{"id":"6b8d700f-5b5c-4840-865e-6065a1180147","arxiv_id":"2507.02453","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Heart rate-driven haptic vibration reduced heart rate and increased subjective restfulness during wakeful rest, with forearm and shoulder placements rated more comfortable and relaxing than the wrist.","lead":"This study tested whether a vibrating wearable that pulses slightly slower than the user's heartbeat can support relaxation during eyes-closed rest. It found that the forearm and shoulder felt most restful and comfortable, while the wrist was easier to notice but less calming.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Without a fixed-rate or sham vibration control, the study cannot separate heart-rate-synchronized biofeedback from mere tactile stimulation; the central attribution of the small HR reduction and higher restfulness to 'biofeedback' is not established.","rationale":"I read the paper as an exploratory within-subject study of body location for vibrotactile relaxation feedback during wakeful rest. The central claim in Section 6 is that biofeedback delivered to the forearm and shoulder effectively reduced heart rate and enhanced subjective restfulness. The reader's weakest_assumption is the adequacy of the no-vibration control, and I agree that this is the most load-bearing concern: the study's unique contribution is the real-time HR-synchronized pacing, but the control condition cannot rule out that any tactile stimulus, or the expectation of a calming device, drives the small HR reductions and higher restfulness scores. The location-specific findings (forearm and shoulder preferred, wrist more recognizable) are better supported because they are within-vibration-condition comparisons, but the biofeedback-specific conclusion remains unverified. I also noticed that Eq. (1) in Section 2.1.2 appears inconsistent with the stated 1.4 s interval and the 30%-below-HR rationale; this is a secondary but concrete internal issue that should be corrected or clarified, as it affects the characterization of the intervention. A fixed-rate vibration control is the direct experimental check that would settle whether the closed-loop adaptation is the active ingredient. Since the paper already earns a CONDITIONAL verdict on these grounds, my stress-test does not shift the reader's assessment, hence UNCHANGED.","tokens_in":13020,"tokens_out":8292,"duration_ms":96216,"concrete_test":"Run a within-subject follow-up (n≥20) at the forearm with four conditions: HR-synchronized biofeedback, fixed-rate vibration matched to the mean biofeedback interval, no vibration, and non-rhythmic sham vibration. If the HR-synchronized condition does not significantly outperform fixed-rate vibration on HR reduction or Stanford Sleepiness Scale ratings, the active ingredient is not the closed-loop signal and the central claim should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that real-time HR-synchronized vibration ('biofeedback') lowers heart rate and increases subjective restfulness relative to no vibration. Every comparison in Section 3.1 uses the no-vibration condition as the baseline, so the design cannot distinguish the closed-loop signal from a non-informative tactile stimulus. A fixed-rate vibration at the same pulse interval, or a sham with unlinked intervals, would be needed to show that the HR-synchronized slowing—the active ingredient named in Section 2.1.2—is what matters. Without it, the observed ~1–2 BPM reductions and higher Stanford Sleepiness Scale ratings are equally compatible with any rhythmic touch, or with demand characteristics. This threatens the biofeedback-specific conclusion in Section 6, though the location-preference results (forearm/shoulder) are more robust. A secondary internal inconsistency compounds the problem: Eq. (1) as printed multiplies by (1−0.30), giving T_vib ≈ 0.65 s at the reported mean HR of 71.4 BPM (≈92 pulses/min), not the stated ≈1.4 s interval nor a 30%-below-HR rhythm; either the equation or the implementation must be corrected before the intervention can be characterized as slow closed-loop biofeedback.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a within-subject study (N=20) comparing heart-rate-driven vibrotactile biofeedback during eyes-closed wakeful rest at four body locations (wrist, hand, forearm, shoulder) against a no-vibration control. Heart rate, ear-EEG alpha power, Stanford Sleepiness Scale restfulness, vibration experience ratings, and preference rankings were collected. The authors report significant heart rate reductions at the wrist, shoulder, and forearm relative to no vibration, no significant alpha-power changes, higher subjective restfulness at the shoulder and forearm, and a preference for the forearm and shoulder. The central claim is that heart-rate-synchronized haptic biofeedback at these locations effectively promotes relaxation, with location-specific effects on subjective experience.","tokens_in":13243,"tokens_out":6792,"duration_ms":74078,"significance":"If the claims held, the study would provide useful design guidance for wearable relaxation feedback, particularly the comparison of four body locations under a standardized wakeful-rest protocol. Strengths include the within-subject Latin-square design, use of a standardized sleepiness scale, real-time closed-loop stimulation based on physiological