ReviewHow is our self related to midline regions and the default-mode network?
Highlights
►Neuronal specificity of self vs. familiarity and non-self. ►Relationship between self and neural activity in the cortical midline structures. ►Relation between self and resting state activity in the default-mode network.
Introduction
The problem of the self has been one of the most pertinent problems in the history of philosophy and has now entered also psychology and neuroscience (Brewer and Weber, 1994, Markus and Kunda, 1986, Metzinger and Gallese, 2003, Northoff and Bermpohl, 2004) and more recently also in psychology (Rogers et al., 1977) and neuroscience (Gillihan and Farah, 2005, Legrand and Ruby, 2009, Northoff and Bermpohl, 2004, Northoff et al., 2006). What is the problem of the self? The problem of the self has both conceptual and empirical dimensions. Conceptually, different concepts of the self like process- and entity-based (Northoff et al., 2006) as well as sensorimotor- and cognitive-based (Legrand and Ruby, 2009) concepts of self can be distinguished. While empirically, i.e., neuronally, the problem of the self consists in the neuronal mechanisms including the kind of regions and networks as for instance cortical midline structures have often been highlighted to be specific for the self (Gillihan and Farah, 2005, Legrand and Ruby, 2009, Northoff and Bermpohl, 2004).
Although numerous studies have indicated that cortical midline structures were involved in the self-processing (Kelley et al., 2002, Mitchell et al., 2005, Northoff and Bermpohl, 2004, Northoff et al., 2006, Platek et al., 2006, Uddin et al., 2007, Yaoi et al., 2009, Zhu et al., 2007), the consideration of confounding factors such as familiarity (Gillihan and Farah, 2005) and tasks effect (Legrand and Ruby, 2009) in studies about self-processing means that the neural basis of the self is still unclear. Since studies have also indicated an apparent overlap between the self and default-mode network (DMN), the relationship between the self and default-mode network (or resting state) draws more attention recently.
The aim of the present study was twofold. First, we aimed to detect the brain regions involved in self-specificity, compared with personal familiar people and strangers. Our hypothesis was that self-specific stimuli recruit neural activity in cortical midline structures and predominantly in anterior ones. Second, we aimed to compare the regions implicated in self-specificity with the ones showing high resting state in the DMN. Based on previous studies (D'Argembeau et al., 2005, Schneider et al., 2008, Whitfield-Gabrieli et al., in press), we hypothesized that there would be strong regional overlap between self-specificity and DMN resting state activity in especially the anterior cortical midline structures. In order to pursue these aims, we performed a coordinate-based meta-analysis (Multilevel Kernel Density Analysis (MKDA) (Wager et al., 2009)) to detect the brain regions involved in self-specificity, compared with personal familiar people and strangers, and to detect the relationship between self-specificity and resting state in the DMN. This MKDA is a standard method which has often been successfully used in various types of meta-analysis of imaging studies (Fan et al., 2011, Salimi-Khorshidi et al., 2009, Wager et al., 2009, Wang et al., 2010).
The self-reference effect was demonstrated in a key behavioral study in which words related to the self were shown to be better remembered than other non-self-related words (Rogers et al., 1977). Comparing self- vs. non-self-specific stimuli, brain imaging studies observed neural activity changes in various medial cortical regions like the perigenual anterior cingulate cortex (PACC), medial prefrontal cortex (MPFC) and the posterior cingulate cortex (PCC) (Kelley et al., 2002, Mitchell et al., 2005, Northoff and Bermpohl, 2004, Northoff et al., 2006, Platek et al., 2006, Uddin et al., 2007, Yaoi et al., 2009, Zhu et al., 2007). This has led to the assumption that self-related processing may be specifically mediated by cortical midline structures (CMS) (Northoff et al., 2006). Furthermore, many studies have demonstrated a predominant involvement of the anterior CMS (e.g., ACC) in the processing of self-specific stimuli (D'Argembeau et al., 2005, D'Argembeau et al., 2007, Feinberg et al., 2010, Feinberg, 2009, Gusnard and Raichle, 2001, Han et al., 2009, Modinos et al., 2009, Ochsner et al., 2005, Zhu et al., 2007). Most of the previous brain-imaging results were obtained by comparing activations related to the presentations of self-specific stimuli to those of familiar or unfamiliar-non-self (i.e., other) stimuli. In the present study, the concept of familiarity refers to the use of stimuli related to people that the participants personally knew, such as the voices or faces of family, friends, or colleagues. While stimuli related to widely-known/famous/unfamiliar people may activate semantic memories, those related to personally familiar people may involve the triggering of more autobiographical memories and/or emotions (Gillihan and Farah, 2005). Based on this, the condition termed ‘other’ in the present study includes those stimuli related to the former group (i.e., non-self–non-familiar).
