Interoceptive Awareness — D/A/U Analysis Project
Purpose
Section titled “Purpose”This project prepares semi-quantitative evidence from existing Baseworks participant accounts to support the claim that training interoceptive discrimination (D) produces emergent appraisal (A) and self-regulatory use (U) outcomes — even when training explicitly targets only D.
The immediate application is the BRNet 2026 poster (Padua, June 8–9, 2026), where interoceptive awareness is the third component in the three-component body representation framework. The two Baseworks training principles most directly responsible for interoceptive development are intensity modification and natural breathing.
The analysis is distinct from the LPA and spatial awareness components: interoceptive U-level outcomes appear early in training (visible even in Primer students), whereas LPA and SA require longer development timelines.
Theoretical Frame
Section titled “Theoretical Frame”The D/A/U Dimensions
Section titled “The D/A/U Dimensions”These three dimensions apply within each perceptual skill domain (Localized proprioceptive awareness, spatial awareness, and interoceptive awareness )
| Dimension | What it is | What it is NOT |
|---|---|---|
| D — Discrimination | Detecting and localizing internal sensations; noticing breathing changes, effort signals, discomfort thresholds | Evaluating or acting on those signals |
| A — Appraisal | The relationship to internal sensations — trust, safety, acceptance, equanimity, internal vs. external locus of evaluation; not catastrophizing or avoiding sensation | Using sensory info to make a behavioral decision |
| U — Use / Skill | Using available sensory information to make self-regulatory decisions — modifying intensity, pacing, emotional regulation, behavior change in daily life | The sensory signal itself |
Key distinction: A vs. U
Section titled “Key distinction: A vs. U”This is the easiest boundary to blur. A is closest to what MAIA measures — e.g., “I trust my body sensations,” “I feel my body is a safe place,” “I am at home in my body.” It concerns the attitude toward internal experience. U is active regulation — making a decision or changing behavior based on what is sensed.
Examples from the dataset:
- A: Yasuko moving from external rule (“do I have to do this?”) to internal evaluation — restructuring the authority relationship with her own sensations; Clementine Morrigan: “trusting myself to adapt” instead of avoidance; Daria: “a new acceptance around not being able to do things”
- U: Cheri deciding not to push today based on how she feels; Katharina anticipating emotional difficulty and stepping back; Mimi Robert Trudel using practice tools as a nervous system regulation strategy on the metro; Niki becoming able to wait instead of reacting with anger
Why D is trained, A and U are emergent
Section titled “Why D is trained, A and U are emergent”Baseworks training explicitly targets D through intensity modification (real-time interoceptive monitoring of effort) and natural breathing (breath as a barometer of overload). A and U are not directly instructed — they are reported spontaneously by participants as consequences of improved D. This is the core empirical claim this project supports.
Dataset
Section titled “Dataset”Source 1 — Testimonials
Section titled “Source 1 — Testimonials”Location: 02-areas/method-admin/audience/testimonials/
Excluded: origin-language-transcripts/ subfolder (source-language versions only)
Filter: Training: public classes OR Training: private sessions
N = 33 participants from original filter (+ magali-fr-en as exception: Training: study group/primer, added by Asia’s explicit request — counted once in total N=39 via combined testimonial+segment-feedback row)
All testimonials are English translations of video or written accounts collected across multiple years and countries (Japan, Canada, Europe, Latin America, China).
