Investigating proxemics behaviors towards individuals, pairs, and groups in virtual reality

ORIGINAL ARTICLE Vi ual Reali y (2025) 29:58 h ps://doi.o g/10.1007/s10055-025-01127-y become. As a esul , iola ing hese dis inc zones can cause discom o (O e b ing e al. 2021). Fo ins ance, in e ac- ions wi h close iends o amily ypically occu wi hin he pe sonal zone. In con as , in e ac ions wi h acquain ances a e gene ally elega ed o he social zone, while in e ac ions wi h s ange s will occu in he public zone. The e o e, in es iga ing when he di e en spaces a e kep and when hey a e iola ed allows us o model which ac o s can shape his aspec o human non- e bal communica ion. 1.2 P oxemics in i ual en i onmen In e es ingly, human-human p oxemic beha io in he phys- ical wo ld can be eplica ed in an imme si e i ual en i- onmen (VR) popula ed wi h human-like a a a s (Hech e al. 2019; Ruggie o e al. 2019). By hei human-like appea ance, hese a a a s exe a social p esence, a ac - ing in e ac ions and engagemen s om human use s (Kim e al. 2018). Howe e , when an indi idual app oaches a human-like a a a oo closely, i.e., in ading hei pe sonal zone, he e is a pe cei ed discom o due o he in asion o he space occupied by he a a a (Hech e al. 2019). Simi- la ly, an app oaching i ual human can c ea e a sense o 1 In oduc ion 1.1 The p oxemics heo y The p oxemics heo y examines how people use he space su ounding hem (Hall 1966), a i al aspec o non- e bal communica ion. I sugges s ha indi iduals keep di e en dis ances om one ano he depending on how amilia and in ima e hey a e (Hall 1966; So okowska e al. 2017). P e- ious esea ch (as seen in So okowska e al. 2017) g ouped in e pe sonal dis ances in o ou dis inc zones: he in ima e (up o 0.46 m), pe sonal (0.46–1.22 m), social (1.22–3.65 m), and public zone (beyond 3.65 m). The close he zone is o a pe son, he mo e amilia and in ima e he in e ac ions Debo a Nol e [email p o ec ed] 1 Ins i u e o Cogni i e Science, Uni e si y o Osnab ück, Wachsbleiche 27, 49090 Osnab ueck, Ge many 2 Depa men o Neu ophysiology and Pa hophysiology, Uni e si y Medical Cen e Hambu g-Eppendo , Hambu g, Ge many Abs ac The p oxemics heo y explains he consis en social bounda ies su ounding indi iduals as epo ed (Hall in The Hidden Dimension, Doubleday, Ga den Ci y, 1966), ye li le is known abou he social bounda ies su ounding pai s o g oups o people. The cu en s udy explo ed in e pe sonal p oxemics beha io in a i ual en i onmen , ocusing on dis ances main ained owa ds indi idual pedes ians, pai s, and g oups. Using i ual eali y o simula e a ci y cen e , pa icipan s eely na iga ed i while hei mo emen s and gazes we e cap u ed. Impo an ly, he ci y was popula ed by pedes ians in di e en social con igu a ions. Eye mo emen s iden i ied in e ac ions de ined by gaze-onse s owa ds a pedes ian’s head. Ou esul s indica e ha pa icipan s app oached indi iduals wi h a median dis ance o 3.18 uni y uni s aligned wi h he social space bounda y as epo ed (Hall in The Hidden Dimension, Doubleday, Ga den Ci y, 1966). Dis ances kep om pai s and g oups we e simila ly cen e ed wi hin he social space, e ealing no signi ican di e ence in app oaching beha io ac oss di e en social con igu a ions. The consis ency in app oaching dis ances sugges s ha pe sonal and social spaces a e no subs an ially al e ed, i espec i e o he social con ex . Keywo ds P oxemics · In e pe sonal dis ance · Vi ual eali y · Non e bal beha io Recei ed: 29 Ap il 2024 / Accep ed: 3 Ma ch 2025 / Published online: 25 Ma ch 2025 © The Au ho (s) 2025 In es iga ing p oxemics beha io s owa ds indi iduals, pai s, and g oups in i ual eali y Debo aNol e1· ReemHjoj1· T acySánchez Pacheco1· AnnHuang1· Pe e König1,2 1 3 Vi ual Reali y (2025) 29:58 discom o in pa icipan s when hei pe sonal space ge s in aded (Bailenson e al. 2003) o will lead o pa icipan s signaling o s op he app oaching i ual human when hey eel gene ally uncom o able as i became oo close (Iachini e al. 2016). In e es ingly, being app oached by o app oach- ing indi iduals caused discom o a simila dis ances (Rug- gie o e al. 2019). Toge he , empi ical wo k in eal and i ual social encoun e s indica ed ha pe sonal space in social in e ac ion is sha ed when indi iduals’ spaces coin- cide. When an indi idual a e ses in VR and encoun e s a i ual human, he indi idual exhibi s simila p oxemic beha io obse ed in he physical wo ld, he eby modula - ing social dynamics. 