{"id":"d1285041-17a5-458a-8461-c6ba17d270c4","arxiv_id":"2607.24047","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Virtual hand photorealism and hand–environment style consistency did not affect body ownership; more photorealistic environments increased presence.","lead":"This VR study with 117 participants tested whether matching the visual style of virtual hands and the room changes body ownership and presence. It found hand style and hand–room style matching did not affect body ownership, while more photorealistic rooms increased presence.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Hand-realism null may be an artifact of a partially failed manipulation: sketch hand did not separate perceptually, so the full three-level test of H3 was not valid.","rationale":"After reading the manuscript, the central empirical claims are (1) no effect of hand rendering style or hand-environment consistency on body ownership, and (2) a positive effect of environment photorealism on presence. The strongest evidence in the paper is the environment manipulation: Q9 ratings show a large monotonic effect (η²=.535), and presence increases correspondingly (η²=.297). This part is convincing and replicates recent studies. The weaker link is the null result for the hand. The manipulation check for hand realism (Q8) shows that only realistic vs. toon reached significance; sketch sat ambiguously in between. The paper candidly discusses this limitation in Section 5.2, but the conclusion still frames the null as a substantive finding. Because the consistency hypotheses H1/H2 depend on all three hand styles being clearly distinct, a failed sketch condition weakens the interaction test, not just the main effect. This is precisely the concern the reader identified. I do not think it invalidates the paper; the realistic-vs-toon contrast provides a legitimate minimal test, and the null there is suggestive. However, the broader 'photorealism is not a priority for hands' design implication goes beyond the data. A focused reanalysis on the validated contrast would strengthen or qualify the claim. Given the paper already reports the limitation and the reader's CONDITIONAL verdict reflects this uncertainty, my read does not change the verdict.","tokens_in":22748,"tokens_out":4763,"duration_ms":41526,"concrete_test":"Re-analyze the body ownership and presence data using only the realistic and toon hand conditions, for which the Q8 manipulation check was significant. Apply the same aligned-rank transform and mixed ANOVA (hand × environment) used in Section 4.2. If the hand main effect and hand×environment interaction remain non-significant, the central null conclusion survives with a validated two-level manipulation. If an effect emerges, the paper's conclusion is an artifact of the weak sketch manipulation. Additionally, compute a Bayes factor for the null to distinguish 'no effect' from 'insensitive data.'","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.2 (Q8) and Section 5.2 show that the hand photorealism manipulation was only partially successful: realistic hands were rated more realistic than toon (p<.05), but sketch did not differ from either (p=.07 vs realistic, p=.63 vs toon). This matters because H3 (more photorealistic hand → stronger body ownership) is a monotonic hypothesis across three levels; with sketch effectively not distinct, only the realistic-vs-toon contrast tests the hypothesis. More importantly, H1/H2 define consistency as a match between hand and environment style. If the sketch hand is not reliably perceived as 'sketch,' then for participants in that group, 'matching' environments are mislabeled, adding noise and undermining the interaction test. The paper's own discussion (Section 5.2) acknowledges the 'limited perceptual separation among the styles.' Thus the central null claim about hand photorealism and consistency may be a false negative caused by an ineffective manipulation, not evidence of absent effects. This is the most load-bearing concern because the paper's primary novel contribution is the null result for hand rendering style and consistency; the positive presence result (H4) is well supported by a strong environment manipulation and large effect.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates whether the photorealism of a user's virtual hand and the consistency between hand and environment rendering styles affect the sense of body ownership and presence in virtual reality. It presents a 3×3 mixed-design remote VR experiment (hand rendering style between-subjects: realistic, toon, sketch; environment rendering style within-subjects: the same three levels) with N=117 consumer HMD users. Participants performed a pointing task and reported body ownership, presence, and perceived realism via in-headset questionnaires. The authors report no significant main effect of hand rendering style and no hand×environment interaction on body ownership, no interaction on presence, and a robust main effect of environment rendering style on presence (F(2,228)=48.08, p<.001, η²p=.297), with more photorealistic environments yielding higher presence. They conclude that photorealism of the virtual hands and its consistency with the environment do not influence body ownership, while environmental photorealism enhances presence, and they draw design implications for VR content creation.","tokens_in":23015,"tokens_out":5721,"duration_ms":57475,"significance":"The study is a useful empirical contribution to an area with inconsistent evidence. Its strengths include a systematic operationalization of photorealism separately from anthropomorphism and truthfulness, a relatively large sample for a VR experiment, use of the same shaders for hand and environment in the non-realistic conditions, pre-specified data-quality screening for remote data collection, and candid reporting of null effects and limitations. The environment→presence result is strong and consistent with some recent work. However, the main novel null claim—that hand photorealism and hand-environment consistency do not affect body ownership—is weakened by the paper's own manipulation check, which shows only partial perceptual separation of the hand styles. The paper should be revised to limit its claims to the range actually tested and to provide precision/equivalence information for the null effects.","major_comments":[{"comment":"The hand-realism manipulation was only partially successful: only the realistic and toon hand conditions differed significantly (p<.05); the sketch hand did not differ from realistic (p=.07) or toon (p=.63). Since H3 is a monotonic hypothesis across three levels, the failure of the sketch condition means the three-level test does not actually test the full monotonic hypothesis. More importantly, H1/H2 rely on participants perceiving the hand style categories; if the sketch hand was not reliably perceived as 'sketch,' then for participants in that group the 'matching' environment conditions are mismatched perceptually, which adds noise to the interaction tests. The abstract, §5.1, and §6 state categorically that hand rendering style did not influence body ownership; this is not supported by the manipulation check. I request a reanalysis restricted to the successfully separated realistic v","section":"§4.2 (Q8) and §5.2"},{"comment":"The paper reports null results for H1, H2, and H3 without any power analysis, equivalence test, or sensitivity analysis. With approximately 38–40 participants per between-subjects hand condition, the absence of a significant main effect or interaction does not license the conclusion that 'no effect exists'; the design may have low power to detect small-to-moderate effects, especially interactions. Please report effect sizes with confidence intervals for the non-significant effects and, if the goal is to support a null claim, provide an equivalence bound or a sensitivity analysis showing what effect sizes the design could detect. The design-implication conclusion that rendering styles can be optimized 'independently' is stronger than the data justify without such analysis.","section":"§3.1, §4.2"},{"comment":"After item exclusions, Cronbach's alpha for body ownership is 0.70 (two items) and for presence is 0.59 (three items). The presence value is below customary reliability thresholds, and the body-ownership value is marginal. Since all conclusions rest on these questionnaire scores, low reliability attenuates statistical power and increases the risk of false-negative results. Please discuss this explicitly and, ideally, report item-level descriptive statistics or a robustness analysis (e.g., treating items as ordinal indicators).","section":"§4.2 (scale reliability)"}],"minor_comments":[{"comment":"The author name 'S. Frlston' appears to be a typo for 'S. Friston'.","section":"References [58]"},{"comment":"The sentence 'the participants (N=117) completed 35.15±12.29 items on average per experiment' is confusing: 35.15 appears to be the total across the three 90-s conditions, not per condition. Please clarify the wording.","section":"§3.4"},{"comment":"Q6 and Q7 were inverted as an attention check, but after excluding Q6 the final presence scale differs from the original SUS construct. Please clarify what the remaining three items are intended to measure and discuss the consequence for construct validity.","section":"Table 1, §4.2"},{"comment":"The statement that possible tracking issues 'would not significantly influence' the conclusions because conditions were randomized addresses systematic bias but not statistical power: tracking failures could add noise and reduce sensitivity of the ownership measures in all groups. Please soften or support this claim.","section":"§5.2"}],"recommendation":"major_revision","confidential_remarks":"The core issue is the partially successful hand-realism manipulation. The authors already acknowledge the limited perceptual separation, but their abstract and conclusions overstate the null claim. If they reframe the claims to the range actually tested, provide the requested sensitivity/equivalence analysis, and address the reliability of the questionnaires, the paper would be a solid empirical contribution. The strong environment→presence result is not in question."