| File slug | Training | Notes |
|---|---|---|
| akiko-jp-en | public classes | |
| althea-en-en | private sessions | |
| anna-jp-en | public classes | |
| cheri-en-en | public classes | |
| chieko-jp-en | public classes | |
| chikara-jp-en | public classes | |
| chisa-jp-en | public classes | |
| daria-en-en | public classes | |
| hirotaka-jp-en | public classes | |
| hisako-jp-en | public classes | |
| jiletta-en-en | public classes | |
| kangying-en-en | public classes | |
| katharina-en-en | public classes | |
| kazuko-en-en | public classes | |
| krister-en-en | public classes | |
| lin-en-en | public classes | |
| ling-en-en | public classes | |
| lionel-en-en | public classes | |
| mari-jp-en | public classes | |
| mayumi-jp-en | public classes | |
| megumi-jp-en | public classes | |
| miki-jp-en | public classes | |
| mirjana-en-en | public classes | |
| morten-en-en | public classes | |
| nana-jp-en | public classes | |
| niki-jp-en | public classes | professional dancer/instructor |
| risa-jp-en | public classes | |
| satomi-jp-en | public classes | |
| stacey-en-en | public classes | |
| tiffanie-en-en | public classes | |
| yasuko-jp-en | public classes | history of anorexia, noted for A-domain |
| yuka-jp-en | public classes | |
| yuna-jp-en | public classes | professional dancer/instructor |
| magali-fr-en | study group / primer | exception: filter normally excludes study group/primer; included by Asia’s request |
Source 2 — Segment Feedback
Section titled “Source 2 — Segment Feedback”Location: 02-areas/method-admin/audience/segment-feedback/
Filter: Specific respondents only (see below)
N = 6 respondents (multiple files each, covering different curriculum segments)
Program: All are Primer (early learners)
| Respondent | Files |
|---|---|
| Any Guay | s02–s10 |
| Clementine Morrigan | s02–s05 |
| Luchida Michel | s02–s10 |
| Magali Lalonde | s02–s06 |
| Mimi Robert Trudel | s02–s10 |
| Nathalie Dore | s02–s08 |
Note: Segment feedback is collected during early training (Primer course). The presence of U-level outcomes in this group is theoretically significant — it suggests behavioral self-regulation generalizes quickly from interoceptive training, unlike LPA and SA which require longer development.
Analysis Strategy
Section titled “Analysis Strategy”D/A/U Prerequisite Analysis — Results (Session 3)
Section titled “D/A/U Prerequisite Analysis — Results (Session 3)”Statistical analysis of the prerequisite hypothesis is complete. Full methods, script, and results in dau-nca.
Hypothesis tested: D is a necessary condition for both A and U; A and U are independent of each other.
Pattern distribution (N = 39): 38/39 participants (97.4%) showed patterns consistent with the prerequisite hypothesis. The single inconsistent case (0|1|0, Morten) is a likely ascertainment artifact — the account is cognitively framed and unlikely to surface interoceptive D language in a free-response format.
Necessary Condition Analysis (Dul 2016):
| Hypothesis | Cases tested | Violations | Consistency | p-value | Verdict |
|---|---|---|---|---|---|
| H1: D necessary for A | 17 (A=1) | 1 (5.9%) | 0.941 | 0.095 | Supported |
| H2: D necessary for U | 26 (U=1) | 0 (0%) | 1.000 | < 0.001 | Fully supported |
| H3: A and U independent given D=1 | n = 32 | — | — | 0.654 (Fisher’s) | Independent |
Consistency ≥ 0.90 is the accepted threshold for claiming a necessary condition (Dul 2016). H1’s permutation p = 0.095 reflects limited power at N=39 with a single violation; the consistency score is the more interpretable metric. H2 is perfect and highly significant. H3: Fisher’s exact OR = 0.43, p = 0.654 — no association, consistent with A and U being independent dimensions. U is more prevalent than A among D=1 participants (81% vs. 50%), consistent with U being an earlier emergent outcome.
Unit of analysis
Section titled “Unit of analysis”Participant level (not extract level). For each participant, binary coding: did this person’s account contain evidence of this domain/dimension? This gives prevalence rates that are meaningful at N=39.
Step 1 — Outcome domain coding (for Figure 1)
Section titled “Step 1 — Outcome domain coding (for Figure 1)”Eight binary columns (0/1 per participant):
| Column | What counts |
|---|---|
| Stress reduction / relaxation | Reports of calming, release of tension, feeling lighter, mental rest |
| Intensity management / knowing limits | Modifying effort, knowing when to stop or reduce, respecting body signals |
| Emotional regulation | Changes in mood, emotional reactivity, acceptance, resilience |
| Body awareness — explicit mention | Explicitly states improved awareness of internal body states |
| Daily life behavior change | Applying practice-derived awareness to behavior outside of practice |
| Pain / tension reduction | Reports of reduced pain, chronic tension, or injury prevention |
| Nervous system / energy regulation | Explicit references to nervous system, energy, vitality, recovery |
| Sleep | Any mention of sleep quality change |
Coding is manual binary — keyword search would miss too much nuance at this sample size.
