NOTE-1434: Metzinger — Why are dreams interesting for philosophers?

Contribution

It revises the 2009 definition of minimal phenomenal selfhood (MPS) on the evidence of bodiless experience. Blanke and Metzinger had three defining features: global identification with the body, self-location, and a weak first-person perspective (1PP) (p. 3). Here identification is argued to collapse into self-location when the body drops out, which gives a more parsimonious MPS: transparent self-location in space and time (p. 7). The paper then sets out a list of questions for dream research, among them the case that would test the transparency condition.

Key insight

What is minimally experienced as “I” is wherever the system’s most invariant, transparent self-location is. Strip away the body, as some dreams and out-of-body experiences do, and a self remains at a point; so the body is not necessary, and what is left is location under transparency. The question that would test the account is then a sharp one: is there ever a self-location that is experienced as a representation, and if so, is there still a self?

Assumptions

• The self-model theory. Transparency is a property of conscious representations, “namely, that they are not experienced as representations”; opacity is the converse, as “in pseudo-hallucinations or lucid dreams”; the two form “a graded spectrum” (p. 2, note 2). Cited to Being No One and other works, not argued here.
• Explanatory parsimony. Begin with “those conditions that are not only causally enabling, but strictly necessary” for the minimal phenomenon (p. 2).
• Reports as heterophenomenological constraints (p. 4, note 6), with the caveat (pp. 13–14) that dreams show unnoticed cognitive corruption, answered by an inference to the best explanation that reports gathered under ideal conditions “will be veridical” (Windt 2013).
• The survey categories mean what they say. The prevalence figures for “bodiless” out-of-body experiences are taken as evidence of bodiless selfhood, although the paper itself doubts what subjects understood by the questions (p. 6).

Key results

The paper has no data of its own. Its claims, in order:

• MPS, original hypothesis (p. 3). “Central defining features are (1) a globalized form of identification with the body as a whole (as opposed to ownership for body parts), (2) spatiotemporal self-location, and (3) a 1PP.” MPS is necessary for the strong (attentional, cognitive) 1PP; the weak, geometrical 1PP is a constituent of MPS. “Because our conscious body model is transparent … we fully identify with its representational content.”
• Difficulty 1 (pp. 3–4). Alsmith’s objection that “full-body illusions” may be partial: there is “no measure of the extent to which the effect is localized to particular parts”. Metzinger separates the empirical underdetermination from a possible phenomenal indeterminacy of the object of identification, and asks “What exactly is the minimal unit of identification?”
• Difficulty 2: asomatic out-of-body experiences (pp. 4–5). Alvarado’s review: experiences in another body (10 studies, mean 46%), as “pure consciousness” (6 studies, mean 31%), or as “balls of light”, “points in space” or “clouds” (6 studies, mean 29%); Wolfradt: 18.8% bodiless, 34.4% in a body like their own, 21.9% “in a cloud, fog, or at a point in space”.
• Three conclusions from those reports (p. 5). “Explicit body representation is not necessary, because an extensionless point in space is sufficient for phenomenal self-identification”; bodily agency is not necessary either; and since selfhood seems robust with no motion and no goal-directed mental activity, “agency is not a constitutive element for the minimal unit of identification”.
• The UI-theory (p. 5). Identification is a domain-general mechanism that attaches to “a given region of maximal invariance”; the unit of identification (UI) can change within an episode. A minimal UI is the point of self-location, a maximal UI is “phenomenality per se” (as in some meditation reports).
• Ambiguity of the evidence (pp. 5–6). It is “not entirely clear” how many of the “pure consciousness” subjects were in fact located at a point, “because it is not exactly clear what subjects (and investigators) actually understood or meant in the questionnaire”.
• Bodiless dreams (pp. 6–7). Defined by minimized spatial content, “phenomenal epistemicity” (the experience of being a knowing self, an “epistemic agent model”, EAM) and self-identification. “Most bodiless dreams occur during lucid dreams.” One example report is given (an exam book, nothing else). In such dreams “the abstract unit of identification and the unit of self-location coincide”, so the identification condition “would then be reduced to the self-location condition”, giving MPS as “transparent self-location in a spatiotemporal frame of reference” (Windt 2010, p. 7).
• Two research questions (p. 7, repeated p. 14). Is the epistemicity of the EAM necessary for robust selfhood, or are transparency and self-location sufficient? And: “are there fully lucid, phenomenally opaque, and bodiless dreams for which it is true that spatiotemporal self-location is itself experienced as a form of (mis)representational content?”
• Why dreams (pp. 7–8). Asomatic out-of-body experiences and meditative “pure consciousness” are rare and ideologically coloured; virtual reality starts from a rich waking self-model; dreams happen nightly and thousands of reports exist to be re-analysed.
• Part 2 (pp. 8–14). Four contact points: the 1PP as graded (attentional agency is “fully transparent”, p. 8; lucid gaze signalling as a grounding relation, p. 9); mind wandering and the recurrent loss of “mental autonomy” (pp. 9–10); embodiment, where the general “functional disembodiment” of dreams “has now been refuted” by evidence of stimulus incorporation (p. 11); and the “argument from cognitive corruption” (p. 13).
• The prediction (p. 14). “One empirical prediction the current proposal makes is that transparent spatiotemporal self-location is necessary and sufficient for bringing about a minimal, consciously experienced sense of selfhood.” Then: “Do we know any cases of phenomenally opaque self-location …? Is MPS instantiated in these cases?”