1.3 P oxemics beha io o g oups While subs an ial e o s ha e been in es ed in o be e unde s anding p oxemics o indi iduals, he app oach- ing beha io owa d g oups o people needs o be in es i- ga ed and be e unde s ood. The main ocus o his a ea o esea ch hus a is he s udy o he discom o when being app oached by a g oup compa ed o an indi idual. Subjec s el uncom o able a g ea e dis ances when app oached by a g oup han an indi idual (Bönsch e al. 2018). They displayed a signi ican inc ease in nega i e a ec and eac- i e beha io wi h inc easing c owd densi y (Dickinson e al. 2019). Simila ly, being app oached by a g oup compa ed o indi iduals esul ed in highe le els o a ousal, which can be unde s ood as discom o (Llobe a e al. 2010). Taken oge he , p e ious s udies indica e a sys ema ic di e ence in how passi e indi iduals espond o being app oached by g oups as opposed o indi idual pedes ians. Con e sely, while app oaching o being app oached by indi iduals esul ed in simila dis ances kep (Ruggie o e al. 2019), i emains o be in es iga ed whe he subjec s main ain a mo e conside able minimum dis ance when ac i ely app oaching a g oup compa ed o indi idual pedes ians. The cu en s udy aims o expand on p e ious indings by in es iga ing he app oaching beha io owa ds indi iduals and g oups. Speci ically, we we e in e es ed in he p oxemics beha - io in a na u alis ic en i onmen when subjec s ac i ely app oached pedes ians in di e en social dynamics: indi- iduals, pai s, and g oups. To achie e his, we analyzed he subjec s’ beha io eco ded du ing a ee-explo a ion s udy in a i ual eali y ci y cen e . Moni o ing he subjec s’ p ox- emics beha io allowed us o explo e how subjec s beha ed when na u ally and ac i ely app oaching indi iduals and g oups o pedes ians in di e en social dynamics. 1.4 The cu en s udy We p oposed h ee hypo heses o unde s and p oxemics beha io ac oss a ious social dynamics in a i ual eali y con ex . We expec he subjec s o espec he pe sonal space when in e ac ing wi h indi idual pedes ians (Hall 1966). Speci ically, we hypo hesize ha subjec s will keep a social dis ance o 1.22–3.65 m om he i ual indi idual pedes- ians. To ou knowledge, p oxemic beha io owa ds pai s o g oups, and speci ically indi iduals wi hin hese pai s o g oups, has no been sys ema ically in es iga ed. In o de o add ess his gap, we conside ed wo po en ial ideas o desc ibe space based on se heo y (see Fig. 1 o a isual desc ip ion o hese ideas). On he one hand, subjec s migh espec he pe sonal space o each membe o he pai o g oup (Fig. 1A). On he o he hand, hey migh conside he pe sonal space o he g oup o be he join social space o all membe s in he pai o g oup (Fig. 1B), ea ing he pe sonal space o a pai o g oup as he union o he indi idual social spaces. Based on hese conside a ions, we de i ed wo compe ing hypo heses o es empi ically. Hypo hesis 2A, co esponding o he null hypo hesis, assumes ha belong- ing o a pai o g oup does no in luence he dis ance kep owa ds his indi idual. In con as , hypo hesis 2B assumes ha being pa o a pai o g oup changes he minimal dis- ance kep owa ds an indi idual wi hin ha pai o g oup. To es his hypo hesis, we employed Bayes ac o analy- sis because i allows o a di ec assessmen o he e idence in a o o he null hypo hesis, unlike adi ional equen is me hods, which p ima ily ocus on ejec ing he null. The esul s a i med hypo hesis 1, demons a ing ha pa ici- pan s app oached indi idual pedes ians in VR a dis ances simila o hose epo ed in p e ious eal-wo ld s udies (Bailenson e al. 2003; Hall 1966; Hech e al. 2019). Fu - he mo e, he lack o a signi ican di e ence in app oaching dis ances be ween indi iduals, pai s, and g oups con i med ou null hypo hesis 2A and p o