[1]
Here I will focus on the minimal sense of self-as-subject, which means that the sense of self-as-subject does not require exercising conceptual capacities and can be transient. It is contrasted with the “narrative self” or “autobiographical self”, which involves episodic memory and persists through time (Gallagher 2000).
[2]
Proprioceptive drift is another aspect frequently associated with RHI: many subjects judge (by proprioception) their real hand as being located closer to the rubber hand, rather than as where it really is. But Rohde et al. (2011) have recently shown that this aspect can be dissociated from the feeling of the rubber hand as one’s own.
[3]
Cf. Ehrsson (2007) for a different OBE experiment.
[4]
The relationship between body-part and whole-body representations for body ownership is a controversial issue. Clearly they are not the same. The issue is: are they fundamentally different? Or is the difference only a matter of degree? As an anonymous reviewer points out, during the rubber-hand illusion, one’s self-location and global body ownership are unaffected. However, during full-body illusions these aspects are affected and misrepresented because they concern the whole-body. Some researchers might therefore think that there exist some fundamental differences between body-part and whole-body representations for body ownership. One can also reasonably hypothesize that the neural mechanisms that are responsible for hand ownership do not need to involve brain regions that process leg or trunk representations. However, in my opinion more interdisciplinary studies would be required to really solve this issue. My current position is that, regarding the sense of body ownership, the difference between body-part and whole-body representations is a matter of degree. First, conceptually speaking, there doesn’t seem to be a sharp distinction between body-part and whole-body representations. Second, if we consider the experimental set-ups of the rubber hand illusion and of the full-body illusions (either Lenggenhager’s version or Ehrsson’s), the differences between them seem to be a matter of degree as well. Of course, these are not arguments yet. I have recently designed a set of experiments precisely to deal with this issue, and I hope to be able to say something about it soon.
[5]
Are there borderline cases between the first-personal and the third-personal experiences of one’s own body? I think so. For example, to use the above example again, if one of my legs suddenly hurts a little bit, I may be able to continue my conversation without disruption, but I have to pay attention to proprioception in order to walk normally. In this case, I submit, the distinction between the first-personal and the third-personal senses of body ownership is not sharp. However, this will not affect my proposal below regarding the relationship between the sense of full-body ownership and the sense of self-as-object.
[6]
Both the first-personal and the third-personal senses of body ownership are involved in RHI and OBE. On the one hand, the fake hand or the virtual body that the subject sees is the object of visual awareness, which is experienced as standing apart from their visual perspective. In addition, by filling in the questionnaires after the experiment, the subject makes explicit judgments about body ownership. This is the third-personal sense of body ownership. On the other hand, during the experiment, the synchronous touch and proprioception causes the subject to feel as if “it is my body that is being touched”. This is the first-personal sense of body ownership, which can be indirectly measured by skin conductance response (SCR). In RHI and OBE, both the third-personal and the first-personal senses of body ownership are prone to misrepresentation.
[7]
Here, “physical body” is broadly construed such that it can refer not only to a physical object but also to a biological organism or a flesh-and-blood person.
[8]
A Cartesian dualist might say that, although I experience a particular body as mine, I fundamentally conceive of myself as a thinking being rather than as a physical body. For the purpose of this paper, we can set Cartesianism aside.
[9]
The idea is that we know how to conduct empirical research in order to study the sense of full-body ownership which, as Blanke and Metzinger suggest, is connected with the following features: (i) the global sense of identification with a physical body as a whole (self-identification); (ii) the sense of being situated in a specific place (self-location); and (iii) the sense of possessing “a point of projection functioning as its origin in sensory and mental processing (weak 1PP)” (2009, pp. 7–8). Together, these features characterize what Blanke and Metzinger call minimal phenomenal selfhood (MPS), defined as “the conscious experience of being a self” (2009, p. 7). It is my view that these three features articulate what it is like to be a self as a physical body. In this regard, the sense of full-body ownership helps us to understand the sense of self-as-physical-body. Also, thanks to the recent findings of the RHI and the OBE experiments, we have now better ideas regarding how misrepresentation may occur in the sense of body ownership. This, in turn, suggests that the sense of self-as-physical-body can involve misrepresentation as well.