Step 2 — D/A/U coding (for Figure 2)
Section titled “Step 2 — D/A/U coding (for Figure 2)”Applied only to extracts that concern interoceptive content (internal-state sensing: effort, load, breath, fatigue, discomfort, emotional state in the body — not muscle location or spatial configuration).
Two approaches under consideration:
- Option A (preferred for rigor): Add a component column (IA / LPA / SA) alongside D/A/U, code all extracts, then filter to IA for the interoceptive analysis. Keeps the full dataset usable for future LPA/SA analysis.
- Option B (faster): Code D/A/U at participant level for the whole testimonial, accepting that most D statements in free-text accounts are interoceptive. Defensible given that LPA and SA language is more technical and less likely to appear spontaneously.
Decision pending. For poster purposes, Option B is probably sufficient with appropriate caveats.
D/A/U boundary notes for coding:
- If the person reports noticing a body signal → D
- If the person reports a changed relationship to body signals (trust, acceptance, not avoiding, not catastrophizing, internal vs. external authority) → A
- If the person reports using body signals to make a decision or change behavior → U
- Many statements will code as D+U (discrimination enabling use); pure A is likely rare in this dataset — itself a finding
Step 3 — Spreadsheet structure
Section titled “Step 3 — Spreadsheet structure”Participant | Source | Training context | Program stage | Full text |[8 outcome domain columns] |D | A | U | Notes- Source: testimonial / segment-feedback
- Training context: public classes / private sessions / primer
- Program stage: indicates early vs. established training
Planned Figures
Section titled “Planned Figures”The poster covers all three body-representation components (IA, SA, LPA). The INT-D-A-U dataset contributes the IA-specific evidence (Figures 1 and 3); Figures 4 and 5 reuse data from earlier presentations. Figure ordering reflects the audience — IA is more familiar to a body representation / interoception audience and leads; SA and LPA follow as the novel contributions.
Figure 1 — Conceptual framework (3×3 matrix)
Section titled “Figure 1 — Conceptual framework (3×3 matrix)”Rows: LPA, SA, IA. Columns: D (Discrimination), A (Appraisal), U (Use/Skill). Each D cell names the Baseworks training principles that directly target it: DA/MM → LPA-D; GS/FSA → SA-D; NB/IM → IA-D. A and U cells are marked as emergent — not directly instructed, reported spontaneously. Optional bottom row maps components to existing body representation vocabulary (body schema, body image).
The core conceptual claim: D is trained; A and U are parallel emergent pathways from D, not sequential stages.
Figure 2 — Questionnaire item mapping
Section titled “Figure 2 — Questionnaire item mapping”Item-level coverage of MAIA-2, BAQ, BPQ, and IAS coded against INT-D / INT-A / INT-U / PRO / SPA / GEN. Draft figure exists (BRNET-sample map.png on desktop). Bar chart shows that MAIA and BAQ are nearly entirely INT; BPQ and IAS mostly GEN with minimal INT; PRO and SPA columns are zero or near-zero across all tools. Item-level grids for MAIA and BAQ show the coupling of INT-D and INT-U within individual items — distinguishing items rarely separate D from U within a single question. INT-A items are sparse.
The argument: existing tools measure interoceptive awareness but have no vocabulary for LPA or SA as trainable sensory capacities.
Figure 3 — IA: NCA results + INT outcomes (two panels)
Section titled “Figure 3 — IA: NCA results + INT outcomes (two panels)”Panel a — D/A/U pattern distribution and NCA: The D|A|U pattern frequency table from dau-nca (8 patterns, N=39; 38/39 consistent with D as prerequisite). Alongside or below: the NCA results for H1 and H2 in a compact two-row table (cases tested, violations, consistency, p-value, verdict). H3 presented as a one-sentence result below the table: “Conditional on D=1, A and U are statistically independent (Fisher’s p = 0.65) — they are distinct emergent outcomes, not a single construct.”
