NOTE-1008: Alsmith & Longo — Where exactly am I?

Contribution

It asked people to locate themselves on their own bodies, directly, with a task that let different answers come out on different trials. Before it, one study narrowed self-location by interview to a single point in the head (Bertossa et al. 2008), and two asked for judgements about a drawn figure (Limanowski & Hecht 2011; Starmans & Bloom 2012). After it, there is a small first-person result: judgements of where “I” am fall on two regions, the upper face and the upper torso, and the context of the task decides between them.

Key insight

Asked to point exactly at themselves, people do not converge on one spot. They stop at whichever of two candidate regions, face or chest, the pointer reaches first. The authors read this as ambiguity in the concept of oneself as a located thing, resolved by context, not as a self spread over the body.

Assumptions

• Pointing at oneself reports self-location. The instruction was to set the pointer “pointing directly at you” (p. 71). That the answer reports where the participant takes the self to be, and not a body part judged salient or a point that is merely easy to aim at, is assumed. The authors move between “the experienced self” (abstract) and “our concept of the self as a spatial entity” (p. 74) without saying which the task measures.
• The egocentre analogy. The task generalises the binocular “egocentre” test of Wells and Hering, in which a rod is turned until it points at oneself and the lines from several directions intersect at a point (p. 71). Here only where the line meets the body was coded, in one plane, so no point inside the body is recovered.
• Region coding by surface landmarks. Each response was coded by where the pointer’s line met the body: lower torso (below the elbow), upper torso (elbow to acromion), neck (acromion to chin), lower face (chin to tip of the nose), upper face (above the tip of the nose) (p. 72). The boundaries were chosen to be visually salient for coding. The upper face therefore includes eyes, forehead and crown, and the task cannot tell the eyes from the middle of the head.
• Context variation probes stability. Height of the pivot and the direction of rotation were varied “to add diversity to the task and probe possible influences” (p. 72); invariance across them would have meant a stable self-location.

Key results

• Design (pp. 71–72). n = 10 (six women, aged 24–48, all right-handed). Participants stood before a pole carrying a pointer that rotated in their sagittal plane. Haptic: blindfolded, they turned the pointer with the right hand. Visual: the experimenter turned it at roughly constant speed and they said when to stop, with corrections allowed. A webcam photographed each answer. Two blocks (modality, counterbalanced), each of eight mini-blocks at four pivot heights spaced at 1/12 of body height around the chin; in each mini-block three trials started pointing straight up (Top) and three straight down (Bottom). 96 trials each. Participants were told the task “was not a test of the ability to judge the direction perpendicular to the pole”.
• Exclusions (p. 72). Because of delays between key press and photo, “A total of 20.7% of trials were therefore excluded from analysis”. An ANOVA on the proportion excluded found no main effects or interactions (all p > .05).
• Region (p. 73). Repeated-measures ANOVA, region × modality × start (height dropped, no effects): main effect of region F(4, 36) = 9.70, p < .0001, “a clearly bimodal distribution of responses, with points towards the upper torso and (especially) the upper face predominating”.
• Start location (p. 73). Start × region F(4, 36) = 24.00, p < .0001. Bonferroni-corrected: more upper-face answers from the top than from the bottom in the visual condition (t(9) = 5.45, p < .002), a trend in the haptic (t(9) = 2.83, p = .079); more upper-torso answers from the bottom in haptic (t(9) = 4.13, p < .02) and visual (t(9) = 3.81, p < .02). Read off Fig. 2: visual-top about 75% upper face; haptic-top about 39% upper face; visual-bottom about 39% and haptic-bottom about 32% upper torso, with the remainder spread over neck and lower face; lower torso near zero throughout.
• Modality (p. 73). Modality × region F(4, 36) = 4.56, p < .005, and a three-way interaction F(4, 36) = 3.35, p < .02, read as “a more focused bimodality in the visual than the haptic modality”, reflecting “reduced precision in the haptic modality, rather than any qualitative difference”.
• Reanalysis without the lower torso (p. 73). “Critically, all the effects reported above remained significant”: region F(3, 27) = 2.90, printed “p = .53”; modality × region F(3, 37) = 4.59, p < .02; start × region F(3, 37) = 24.25, p < .001; three-way F(3, 27) = 3.38, p < .05. F(3, 27) = 2.90 gives p ≈ .053, so the region effect, the bimodality itself, is not significant at .05 once the empty lower-torso region is removed; the start interaction is. The denominator 37 is also inconsistent with 10 participants.
• Conclusion (pp. 73–74). “No single body part is judged as the unique seat of the self.” Answers are “ambiguous between at least two locations, though when forced to judge this ambiguity can be resolved according to contextual factors”. Against laziness: lazy answers from below would have stopped at the lower, not the upper, torso.
• Why these two regions (p. 74). The head houses eyes, ears and the vestibular organ, which Sherrington put at the top of the proprioceptive hierarchy. The torso is “the great continent of the body, relative to which all other body parts are mere peninsulas”, apt to anchor an egocentric frame; the authors cite Blanke’s (2012) hypothesis that full-body illusions shift trunk-centred visuo-tactile receptive fields, and trunk rotation effects in neglect. They offer this against Limanowski & Hecht’s reading of torso answers as the heart in a folk concept of selfhood.
• The concept (p. 74). “To the extent that such self-location judgements tell us anything about our concept of the self as a spatial entity, they tell us that the concept is inherently ambiguous.”

