{"id":"aa817602-11cf-4a68-9bb0-db22e62ed99a","arxiv_id":"2505.00956","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Body-anchored sounds ('audio personas') shift how strangers are perceived: positive sounds increase attractiveness and likability while negative sounds increase threat, though anchoring-dependent effects appeared only in qualitative impressions.","lead":"This paper introduces 'audio personas', body-anchored sounds heard by others through headphones, and tests whether they change social impressions in face-to-face meetings. A preregistered experiment with 64 participants found that people rated someone with positive sounds as more likable and less threatening, but the specific benefit of anchoring sounds to the body was only visible in drawings and writings, not in rating scales.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Body-anchoring effect (H1) is confounded with sound movement: audio personas move with the research assistant while object-anchored controls are stationary, so the observed difference in impression integration may reflect moving-sound novelty rather than body anchoring as such.","rationale":"The reader identified the movement confound as the weakest assumption, and my re-reading of the paper confirms that this is the most load-bearing concern. The headline valence effect—positive audio personas rated as more socially attractive, likable, and less threatening—is supported by preregistered pairwise comparisons within the audio persona group and is not directly undermined by the confound. However, the paper's conceptual contribution is the body-anchored nature of audio personas, and the only direct evidence for that is H1, which compares body-anchored moving sounds with object-anchored static sounds. Because movement is deliberately built into the audio persona condition and deliberately absent from the control condition, the observed difference in impression integration cannot be attributed specifically to body anchoring. The lack of a significant anchoring-by-valence interaction further means the self-report valence effects are not shown to depend on anchoring. These are correctable issues, not fatal flaws: the preregistration, counterbalancing, and neutral-behavior controls are commendable, and the valence effect itself is plausible and consistent with prior work. The appropriate verdict remains conditional acceptance: the authors should add a moving-object control or otherwise deconfound movement from body anchoring before claiming that body anchoring is the mechanism. I therefore agree with the reader's CONDITIONAL verdict and see no reason to change it.","tokens_in":29507,"tokens_out":5142,"duration_ms":53107,"concrete_test":"Run a follow-up experiment with three conditions: (a) body-anchored audio persona (moving with the assistant), (b) static object-anchored audio (as in the current control), and (c) object-anchored audio that moves with the assistant (e.g., a portable speaker carried by the assistant or a remotely controlled platform that follows the assistant at a fixed offset). Keep all other procedures identical. Compare the proportion of participants who reference sounds in impression drawings/writings using a chi-square or logistic regression with planned contrasts. If condition (a) significantly exceeds condition (c), the body-anchoring interpretation of H1 is supported; if (a) and (c) are comparable and both exceed (b), the current H1 result is explained by movement rather than body anchoring.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's main claim has two parts: (1) positive audio personas improve social impressions relative to negative ones (supported by within-group pairwise tests, e.g., social attraction p=0.002), and (2) body anchoring is the key design feature that makes audio personas effective (H1, Finding #1). The evidence for part (2) is the contrast between the audio persona group and the object-anchored control group in how often participants referenced sounds in their impression drawings/writings (62.5% vs. 18.75%, p=0.005/p=0.002). However, Section 4.1.1 states that in the audio persona group, sounds 'shifted in locations corresponding to their movements,' whereas in the control group, sounds were anchored to static speakers and 'the research assistants' movements would not affect the sound.' This means the comparison varies not only body anchoring but also whether the sound moves with the person. A moving sound is more likely to be perceptually bound to the moving person regardless of whether it is anchored to the body per se, so H1 could be explained by dynamic spatial coincidence rather than body anchoring. This confound is load-bearing because Section 4.13 explicitly concludes that 'this contrast highlights body-anchoring as a key design feature that enables audio personas to influence social perception.' The self-report measures do not rescue the anchoring claim: no significant anchoring-by-valence interaction was found for any social impression measure (e.g., social attraction F(1,62)=1.0, p=0.32; threat F(1,62)=1.56, p=0.22). Thus the central theoretical distinction—body-anchored versus merely co-occurring audio—is not cleanly established, even though the valence effect within the audio persona group remains credible.