Probing the cortical network underlying the psychological refractory period: A combined EEG–fMRI study

Abstract

Human performance exhibits strong multi-tasking limitations in simple response time tasks. In the psychological refractory period (PRP) paradigm, where two tasks have to be performed in brief succession, central processing of the second task is delayed when the two tasks are performed at short time intervals. Here, we aimed to probe the cortical network underlying this postponement of central processing by simultaneously recording electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) data while 12 subjects performed two simple number-comparison tasks. Behavioral data showed a significant slowing of response times to the second target stimulus at short stimulus-onset asynchronies, together with significant correlations between response times to the first and second target stimulus, i.e., the hallmarks of the PRP effect. The analysis of EEG data showed a significant delay of the post-perceptual P3 component evoked by the second target, which was of similar magnitude as the effect on response times. fMRI data revealed an involvement of parietal and prefrontal regions in dual-task processing. The combined analysis of fMRI and EEG data—based on the trial-by-trial variability of the P3—revealed that BOLD signals in two bilateral regions in the inferior parietal lobe and precentral gyrus significantly covaried with P3 related activity. Our results show that combining neuroimaging methods of high spatial and temporal resolutions can help to identify cortical regions underlying the central bottleneck of information processing, and strengthen the conclusion that fronto-parietal cortical regions participate in a distributed “global neuronal workspace” system that underlies the generation of the P3 component and may be one of the key cerebral underpinnings of the PRP bottleneck.

Introduction

Despite the primate brain's massively distributed processing architecture (Felleman and Van Essen, 1991), reminiscent of the multiple-processor design of parallel computers (Nelson and Bower, 1990), human performance exhibits surprisingly strong limitations in multi-tasking. In one of the simplest multi-tasking experiments (Fig. 1), two target stimuli (T1 and T2, e.g., two tones) are presented in brief succession, and subjects' responses (R1, R2) to both targets are recorded. Under these simple dual-task conditions, response times to the second stimulus (RT2) show a significant increase when the stimulus-onset asynchrony (SOA) between the two tasks is shortened; response times to the first stimulus (RT1), however, remain largely unaffected by SOA. This classic and widely replicated finding has been dubbed the “psychological refractory period” (PRP), in analogy to post-stimulation refractory phenomena observed in nerves (Telford, 1931). The original hypothesis put forward to explain it was that the “central organizing times” of stimulus processing cannot overlap for two stimuli, and thus have to unfold strictly serially, one after the other (Welford, 1952). This notion of non-overlapping “organizing times”, or in other words, a serial processing stage that acts as a bottleneck of information processing, remains a central ingredient of modern theories of the PRP. The central bottleneck model (Pashler, 1994, Pashler, 1998), which emerged from numerous behavioral experiments, involves three stages of processing: a perceptual (P), a central (C), and a motor (M) stage. According to the model, P and M stages can occur in parallel, while the C stages of two tasks cannot overlap and have to be processed serially. Thus, at short SOAs, central processing for T2 is deferred, or passively queued, until central processing for T1 is completed, and RT2 is increased (Fig. 1).

Various behavioral experiments have associated the central processing stage to response selection, i.e., the mapping between sensory information and motor action (De Jong, 1993, Pashler and Johnston, 1989). Recently, it has been proposed that the C stage can be characterized as a decision-making process based on the noisy integration of evidence (Sigman and Dehaene, 2005). Alternative models argue against a structural bottleneck of stimulus processing, which invariably results in passive queuing of the second stimulus, and instead propose a strong influence of executive control in the strategical monitoring of the two tasks (Logan and Gordon, 2001, Meyer and Kieras, 1997a, Meyer and Kieras, 1997b), or the ability to share processing capacity between them (Navon and Miller, 2002, Tombu and Jolicoeur, 2003). According to shared capacity models, response selection can occur in parallel, but with limited processing resources differentially weighted for one task over the other, resulting in a lag between RT1 and RT2 typical for the PRP. Several dual-task studies have reported behavioral congruency effects that poses a major challenge for the bottleneck model, namely the dependence of RT1 on the response that is required for the second stimulus, referred to as backward crosstalk (as opposed to crosstalk from T1 on RT2). Backward crosstalk effects, which are observed when both tasks are similar [(Logan and Delheimer, 2001, Logan and Schulkind, 2000), but see (Miller, 2006)], are difficult to reconcile with the strictly serial bottleneck model, because they provide evidence that central processing for T2 may start before the C stage for T1 is complete.