Claims

id	claim	strength	support
C1	Judgements of where one is, pointing at one’s own body, fall mostly on the upper face and the upper torso	moderate as description, for this sample	10 participants, about 760 retained trials; Fig. 2; region F(4, 36) = 9.70
C2	Which of the two is chosen depends on the pointer’s starting position	moderate	start × region F(4, 36) = 24.00, survives reanalysis; three of four follow-up tests significant
C3	Visual and haptic self-location judgements do not differ in kind	weak	no condition separates lower precision from a different judgement; the haptic answers fill the neck and lower face
C4	The self is judged neither spread homogeneously over the body nor at a single point	weak to moderate	follows from C1 and C2, but the region effect without the lower torso is p ≈ .053, misreported
C5	The concept of the self as a spatial entity is inherently ambiguous	interpretation	rests on C1–C2; the task does not separate concept, experience and aiming strategy
C6	Head and torso are chosen because of their functional roles in perception, posture and egocentric space	speculation	cited physiology and neglect studies; nothing tested

Concepts

• Egocentre — the point inside an observer from which visual direction is judged, found by having a rod pointed at oneself from several directions and intersecting the lines (p. 71); also “cyclopean eye”, “binoculus”.
• Self-location judgement — here, where on one’s own body a pointer aimed “directly at you” meets it.
• Projected self-location — the authors’ term for judgements made about a depicted figure rather than one’s own body, as in Starmans & Bloom and Limanowski & Hecht (p. 71).

Connections

It is built on the egocentre literature (Howard & Templeton 1966; Mitson et al. 1976) and set against three studies of where the self is located: Bertossa et al. 2008 (head, by interview, also in blind participants), Limanowski & Hecht 2011 (head and chest on an outline), and Starmans & Bloom 2012 (eyes, by judgements about cartoon characters), read alongside in NOTE-995. It calls the last “provocative” and calls both drawing studies projections onto a depicted subject. Its torso discussion reaches toward Blanke’s account of full-body illusions, the account THEORY-768 is filed from. Longo is a co-author of Longo et al. 2008 (LIT-1275) and signs Ehrsson et al. 2022 (LIT-1264).

Bearing on the record

• THEORY-768, the self-location dimension. Blanke & Metzinger (LIT-1274, NOTE-1002) define self-location as “a determinate volume in space, normally localized within the bodily boundaries as represented”, within which the origin of the weak first-person perspective lies. This study fits a volume and not a point, and it fits “within the bodily boundaries”: no answer left the body, and almost none fell on the lower torso. Two things are new for the record. First, it is a measure of self-location other than the walking-back drift, which the THEORY’s promotion test says cannot measure self-location alone after Rohde et al. (THEORY-766). Second, that measure is moved by a task factor, the start of the pointer, as strongly as by anything else, so it inherits the same kind of problem from a different source. The head answers fit the weak first-person perspective’s origin at the eyes; the torso answers do not, which is consistent with Blanke & Metzinger’s allowance that self-location and the perspective’s origin can come apart, and does not test it. It neither meets nor bears against the THEORY’s promotion test: it involves no rubber hand, no full-body illusion and no one without an agency history. Proposal only: the promotion test could say that a pointing measure of self-location, if used, must vary the starting direction and report it, since this study shows the start can decide the answer.
• THEORY-766, THEORY-769, THEORY-767. No bearing. Nothing is owned, stroked or synchronised, and no interoceptive signal is measured.
• THEORY-458. No bearing. Participants acted (in the haptic condition they moved the pointer), but agency was neither manipulated nor measured.
• TERM-045 and ADR-024. The “I” pointed at is TERM-045’s sense 1, the subject, as located: the self the self topic keeps under ADR-024. The instruction does not separate that self from the body or the person: “pointing directly at you” invites any of them.
• QUESTION-048. A worked case of what a report shows. The answer is a report of a judgement under instruction, and it moved with where the pointer started. It establishes where people stop when asked to point at themselves, and that this is not one fixed place. It does not establish where anyone experiences themselves to be, and the authors’ own last sentence retreats from “the experienced self” to “our concept”.
• LIT-1265 (Starmans & Bloom 2012). The two disagree, and the disagreement is partly about method: first person against third, own body against cartoon, a surface region that includes the eyes against a design that separates eyes from head. Both find the head favoured; only this one finds the torso.
• No instruction for machine-learning practice; nothing for the anthology.

Limitations

• Ten participants, all right-handed, ages 24–48, recruitment not stated.
• A fifth of trials lost to a camera fault, shown not to differ by condition but reducing an already small sample of trials.
• Statistical slips. The reanalysis prints “p = .53” for F(3, 27) = 2.90 (p ≈ .053) and calls it significant, and gives denominators of 37 in a design with 10 participants. The claim that “all the effects reported above remained significant” is false for the region effect.
• No baseline start. The pointer started only straight up or straight down. A start pointing at the body’s middle, or a random start, would show whether the two modes survive without a direction of approach. The authors’ answer to the laziness worry rules out stopping at the first contact with the body, not stopping at the first acceptable place.
• One plane, surface coding. Only the sagittal plane was sampled and only the intersection with the body surface coded. “Upper face” cannot separate the eyes from the middle of the head, which is the contrast between Starmans & Bloom and Bertossa et al.
• The visual condition is someone else’s movement. Participants stopped a pointer the experimenter moved, so reaction time and the experimenter’s speed enter the answer.
• No figures for the means. The distribution is given only in Fig. 2.

Open questions

• Do the two modes survive a start at the body’s middle, or random starts in several planes, so that lines can be intersected as in the egocentre task and a point or volume inside the body recovered?
• Does a full-body illusion, which moves self-location in Lenggenhager et al.’s drift measure, move the answers on this task, and toward which region?
• Is the torso mode a judgement about the self or about the body’s centre of mass and orientation? A condition asking participants to point at their body, not at themselves, would separate them.