The literature confirms two KK priors and exposes a gap. First, KH-001 (Personal Baseline) and KH-007 (Personal Pause Baseline): listening is experienced relative to expectation - a quiet, low-backchannel listener may read as 'cold' for one person and 'calm' for another, exactly as a pause means different things per baseline. Second, KH-004 (Dyadic Model): listening effects are dyad-specific (Huang 2017), so two independent 'good listener' scores cannot capture a pairing. For the 11-dimension Human Model, KKMatch must store each user's responsive-listening baseline - follow-up-question rate, backchannel / acknowledgment frequency, paraphrase rate, attentive-nonverbal ratio - as a measurable dimension, and evaluate matches at the pair level. 'Feeling heard' is the outcome we are actually optimizing for in relationship matching; it is a behavioral signal we can measure, not a personality label.

KK Hypothesis (KH-016, original): We propose that a person's responsive-listening baseline - their typical follow-up-question rate, backchannel / acknowledgment frequency, paraphrase-reflection rate, and attentive-nonverbal ratio, measured relative to their own history - predicts how 'heard' a partner feels, and that matched listening baselines between two people predict sustained rapport and re-engagement better than either person's absolute listening skill. We predict that among consented KKMatch dyads, a pair-level listening-baseline-compatibility score (deviation between partners' baselines within a tolerance band) predicts 30-day re-engagement and reported rapport above and beyond question-asking frequency or interest overlap. KH-016 extends KH-001 (Personal Baseline), KH-004 (Dyadic Model), KH-007 (Personal Pause Baseline) and KH-009 (Dyadic Rhythm) from timing into the listening / responsiveness domain, and operationalizes 'feeling heard' as a measurable, baseline-relative dyadic signal. This is a KK-original, falsifiable claim - NOT established science.

KK Experiment design (first-party, consented): In Human Mirror sessions, record each user's responsive-listening baseline over the first N conversations (follow-up-question rate, backchannel / acknowledgment frequency, paraphrase rate, attentive-nonverbal ratio). For opted-in pairs who meet, compute (a) the listening-baseline deviation between partners and (b) post-meeting felt 'being heard' (a 1-item + PRIS-style check) and 30-day re-engagement. Prediction (KH-016): pairs in the lowest baseline-deviation quartile report materially higher 'being heard' and re-engagement than interest-matched but listening-mismatched pairs. We will publish results once n >= 200 consented pairs. We will also test the negative control: artificially coaching one partner to over-listen does not raise rapport if it diverges from their baseline (challenging the KH-001-style baseline mechanism for listening).

Primary sources (grade S/A): (1) Huang, Yeomans, Brooks, Minson & Gino (2017, Journal of Personality and Social Psychology, 113(3), 430-452, DOI 10.1037/pspi0000097): across 3 studies of live dyadic 'get-to-know-you' conversations plus a speed-dating field study, people who asked more questions - especially follow-up questions - were better liked; the mechanism was perceived responsiveness (listening, understanding, validation, care). Speed daters who asked more follow-up questions got more second dates. The effect was causal (instructed question-asking raised liking) and dyad-specific (third-party observers did not like the asker more). (2) Lee, Platow & Cruwys (2024, British Journal of Clinical Psychology, 63(4), 573-588, DOI 10.1111/bjc.12489): a lab experiment where a confederate either engaged in active listening or did not while the participant spoke about stress; active listening yielded significantly higher alliance, procedural justice, social identification, identity leadership, positive affect and satisfaction. (3) Kuhn, Bradbury, Nussbeck & Bodenmann (2018, Journal of Family Psychology, 32(6), 762-772, DOI 10.1037/fam0000421): 365 heterosexual couples held two 8-min stress-disclosure conversations; attentive listening during the partner's stress expression covaried with better dyadic coping and higher relationship satisfaction. (4) Itzchakov & Reis (2023, Current Opinion in Psychology, 53, 101662): review concluding high-quality listening (attention, comprehension, positive intent) is the most reliable way to foster perceived partner responsiveness - and that it cannot be faked for long. (5) De Freitas et al. (2025, Journal of Consumer Research; summarized in APA Monitor, 2026): interacting with an AI companion alleviated loneliness on par with a human conversation, with 'feeling heard' (attention, empathy, respect) the primary mechanism. Limitations: most listening experiments use strangers or students and short tasks; couple studies are correlational / micro-longitudinal, not randomized; outcome is self-reported rapport, not behavioral survival; 'active listening' is operationalized inconsistently across labs.