ided new insigh s in o app oaching beha io owa ds pai s and g oups. 2 Me hods 2.1 Subjec s We eco ded a o al o 59 subjec s, 44 o whom we e included in he analysis o his pape (mean age: 23 ± 3.10; 27 emales, ze o di e se). Subjec s who expe ienced sub- s an ial mo ion sickness, so wa e e o s, ailed o ollow ask ins uc ions, and one subjec who did no ha e ials o indi iduals we e excluded om he s udy (N = 15). All subjec s we e s uden s a he Uni e si y o Osnab ück. Pa - icipa ion was olun a y, and all subjec s p o ided w i en 1 3 58 Page 2 o 11 Vi ual Reali y (2025) 29:58 in o med consen . Recognizing hei ime, hey we e o e ed cou se c edi s o mone a y compensa ion. The esea ch p o- ocol ecei ed app o al om he E hics Commi ee o he Uni e si y o Osnab ück. 2.2 Expe imen al se up and p ocedu e The da a was eco ded as pa o a mo e ex ensi e s udy; he expe imen al se up is explained in de ail by (Nol e e al. 2024). B ie ly, he expe imen was designed using Uni- y3D (www.uni y.com; e sion 2019.4.21 1) o ep esen a ci y cen e popula ed wi h houses, g eene y, objec s such as Fig. 1 The wo compe ing ideas o he p oxemics beha io owa ds pai s and g oups. a Each g oup membe ’s pe sonal space will be unal- e ed. As a esul , he app oaching beha io o pai s and g oups will be iden ical o ha o indi iduals. Thus, he minimum dis ance dis i- bu ions o indi iduals, pai s, and g oups will be s a is ically indis- inguishable. b The pe sonal space o he g oup will be he sum o he social spaces o each membe o he pai o g oup; he e o e, he dis ances kep o indi idual membe s o a pai o g oup will be mo e conside able han and s a is ically dis inguishable om he dis ances kep o indi iduals 1 3 Page 3 o 11 58 Vi ual Reali y (2025) 29:58 h oughou he ci y in a one-minu e ial session ha dou- bled as a mo ion sickness es . A e comple ing his one- minu e es , subjec s who did no epo mo ion sickness we e admi ed o he main s udy. Du ing his amilia iza ion phase, subjec s could po en ially see i ual pedes ians in he dis ance bu we e no able o inspec o app oach hem closely. Du ing he main s udy, pa icipan s we e di ec ed o explo e a speci ied a ea, dema ca ed by beige loo iles ( e e o Figs. 2 and 3), as hough awai ing a iend, empha- sizing mimicking eal-li e beha io . The en i e expe imen al session, including he EEG p epa a ions, las ed abou wo and a hal hou s. Fo hal o he pa icipan s, an un ela ed s udy ollowed he cu en one, aking an addi ional i e o en minu es. Pa icipan s we e allowed 30 min o eely explo e he i ual ci y, wi h pauses o 5-poin alida ion checks and possible ecalib a ions o addi ional b eaks as needed. Du ing alida ion checks e e y i e minu es, pa - icipan s we e elepo ed o a sepa a e space wi hin he i - ual en i onmen . I necessa y, an in e nal calib a ion and a subsequen 5-poin alida ion we e s a ed au oma ically. A e a success ul alida ion, pa icipan s we e e u ned o hei las posi ion wi hin he i ual ci y cen e . 2.3 The pedes ians O e all, 140 pedes ians aken om he Adobe Mixamo collec ion (Mixamo, 2008, h ps://www.mixamo.com) popula ed he i ual ci y. We selec ed all ealis ic-look- ing pedes ians om his collec ion ha we e easonably benches, and pedes ians (see Sec . 2.3 The Pedes ians o mo e de ail). The dimensions o he i ual ci y we e buil o ma ch hose o he eal wo ld - one uni y uni equals one me e . The i ual ci y was displayed wi h a cons an ame a e o 90 Hz using he HTC Vi eP oEye HMD (110° ield o iew, 1440 × 1600 pixels pe eye; h p s : / / b u s i n e s s . i e . c o m / u s / p o d u c / i e - p o - e y e - o i c e / ), he HTC Vi e L i g h h o u s e 2.0 acking sys em ( h p s : / / w w w . i e . c o m / e u / a c c e s s o y / b a s e - s a i o n 2 /) and he HTC Vi e con olle s 2.0 ( h p s : / / w w w . i e . c o m / e u / a c c e s s o y / c o n o l l e 2 0 1 8 /). Eye m o e m e n s we e cap u ed wi h he help o he Eye-T acking SDK SRanipal ( 1.1.0.1; h p s : / / d e e l o p e - e x p e s s . i e . c o m / e s o u c e s / i e - s e n s e / e y e - a n d - a c i a l - a c k i n g - s d k / ). The e x p e i m e n was played on an