In contrast to these studies, recent review papers of the self (Gillihan and Farah, 2005, Legrand and Ruby, 2009) point out that processes other than self-related processing may account for neural activity changes in cortical midline structures during the presentation of self-specific stimuli. Processes like familiarity (Gillihan and Farah, 2005) or task-related requirements like judgment or some general evaluation (Legrand and Ruby, 2009) that are supposedly implicated in self-specific stimuli are instead suggested to underlie neural activity changes in CMS. However, these hypotheses (i.e. self as familiarity or judgment/general evaluation) remain to be tested empirically.
In the cortical midline structures, neural processing of self-specific stimuli has also been associated with regions implicated in the default-mode network (DMN) (Buckner et al., 2008, Buckner and Vincent, 2007, Raichle et al., 2001). Compared with resting state activity, cognitive tasks typically induce deactivation in the DMN. This brain network is believed to reflect one default mode of brain function (Raichle et al., 2001). The DMN includes cortical midline structures (e.g., ACC and PCC) and lateral cortices (e.g., TPJ) that are functionally connected with each other during resting state conditions (Fox et al., 2005). Due to the overlap in activity between regions that are suggested to be involved in self-relatedness processing and DMN regions (D'Argembeau et al., 2005, Schneider et al., 2008), some even speak of a so-called “default self” arguing that the self may be more or less identical with the resting state activity observed in DMN regions (Beer, 2007, Boly et al., 2008, Christoff et al., 2003, David et al., 2007, Golland et al., 2007, Gusnard et al., 2001, Wicker et al., 2003).
Previous studies showed that the resting state activity in the DMN could affect the subsequent stimulus-induced activity in corresponding sensory cortices, so called “rest–stimulus interaction” (Northoff et al., 2010). Regarding the role of the DMN in rest–stimulus interactions and the overlap seen with self-specific processing, self-specific stimuli could be hypothesized to induce a special type of rest–stimulus interaction when generating our sense of the self (compared to non-self-related stimuli). However the special form of rest–stimulus interaction between self-specific stimuli and resting state activity in the DMN remains to be investigated.
The general aim of our study was to investigate the relationship between brain activity related to the processing of self-specific, personally familiar, and other (non-self and non-familiar) stimuli. This also included the investigation of general non-specific task-related requirements as well as their relationship to neural networks showing high resting state activity, i.e. the DMN. Our specific aims were thus twofold. The first aim consisted in detecting the possible regional differences and overlap between self-specific, personally familiar and other stimuli (non-self-specific and non-personal-familiar) while at the same time accounting for unspecific task- and stimulus-related effects. Following the results from previous studies, we hypothesized that anterior cortical midline structures may be involved in self-related processing when compared to the regions recruited during the perception of personally familiar and other stimuli. The second aim consisted in detecting the relationship between self-specificity and resting state in DMN. Based on the previous results, we hypothesized an overlap in neural activity in anterior midline regions like the PACC between stimulus-induced activity during self-specificity and resting state activity in the DMN. We hypothesized this to be special for self-specific stimuli as distinguished from familiar and other (e.g., non-self and non-familiar) stimuli.