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one. It's a well-executed remote VR experiment (N=117, prespecified exclusions) with a genuinely new 3x3 factorial: hand rendering style (realistic/toon/sketch, between) crossed with environment rendering style (same three, within). Previous work either varied hand style without controlling the environment or changed the whole scene at once, so the independent manipulation and the test of a consistency interaction are new. The paper also does a good job separating photorealism from anthropomorphism and truthfulness, drawing on Garau's taxonomy.\n\nThe strong result is H4: more photorealistic environment → higher presence, F(2,228)=48.08, partial eta^2=.297, all pairwise post-hocs significant. That is a large, clean effect and worth citing. The null results for hand photorealism and consistency are plausible and partly consistent with Lin and Jörg, but they are not as solid as the abstract implies. The manipulation check for hand realism only separated realistic from toon; sketch did not differ significantly from either (p=.07 vs realistic, p=.63 vs toon). So the monotonic H3 was not fully tested, and the consistency manipulation is muddier because for participants in the sketch-hand group the environment/hand match is not reliably perceived. The paper acknowledges this in Section 5.2, which I appreciate, but the design implications in Section 5.1 (\"photorealism is not a priority for hands\") overstate what the data can support.\n\nOther soft spots: final presence scale alpha is only .59 after deleting an item; the sample is 92.7% male HMD owners; data and code are not available. None of these are fatal. The environment effect is robust enough that H4 survives; the hand null just needs more cautious framing.\n\nThe stress-test note is basically right. It's a real limitation, but not a takedown. I'd send this to a serious referee. It is exactly the kind of empirical paper that should see the light of day, with requested revisions: soften the hand-photorealism conclusions, report the manipulation check more prominently, and ideally share the anonymized data. For a VR/HCI reader, the paper is useful; for researchers working on body ownership, the failed hand manipulation is itself a lesson.","headline":"A useful but overstated null: the new 3x3 design is solid and the environment→presence effect is real, but the hand-photorealism null rests on a partly failed manipulation and the design recommendations should be softened.","tokens_in":23459,"tokens_out":2995,"would_cite":true,"duration_ms":28433,"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":"This paper reports that the photorealism of a virtual hand, and the match between hand and environment, does not change the sense of body ownership in VR, while the environment's photorealism strongly increases presence.","keywords":["virtual reality","body ownership","virtual hand illusion","photorealism","rendering style","presence","avatar realism","remote experiment"],"falsifier":"Run the same ownership questionnaire with a hand-realism manipulation whose three levels are each clearly perceptually distinct—for example a textured realistic hand, a flat-shaded toon hand, and a line-only or wireframe hand rated well below both—and compare ownership scores. If ownership no longer stays equal across all three levels, the paper's central null is overturned; if ownership remains flat while environment realism still raises presence, the claim survives.","tokens_in":22634,"feed_emoji":"🥽","tokens_out":5661,"duration_ms":54840,"temperature":0.7,"pith_summary":"The paper isolates photorealism (how photograph-like an object appears) from two other realism dimensions—anthropomorphism (human-likeness) and truthfulness (resemblance to the user)—and tests its effect on the body ownership illusion. In a remote VR experiment with 117 participants, each person saw one of three hand renderings (realistic, toon, sketch) across three environment renderings of the same styles. Body ownership stayed high regardless of hand style or of whether the hand matched the environment, and rendering-style consistency did not affect presence either. Only the environment's photorealism mattered, raising presence with a large effect size. A sympathetic reading is that realism is not a single lever: anthropomorphism shapes ownership, photorealism shapes presence, and designers can choose hand styling independently of the environment.","feed_headline":"VR hand photorealism doesn't change body ownership","feed_subtitle":"Hand style doesn't affect ownership; only environment realism lifts presence, so stylized avatars are safe.","key_machinery":"The load-bearing device is a 3x3 mixed factorial design that independently crosses hand rendering style (between participants) with environment rendering style (within participants), built from the same