Neurons in different brain regions may exhibit rhythmic firing patterns. This is called neural oscillation, the frequency of which can be recorded by an electroencephalogram (EEG). When a group of neurons fire together with the same oscillation pattern, they are in synchrony. Neural synchrony is considered to be a central mechanism of many cognitive functions. In the case of conscious perception, multifarious types of visual information are processed in different brain regions, which need to be combined in order to produce coherent percepts. Many researchers suggest that transient synchronization in the visual system provides such a binding mechanism (Engel & Singer 2001; Singer 2004; Singer 2007; Koch 2004). In addition to vision, synchronization in the beta and gamma ranges is also found in the olfactory, auditory, and somatosensory systems, as well as in other brain areas that influence (or are influenced by) perception, such as the pre-frontal cortex, the motor cortex, and the hippocampus (Singer 2007).
However, if this is all there is to neural synchrony, it would not explain the sense of self-as-subject at all. What we are looking for is not the mechanism that explains what I consciously perceive, but the mechanism that produces the sense that I, rather than someone else, am the subject of these perceptions.[30] Thus, information integration by neural synchrony may explain the content of consciousness without explaining the sense of experiential ownership, i.e., it explains what one experiences rather than who the subject of that experience is. In the following I consider three recent developments that connect neural synchrony more closely with self-consciousness.
(1) Uhlhaas et al. (2009) recently suggested that there are high correlations between disorders of self-consciousness and abnormalities in neural synchrony. Symptoms of schizophrenia, epilepsy, autism, Alzheimer’s disease, and Parkinson’s disease are related to dysfunctions of synchronization. For example, correlations have been suggested between reduced or abnormal alpha- or gamma-band oscillations, on the one hand, and impaired visual binding, auditory hallucination in schizophrenia, and impaired linguistic and auditory performance in autism, on the other. The problem is that the sense of experiential ownership is not itself targeted in these studies. Researchers measured how abnormal neural synchrony relates to impaired cognitive performance, tather than to who the subject of the experience is.
(2) Lou et al. (2010) used transcranial magnetic stimulation (TMS) to show that a medial paralimbic network is crucial for minimal self-consciousness.[31] This network may “bind conscious experiences with different degrees of self-reference through synchrony of high frequency oscillations” (2010, p. 185). They tested three conditions that represent different degrees of self-reference: maximal (“Self”), intermediate (“Franz”), and minimal (“Syl”). In each condition a set of adjectives were sequentially presented on a screen.[32] In the “Self” condition, the subject’s task was to make personal judgments concerning how well each adjective fitted him or herself. However, none of these conditions are about the sense of experiential ownership. Whether it was “I” who looked at the screen and made the judgments was not in question. Hence, the sense of self-as-subject was not measured by the reported patterns of synchronization.
(3) Kanayama et al. (2009) used EEG to investigate the rubber hand illusion (RHI), and found high correlation between the visual-tactile integration process and gamma-band synchrony in the parietal cortex. The stronger the subjects experienced the illusion, the higher the synchrony was. The authors suggested that RHI is caused by gamma band synchrony. In addition, a study of the full-body illusion by Lenggenhager et al. (2011) found high correlation between alpha-band oscillations in the sensorimotor cortex and the medial prefrontal cortex, on the one hand, and subjects feeling themselves to be located in space, on the other. Unfortunately, these studies do not really tell us about the sense of self-as-subject. In these experiments, what was misrepresented was the sense of ownership of a body part or a whole body. Whether “I” was the one who was experiencing the illusions was not in question. The synchronization reported by these studies can help explain the sense of body ownership, but not the sense of self-as-subject.
As far as I know, no empirical study on neural synchrony really targets the sense of self-as-subject. We cannot explain the sense of experiential ownership simply by describing the mechanisms of content of conscious perception, cognitive deficits, or body ownership. The lesson here is that we need first to ascertain that the neural information being integrated by synchrony is about the sense of self-as-subject, and not just about representation of the organism’s bodily condition. Unless we know exactly how the integrating processes bring about that one represents oneself as the subject of phenomenal or conscious states, we cannot say that the mechanisms of the sense of self-as-subject have been found. As I will suggest below, the key here is to identify the right research question. And this is where philosophy can make contributions to neuroscience.
The second proposal regarding the mechanisms of the sense of self-as-subject, suggested by Panksepp & Northoff (2009), is self-related processing implemented in the subcortical-cortical midline system (SCMS). This mechanism is notably related to the so-called resting state and the default mode network. Researchers have found that some brain areas are highly activated in the resting state, i.e. when the subject is not actively engaging with its environment (e.g. lying quietly in a scanner with eyes closed but awake) (Raichle et al. 2001). Interestingly, the activations decrease significantly when the subject performs tasks that involve focusing on the external world. These brain areas constitute what is now called the default mode network.