Panel b — INT outcomes by population: Horizontal bar chart. Three overlapping populations: Fitness (n=9), D→A (n=16), D→U (n=26). Seven outcome domains (sleep removed; too low to be informative). Ordered to emphasize the sharpest contrasts: emotional regulation is the clearest differentiator between D→A and D→U populations; intensity management / knowing limits and daily life behavior change show the starkest Fitness vs. D/A/U separation.
Note on population framing: D→A and D→U are overlapping sets (participants with both A=1 and U=1 appear in both bars), not mutually exclusive groups. This is justified by the A–U independence finding and is labeled accordingly on the figure.
N|Group|Stress reduction / relaxation|Intensity management / knowing limits|Emotional regulation|Body awareness|Daily life behavior change|Pain / tension reduction|Nervous system / Energy regulation|Sleep 9|Fitness|44%|33%|0%|33%|44%|44%|11%|0% 16|D→A|50%|81%|50%|100%|69%|63%|31%|13% 26|D→U|50%|96%|27%|100%|88%|73%|27%|8%
Figure 4 — SA-D: Star-Tilt (reuse from Neuro 2026)
Section titled “Figure 4 — SA-D: Star-Tilt (reuse from Neuro 2026)”Two sub-panels: (a) trained vs. untrained cross-sectional comparison using shear angle metric — large effect sizes confirming systematic failure in untrained individuals; (b) longitudinal Tilt performance over 7 sessions (8.3% → 36.4% correct), showing slow but consistent improvement and confirming that SA-D is acquirable but requires dedicated training not provided by standard movement contexts.
Single task used across both panels for coherence. Source data: Neuro 2026 Abstract.
Figure 5 — LPA-D: Muscle sensations survey (reuse from Neuro 2025)
Section titled “Figure 5 — LPA-D: Muscle sensations survey (reuse from Neuro 2025)”Individual variability in awareness of muscular sensations at rest and during exercise (n=36 non-practitioners). Key findings: ~50% report no awareness at rest; high inter-individual variance; athletic activity is not a predictor of resting awareness. Confirms that LPA-D is systematically undertrained and not captured by conventional exercise exposure.
Source data: Neuro 2025 Abstract.
Tentative / Holding
Section titled “Tentative / Holding”Illustrative quotes (D/A/U boundary examples)
Section titled “Illustrative quotes (D/A/U boundary examples)”Short participant quotes illustrating the D/A/U distinctions may be used as figure captions, sidebar annotations, or verbal explanations during the poster session. Framing is D→{A, U} as parallel pathways — not D→A→U as a sequential chain. The sequential chain framing was dropped following statistical confirmation that A and U are independent.
Candidate quotes:
- D: Chisa — “I have become able to catch the state of discomfort before I feel pain or fatigue”
- A: Yasuko — shift from external rule (“do I have to do this?”) to internal evaluation; restructuring authority relationship with own sensations
- A: Clementine Morrigan — “trusting myself to adapt” instead of avoidance
- U: Cheri — deciding not to push today based on how she feels
- U: Mimi Robert Trudel — using practice tools as a nervous system regulation strategy on the metro
- U: Niki — becoming able to wait instead of reacting with anger
These quotes can anchor the D/A/U boundary definitions in the framework figure (Figure 1) or accompany the NCA panel (Figure 3a).
Dropped analysis approaches (and why)
Section titled “Dropped analysis approaches (and why)”| Approach | Reason dropped |
|---|---|
| Before/after contrast analysis | Promotional register; implicit in thematic frequency data already |
| Intensity language coding | Confounded by personality; not a valid proxy for effect size without personality data |
| Figure 4 heatmap (per-participant coding matrix) | Too dense for poster; may be useful as internal organizing tool before building spreadsheet |
| D→A→U sequential chain framing | Dropped after NCA confirmed A and U are statistically independent (Fisher’s p = 0.65); A and U are parallel emergent pathways from D, not stages |
| Four-group breakdown (Fitness / D→A-only / D→A&U / D→U-only) | n=4 in D→A-only group is too small to interpret; three-group overlapping format carries the same message more reliably |
| Early vs. established group comparison as a standalone figure | n=6 early group too small for reliable comparison; groups combined into N=39 and broken down by D/A/U population instead |
Session Log
Section titled “Session Log”Session 1 — 2026-06-02
Section titled “Session 1 — 2026-06-02”Completed:
- Scanned testimonials directory; confirmed Source breakdown (form submission vs. video) and applied project filter (Training: public classes or private sessions). 11 files excluded (workshop, study group/primer).