[10]
In my account, “the sense of self-as-physical-body” serves as a conceptual bridge between “the sense of full-body ownership” and the “sense of self-as-object”. Experientially, the sense of full-body ownership and the sense of self-as-physical-body are closely related. I deliberately leave open whether these two notions denote the same or different experiences. I think more interdisciplinary work will be required to fix this issue.
[11]
My proposal here is very different from what might be called the Pre-reflective Account of self-consciousness (Legrand 2006, 2007, 2010, 2011; Gallagher 2005; Zahavi 2005). According to this account, self and body are constitutively tied together, and body can provide a sense of self-as-subject, i.e., one can experience one’s body-as-subject. Pre-reflectively experiencing the self as a physical body would correspond to the sense of body-as-subject rather than as-object. The difference between my view and this account centers on whether the notion of object in “self-as-object” is construed as a physical body or as an “intentional object of consciousness”. I contend that the sense of self-as-subject is different from the sense of body-as-subject. Experiencing the self as the subject of experiences is not the same as experiencing the self as a perceiving or acting body. I address these issues in another paper.
[12]
Just before this, Wittgenstein says: “Let us now ask: ‘Can a human body have pain?’ One is inclined to say: ‘How can the body have pain? The body in itself is something dead; a body isn’t conscious!’ And here again it is as though we looked into the nature of pain and saw that it lies in its nature that a material object can’t have it. And it is as though we saw that what has pain must be an entity of a different nature from that of a material object; that, in fact, it must be of a mental nature. But to say that the ego is mental is like saying that the number 3 is of a mental or an immaterial nature, when we recognize that the numeral ‘3’ isn’t used as a sign for a physical object” (1958, p.73).
[13]
For other views about the for-me aspect, cf. Kriegel (2009) and Legrand (2007).
[14]
The method used here is called hyperscanning; cf. Montague et al. (2002).
[15]
Also, when specifying the “I”-as-subject, Shoemaker remarks that “not every self-ascription could be grounded on an identification of a presented object as oneself” (1968, p. 561). Because identification of the self requires that when ascribing a mental state to oneself, e.g. “a is F”, one needs to demonstrate both “b is F” and “a = b.” But “b is F” would in turn require both “c is F” and “b = c”, and hence generates an infinite regress. This, Shoemaker argues, shows that the sense of self-as-subject must be identification-free.
[16]
Although disputed (Coliva 2006), many still consider this distinction useful.
[17]
Other varieties of IEM have been proposed in the literature. For example, Shoemaker (1968) distinguishes between circumstantial and absolute immunity, and between de facto and logical immunity (Shoemaker 1970; cf. also Coliva 2006). Pryor (1999) distinguishes between relative and absolute immunity. The former refers to immunity relative to certain rational grounds G, and the latter immunity by every possible ground. Regarding judgments and beliefs, Coliva (2006) suggests a distinction between immunity relative to the subject’s own rational grounds and immunity relative to background presuppositions.
[18]
Test (ii) was conducted for four sessions, and FB reported feeling touches in 70%, 70%, 100%, and 80% of the trials respectively. As Bottini et al. observe: “her tactile imperceptions dramatically recovered” (2002, p. 251). To test if FB was just guessing, she was again blindfolded and told that her right hand (which is normal) would be touched. But actually the researcher did not touch her right hand. The result was that FB never reported feeling sensations—i.e., she passed the catch trials.
[19]
Shoemaker describes IEM as follows: ‘The statement ‘I feel pain’ is not subject to error through misidentification relative to ‘I’: it cannot happen that I am mistaken in saying ‘I feel pain’ because, although I do know of someone that feels pain, I am mistaken in thinking that person to be myself’ (1968, p. 557). Based on this description, some might insist that the self-ascriptions involved in IEM must be propositional in form, i.e. judgments, beliefs or statements. However, I contend that this restriction is unnecessary. What is crucial for IEM is that the self-ascriptions are based on first-personal grounds such as introspection, somatosensation, and proprioception, etc. As Bottini et al. have stated: “The patient was blindfolded and instructed to say ‘yes’ when she felt a touch and ‘no’ when she did not feel any touch” (2002, p. 251). So when FB said “yes” during test (ii), there is no reason why this wouldn’t count as a self-ascription. Applying Shoemaker’s description to FB’s case: I am mistaken in reporting ‘yes’ during test (ii) because, although I do know of someone that feels the sensations (via first-personal access), I am mistaken in my thinking about who that person is. Shoemaker’s IEM can be violated.