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the concept of 'Audio Personas'—body-anchored sounds rendered in audio augmented reality—and evaluates it through a proof-of-concept system and two empirical studies. Study #1 (n = 64, preregistered, mixed-factorial) compares an audio-persona condition (sounds anchored to research assistants) with an object-anchored control condition (sounds anchored to fixed speakers), crossing positive/negative audio valence within participants. The primary findings are that participants in the persona group were more likely to include sound in their impression drawings/writings (H1), and that within the persona group, positive personas increased social attraction, likability, and reduced perceived threat relative to negative personas (H2). The anchoring-by-valence interaction was not significant on any self-report measure, and no effects were found on perceived emotional states or behavioral measures. Study #2 (n = 8 audio designers) explores use cases and design patterns, reporting that audio personas are preferred in public and semi-public-private spaces for impression management and state signaling. The central claim is that audio personas can shape social perception and that body-anchoring is a key design feature for this effect.","tokens_in":29773,"tokens_out":6258,"duration_ms":64298,"significance":"If the claims hold, this paper opens a novel design space for augmenting face-to-face social interactions through body-anchored audio, connecting audio AR with impression management. The work has tangible strengths: the study is preregistered, uses a controlled mixed-factorial design with counterbalancing, and combines self-report, behavioral, and open-ended drawing/writing measures. The proof-of-concept system is a concrete technical contribution, and the design study provides convergent qualitative insights. However, the strongest causal claim—that body-anchoring, specifically, enables the effect—is undercut by a confound in the H1 contrast and by the absence of a significant anchoring-by-valence interaction on the self-report measures. The paper contributes a useful exploratory prototype and a set of findings that are promising but not conclusive regarding the role of body-anchoring per se. For the field, the concept itself is interesting enough to warrant further investigation, but the current evidence does not fully support the abstract's strong framing.","major_comments":[{"comment":"","section":"§4.1.1, §4.13"},{"comment":"This comment covers the overinterpretation of non-significant interaction and exploratory pairwise tests.","section":"§4.10, §6.1"},{"comment":"","section":"§4.2.3"}],"minor_comments":[{"comment":"","section":"§4.5.1"},{"comment":"","section":"§4.1.2"},{"comment":"","section":"§4.9"},{"comment":"","section":"§6.1"},{"comment":"","section":"§1"}],"recommendation":"major_revision","confidential_remarks":"This is a promising work with a well-executed system and a preregistered study, but the central claim about body-anchoring being the key mechanism is not fully supported by the current design. The H1 contrast is confounded with sound movement, and the familiarization asymmetry adds further ambiguity. I recommend major revision rather than rejection because the concept is valuable and the issues are addressable through additional analyses or a carefully designed follow-up. The authors should also be mindful of the strong framing in the abstract relative to the non-significant interaction; reframing the findings as exploratory would improve the paper's inferential honesty."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Good paper to know about. The concept—body-anchored audio cues as a dynamic social signal in face-to-face interaction—is genuinely new and worth building on. The proof-of-concept system is real, and the preregistered study is carefully designed: mixed-factorial, counterbalanced, standardized RA behavior, and multiple measurement channels. The valence effect inside the audio persona group is credible and consistent across social attraction (p=.002), threat (p=.008), likability (p=.03), and sentiment of open-ended writings. For the central abstract claim—positive audio personas make people seem more attractive, likable, and less threatening—the evidence holds up.\n\nThe soft spot is the body-anchoring-specific claim. The comparison that supports it (H1: more sound references in drawings/writings in the audio persona group vs. the object-anchored control) confounds body anchoring with sound movement: in the persona condition the sound travels with the research assistant, while in the control condition it comes from static speakers. A moving sound is more likely to be perceptually bound to the moving person regardless of anchoring mechanism. The paper's design recommendation that body-anchoring is the key feature rests on that contrast. The self-report measures do not rescue it: there is no anchoring-by-valence interaction on any social impression scale, and the paper is transparent about that. The likability scale also had items dropped post hoc; that is disclosed and reliability is acceptable after the drop, so it is a minor issue.