In order to investigate our aims, we conducted a quantitative meta-analysis (Multilevel Kernel Density Analysis (MKDA) (Wager et al., 2009)) based on the coordinates of the peak voxels reported in the previous studies. The present study focused on the recent brain imaging studies using self-specific, personally familiar, and other stimuli and those on resting state activity in the DMN. Our meta-analysis focused first on the regions implicated in self-specific, familiar and other stimuli, along with DMN activity, followed by direct statistical comparison between the four conditions. In order to exclude unspecific task- and stimulus-related effects, we also controlled for task- and stimulus-dependent effects independent of the four conditions. Two kinds of tasks were investigated in the present study: one involving evaluation of whether the stimuli were self- or non-self-specific; the other pertaining to the strict recognition of stimuli related to faces, names, bodies and/or voices. Two kinds of stimuli were investigated, word stimuli (e.g. trait adjectives), and face stimuli (e.g. one's own face, or other familiar or unfamiliar faces or). Finally, as a first step towards investigating the potential interaction between stimulus-dependent conditions (self-specificity, familiar and other) and resting state activity in the DMN, we plotted the regional activation clusters during the self-specific, familiar and other stimuli against those from the resting state activity.
Section snippets
Operationalization of the concept of self
Gillihan and Farah (2005) define the self on the concept of self-knowledge, dividing the self into two branches: physical and psychological. The concept of the physical self reflects the knowledge of the body, including its separate parts (e.g. face, arms), as well as the body as a whole. The concept of the psychological self reflects the knowledge ingrained in episodic memories (past experiences), semantic memories (traits about oneself), and the first-person perspective experience. In sum,
Regional characterization of the four conditions: self, familiarity, other, DMN
In the first step, we searched for those regions implicated in each of the four conditions of interest. During the self condition, the meta-analysis yielded activated clusters in several midline regions (including the PACC, MPFC and PCC), as well as in regions such as the left anterior insula and right inferior frontal gyrus (IFG) (see Fig. 1A and Supplementary Table 2). To rule out a possible bias in the selection and grouping of studies on the self, we also conducted the same analysis with
Discussion
Our first main finding showed that the self condition recruited the PACC when compared to the familiarity and other condition. Concerning the other midline regions, there was either regional overlap of activations related to the self and familiarity conditions, as in the MPFC, or with both the familiarity and other condition, as in the PCC. This, in part, confirms our first hypothesis that the PACC, as an anterior midline region, is important for the self. Our second main finding concerns the
Conclusions
We here report findings from a meta-analytic study on imaging studies of self, familiarity, other and resting state in DMN that allowed the direct comparison of the four conditions with each other without some of the methodological limitations inherent in single studies. Our results contribute to the current knowledge of the self in two ways. First, we here address for the first time the relationship between self and familiarity in a meta-analytic way. This is of major importance given that
Acknowledgments
The work on this paper was supported by grants from the CRC, the CIHR, and the CIHR-EJLB to GN. We also give our thanks to Dr. Dave Hayes and Niall Duncan for their useful input.
References (85)
The default self: feeling good or being right?
Trends Cogn. Sci.
(2007)- et al.
Unrest at rest: default activity and spontaneous network correlations
Neuroimage
(2007) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages
Comput. Biomed. Res.
(1996)Human feelings: why are some more aware than others?
Trends Cogn. Sci.
(2004)- et al.
Self-referential reflective activity and its relationship with rest: a PET study
Neuroimage
(2005) - et al.
The extrastriate cortex distinguishes between the consequences of one's own and others' behavior
Neuroimage
(2007) - et al.
Is there a core neural network in empathy? An fMRI based quantitative meta-analysis
Neurosci. Biobehav. Rev.
(2011) - et al.
The neural basis of mentalizing
Neuron
(2006) Neural activity that predicts subsequent memory and forgetting: a meta-analysis of 74 fMRI studies
Neuroimage
(2011)- et al.
Functional grouping and cortical–subcortical interactions in emotion: a meta-analysis of neuroimaging studies
Neuroimage
(2008)