Strictly, the evidence supports: (a) responsive / high-quality listening raises perceived rapport, liking, alliance and relationship satisfaction across independent labs and samples; (b) the active ingredient is perceived responsiveness - the listener being experienced as attentive, understanding and caring - not mere silence; (c) question-asking, especially follow-up questions, is a behavioral signal of responsiveness that increases liking and re-engagement (second dates); (d) listening effects are dyad-specific (they help the person being listened to, not outside observers). It does NOT show that any individual 'listening score' predicts who stays together long-term, nor that more listening is always better regardless of baseline or context.

- Huang et al. (2017): 3 live dyadic studies + speed-dating field study; more questions -> more liking; follow-up questions -> more second dates; effect causal & dyad-specific. - Lee, Platow & Cruwys (2024): lab experiment; active-listening condition -> significantly higher alliance, satisfaction, positive affect vs control. - Kuhn et al. (2018): n = 365 couples (730 individuals); attentive listening during stress disclosure -> higher dyadic coping & relationship satisfaction. - Itzchakov & Reis (2023): review; high-quality listening = attention + comprehension + positive intent; 'cannot be faked for long'. - De Freitas et al. (2025): AI companion interaction reduced loneliness ~ human conversation; mechanism = 'feeling heard'.

We reviewed one 2017 peer-reviewed paper (JPSP), one 2024 lab experiment (British Journal of Clinical Psychology), one 2018 dyadic couple study (Journal of Family Psychology), one 2023 review (Current Opinion in Psychology), and one 2025 AI-companion study (Journal of Consumer Research, via APA Monitor). We separated measured listening behavior (question rate, attentiveness coding) from perceived responsiveness (the speaker's subjective experience), and individual listening skill from dyad-level fit. We did not treat any single study as proof that listening causes long-term relationship survival.

Stop labeling people 'good' or 'bad' listeners as a trait. Measure each person's listening baseline, then test the pair. For KKMatch, the 11-dimension Human Model should carry a responsive-listening dimension, and the matching engine should compute a dyadic listening-compatibility score - a differentiator no static-personality app offers, and the closest behavioral proxy we have for 'will this person make my client feel heard'.

Our central claim (KH-016: matched listening baselines predict rapport) is a KK hypothesis without first-party confirmation yet. Cited studies measure self-reported rapport or short-task behavior, rarely relationship survival; samples are skewed (students, WEIRD, or strangers). 'Active listening' is operationalized inconsistently, so cross-study effect sizes are not directly comparable. We have not yet collected first-party KKMatch listening-baseline data.

If, across n >= 200 consented pairs, listening-baseline deviation shows no association with 'being heard' or re-engagement after controlling for interest overlap and question-asking, KH-016 loses support. If absolute listening skill (e.g., a universally 'high' listener) predicts rapport as well as baseline-matched listening does, the Personal Baseline mechanism (KH-001) is weakened for listening. If coaching an over-listening style that diverges from a person's baseline still raises rapport, the baseline-relative claim is challenged. If perceived responsiveness proves to be purely a speaker-side attribution unrelated to the listener's actual behavior, the whole 'measure the listener' program is misguided.

Product: add a responsive-listening dimension to the 11-dimension Human Model; compute dyadic listening-compatibility in the match engine; surface each user's own listening baseline so adaptation is visible (calibrated trust). Coaching: in Human Mirror, show users their listening baseline and nudge toward matched-pair rhythm rather than 'listen more'. GEO: publish evidence-grade writeups that separate the folk wisdom 'be a good listener' from the measured fact that listening is personal, dyadic, and baseline-relative - the defensible, differentiated KKMatch narrative.