Cortical networks for working memory and executive functions sustain the conscious resting state in man
Introduction
A large part of our daily mental activities are internalized, id est performed without external input or motor output, and not goal directed. During this particular state of consciousness, that is not to be confounded with arousal or perceptual consciousness, one is monitoring both somesthesic and vegetative information, such as sensations and body position, and experiencing association of free thoughts that deal with the recollection of past experiences, inner speech, mental images, emotions, planning of future activities, etc. This mental state has been referred to as a Random Episodic Silent Thinking (REST) state by previous authors [3], thereby emphasizing the unconstrained nature of this kind of thoughts, as opposed to a more focused and constrained sort of episodic memory activity when one is driven to search into his past history.
Despite the fact that assessing the exact mental content of a subject during REST is by essence difficult and must rely on introspection, which clearly constitutes a scientific limitation, there is a considerable interest in the study of the neural bases of the conscious resting state.
First, from a strict cognitive neuroscience point of view, it is important to try to establish whether or not a network of brain areas is specifically engaged during this mental state brain. If so, this would provide a strong argument favoring the “computational” nature of brain activity during REST [7] and, depending on the locations of these areas, insights regarding the major cognitive processes involved.
The answer to this question is also of primary importance for cognitive neuroimaging experiments, because conscious rest has been (and is still) widely used in the cognitive neuroimaging community as a reference state to which better controlled cognitive processes are contrasted. This is explicitly the case for positron emission tomography (PET) or functional magnetic resonance imaging (FMRI) experiments, in which local neural activity during conscious rest serves as a baseline for assessing hemodynamic variations during cognitive process execution. To some extent, it is also implicitly the case for event-related electroencephalography or magnetoencephalography experiments, because electrical/magnetic activity is generally compared to a baseline activity measured over the several tens/hundreds of milliseconds that precede stimulus presentation when the subject is in a state that resembles to conscious rest.
Although it is an ill-defined mental state, there are some good reasons for using conscious rest as a reference state in cognitive neuroimaging. First, it is applicable in all imaging experiments, as opposed to high level cognitive control tasks that have been used to test the involvement of a specific cognitive module, with the potential drawback of masking joint activations in both the reference and the task of interest [24]. As a matter of fact, differences in the reference conditions have been put forward in order to try to solve controversies regarding the putative brain network involved in identical tasks [38]. As such, the conscious resting state can serve as a common reference both within and between laboratory experiments. Second, the conscious resting state does not rely on a given sensory modality for stimulus delivery or on a specific kind of behavioral output. Third, its variability both within and across subjects has been shown to be of the same magnitude than any other cognitive task [14].
Studies on the cortical anatomy of the resting state have been first relying on the observation of the regional pattern of the glucose metabolic rate (rCMRGlu) or regional cerebral blood flow (rCBF) 14, 42, 55. These studies revealed brain areas of higher or lower metabolic/hemodynamic activity, provided figures of their local variability, but did not allow to uncover a specific network of brain areas involved during the resting state. A recent PET experiment compared rCMRGlu maps in healthy volunteers at REST to that of patients in a vegetative state [40], showing a widely distributed network of brain areas having higher neural activity during REST. A recent FMRI experiment has focused on the issue of the neural bases of the resting state, contrasting REST to auditory perceptual or semantic tasks [7]. A left hemisphere parieto-frontal network was found of equal activity during REST and the semantic task, but of reduced activity during the perceptual task, leading the authors to propose that mental activity during REST was of “conceptual processing” nature.
However, this claim remains to be formally demonstrated because it relied on simple contrast analysis (REST vs. tone, REST vs. semantic, etc.), concerned task involving the sole auditory modality, with possible unwanted interference of attentional process due to scanner noise, and was based on a somewhat limited sample size, which raises a sensitivity issue. One way of alleviating these limits could be through a meta-analysis of activation protocols in which a REST condition was used. The idea of using a meta-analysis to uncover process common to different tasks has been proposed and previously exploited by others 63, 64, 65, 66. In their approach, Shulman et al. reanalyzed a set of 9 PET experiments contrasting passive vs. active visual tasks, using activation map averaging in order to uncover processes that generalize across tasks. This approach has the potential drawback that it could reveal activation that would not necessarily be common to all protocols but rather driven by the strongest ones, with possibly reduced activity in some others. Actually, conjunction analysis appears as the optimal statistical approach of choice when trying to identify joint activations across protocols [57].