How one should interpret the neural activities in the resting state and the default mode network, and how they relate to self-consciousness, are controversial issues. For example, Gillihan & Farah (2005) point out that different research programs on the self employ divergent methodologies and implicate a wide range of brain areas. Putting all the data together, we do not obtain a specific or unitary picture, because pretty much the entire brain is involved in processing the sense of self. This and other criticisms suggest that we should be cautious when interpreting the alleged empirical evidence about the sense of self-as-subject.[33]
Still, many researchers maintain that resting state activities and the default mode network are closely related to the self (cf. Gusnard 2005; D’Argembeau et al. 2007). Northoff et al. (2006) reviewed a vast number of imaging studies, and compared the processing of what they call self-related tasks and non-self-related tasks.[34] They found that the data indicate the same group of brain areas, including “the medial orbital prefrontal cortex (MOFC), the ventromedial prefrontal cortex (VMPFC), the sub/pre- and supragenual anterior cingulate cortex (PACC, SACC), the dorsomedial prefrontal cortex (DMPFC), the medial parietal cortex (MPC), the posterior cingulated cortex (PCC), and the retrosplenial cortex (RSC)” (2006, pp. 441–442). These areas constitute the cortical midline structures (CMS), i.e. the cortical parts of the SCMS. Compared with non-self-related tasks, when subjects perform self-related tasks their CMS reveal high activation across all domains (2006, p. 450). The authors suggest that the CMS correspond to the default mode network,[35] and that neural activity in the CMS constitutes “an experiential self that mediates ownership of experience” (2006, p. 441). “Ownership”, they claim, “describes the sense that I am the one who is undergoing an experience” (2006, p. 448), which makes this account directly relevant to our investigation.
Legrand & Ruby (2009) argue against Northoff et al. that the CMS are at most self-related, i.e. related to the self only to some extent, but not self-specific, i.e., not specific enough to capture the sense of self-as-subject.[36] Partly because of this criticism, but more because of new findings by his own group, Northoff’s view has changed significantly in recent times. First, Qin et al. (2010) recently studied the CMS in patients who are in a vegetative state. Surprisingly, by showing the patients their own names, various regions in their CMS were activated. Assuming that vegetative patients have lost the capacity to experience themselves as subjects, this finding undermines Northoff’s previous claim that the CMS constitutes an “experiential self that mediates ownership of experience.” In fact, Northoff now agrees that the neural processing in the CMS is at most a necessary condition for the experiential self.[37]
Second, after conducting a meta-analysis on eighty-seven imaging studies covering 1433 participants, Qin & Northoff (2011) suggest that self-related processing involves far fewer areas in the CMS. It is the perigenual anterior cingulate cortex (PACC), rather than the medial prefrontal cortex (MPFC) or posterior cingulate cortex (PCC), that is specifically involved in self-processing. This indicates that they have become more cautious about interpreting data. However, they still maintain that there exists a strong connection between the PACC and the sense of self. They argue that “our sense of self may result from a specific kind of interaction between resting state activity and stimulus-induced activity, i.e., rest–stimulus interaction, within the midline regions” (2011, p. 1221). That is, a narrower network within the CMS is not just necessary but indeed sufficient for “generating our sense of the self” (2011, p. 1222). I will comment on this last claim below.
Whether or not Qin and Northoff take their notion of “sense of self” to include the sense of self-as-subject, I argue that their meta-analysis does not capture the sense of self-as-subject. They describe the operational criteria as follows: “the specificity of the self (e.g. hearing one's own name, seeing one's own face) was tested and compared across familiar (using stimuli from personally known people) and other (non-self–non-familiar, i.e. strangers and widely-known figures) conditions” (2011, p. 1211). The tasks in the “self condition” include “trait adjective judgment, retrieval of personality traits, face recognition, body recognition, personal thinking, name perception, autobiographical memory, own feeling, self-administered pain, person perspective tasks and agency tasks” (2011, p. 1224). All these tasks are about participants making judgments about whether a certain property may be suitably attributed to themselves. From the first-person point of view, the participants are judging whether the contents of the stimuli accurately characterize themselves. But again, whether “I” am the one who is experiencing the stimuli and making the judgments is really not in question, and hence not reflected in the data. Once again, the sense of self-as-subject is not measured by Qin and Northoff’s most recent study.
I conclude that Damasio, Panksepp, and Northoff have all failed to explain the mechanisms of the sense of self-as-subject. A theoretical gap exists between neural synchrony and the SCMS, on the one hand, and the sense of self-as-subject, on the other. But it is important to see exactly where the shortcoming is. It is not that neural synchrony and the SCMS are completely irrelevant to the sense of self-as-subject. Rather, the failure is that why and how they are relevant have not really been explained. This is because the neuroscientists have not clarified and captured the sense of self-as-subject well enough, such that they over-interpret data and make unjustified claims about this target phenomenon. In this regard, my proposals in sections 2 and 3 have provided the required clarification.