Alienwa e Au o a Ryzen compu e using a Windows 10 sys em (64-bi , build- e sion 19044; 6553 MB RAM), an N idia RTX 3090 GPU (d i e e - sion 31.0.15.2698), and an AMD Ryzen 9 3900 × 12-Co e CPU. The da a was eco ded using LabS eamingLaye (LSL; h p s : / / g i h u b . c o m / s c c n / l a b s e a m i n g l a y e ) on a Dell Inc. P ecision 5820 Towe compu e (Windows 10; 64-bi , Build 19044; an N idia RTX 2080 Ti GPU; d i e e sion 31.0.15.1694; In el Xeon W-2133 CPU). We addi ionally eco ded EEG da a, bu since i is i ele an o he cu en pape , i will no be discussed u he . The s udy led pa icipan s o explo e he i ual en i on- men eely. The pa icipan s’ mo emen was implemen ed o mimic eal-li e mo emen using small displacemen s con olled ia a HTC Vi e con olle . Each session s a ed wi h amilia iza ion wi h he equipmen and mo emen Fig. 2 The mo emen pa h o an indi idual subjec plo ed on op o a bi ds-eye iew o he walkable ci y cen e a ea 1 3 58 Page 4 o 11 Vi ual Reali y (2025) 29:58 g oups. Ten pai s we e accessible om he cen al squa e, wi h a ied con igu a ions: i e pai s sea ed, one pai wi h a s anding and a sea ed pedes ian, and h ee pai s wi h bo h pedes ians s anding. Addi ionally, wo g oups, each com- p ising h ee pedes ians, we e p esen . One g oup had all sea ed indi iduals, while he o he had s anding ones. All s anding pedes ians had an anima ion a ached o hem and we e mo ing in place. This se up amoun s o a ealis ic mix- u e o pedes ians one migh encoun e walking in a ci y. 2.4 De ining ials Ou s udy allowed subjec s o mo e eely h oughou a i ual ci y cen e o 30 min. The mo emen pa h o one subjec is shown in Fig. 2, and a sho ideo displaying i s - pe son imp essions o he i ual eali y ci y can be seen d essed o appea in a ci y cen e . Fo his s udy, we ca ego- ized he pedes ians as indi iduals, e e ing o pedes ians s anding o si ing alone, pai s o pedes ians ha consis ed o wo people nea each o he , and g oups o pedes ians, unde s ood as a g oup o h ee o mo e indi iduals close o one ano he . An example o each ca ego y can be seen in Fig. 3. A subse o he 140 pedes ians was selec ed o he cu en analysis. Fi s , only s a iona y pedes ians we e con- side ed o mi iga e biases om di e ing mo emen speeds. Fu he mo e, pedes ians we e only included in he inal da a analysis i hey we e posi ioned in he walkable a ea o he expe imen . O e all, i e s a ic indi iduals posi ioned a ound cen al ix u es such as ables, benches, o s a ues we e included in he analysis. Nex o s a ic indi iduals, we analyzed he app oaching beha io owa ds pai s o wo, om now on simply pai s, and g oups o h ee, in sho Fig. 3 P oxemics beha io o an indi idual pa icipan owa ds indi- iduals, pai s, and g oups. The uppe panel displays h ee examples o pedes ians isible in he VR scene: a an indi idual pedes ian, b a pai o pedes ians, and c a g oup o h ee. The bo om panel displays he di e en mo emen pa hs a ound he pedes ians in each condi ion in he example o a single subjec . The subjec shown in his plo is iden ical o ha shown in Fig. 2. Fo easy isualiza ion, he loca ion o he pedes ians is se o be he o igin. Fo he condi ion o pai s b and g oups c, he i s gazed-a pedes ian o he pai /g oup is he g aph’s o igin. The o he pedes ians o ha pai o g oup a e le ou o his display. The da k blue ci cle in each g aph co esponds o he pe sonal space o he app oached indi idual, and he ligh blue a ea co esponds o he social space 1 3 Page 5 o 11 58 Vi ual Reali y (2025) 29:58 o e alua e he e ec s o a a a g oup size and pa icipan gende on minimal dis ance. Speci ically, we u ilized he gene alTes BF unc ion om he BayesFac o package in R (Mo ey and Roude 2018), which allowed us o accoun o he nes ed na u e o he da a by ea ing pa icipan s as a epea ed measu e. This analysis enabled us o compa e models and de e mine whe he he da a p o ided e idence o he null hypo hesis (H0: no e ec o g oup size and gen- de ) ela i e o he al e na i e hypo hesis (H1: he e is an e ec ). Impo an ly, in his s ep, we included only he dis- ances be ween he subjec and he i s gazed-a pedes ian o a pai