\n\nThe authors are honest about some of this in the discussion, noting the subtlety of the manipulation and the possibility that temporal binding alone elicits similar affect. That honesty is a point in their favor. Still, the abstract and conclusion claim a bit more than the design can cleanly support. A revision should either add a control where the sound moves but is not body-anchored, or explicitly frame the contribution as “body-anchored and mobile sound” rather than body-anchoring per se.\n\nWho this is for: HCI researchers in audio AR, mixed reality, social perception, and wearable expression. It deserves a serious referee and likely publication after conditional revisions. The concept is strong enough that the field benefits from having it out there with the limitation clearly marked.","headline":"A genuinely new concept and a carefully run preregistered study, but the claim that body-anchoring specifically drives the effect is weaker than the abstract suggests because the control condition also varies whether the sound moves.","tokens_in":30380,"tokens_out":1990,"would_cite":true,"duration_ms":20890,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A person's body-anchored audio cues—their audio persona—change how strangers rate their attractiveness, likability, and threat, and the paper shows that anchoring sound to the body, not just playing it nearby, is what makes the cue stick…","keywords":["audio personas","audio augmented reality","social perception","impression formation","body-anchored audio","multisensory integration","spatial audio","sound valence"],"falsifier":"Run the same 64-person study with a third control condition in which the research assistant carries a sound source, such as a handheld speaker, that moves with them but is visibly not part of their body; if participants incorporate the sound into impressions at the same rate as the audio-persona condition, body attachment itself is not what drives the effect.","tokens_in":29302,"feed_emoji":"🔊","tokens_out":6574,"duration_ms":61392,"temperature":0.7,"pith_summary":"The paper introduces audio personas: sounds anchored to a person's body and played to nearby people through headphones, like a wearable sonic outfit. It claims that these body-anchored cues get incorporated into other people's impressions of the wearer, and that the emotional tone of the cue shifts those impressions. In a preregistered lab study with 64 participants, a person wearing a positive audio persona was rated as more socially attractive, more likable, and less threatening than a person wearing a negative one, while object-anchored sounds produced the same trend without reaching significance. A second study with eight audio designers maps out intended uses, with public and semi-public spaces seen as the natural setting for managing impressions and signaling current states.","feed_headline":"Body-anchored sounds shape others' first impressions","feed_subtitle":"In a 64-person test, upbeat body-anchored sounds made people seem more attractive, likable, and less threatening.","key_machinery":"The load-bearing mechanism is body anchoring: an audio source whose spatial position tracks the wearer's body in real time, so the sound moves when the person moves and stays within the wearer's social space. The proof-of-concept uses head-worn trackers, a game engine for 3D audio rendering with distance-based volume rolloff, and a multi-listener audio streaming pipeline so each user both broadcasts a persona and hears others'. This spatial coincidence between sound and person is what, under multisensory cue integration, makes the sound bind to the person rather than the environment.","core_discovery":"The central claim is that body-anchored audio cues—audio personas—change how strangers form impressions of a person in face-to-face interaction. The paper argues that spatially anchoring a sound to the body aligns the auditory cue with the person, making the sound available information about that person, whereas the same sound anchored to an object stays background. The supporting evidence is that 62.5% of participants in the audio-persona condition included the sound in impression drawings or writings, versus 18.75% in the object-anchored control, and that positive audio personas produced significantly higher social attraction, higher likability, and lower perceived threat than negative audio personas within the audio-persona group. The valence effects replicated as non-significant trends in the control group, which the paper reads as evidence that body anchoring strengthens, rather than creates, the effect.","pith_inferences":["A cleaner test of body anchoring would replace the object-anchored control with a moving sound source carried by the research assistant; if incorporation rates stay high, motion rather than body attachment would be doing the work.","Because valence effects reached significance only in the body-anchored group, audio personas may matter most in multi-person settings, where an object-anchored sound cannot be attributed to any one person; the paper did not test that case.","The null behavioral results suggest the shift may be confined to explicit ratings rather than approach or avoidance; a social-distance or implicit-association measure could reveal whether the effect reaches behavior.","A shared vocabulary problem follows: if audio personas rely on valence