The cortical generators of P3a and P3b: A LORETA study

Abstract

The P3 is probably the most well known component of the brain event-related potentials (ERPs). Using a three-tone oddball paradigm two different components can be identified: the P3b elicited by rare target stimuli and the P3a elicited by the presentation of rare non-target stimuli. Although the two components may partially overlap in time and space, they have a different scalp topography suggesting different neural generators.

The present study is aimed at defining the scalp topography of the two P3 components by means of reference-independent methods and identifying their electrical cortical generators by using the low-resolution electromagnetic tomography (LORETA).

ERPs were recorded during a three-tone oddball task in 32 healthy, right-handed university students. The scalp topography of the P3 components was assessed by means of the brain electrical microstates technique and their cortical sources were evaluated by LORETA.

P3a and P3b showed different scalp topography and cortical sources. The P3a electrical field had a more anterior distribution as compared to the P3b and its generators were localized in cingulate, frontal and right parietal areas. P3b sources included bilateral frontal, parietal, limbic, cingulate and temporo-occipital regions.

Differences in scalp topography and cortical sources suggest that the two components reflect different neural processes. Our findings on cortical generators are in line with the hypothesis that P3a reflects the automatic allocation of attention, while P3b is related to the effortful processing of task-relevant events.

Introduction

The so-called “P300” (or “P3”) is probably the most well known component of the brain event-related potentials (ERPs). Recently it has gained interest as an endophenotype for research into genetic predisposition to psychosis [6] and several studies have reported an association between P300 amplitude and the COMT val/met polymorphism [27], [93], [40], [7], [19].

According to several experimental findings, P300 is independent of the stimulus physical properties and reflects attention/memory processes related to changes in the neural representation of the environment induced by new sensory inputs (“context-updating” theory) [38], [17], [77], [65], [30], [39]. The “oddball” paradigm has frequently been used to record the P3: a subject is asked to discriminate between two different stimuli by responding (overtly or covertly) to the “target” stimulus, which usually occurs less frequently, while ignoring the standard (non-target) stimulus, which occurs more frequently. The stimuli are presented in a random series and vary on some dimensions (e.g., physical properties). The P3 potential is generally recorded across the scalp and has its maxima over the midline central and parietal leads, at about 300 ms after the onset of the rare target stimulus. A modified version of the paradigm includes an additional stimulus, a rare non-target, inserted into the sequence of target and frequent standard stimuli. The P3 obtained using this paradigm includes two different components [88], [14], [82]: the so-called P3b, elicited by target stimuli, with a maximum over the parietal regions, and the P3a, elicited by rare non-target stimuli, with a more anterior distribution than the P3b. The rare non-targets used to elicit the P3a include both “novel”, unrecognizable, and common, easily discriminable stimuli [12], [39], [11], [82], [87], [28]. Several studies have reported that the P3 components elicited by the two types of infrequent non-target stimuli have similar topographies, show an amplitude reduction with stimulus repetition (habituation) and are similarly influenced by the stimulus context [82], [87], [28]. They are believed to be the same component, generated by the same neural network [23], [28].

The scalp topography of the P3b is consistently determined in both the standard and three-tone oddball tasks, the topography of the P3a is more variable, depending on the number of standards preceding the rare non-target stimuli and their discriminability from the targets, as well as the number of repetitions of the rare non-target stimuli (novel stimuli are never-repeated patterns while other rare non-target stimuli used in several seminal studies have the same or lower repetition rates with respect to the targets) [13], [76], [29], [39], [82], [28]. All the studies, but one [76], have reported a more anterior distribution of the P3a with respect to the P3b [13], [29], [39], [82], [28]. The P3a is thought to represent the automatic attentional switch to deviant stimuli or distractors with respect to the ongoing task, while the P3b reflects the match between the incoming stimulus and the voluntarily maintained attentional trace of the task relevant stimulus [22], [39], [28].

The different scalp topography and influence of different experimental conditions suggest that the two components have different neural generators. However, traditional peak analyses cannot reliably separate components with a partial overlap in time and space and topographic inferences are severely affected by the reference electrodes used in ERP recording [16], [52], [87], [64]. Alternative, reference-independent methods of identification of ERP components might better determine the time frames of the underlying brain processes [48], [64].

Depth recordings have shown that the P3a generators are located in the anterior cingulate and fronto-parietal cortex and the P3b generators in superior temporal, posterior parietal, hippocampal, cingulate and frontal structures [2], [30], [32], [33]. However, intracranial investigations are not methodologically flawless and, mainly due to the extreme invasiveness of the technique itself, they are not suitable for investigations involving healthy volunteers or psychiatric patients.

In the last decades, new brain imaging techniques became available, which allow direct investigation of brain activity, but do not require invasive procedures. The functional magnetic resonance imaging (fMRI), achieving an acceptable compromise between spatial and temporal resolution [57], [25], has been used in several studies to investigate the neural basis of P300 [61], [62], [10], [18], [44], [89], [42], [36], [43], [4], [66], [56], [90], [5]. Most of them have confirmed the involvement of the frontal, parietal, temporal and cingulate areas in the genesis of this ERP component. However, the contribution of the medial temporal structures and hippocampus cannot be adequately assessed by echo-planar imaging (EPI) sequences currently used in most fMRI experiments [86] and, due to the partial temporal overlap of the two processes [85], [88], the identification of brain generators of P3a and P3b components might be hindered the relatively poor temporal resolution of fMRI techniques [3]. Electrophysiological techniques have a higher temporal resolution. However, due to the so-called “inverse problem”, precise inference on the brain generators of scalp recorded activities cannot be made. In recent years, different algorhythms have been proposed to solve the inverse problem and new methods for reconstructing the current source for a given scalp electrical distribution have been developed [64], [96]. These algorhythms can be divided in equivalent current dipole models [80], [81] and current distributed source models [34], [72], [94]. According to recent comparative studies, the dipole models are suitable only when a single source is expected [96]. Among the distributed source methods, the low-resolution brain electromagnetic tomography [(LORETA) [73] has been proved to present the smallest localization error, particularly when multiple sources and noise are present [74], [96]. This algorithm selects the smoothest spatial source distribution by minimizing the Laplacian of the weighted sources [72], under the main assumption that the activity of neighbor neurons is highly correlated. LORETA software limits the solution space to cortical gray matter and hippocampus, excluding subcortical sources, for which the spatial resolution of the method could be extremely poor. Several studies reported consistency between LORETA and neuroimaging studies: Strik et al. [92] used LORETA to study the electrical generators of the P300 produced during a cued continuous performance test and found frontal activation, in line with previous positron emission tomography and near infrared spectroscopy studies. Worrell et al. [95] used LORETA to identify the electrical sources of ictal EEG discharges and the results were consistent with well-defined symptomatic MRI lesions. Consistency between LORETA findings and MRI results in subjects with schizophrenia was also reported [71], [70]. Mulert et al. [66] used both LORETA and fMRI to investigate the time-course of the activations corresponding to the P300 component and found that LORETA findings were consistent with those provided by fMRI and were in line with previous results obtained by intracranial recordings.

According to these data, LORETA might significantly improve knowledge on neural processes underlying the P3a and P3b components.

In this study, we used a modified (three tones) auditory oddball paradigm to record P3a and P3b components of ERPs in a group of healthy subjects. Advanced reference-independent topographical analysis was used to identify the two P300 components. LORETA was used to characterize the cortical distribution of P3a and P3b electrical generators.