[20]
One might object that the mistake that FB made was about the judgment of experiential ownership, not the sense of experiential ownership. My reply is that since FB was blindfolded, her report was based on first-personal grounds, i.e. on introspection. In addition, FB passed the catch trials mentioned in. As Bottini et al. have stated, FB “did not show any other sign of mental deterioration on the Mini Mental State Examination” (2002, p. 251). Therefore, no evidence suggests that her reports were unreliable. These considerations support the idea that the mistake was FB’s sense of experiential ownership rather than her judgment. For other objections and responses, cf. Lane & Liang (2011).
[21]
Again, one might wonder whether the misrepresentation in this case was about the judgment rather than about the sense of experiential ownership. My reply is that since the subjects were normal college students, their reportability was not in question. So it is plausible to assume that their reports that “I was shaking hands with myself” were based on their subjective phenomenology, and more specifically on their sense of experiential ownership. Hence it was their sense of experiential ownership that committed misrepresentation.
[22]
There are at least two other (possible) cases of misrepresentation of the sense of experiential ownership. One is voice ownership: an illusion in which a stranger’s voice, when presented as the auditory concomitant of a participant’s own speech, is perceived as a modified version of one’s own voice. “It felt as if the voice I heard was my voice” (Zeng et al. 2011). The other is perception ownership: A twenty-three-year-old male (DP) suffered from right inferior temporal hypometabolism (Zahn et al. 2008). The authors of a study on this male described his sensations as follows: “It appeared to him that he was able to see everything normally, but that he did not immediately recognize that he was the one who perceives and that he needed a second step to become aware that he himself was the one who perceives the object.”
[23]

Let me briefly compare my position with other views. First, following Shoemaker, Coliva (2000) states that “If a subject is introspectively aware of pain, this just means that she is feeling pain […] it is a matter of conceptual truth that if a subject is introspectively aware of a certain mental state, then she herself is having it and, therefore, that mental state is her own” (my emphasis). In contrast to Coliva, my account rejects IEM as a conceptual truth. From the fact that a subject experiences a mental state it does not necessarily follow that the subject represents herself as the one who experiences that state. I take the possibility of misrepresentation to be an important feature of the sense of experiential ownership.

Second, Legrand (2007) emphasizes that consciousness of self-as-subject is pre-reflective, meaning that it is not an object of intentional consciousness. She says that the self-as-subject “is neither an external object (for example, it is not my body that I can observe in the mirror) nor an internal object […] I am simply looking outside at the external world, and within this single act of consciousness I pre-reflectively experience myself-as-subject” (2007). I agree that the sense of self-as-subject is often implicit rather than explicit. But Legrand’s view neglects the distinction that I draw between the fact and the sense of experiential ownership. This is indicated by the fact of her embracing IEM. The fact of experiential ownership can be secured simply by looking outside at the external world, but whether one’s sense of experiential ownership is correct is another issue.

[24]
What is the relationship between the sense of body ownership and the sense of experiential ownership? The short answer is that the former presupposes the latter, but a full treatment would require another paper. Here, let me draw on Metzinger’s Self-model Theory of Subjectivity (2003, 2008) to briefly address this issue. According to this theory, PMIR (phenomenal model of the intentionality relation) is a phenomenal experience that represents the relation between a subject and an object component. For example, I take a bite of an apple. The PMIR contains a subject component (I), a relation component (tasting), and an object component (the apple). But I want to propose a revised version of PMIR. Since the PMIR is a complex phenomenal property experienced by a subject, it would sometimes be legitimate to ask who is experiencing this particular PMIR. Does the subject attribute the sense of experiential ownership of this PMIR to him or herself? My proposal is that PMIR consists of three components: (1) the sense of experiential ownership; (2) intentional relations; and (3) an object component. On this view, PMIR already involves the sense of experiential ownership as the subject component, which is distinct from intentional relations and the object component. This revised version of PMIR helps to unpack the phenomenological structures of the sense of body ownership as follows. The subject component is served by the sense of experiential ownership. The object component can be one of the following: my hand, a rubber hand, someone else’s leg, my whole-body, or a virtual body, etc. The intentional relations include vision, touch, proprioception, location, motion, introspective awareness, affective feelings, and so on. Four quick remarks are relevant here. First, the sense of body ownership is itself a phenomenal state, about which (2) and (3) specify the what-component. The who-component of the sense of body ownership is characterized by (1) the sense of experiential ownership. Hence, the sense of body ownership presupposes the sense of experiential ownership. Second, it is (1) and (2) that generate the sense that (3) is part of my body. Third, the difference between the sense of body-part ownership and the sense of full-body ownership lies in (3), while (1) and (2) may remain the same. Finally, based on my proposal in section 1, the sense of self-as-physical-body can be understood in terms of the following structure of PMIR: (1) the sense of experiential ownership; (2) intentional relations; and (3) a whole body. And the sense of self-as-object can be understood in terms of the same structure of PMIR as well.