o g oup. This app oach ensu ed ha he s a is ical analysis was ancho ed o a subjec -de ined e en , s a ing wi h hei i s isual engagemen wi h he pai o g oup. By doing so, we a oided ea ing he app oaching dis ances o all pedes ians in a pai o g oup as sepa a e e en s, as hese a e no independen measu es. The minimum dis ances owa ds he emaining pedes ian o a pai and he mean minimum dis ance owa ds he emaining wo pedes ians o a g oup we e assessed (see Fig. 4) and compa ed o he dis ibu ion o he i s gazed-a pedes ian, showing ha hey emained wi hin he ange o he minimum dis ance kep owa d he gaze a pedes ian. 3 Resul s 3.1 Numbe o in e ac ions Due o he na u e o ial de ini ions, a di e ging numbe o e en s we e obse ed ac oss he di e en condi ions. On a e age, ou subjec s app oached indi iduals i e imes (IQR = 4), pai s an a e age o 13 app oaches (IQR = 8), and g oups ou imes (IQR = 2). An example o in e ac ions can be seen in Fig. 3. The uppe panel displays pedes ians isible in ou i ual eali y scene, while he bo om panel shows he di e en mo emen pa hs owa d pedes ians o one subjec . The di e en numbe o g ay lines co espond o he di e en numbe o app oaches. I is isible ha o his pa icula subjec , mos in e ac ions we e had wi h pai s o pedes ians, while only h ee di e en in e ac ions could be ound among he g oup o h ee. 3.2 Dis ance kep pe condi ion Be o e a comp ehensi e compa ison, we analyzed he dis- ances main ained by pa icipan s ela i e o he social con- s uc s: indi iduals, pai s, and g oups. We compa ed he medians (o e ials) o he minimum app oaching dis ance (o e ime) pe condi ion wi hin each subjec . Indi idually, 16 subjec s app oached singula pedes ians mo e closely han pai s o g oups. Con e sely, 18 subjec s kep a bigge in he supplemen a y ma e ial. As he s udy did no con ain p ede ined ials, we had o de ine ials o line using he explo a ion beha io o he subjec s. The ials we e cen- e ed a ound subjec s looking a pedes ians’ heads. As a esul , we classi ied he da a in o gazes, all eye-s abiliza- ion mo emen s, and saccades, u ilizing he eye mo emen classi ica ion algo i hm desc ibed in (Nol e e al. 2024). In sho , he eye mo emen classi ica ion employed a eloci y- based algo i hm wi h an adap i e h eshold o di e en ia e be ween gazes and saccades (Nol e, e al. 2024). The clas- si ica ion was based on e sions o he REMoDNaV (Da e al. 2021) and MAD saccade (Voloh e al. 2020) algo i hms, adjus ed o h ee-dimensional da a (Kesha a e al. 2023) while accoun ing o ansla ional mo emen s o he playe wi hin he i ual scene (Nol e e al. 2024). Subsequen ly, gazes speci ically a ge ing pedes ian heads we e isola ed. Each ial was dema ca ed a ound a single gaze e en , encapsula ing a pe iod om 50 s p io o 150 s pos -gaze ini ia ion. This ial du a ion was chosen o ha e an in e al longe han he ypical du a ion o an encoun e , cap u ing he comple e app oach and depa u e phases. Fu he mo e, he in e al is much sho e han he o al expe imen al du a- ion o dis inguish be ween encoun e s in ol ing he same a a a s, e en when o he in e ac ions occu . Fo he pai s and g oups, he pedes ian unde in es iga ion was de ined as he one gazed a i s . Howe e , ano he pedes ian ou o ha pai o g oup could be unde in es iga ion in a subse- quen ial. The ials we e ca ego ized in o h ee condi ions: app oaching indi iduals, pai s, o g oups. I is impo an o no e ha he numbe o ials pe condi ion and subjec di - e ed due o he subjec s’ a ying explo a ion beha io s. De ining ials based on he explo a ions allowed us o ana- lyze he p oxemics’ beha io in a na u alis ic se ing. 2.5 Calcula ing dis ances We assessed he minimum dis ance be ween he pa icipan and he i ual pedes ians du ing each ial o e alua e p oxemics beha io ac oss indi iduals, pai s, and g oups. The minimum dis ances we e de e mined by calcula ing he Euclidean dis ances be ween he subjec and pedes ian posi ions h oughou he ial and hen compu ing he mini- mum. The Euclidean dis ances we e calcula ed using wo dimensions only, he x-and z-coo dina es, as he ele a ion o he g ound in ou en i onmen was