input, and careful reporting of effect sizes. The location-preference results (forearm and shoulder favored over wrist and hand) are a plausible and potentially actionable contribution. However, the central causal attribution to heart-rate-synchronized biofeedback, as opposed to mere tactile stimulation or expectation effects, is not supported by the experimental design, and the reported vibration-timing formula contains an internal inconsistency. These issues are load-bearing for the abstract and conclusion claims, which currently overstate what the data can establish.","major_comments":[{"comment":"The only control condition is 'without vibration'; there is no sham or fixed-rate vibration condition matched in pulse timing. As a result, the comparisons in Section 3.1 (wrist, shoulder, and forearm vs. no-vibration control) cannot separate the effect of heart-rate-synchronized biofeedback from the effect of any rhythmic tactile stimulation or from demand characteristics. This is load-bearing for the Section 6 conclusion that 'biofeedback delivered to the forearm and shoulder effectively reduced heart rate'; at most, the data support the claim that 'vibrotactile stimulation at these sites reduced heart rate.' The authors should either add a sham or fixed-rate condition (if feasible) or explicitly restrict all causal claims about the closed-loop adaptation, and revise the abstract and Section 6 accordingly.","section":"Section 2.1 and Section 3.1"},{"comment":"The timing formula as printed is not dimensionally consistent: it reads 'T_vib = 60 max(40, min(HR, 65))·(1−0.30)' with no division, and no natural interpretation of it yields the stated example of 'approximately 1.4 seconds' for a mean HR of 71.4 BPM (for instance, 60/(clamped HR · 0.7) gives 1.32 s, while (60/clamped HR) · 0.7 gives 0.65 s). Please provide the exact formula with unambiguous parentheses and verify the reported example against the actual implementation; this matters because the intervention is defined by its pulse-timing rule.","section":"Section 2.1.2, Eq. (1)"},{"comment":"Each of these sections reports multiple post hoc Wilcoxon signed-rank tests (10 for heart rate, 10 for restfulness, 6 per vibration-experience scale, and 10 for preference) with no correction for multiple comparisons at alpha = .05. Under these conditions, several of the reported p-values in the .01–.05 range would not survive a familywise correction, and the binary 'significant/not significant' language is therefore fragile. The authors should report adjusted p-values or clearly label these analyses as exploratory, especially for the less-strongly-hypothesized comparisons among body locations.","section":"Sections 3.1, 3.4, 3.5, and 3.6"},{"comment":"The statistically significant heart rate reductions are small in absolute terms (approximately 1–2 BPM, e.g., no-vibration 72.99 BPM vs. forearm 70.72 BPM). The manuscript does not discuss whether changes of this size are practically meaningful for relaxation or well-being, nor does it reconcile the large rank-biserial effect sizes with the small raw differences. A brief discussion of clinical or experiential significance is needed before concluding that the forearm and shoulder are 'ideal' locations for relaxation feedback.","section":"Section 3.1 and Table 1"}],"minor_comments":[{"comment":"There are formatting errors: 'reached significance.Recognizability' is missing a space, and an isolated 'M=3.75' appears after Figure 6.","section":"Section 3.5"},{"comment":"The caption reads '((A) Mean restfulness ratings...' with a doubled opening parenthesis; this should be corrected.","section":"Figure 5 caption"},{"comment":"The exclusion of participant P3 from EEG analyses is reported, but Table 1 does not state whether the alpha PSD values are based on N=19. Please clarify the sample size for each reported analysis.","section":"Section 3.2"},{"comment":"The paper does not report a power analysis or sample-size justification for N=20. A brief statement would help the reader assess the null results (e.g., alpha power, hand heart rate).","section":"Section 2.1"},{"comment":"The reference to 'Azevedo et al. [53]' is correct in content, but the citation numbering places this reference late in the list, making it less convenient to locate; consider reordering or adding a cross-reference to the bibliographic entry.","section":"Section 2.1.2"}],"recommendation":"major_revision","confidential_remarks":"The paper's more defensible contribution is the systematic location comparison for vibrotactile stimulation; the biofeedback-specific causal claim is not supported by the current design. A revision that reframes the conclusions and corrects the technical inconsistencies would make the manuscript suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the haptic biofeedback paper. The headline finding—that HR-synchronized vibration lowers heart rate and improves restfulness—is not actually established by the data as presented. The design compares each location against a no-vibration control, so any rhythmic tactile stimulation, or participant expectation, could explain the ~1–2 BPM drops. The location-preference results (forearm/shoulder over wrist/hand) are more convincing, because they rest on subjective ratings that are internally consistent and align with preference rankings. That part is worth taking seriously.