- Established 4-pass plan: Pass 1 = form submissions (n=17), Pass 2 = video batch A (n=8), Pass 3 = video batch B (n=8), Pass 4 = segment feedback (n=5 respondents, multi-file).
- Created
[dau-analysis.csv](/dau-analysiscsv/)with full column structure including Mixed column. - Pass 1 complete: All 17 form submission testimonials coded (outcome domains + D/A/U + Mixed).
- Pass 2 complete: Video batch A coded (akiko, althea, cheri, chieko, daria, hirotaka, katharina, krister).
- Pass 3 complete: Video batch B coded (megumi, mirjana, morten, nana, niki, tiffanie, yasuko, yuna).
- All coding decisions reviewed and confirmed with Asia. Key corrections applied during review: Lin NS/energy corrected to 0 (metaphorical language); Morten A corrected to 1 (attitudinal shift re: beginner comfort, confirmed for competitive athlete).
Pending:
- Figure construction (Figures 1, 2, 3) — planned for Session 3.
Session 2 — 2026-06-02
Section titled “Session 2 — 2026-06-02”Pass 4 — initial (5 respondents, original files):
- Coded all 5 original segment-feedback respondents: any-guay, clementine-morrigan, luchida-michel, magali-lalonde, mimi-robert-trudel.
- Discovered Magali Lalonde s04 exists in the directory (project outline had listed s02/s03/s05 only); s04 content included — pain/tension=1 added from postural correction.
Dataset expansion:
- Added Magali Lalonde testimonial (magali-fr-en) as a filter exception (Training: study group/primer; added by Asia’s explicit request). Removed from excluded list.
- Added Nathalie Dore as 6th segment-feedback respondent (s02–s08).
- Extended existing respondents to their full file ranges: Any Guay s02–s10, Luchida Michel s02–s10, Mimi Robert Trudel s02–s10, Magali Lalonde s02–s06.
- Magali’s two rows (testimonial + segment-feedback) merged into one combined row with
Source: testimonial — video; segment-feedback. Counted once in total N. - Total dataset: N=39 (33 testimonials from original filter + 6 segment-feedback respondents; Magali counted once via combined row).
Coding updates from extended data:
- Luchida Michel NS/energy: 0→1 (s07: “I stopped and went back to standing with open palms, to regulate my nervous system. It worked”)
- Luchida Michel A: 0→1 (s08: shift from performance-focus to observation; “it’s ok if I don’t go deep as long as I do some movement with the capacity I have — NOW I UNDERSTAND”)
- Mimi Robert Trudel A: 0→1 (s08: “peut-être pas. Ce n’est pas grave. J’ai marché lentement et c’est parfait comme ça. Un autre jour, les choses seront différentes” — acceptance of current capacity)
- Magali D/A/U: updated to D=1, U=1, A=0, Mixed=1 — per Asia’s clarification that D/A/U applies across all three body-representation domains (not only interoceptive). Magali’s D and U are spatial/proprioceptive domain; Mixed=1 flags this for Figure 2 interpretation.
Conceptual clarification recorded:
- D/A/U dimensions apply across all three body-representation domains (interoceptive awareness, LPA, SA). The D/A/U columns in this dataset reflect whichever domain the evidence comes from. Mixed=1 flags cases where the evidence is spatial/proprioceptive rather than interoceptive, or where both are present — requiring a domain qualifier for Figure 2.
Flags resolved (end of Session 2):
- Any Guay U=1 — confirmed. Interoceptive skill transfer present; retrospective signal counts.
- Luchida Michel Mixed=1 — A=1 and U=1 confirmed as interoceptive (accepting not having to push hard; intensity modification in daily practices). Spatial domain struggles noted but not the relevant evidence base. Mixed=1 retained as Figure 2 domain qualifier.