cons an . Then, o each pa icipan , he median (o e ials) o he minimum app oaching dis ance (o e ime) was calcula ed o each o he h ee condi ions: app oaching indi iduals, pai s, and g oups. To compa e he h ee condi ions, we compu ed he median (o e subjec s) o ha median (o e ials) o he minimum dis ance (o e ime) o each condi ion ac oss subjec s. A se ies o Bayesian analyses we e conduc ed 1 3 58 Page 6 o 11 Vi ual Reali y (2025) 29:58 Fig. 4 Dis ances ac oss subjec s o all h ee condi ions. a The plo displays he p obabili y densi y dis ibu ion o all subjec s’ median minimal app oaching dis ances o he indi idual, pai , and g oup con- di ions. The shaded a eas in he backg ound co espond o he di e - en spaces a ound he pedes ians. Da k blue ep esen s he pe sonal space, ligh blue is he social space, and whi e ep esen s he public space. b The dis ibu ions o all subjec s’ median minimal app oaching dis ances a e addi ionally displayed using ba plo s. The x-axis o his plo is iden ical o he x-axis o Figu e a. We addi ionally plo ed he dis ibu ions o all subjec s’ median minimal app oaching dis ances owa ds he o he pedes ian o a pai (no he i s gazed a ) o he mean dis ance o he o he wo pedes ians o a g oup 1 3 Page 7 o 11 58 Vi ual Reali y (2025) 29:58 and g oups iden ically o indi idual a a a s, co esponding o ou null hypo hesis 2A. 4 Discussion This s udy in es iga ed in e pe sonal dis ances kep owa ds pedes ians in di e en social dynamics, indi idual pedes- ians, pai s o pedes ians, and g oups o pedes ians. In de ail, we analyzed app oaching beha io in a li ely i - ual eali y ci y cen e ha subjec s could explo e eely. When e alua ing he dis ances kep owa d indi iduals, we obse ed a median dis ance co esponding o he social space o he indi iduals (Hall 1966). This esul subs an i- a es ou main hypo hesis, sugges ing ha in casual i ual in e ac ions, pa icipan s main ain a dis ance akin o Hall’s delimi a ion o social space (Hall 1966). Such a pa e n s ands in acco dance wi h ecen li e a u e (Bailenson e al. 2003; Hech e al. 2019; Huang e al. 2022), concluding ha pe sonal space and non- e bal in e ac ions in i ual eali y ollow he same implici ules as in eal-wo ld se ings. Con- side ing he app oaching beha io owa ds pai s o g oups o pedes ians, we ound e idence o no di e ence be ween he di e en social condi ions. This esul suppo ed ou null hypo hesis 2 A: The e is no di e ence be ween he app oaching beha io owa ds indi iduals and ha owa ds pedes ians in pai s o g oups. Ou obse a ions sugges ha p oxemic beha io owa ds indi iduals emains consis en , i espec i e o hei engagemen in di e se social dynam- ics. Thus, hypo hesis 2B mus be disca ded as he minimal equi emen ha belonging o a pai o g oup esul s in di - e en minimal dis ances han being an indi idual is no me . O e all, we showed ha unknown pedes ians in a public se ing a e being app oached up o and including hei social space bu a ely en e hei pe sonal space ega dless o he social si ua ion. In e es ingly, as opposed o app oaching beha io owa ds indi idual subjec s, he dis ances kep owa ds pai s and g oups o people we e somewha su p ising in ligh o cu en esea ch indings. S udying p oxemics beha io wi h a s op-dis ance- ask e ealed highe le els o discom- o (Llobe a e al. 2010) and he wish o keep a bigge dis ance (Bönsch e al. 2018) when app oached by a g oup ins ead o a single pe son. The p esen esul s oppose hese indings o some deg ee, demons a ing ha subjec s in e - ac wi h pai s o g oups o pedes ians a he same dis ance as indi iduals. Se e al ac o s migh elucida e he obse ed dispa i y be ween ou indings and p e ious s udies. P ima - ily, di e ences in s udy designs could be a con ibu ing ac- o . Con a y o s udies whe e s a iona y pa icipan s we e app oached by pedes ians (as seen in Bönsch e al. 2018; Llobe a e al. 2010), ou esea ch emphasized in e pe sonal dis ance om indi idual pedes ians, while en subjec s displayed a mixed pa e n; h ee had he smalles median dis ance when app oaching g oups and he mos ex ensi e dis ance when app oaching pai s, while se en subjec