The present study was designed to alleviate some of the concerns raised by previous studies dealing with the neural bases of the conscious resting state. It combines the intrinsic power of meta-analysis, the specificity of conjunction analysis, and a variety of cognitive tasks to be contrasted to REST. Accordingly, a set of nine PET activation protocols performed in our laboratory over the past 3 years, all including REST as a control task, have been reanalyzed using conjunction analysis, searching for joint activation during REST as compared to a variety of cognitive tasks, including different stimulus presentation modalities, cognitive processes, and behavioral outputs.
Section snippets
Subjects
Sixty-three young healthy male volunteers participated to the study (22.0 ± 2.0 years, mean ± SD). All were right-handed as assessed by the Edinburgh questionnaire (88.1 ± 13.7; range, 50–100) and were free of brain abnormalities as assessed on their T1-weighted 3D MRI scan. All subjects gave their informed consent. The study was approved by the Basse-Normandie Ethic Committee.
Cognitive tasks
Each of the 63 subjects participated in one and only one of nine different activation protocols (see Table 1). Each
Behavioral observations
Among the 63 subjects that were included in the study, 41 properly answered the questionnaire dealing with their REST condition mental activity. Overall, a majority of subjects (56%) reported that mental events were partly dealing with autobiographic reminiscences, either recent or ancient, consisting of familiar faces, scenes, dialogs, stories, melodies, etc.
Inner speech and mental images were the most frequently reported types of mental events during REST. Ranking the occurrence of such
Discussion
The present meta-analysis uncovered a parieto-frontal network of brain areas that was more active during the resting condition than during a variety of cognitive tasks, independent of the modality in which the stimuli were delivered, of the type of cognitive activity, and of behavioral output. In the following we will, (1) argue that this network sustains processes active during REST rather than reflects deactivation processes common to all tasks, (2) discuss the similarities between this
Conclusion
The conscious resting state in humans is sustained by a large scale network of heteromodal associative parietal and frontal cortical areas, that can be further hierarchically organized in an episodic working memory parieto-frontal network, driven in part by emotions, working under the supervision of an executive prefrontal network. The bilateral angular gyrus and the anterior precuneus, parts of the parieto-frontal sub-network, appears to be specifically involved during REST and may reflect the
Acknowledgements
The authors are indebted to the Cyceron Cyclotron (P. Lochon, O. Tirel) and PET camera staff (V. Beaudoin, G. Perchey) for their invaluable help in acquiring the PET activation studies. The authors would also like to thank A. Mazard for her help in some of data acquisition. Parts of this work have been supported by grants from the GIS Sciences de la Cognition.
This work has been presented in part at the 5th International Conference on Functional Mapping of the Human Brain (Dusseldorf, Germany,
References (72)
- et al.
PET studies of memoryNovel and practised free recall of complex narratives
Neuroimage
(1995) - et al.
Regional brain activity during tasks devoted to the central executive of working memory
Cogn. Brain Res.
(1999) - et al.
The mind’eye—Precuneus activation in memory-related imagery
Neuroimage
(1995) - et al.
The trouble with cognitive subtraction
Neuroimage
(1996) Linkage at the top
Neuron
(1998)- et al.
The large-scale neural network for spatial attention displays multifunctional overlap but differential asymmetry
Neuroimage
(1999) - et al.
Impaired effective cortical connectivity in vegetative statePreliminary investigations using PET
Neuroimage
(1999) The retrosplenial cortex and emotionNew insights from functional neuroimaging of the human brain
TINS
(1999)- et al.
Reopening the mental imagery debateLessons from functional anatomy
Neuroimage
(1998) - et al.
A common language network for comprehension and productionA contribution to the definition of language epicentres with PET
Neuroimage
(2000)