\n\nWhat's genuinely new: the combination of dynamic HR-based vibrotactile feedback during eyes-closed wakeful rest, with a four-location comparison and ear-EEG alpha. The system is described in enough detail to reproduce, and the authors are appropriately cautious about the null alpha result.\n\nThe soft spots are real. The missing sham/fixed-rate condition means the central attribution to \"biofeedback\" is unsupported. Multiple Wilcoxon comparisons without correction inflate the risk of false positives. And there's a concrete internal inconsistency: Eq. (1) as printed multiplies by (1−0.30), which at the clamped 65 BPM gives a cycle of ~0.65 s, not the stated ~1.4 s, and it speeds the vibration up rather than slowing it 30%. Either the equation or the implementation description is wrong. The authors need to fix that before we can trust the intervention is what they say it is.\n\nWho this is for: researchers working on wearable haptics for relaxation or well-being. The location-comparison data will be useful to them, even if the physiological effect remains unproven. I'd want a serious referee to see it, mostly to push the authors on the control condition and the equation. With those fixed, the paper could be a decent contribution.\n\nRecommendation: send it to peer review, but expect revisions.","headline":"A plausible location-comparison study whose core biofeedback claim is undercut by a missing control condition and a misprinted timing equation.","tokens_in":13775,"tokens_out":2486,"would_cite":false,"duration_ms":26672,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Heartbeat-paced vibration at the forearm or shoulder lowers heart rate and increases restfulness during quiet rest, while the wrist feels more noticeable but less calming.","keywords":["Haptics and Actuators","Body Placement in Wearables","Vibrotactile Biofeedback","Relaxation Interventions","Heart Rate Modulation","Alpha Wave Activity","Wakeful Rest"],"falsifier":"Run the same five-minute rest protocol with three additional arms—closed-loop vibration at 30% below live heart rate, fixed-rate vibration at the same average interval, and no vibration—and compare heart-rate change and Stanford Sleepiness Scale ratings; if the fixed-rate arm matches the closed-loop arm, the claim that real-time heart-rate coupling is the active ingredient is falsified.","tokens_in":12807,"feed_emoji":"🧘","tokens_out":6601,"duration_ms":67000,"temperature":0.7,"pith_summary":"This paper asks whether where on the body a wearable delivers heartbeat-paced vibration changes how well it relaxes a person resting with eyes closed. Across twenty participants, slow vibration pulses timed to 30% below the wearer's real-time heart rate lowered heart rate compared to no vibration at the wrist, shoulder, and forearm, while the hand did not produce a reliable effect. Location did not change the size of the heart-rate effect among the three responsive sites, but it strongly shaped experience: the forearm and shoulder were rated more restful, comfortable, and relaxing, and were preferred, while the wrist was the most noticeable but least calming. The paper concludes that forearm and shoulder are the better locations for unobtrusive relaxation biofeedback, and that wrist-worn designs need improvement for restful use.","feed_headline":"Vibration paced to heartbeat works best on forearm and shoulder","feed_subtitle":"In a 20-person rest study, these spots lowered heart rate and scored highest on comfort, relaxation, and preference.","key_machinery":"The heart of the system is a closed-loop pacing loop. Every second the controller takes the current heart rate $HR$, clamps it to 40–65 BPM, and sets the vibration-pulse interval to $T_{\\mathrm{vib}} = \\frac{60}{\\max(40,\\min(HR,65))}(1-0.30)$, so the tactile pulses arrive 30% slower than the real-time heartbeat. The study tests this same loop at four body locations—shoulder, forearm, wrist, and hand—against a no-vibration control, with chest-strap heart rate, in-ear EEG $\\alpha$ power, and Stanford Sleepiness Scale ratings as the outcomes.","core_discovery":"The paper's central claim is that heart-rate-synchronized haptic feedback can shift people into a deeper rest during eyes-closed wakeful rest, and that the body location carrying the vibration matters mainly for how the intervention feels rather than for how much the heart rate drops. Wrist, shoulder, and forearm stimulation all lowered heart rate significantly relative to no vibration, with the forearm showing the largest effect size, and hand stimulation showing no significant reduction. The forearm and shoulder produced the highest subjective restfulness, comfort, and preference, while the wrist was the most recognizable but least relaxing, suggesting designers should put unobtrusive relaxation feedback on the forearm or shoulder and reserve the wrist for salient alerts. Alpha power at the ear did not change, and physiological and subjective measures were only weakly linked, so the paper treats body placement as a key design decision while treating EEG alpha as insensitive to this short intervention.","pith_inferences":["Because there was no sham condition, this design cannot separate the closed-loop slowing from the mere presence of rhythmic vibration or from participants' expectations; a fixed-rate or random vibration arm would isolate the active ingredient.","The