s kep he smalles dis ance owa ds pai s and he la ges one owa ds g oups. Compa ing minimal dis ance o indi idual subjec s, no o e a ching pa e n eme ged. Fo close analysis, we compa e he median (o e i- als) o he minimum dis ances (o e ime) kep owa d indi idual a a a s ac oss all subjec s. The da a a e isual- ized in Fig. 4. The median (o e subjec s) o he median (o e ials) minimal (o e ime) dis ance was 3.18 Uni y uni s (IQR = 2.32) kep om indi idual pedes ians. Com- pa ing indi idual subjec s, he smalles dis ance kep was 0.79, while he maximum dis ance eached 6.94 uni y uni s. O e all, wo subjec s had a median dis ance wi hin he pe - sonal space o he app oached pedes ians, 23 subjec s kep a dis ance aligned wi h he social space o he pedes ians, and 19 subjec s had a median dis ance g ea e han he social space. Thus, he dis ibu ion o dis ances kep o indi idual a a a s is ully compa ible wi h p e ious wo k, suppo ing ou i s hypo hesis. We now u n o analyzing app oaches owa ds he gazed-a pedes ians o pai s and g oups o a a a s. This da a is isualized in Fig. 4. The median (o e subjec s) o he median (o e ials) minimal (o e ime) dis ance kep owa ds pai s was 2.7 uni y uni s (IQR = 1.72) and owa ds g oups 2.7 uni y uni s (IQR = 2.03). Fo pai s, one pe son had a median wi hin he pe sonal space, 31 subjec s kep a median dis ance aligned wi h he social space, and 12 had a dis ance bigge han ha . Fo g oups, 32 subjec s app oached up o and including he social space o he pedes ians, while 12 subjec s kep a dis ance bigge han ha . S a is ical di - e ences be ween he condi ions we e de e mined using a Bayes ac o app oach. The analysis indica es mode a e e idence suppo ing he null hypo hesis o he e ec s o a a a g oup size (BF01 = 9.49) and s ong e idence o null in he in e ac ion be ween a a a g oup size and pa icipan gende (BF01 = 24.66). This e idence sugges s ha a a a g oup size and i s in e ac ion wi h pa icipan gende do no signi ican ly impac minimal dis ance. The e is sligh e idence sugges ing a po en ial e ec o pa icipan gende (BF01 = 0.41), bu i is no s ong enough o be conclusi e. The inclusion o andom in e cep s o epea ed measu es is c ucial, as e idenced by he o e whelming Bayes ac o s a o ing models ha accoun o indi idual di e ences (e.g., epea ed measu es alone: BF01 = 4.43e − 13). We also isu- ally examined he minimum dis ances kep owa d he o he pedes ians in a pai o g oup (see Fig. 4B). The dis ibu- ions we e wi hin he same ange as hose o he i s gazed- a pedes ians, excluding po en ial conce ns o con ounding ac o s. These esul s show ha subjec s app oached pai s 1 3 58 Page 8 o 11 Vi ual Reali y (2025) 29:58 app oaching is de e mined by he pa icipan s and, as a esul , a dependen a iable, he p esen s udy is no sui able o examine his e ec . Addi ionally, as he cu en da a we e collec ed as pa o a mo e ex ensi e s udy, o he in luencing ac o s we e no con olled, including undamen al cul u al di e ences (Hall 1966; Li le 1968; So okowska e al. 2017) and en i onmen al ac o s (Li le 1968; So okowska e al. 2017). Con olling en i onmen al ac o s in a VR se up is challenging; o ins ance, he i ual scene migh di e om he ac ual oom empe a u e expe ienced by subjec s wea - ing he head-moun ed display. O e all, while ou esul s con ibu e o unde s anding p oxemics beha io owa ds pai s and g oups, addi ional in luen ial ac o s wa an u - he explo a ion. Se e al aspec s o he cu en s udy design could be ex ended o e ined in u u e esea ch. Fo ins ance, he a iabili y in spa ial placemen , iming, and sequence o di - e en pedes ian con igu a ions may ha e in oduced a i- abili y ha in luenced ou esul s. Mo e consis en con ol o hese a iables could enhance he obus ness o he indings. Addi ionally, con olling ac o s such as age, gende , and gene al likabili y o he pedes ians and including espon- si eness owa ds he pa icipan s could p o ide deepe insigh s in o app oaching beha io owa ds g oups in u u e s udies. Simila ly, u u e s udies could con ol he app oach- ing di ec ion, i.e., i pedes ians a e being app oached om he on o back, o