To further describe and anatomically locate the cognitive processes underlying the PRP, a rich body of evidence, including data from studies using event-related potentials (ERPs), functional magnetic resonance imaging (fMRI), and recently computational modeling (Zylberberg et al., 2010), has been accumulated. Studies using fMRI have reported various frontal and parietal regions associated with the PRP (Marois and Ivanoff, 2005), but the results do not appear to converge and strongly depend on the statistical approaches used to test for dual-task-specific effects (Szameitat et al., 2011). Isolating PRP related activity by contrasting dual-task against single-task activity, or alternatively, short SOA against long SOA trials, has highlighted different sets of regions in the lateral frontal, medial frontal, premotor and parietal cerebral cortex. Recently, studies using time-resolved fMRI have reported delayed peaks of activation in the left posterior lateral prefrontal cortex associated with the PRP (Dux et al., 2006), and PRP related temporal variations of activity in the bilateral parietal and frontal regions, respectively (Sigman and Dehaene, 2008). However, the sensitivity and interpretation of these studies also suffers from the low temporal resolution of fMRI.

A number of ERP studies investigating the PRP effect have targeted the amplitude and latency of the P3 (or, P300) component, which is characterized by a positive deflection broadly distributed over the scalp, but with a focus over parietal electrodes (Picton, 1992, Sutton et al., 1965). Recently, it has been proposed that the P3, which has been linked to post-perceptual processes such as the context-updating of working memory (Coles et al., 1985, Donchin and Coles, 1988, Verleger et al., 2005), may be related to the access of a target stimulus to a global neuronal workspace associated with conscious report (Del Cul et al., 2007, Sergent et al., 2005). Based on the delay of the P3 evoked by the second target (T2-P3) some ERP studies have proposed an overlap between the cognitive processes mediating the PRP effect and P3-related processes (Dell'Acqua et al., 2005, Sigman and Dehaene, 2008), while the evidence from other studies, showing a large discrepancy between RT2 and T2-P3 latency modulations, suggests independent sources for PRP and P3 effects (Arnell et al., 2004, Luck, 1998). The latencies of earlier sensory ERP components, such as the P1 and N1, have been reported to remain stimulus-locked to both targets and show no postponement related to the PRP (Brisson and Jolicoeur, 2007, Sigman and Dehaene, 2008).

In the present study, our main aim was to probe the cortical network underlying the PRP by using a combination of simultaneously recorded high-temporal resolution EEG and high-spatial resolution fMRI responses. Due to recent advances in the combination of both neuroimaging methods (Herrmann and Debener, 2008, Laufs et al., 2008), fluctuations in EEG can be correlated to the simultaneously recorded fMRI data on a trial-by-trial basis, thus helping to identify the cerebral networks underlying dynamic changes in ERPs. Based on previous research (Dell'Acqua et al., 2005, Sigman and Dehaene, 2008) which showed a close link between the P3 component and the PRP, we hypothesized that the P3 would covary with the fMRI-BOLD signal in the dual-task situation of our PRP paradigm. By correlating the single-trial amplitudes of the P3 component with the single-trial blood-oxygen-level dependent (BOLD) fMRI signals, we were able to isolate a set of two bilateral homotopic regions in the precentral gyrus and inferior parietal lobe. The PRP paradigm used in our study consisted of two identical number-comparison tasks, thus allowing for an additional analysis of behavioral crosstalk effects (Logan and Delheimer, 2001, Logan and Schulkind, 2000).