hand's null result suggests a perceptual or neural threshold—possibly related to tactile sensitivity or C-tactile afferent density—that could be tested by varying vibration amplitude or pulse duration at the same location.","If replicated, the forearm and shoulder advantage would motivate adaptive placement in real-world wearables and could be extended to outcomes such as sleep-onset latency or heart-rate-variability coherence.","Individual traits like interoceptive awareness and tactile sensitivity, which the paper names as future work, might explain the large effect sizes and between-subject variability seen here, suggesting personalized placement recommendations."],"forward_implications":["Wearable relaxation devices intended for quiet, eyes-closed rest should place the actuator on the forearm or shoulder rather than the wrist.","For the wrist to serve rest scenarios, designers must address comfort and recognizability, since it reduced heart rate but scored poorly on relaxation and comfort.","Ear-based alpha power is not a sensitive marker for this intervention, so EEG relaxation signatures will likely require richer haptic parameters or multimodal stimulation.","Subjective restfulness should be measured alongside physiology, because the two were only weakly correlated; a device may calm the body without participants feeling sleepier, or vice versa.","Body placement should be matched to the intervention goal: forearm and shoulder for calmness, wrist for clearly noticeable feedback."],"supporting_citations":[{"why":"Supplies the 20–30% below-baseline pacing rule and the smartwatch implementation that motivates this study's 30% slowed vibration timing.","marker":"[9]"},{"why":"Shows that vibration paced slightly below resting heart rate reduces electrodermal activity and anxiety, the empirical basis for slowing the pulse.","marker":"[53]"},{"why":"Prior comparison of haptic heartbeat feedback across four body placements that this study extends to a restful, non-stress context.","marker":"[55]"},{"why":"Provides the wearable placement rationale (hand, wrist, arm) used to select the four stimulation sites.","marker":"[63]"},{"why":"Defines the Stanford Sleepiness Scale used as the study's subjective restfulness outcome.","marker":"[48]"},{"why":"Describes the hardware platform used for ear-based EEG acquisition and vibration-motor control.","marker":"[29]"}],"fun_headline_variants":["Forearm and shoulder best for heartbeat-paced rest vibes","Heartbeat haptics: location matters, forearm and shoulder win","Skip the wrist for relaxing vibration feedback","Restful haptics: forearm and shoulder beat wrist","Where to feel restful heartbeat vibes: not the wrist"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The study assumes that the no-vibration condition is a complete baseline, so that the heart-rate reductions and restfulness increases seen in the vibration conditions are caused by the biofeedback loop rather than by the mere presence of tactile stimulation or by participants' expectations.","fun_headline_variants_meta":{"raw":{"variants":["Forearm and shoulder best for heartbeat-paced rest vibes","Heartbeat haptics: location matters, forearm and shoulder win","Skip the wrist for relaxing vibration feedback","Restful haptics: forearm and shoulder beat wrist","Where to feel restful heartbeat vibes: not the wrist"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000187,"raw_usage":{"total_tokens":1316,"prompt_tokens":920,"completion_tokens":396,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":536,"completion_tokens_details":{"reasoning_tokens":316}},"tokens_in":536,"tokens_out":396,"duration_ms":6478,"temperature":1.0,"reasoning_tokens":316,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:28:36.738414+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same five-minute rest protocol with three additional arms—closed-loop vibration at 30% below live heart rate, fixed-rate vibration at the same average interval, and no vibration—and compare heart-rate change and Stanford Sleepiness Scale ratings; if the fixed-rate arm matches the closed-loop arm, the claim that real-time heart-rate coupling is the active ingredient is falsified.","supporting_citations":[{"cited_title":"Jung, and Tanzeem Choudhury","cited_arxiv_id":null,"evidence_quote":"Supplies the 20–30% below-baseline pacing rule and the smartwatch implementation that motivates this study's 30% slowed vibration timing."},{"cited_title":"Azevedo, Nell Bennett, Andreas Bilicki, Jack Hooper, Fotini Markopoulou, and Manos Tsakiris","cited_arxiv_id":null,"evidence_quote":"Shows that vibration paced slightly below resting heart rate reduces electrodermal activity and anxiety, the empirical basis for slowing the pulse."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior comparison of haptic heartbeat feedback across four body placements that this study extends to a restful, non-stress context."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the wearable placement rationale (hand, wrist, arm) used to select the four stimulation sites."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Stanford Sleepiness Scale used as the study's subjective restfulness outcome."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the hardware platform used for ear-based EEG acquisition and vibration-motor control."}],"review_version":1}