examine i his e ec in luences he app oaching beha io owa ds g oups. Mo eo e , di ec ly compa ing ac i e app oaches e sus passi e encoun e s could help esol e disc epancies wi h p e ious s udies. While hese me hodological conside a ions a e no essen ial o he alidi y o ou cu en conclusions, hey ep esen impo an s eps owa d a mo e comp ehensi e unde s and- ing o app oaching beha io in a ious social con ex s. 5 Conclusion The cu en s udy in es iga ed in e pe sonal dis ances in a ee explo a ion s udy in a i ual eali y en i onmen . Ou indings co obo a e he p inciples o he p oxemics heo y (Hall 1966), ein o cing he c edibili y o conduc ing such explo a ions in a i ual domain. Addi ionally, we demon- s a ed ha social dynamics do no modula e his e ec , con- a y o ecen li e a u e indings. This insigh illumina es an unde - esea ched ace o app oach beha io , emphasizing he in luence o g oup dynamics on indi idual in e pe sonal dis ances. Gi en he di e gence om exis ing li e a u e, u he esea ch is impe a i e o comp ehensi ely unde - s and his nuanced non- e bal communica ion dimension. Supplemen a y In o ma ion The online e sion con ains supplemen a y ma e ial a ailable a h p s : / / d o i . o g / 1 0 . 1 0 0 7 / s 1 0 0 5 5 - 0 dis ances wi hin a ee explo a ion amewo k. Allowing o sel -mo ion could ha e esul ed in less anxie y and led o consis en p oximi ies be ween subjec s and indi iduals o g oups. Addi ionally, he design and beha io o ou i - ual pedes ians migh ha e in luenced pa icipan beha io . Speci ically, he a a a s in ou i ual en i onmen we e passi e, lacking esponsi eness o pa icipan s, and a oided di ec eye con ac - a ac o p e iously highligh ed o i s impac on in e pe sonal dis ances (Bailenson e al. 2003). Howe e , as indi iduals we e app oached a simila dis- ances, as shown in he li e a u e (Bailenson e al. 2003; Hall 1966; Hech e al. 2019; Huang e al. 2022), his seems a less likely possibili y. Taken oge he , how a being ac i e in he app oach owa ds pai s and g oups in luences p e- e ed in e pe sonal dis ances dese es a en ion in u u e wo k. Vi ual eali y en i onmen s o en elici social beha io s ha closely mi o hose obse ed in eal-wo ld in e ac ions, wi h use s ins inc i ely adhe ing o amilia social no ms and d awing in e ences based on o he s’ appea ances. As a esul , ac o s such as clo hing s yle (Wang and Yeh 2013), pe cei ed a ac i eness (Waddell and I o y 2015; Welsch e al. 2020), and o he con ex ual ac o s (O e b ing e al. 2022) may in luence app oach beha io s in VR, simila o hei e ec s on physical spaces, shaping use s’ pe cep- ions and a ec ing hei p oximi y p e e ences. In addi ion o hese appea ance-based in luences, se e al o he ac o s known o a ec p oxemic beha io could ha e impac ed he p esen esul s. Pe sonal ac o s such as age (Iachini e al. 2016; Webb and Webe 2003; Willis 1966) and gen- de (Bailenson e al. 2003; Hech e al. 2019; Huang e al. 2022; Iachini e al. 2016; Webb and Webe 2003) o bo h he subjec and pedes ians ha e been shown o in luence p ox- emics beha io . While he na ow ange o he age dis ibu- ion o he subjec s p ecluded an assessmen o age e ec s, ou esul s p o ided limi ed e idence o a gende e ec . The gende e ec was a he small o be conclusi e on i s own bu no incompa ible wi h p e ious li e a u e. No ably, no in e ac ion be ween gende and social con igu a ion was ound. Thus, o he esea ch ques ion in he ocus o he p esen s udy, gende does no in luence he minimal dis- ances kep as a unc ion o indi idual/pai /g oup. Howe e , he absence o con ol o eco ding o pedes ians’ pe - cei ed gende and age limi ed ou explo a ion o p e iously epo ed pedes ian e ec s (Bailenson e al. 2003; Huang e al. 2022; Iachini e al. 2016). Mo eo e , he o ien a ion o he pe son being app oached can a ec he dis ance main- ained owa ds hem (Bailenson e al. 2003; Hech e al. 2019), wi h smalle dis ances obse ed when app oaching om behind compa ed o he on (Bailenson e al. 2003; Hech e al. 2019), a ac o no conside ed in his s udy. As he pedes ians being app oached and hei o ien a ion while 1 3 Page 9 o 11 58