[25]
Both Damasio and Panksepp distinguish between emotions and their neural substrates, on the one hand, and feelings (Damasio) or affective feelings (Panksepp), on the other. Emotions refer to innate patterns of neural and physiological responses to environmental events. Feelings (or affective feelings) refer to phenomenal consciousness of emotions (Damasio 1999, p. 42, p. 55; Damasio 2010, pp. 108–110; Panksepp 1998, pp. 48–49; Panksepp 2005, p. 32). The emotion-systems closely interact with the self-systems to regulate and manage homeostasis.
[26]
For Damasio, the cortical areas that are important for the core self include insular and somatosensory cortices (2010, pp. 205–209).
[27]
According to Panksepp, emotions and affective feelings are internally generated by neuronal mechanisms to respond to life-challenging events. The neural systems of emotions compute and monitor homeostasis by evaluating an organism’s adaptation to the environment. Each emotion system refers to a specific neural network, mainly in the subcortical areas.
[28]
Panksepp and Northoff prefer to use the expressions “proto-SELF” and “core-SELF” to emphasize neural mechanisms rather than mental phenomena, but this emphasis need not concern us here. They describe the relation between core-SELF and affective consciousness as follows: “What is subjectively experienced here is the relation of one’s body to the incentives in the environment as well as internally generated emotional arousals—the core-SELF thus enables the organism to access this relation in terms of subjective experience, e.g., a primitive form of phenomenal consciousness, which at this level is essentially affective” (2009, p. 196).
[29]
Cf. Legrand (2007) for a slightly different criticism of Damasio.
[30]
The sense of experiential ownership is not studied in Singer’s work on neural synchrony at all.
[31]
Louet al. (2010) suggest that this network includes the anterior cingulate, medial prefrontal and posterior cingulate, and the medial parietal cortices, connected via the thalamus.
[32]
In the ‘Franz’ condition, the subject judged how well each adjective fitted a well-known German football star Franz Beckenbauer. In the ‘‘Syl’’ condition, the subject’s task was to decide whether each of the different sets of adjectives had an even or odd number of syllables.
[33]
Another criticism is that, when the subject is interacting with the world, the neural activity in the default mode network is not totally extinguished. Some studies show that it is “reorganized in response to the working memory task” (Fransson 2006). Others have suggested that it could “function to support exploratory monitoring of the external environment when focused attention is relaxed” (Buckner et al. 2008).
[34]
Many of these studies used a “judgment paradigm”. Subjects made explicit evaluative judgments about first- vs. third-person perspectives, own vs. others’ judgments, self vs. others’ decisions, own vs. others’ personality traits, etc. The domains that Northoff et al. (2006) reviewed include verbal, spatial, memory, emotional, facial, agency, ownership of movements, and social tasks.
[35]
CMS show a high level of neural activity during the resting state. Non-self-referential tasks elicit large signal decreases in the CMS (Northoff et al. 2006, p. 450).
[36]
Legrand and Ruby indicated that the CMS are involved not only in self-related tasks, but also in several cognitive tasks that are not related to self-consciousness at all. For example, their review showed that some areas in the CMS are activated in others’ mind reading, inductive and deductive reasoning, resting state, and memory recall. Moreover, these areas are “sometimes more activated for the self than for others and sometimes more activated for others than for self” (Legrand & Ruby 2009, p. 258).
[37]
Northoffet al. tell us that “the neural mechanisms underlying SRP [self-related processing] may only be considered a necessary condition which is not sufficient by itself to constitute a self with its self-specific contents” (2011, p. 55).