base 3D models so that anthropomorphic shape stays fixed across styles. Photorealism is manipulated only through materials, textures, and shading, yielding realistic, toon, and sketch levels. The measurement relies on the virtual hand illusion, driven by visuomotor synchrony from consumer hand tracking, and on seven-point Likert questionnaire items for body ownership and presence; the design would reveal a consistency effect (H1/H2) as an interaction between hand and environment factors, and","core_discovery":"The central claim is negative but delimited: holding the hand's shape and personal resemblance constant, changing its rendering style from a textured realistic skin to a two-tone toon look to an untextured sketch does not change how strongly people feel the virtual hand is their own, and making the hand and environment share the same rendering style does not strengthen that illusion. Meanwhile, rendering the environment more photorealistically did increase presence, with all pairwise comparisons significant and a large effect size. The authors conclude that dimensions of realism affect different aspects of experience differently, and that prior inconsistencies in virtual-hand realism researc","pith_inferences":["Editorial extension: the null result for hand photorealism is only as strong as its manipulation—the sketch hand was not perceived as significantly less realistic than the toon hand, so the lowest photorealism level may not have registered as distinct.","Editorial extension: real-time visuomotor synchrony may have created a ceiling effect that masked photorealism's contribution; testing with asynchronous hand tracking could reveal whether photorealism matters when bottom-up cues are weaker.","Editorial extension: the realistic hand drew the most negative spontaneous comments yet did not lower ownership, hinting that affective responses (uncanny valley) and body ownership are separable; measuring eeriness alongside ownership could expose a non-monotonic relationship hidden in the means.","Editorial extension: the paper measures hands only, and the authors themselves note the hand-limb discontinuity could make photorealism more consequential for full-body avatars, so the null result should not be generalized to whole-body ownership without further tests."],"forward_implications":["If the null results hold, developers can choose hand rendering independently of environment style without measurably changing users' sense of ownership or presence.","Applications where presence is central—training, therapy, virtual tourism—should invest in environment photorealism, which produced the largest effect in the experiment.","Stylized (toon) hands are a safe choice on resource-constrained standalone headsets: they preserve body ownership and avoid the negative appearance comments that the realistic hand attracted.","The pattern supports treating realism as multidimensional rather than monolithic, and suggests that future virtual-hand studies should report anthropomorphism, photorealism, and truthfulness separately.","The failure of the consistency hypotheses indicates that perceptual unification of body and scene along the photorealism axis is not required for the body ownership illusion in VR."],"fun_headline_variants":["Stylized VR hands fine: only environment boosts presence","Mismatched VR styles don't break body ownership","Photoreal environment, not hands, drives VR presence","Toon VR hands OK; realism only matters for presence"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the three rendering styles are perceptually distinct levels of hand photorealism, so a null hand effect means photorealism truly does not affect ownership; the paper's own realism ratings separated only realistic from toon, leaving the sketch level perceptually ambiguous.","fun_headline_variants_meta":{"raw":{"variants":["Stylized VR hands fine: only environment boosts presence","Mismatched VR styles don't break body ownership","Photoreal environment, not hands, drives VR presence","Toon VR hands OK; realism only matters for presence"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000271,"raw_usage":{"total_tokens":1484,"prompt_tokens":779,"completion_tokens":705,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":523,"completion_tokens_details":{"reasoning_tokens":640}},"tokens_in":523,"tokens_out":705,"duration_ms":7250,"temperature":1.0,"reasoning_tokens":640,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T23:10:13.946034+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same ownership questionnaire with a hand-realism manipulation whose three levels are each clearly perceptually distinct—for example a textured realistic hand, a flat-shaded toon hand, and a line-only or wireframe hand rated well below both—and compare ownership scores. If ownership no longer stays equal across all three levels, the paper's central null is overturned; if ownership remains flat while environment realism still raises presence, the claim survives.","supporting_citations":[],"review_version":1}