Claims

id	claim	strength	support
C1	A stable sense of self is reported with body representation reduced to an extensionless point	moderate as a report	others’ survey figures (Alvarado 2000, Wolfradt); categories ambiguous by the paper’s own account
C2	Neither explicit body representation nor bodily agency is necessary for minimal selfhood	weak	inference from C1
C3	In bodiless dreams self-identification and self-location coincide, so MPS reduces to transparent spatiotemporal self-location	weak; proposal	conceptual argument from C1 and one example dream report; Windt’s definition adopted
C4	Transparent spatiotemporal self-location is necessary and sufficient for minimal selfhood	prediction; untested	stated as the proposal’s empirical prediction (p. 14)
C5	Identification attaches to a region of maximal invariance and its unit can change	speculation	the “UI-theory”, offered as a hypothesis
C6	Generalized functional disembodiment in dreams is refuted	moderate	stimulus-incorporation and dream-enactment studies, cited (p. 11)
C7	Dream reports gathered under ideal conditions will be veridical	weak	inference to the best explanation, after Windt 2013

Concepts

• Minimal phenomenal selfhood (MPS) — “the simplest form of self-consciousness” (p. 1); in 2009, global body identification, self-location and a weak 1PP; proposed here as transparent spatiotemporal self-location (p. 7; Box 1, p. 11).
• Transparency / opacity — a conscious representation not experienced as a representation / one that is; graded (p. 2, note 2).
• Weak 1PP — “a purely geometrical feature of a perceptual or imagined model of reality”, a centre of projection (p. 3).
• Epistemic agent model (EAM) — a conscious self-representation of the system as a knowing self, “in the relation of knowing, thinking, actively guiding attention” (p. 6; Box 1). Lucidity is “the gradual stabilization of an EAM” (Box 2, p. 12).
• Unit of identification (UI) — what the experience of “I am this!” attaches to; minimal (a point of self-location) or maximal (phenomenality as such) (p. 5; Box 2).
• Asomatic / parasomatic / aparasomatic OBE — out-of-body experience with no spatial form, with a second body, or with a double one does not identify with (p. 4).
• M-autonomy — the ability to control one’s own mental functions, with a “veto component” (Box 2, p. 12).

Connections

It builds on Blanke and Metzinger 2009 and adopts Windt’s (2010, 2014) account of dreams as “immersive spatiotemporal hallucinations”. It cites Lenggenhager et al. 2007 (LIT-1279, read in NOTE-999) as showing MPS “can be experimentally controlled” (Box 1) and as dissociating the origin of the 1PP from the body volume (note 5). Its transparency is the précis’s (LIT-568, read in NOTE-460), and its account of the knowing self extends the précis’s phenomenal model of the intentionality relation. Alsmith’s objection to full-body illusions is the same worry the critics of Lenggenhager et al. raised, from another side.

Bearing on the record

• THEORY-064: not met; it restates the test more sharply. The account’s promote_when asks for states in which more of the self-model becomes attentionally available as constructed, with reported selfhood falling in step, or a reliable report of a wholly opaque self-model with selfhood intact; or a measure of attentional availability that covaries with reported selfhood. This paper reports neither. What it does is narrow the prediction to the minimal case, “transparent spatiotemporal self-location is necessary and sufficient” for minimal selfhood (p. 14), name the decisive case (“phenomenally opaque self-location”, sought in “fully lucid, phenomenally opaque, and bodiless dreams”), and say that no such case is yet known to it. Its one empirical contribution bears on the ontological side the promote_when sets aside: reports of selfhood without body representation show the self-model can be minimal, not that it is all a self is. It does add, from the author, that lucid dreams are a standard example of opacity (note 2), which supports promote_when’s choice of lucid dreaming as the place to look.
• THEORY-768 (minimal selfhood as whole-body identification). This paper argues against the body condition in Blanke and Metzinger 2009: “explicit body representation is not necessary” (p. 5), and MPS may reduce to self-location. It bears against THEORY-768’s body clause, on survey-report evidence the paper calls ambiguous.
• THEORY-597 (meditation and “loss of basic self”). The paper reads some meditation reports as identification with the maximal unit, “phenomenality per se”, not as loss of self (p. 5). That is a rival reading of the same reports, offered without data.
• No instruction for machine-learning practice; nothing for the anthology.

Limitations

• No new data. Prevalence figures are others’, and the paper itself says the questionnaire categories cannot tell a point-located self from “pure consciousness” (pp. 5–6).
• One dream report illustrates bodiless dreaming; no corpus is analysed.
• The reduction is conditional. “It may be possible to integrate conditions 1 and 2” (p. 7): the paper proposes it, it does not show it.
• Reports under corruption. The paper’s own Example 4 argues that dream cognition is corrupted without being noticed, which weighs on the reports the reduction rests on; it answers this with an inference to the best explanation, not a test.

Open questions

• Is there any report of a phenomenally opaque self-location, and does a sense of self survive it? The paper names this as the test and does not answer it.
• Is the epistemic agent model necessary for robust selfhood, or are transparency and self-location sufficient?
• Can dream questionnaires be redesigned to pick out the unit of identification reliably?