- Nathalie Dore A=0 — confirmed. Stable baseline as a coach; no attitudinal issues requiring change. D=1 and U=1 confirmed interoceptive.
All coding decisions finalised. Dataset complete at N=39.
Session 3 — 2026-06-03
Section titled “Session 3 — 2026-06-03”D/A/U prerequisite analysis complete:
- Asia analyzed coded data and formulated the prerequisite hypothesis: D is a necessary condition for both A and U; A and U are independent.
- Identified one contradicting case (0|1|0, Morten) as a likely ascertainment artifact of the free-response interview format.
- Statistical methods selected: Necessary Condition Analysis (Dul 2016) for H1/H2; Fisher’s exact test for A–U independence (H3).
- Python script written (
dau-nca.py) and run on full N=39 dataset. - Results: H1 consistency = 0.941 (p=0.095), H2 consistency = 1.000 (p<0.001), H3 Fisher’s p=0.654 — all three hypotheses supported.
- Created dau-nca with full methods, results, limitations, and embedded script.
Figure strategy finalized (Session 3 continued):
- Poster figure set revised to 5 figures covering all three body-representation components.
- Figure order: Framework (Fig 1) → Questionnaire gap (Fig 2) → IA/NCA+INT outcomes (Fig 3) → SA-D Star-Tilt (Fig 4) → LPA-D muscle sensations (Fig 5).
- IA figures lead (Fig 3) because IA is more familiar to the BRNet audience; SA and LPA follow as novel contributions.
- Illustrative vignettes (Katharina, Mimi Robert Trudel) moved to Tentative/Holding section — may be used as figure annotations or verbal examples during poster session.
- D→A→U sequential chain framing removed throughout; replaced with D→{A, U} parallel pathways, consistent with statistical findings.
- Questionnaire figure (Fig 2) confirmed from existing draft (BRNET-sample map.png); item-level grids show INT-D/INT-U coupling within questionnaire items, reinforcing conceptual gap.
Next:
- Figure assembly — Asia has data in spreadsheet; figures need layout/formatting for poster.
- Poster print: print shop in Italy identified; timeline is tight (conference June 8–9).
Analysis Passes — File Breakdown
Section titled “Analysis Passes — File Breakdown”Scanning the testimonials directory confirmed the following Source breakdown. The project filter (Training: public classes OR private sessions) excludes 11 files with Training: workshop or study group, primer.
Source 1 — Testimonials (N=33 in scope)
Section titled “Source 1 — Testimonials (N=33 in scope)”Form submission — 17 participants (Pass 1): anna-jp-en, chikara-jp-en, chisa-jp-en, hisako-jp-en, jiletta-en-en, kangying-en-en, kazuko-en-en, lin-en-en, ling-en-en, lionel-en-en, mari-jp-en, mayumi-jp-en, miki-jp-en, risa-jp-en, satomi-jp-en, stacey-en-en, yuka-jp-en
Video — 16 participants (Passes 2 & 3, ~8 each):
- Batch A: akiko-jp-en, althea-en-en, cheri-en-en, chieko-jp-en, daria-en-en, hirotaka-jp-en, katharina-en-en, krister-en-en
- Batch B: megumi-jp-en, mirjana-en-en, morten-en-en, nana-jp-en, niki-jp-en, tiffanie-en-en, yasuko-jp-en, yuna-jp-en
Excluded (wrong training type) — 10 files: andrea-sp-en, angelica-sp-en, cesar-sp-en, guillermo-sp-en, gustavo-sp-en, guylaine-fr-en, ibalu-sp-en, luz-sp-en, nicole-sp-en, sofia-sp-en (magali-fr-en moved to included — see filter exception note above)
Source 2 — Segment Feedback (N=6 respondents, Pass 4)
Section titled “Source 2 — Segment Feedback (N=6 respondents, Pass 4)”6 respondents × 3–5 files each. Each respondent aggregated across files before coding a single row. See Segment Feedback.
Output file
Section titled “Output file”[dau-analysis.csv](/dau-analysiscsv/) — located in this directory. Built in passes; each pass fills rows for the relevant participant group.