associations, widespread use needs something like an emoji library of sounds, since the same sound can read as positive to one person and negative to another."],"forward_implications":["People can deliberately craft a sonic first impression, adding a dynamic layer to clothing and makeup that can change with mood or context.","The contrast between 62.5% incorporation of body-anchored sounds versus 18.75% for object-anchored sounds indicates that body anchoring itself drives the absorption of the cue into impressions.","The valence findings mean audio personas can work as an implicit signal—buzzing or hurricane sounds read as threatening, brook or forest sounds read as warm—without the wearer saying a word.","The prototype shows that the core technical ingredients—head tracking, spatialized audio with distance rolloff, and multi-user audio streaming—are sufficient to deliver body-anchored personas in a lab setting.","Designer feedback points to public and semi-public spaces as the natural context, suggesting early deployments should target encounters among strangers rather than private gatherings."],"supporting_citations":[{"why":"Supplies the spatial-coincidence principle: sensory cues sharing a location are more likely to bind together, which is the theoretical basis for body anchoring.","marker":"[110]"},{"why":"Provides the cue-integration framework the paper uses to argue that a spatially aligned sound becomes available information about the person.","marker":"[135]"},{"why":"Prior result that music valence shifts attractiveness judgments of faces, which the paper's valence effect extends to body-anchored sounds.","marker":"[92]"},{"why":"Shows that background music changes evaluation of crying faces, supporting the claim that audio valence shapes social impressions.","marker":"[52]"},{"why":"Demonstrates crossmodal transfer of emotion from music to face perception, another pillar for the valence hypothesis.","marker":"[88]"},{"why":"Source of the IADS-2 database from which the study's positive and negative sounds were selected by valence rating.","marker":"[15]"},{"why":"The expanded affective auditory stimulus database used for the second set of positive and negative sound stimuli.","marker":"[133]"},{"why":"Shows that emotional valence and sound source location influence interpersonal distance, guiding the 2-meter social-distance trigger and the neutral-behavior protocol.","marker":"[115]"}],"fun_headline_variants":["Body-anchored sound cues shift social judgment in tests","Positive audio personas lift likability, lower threat in person","Audio outfits: sound auras that recolor first impressions","Wearable audio cues: your sound profile alters how others see you"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"In the comparison that supports body anchoring, the body-anchored sound moves with the research assistant while the object-anchored sound stays fixed in the room, so the measured effect could come from the sound's movement or novelty rather than from being attached to the body.","fun_headline_variants_meta":{"raw":{"variants":["Body-anchored sound cues shift social judgment in tests","Positive audio personas lift likability, lower threat in person","Audio outfits: sound auras that recolor first impressions","Wearable audio cues: your sound profile alters how others see you"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000207,"raw_usage":{"total_tokens":1370,"prompt_tokens":886,"completion_tokens":484,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":502,"completion_tokens_details":{"reasoning_tokens":414}},"tokens_in":502,"tokens_out":484,"duration_ms":5933,"temperature":1.0,"reasoning_tokens":414,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:29:54.571471+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same 64-person study with a third control condition in which the research assistant carries a sound source, such as a handheld speaker, that moves with them but is visibly not part of their body; if participants incorporate the sound into impressions at the same rate as the audio-persona condition, body attachment itself is not what drives the effect.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the spatial-coincidence principle: sensory cues sharing a location are more likely to bind together, which is the theoretical basis for body anchoring."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the cue-integration framework the paper uses to argue that a spatially aligned sound becomes available information about the person."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior result that music valence shifts attractiveness judgments of faces, which the paper's valence effect extends to body-anchored sounds."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates crossmodal transfer of emotion from music to face perception, another pillar for the valence hypothesis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The expanded affective auditory stimulus database used for the second set of positive and negative sound stimuli."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that emotional valence and sound source location influence interpersonal distance, guiding the 2-meter social-distance trigger and the neutral-behavior protocol."}],"review_version":1}