for our task. Consequently, this first article focuses on direct recordings from the occipital lobe and peri-Rolandic regions of epilepsy patients. The companion article, however, is primarily concerned with ‘cognitive’ ERP components recorded from other cortical and subcortical regions. Taken together, all cortical lobes of both hemispheres were sampled, and recordings were made directly from deep (e.g. hippocampal formation, parahippocampal gyrus, cingulate gyrus) as well as from superficial (e.g. cortex) cerebral regions. In this manner, there was precise localization of activity, at least for the regions sampled, together with excellent temporal resolution. By lateralizing stimulus input and response output to either the hemisphere containing a particular recording electrode (i.e. direct condition), or to the opposite hemisphere (i.e. indirect condition), we were able to investigate hemispheric and pathway effects.