{"kind":"task","effective_mode":"full","benchmark":{"kind":"benchmark","effective_mode":"full","slug":"longbench-v2","formal_name":"LongBench v2","introduction":"長い資料の深い理解と推論を、多肢選択問題で評価するベンチマークです。公式紹介では503問を収録し、単一・複数文書の質問応答やコードリポジトリ理解などを扱います。\n\nLongBench v2 evaluates deep understanding and reasoning over long contexts through multiple-choice questions. Its official description lists 503 questions spanning tasks such as single-document and multi-document QA and code-repository understanding.","introduction_ja":"","introduction_en":"","category":"Category not supplied","task_count":null,"acquisition_status":"Acquisition status not supplied","official_url":"https://huggingface.co/datasets/zai-org/LongBench-v2","indexing_mode":"noindex"},"task_id":"366c229d-ae50-527d-a197-48f31386df26","task_key":"train--66ec2e2a821e116aacb1bb9f","task_revision_id":"1","upstream_id":"66ec2e2a821e116aacb1bb9f","short_description":"Why can VR Headset be used to help understand haptic slant adaptation in this…","config":"","split":"train","body":"{\"choice_A\":\"To distinguish between an effect due to a relative static posture adaptation and an effect based on a low-level unimanual adaptation\",\"choice_B\":\"VR devices can provide an unlimited workspace.\",\"choice_C\":\"VR Headset can be used to render virtual slanted surfaces and record the participant’s movement trajectories.\",\"choice_D\":\"Haptic force feedback has been proven to be unnecessary.\",\"context\":\"RESEARCH ARTICLE\\nNo need to touch this: Bimanual haptic slant\\nadaptation does not require touch\\nAbstract\\nIn our daily life, we often interact with objects using both hands raising the question the\\nquestion to what extent information between the hands is shared. It has, for instance, been\\nshown that curvature adaptation aftereffects can transfer from the adapted hand to the non-\\nadapted hand. However, this transfer only occurred for dynamic exploration, e.g. by moving\\na single finger over a surface, but not for static exploration when keeping static contact with\\nthe surface and combining the information from different parts of the hand. This raises the\\nquestion to what extent adaptation to object shape is shared between the hands when both\\nhands are used in static fashion simultaneously and the object shape estimates require\\ninformation from both hands. Here we addressed this question in three experiments using a\\nslant adaptation paradigm. In Experiment 1 we investigated whether an aftereffect of static\\nbimanual adaptation occurs at all and whether it transfers to conditions in which one hand\\nwas moving. In Experiment 2 participants adapted either to a felt slanted surface or simply\\nbe holding their hands in mid-air at similar positions, to investigate to what extent the effects\\nof static bimanual adaptation are posture-based rather than object based. Experiment 3 fur-\\nther explored the idea that bimanual adaptation is largely posture based. We found that\\nbimanual adaptation using static touch did lead to aftereffects when using the same static\\nexploration mode for testing. However, the aftereffect did not transfer to any exploration\\nmode that included a dynamic component. Moreover, we found similar aftereffects both with\\nand without a haptic surface. Thus, we conclude that static bimanual adaptation is of propri-\\noceptive nature and does not occur at the level at which the object is represented.\\nIntroduction\\nIn our daily life we often use both of our hands in many haptic tasks, such as doing the dishes,\\ntyping text using a computer keyboard or playing a musical instrument. When performing\\nsuch tasks, the movements of the two hands are relatively independent, at least at a mechanical\\nlevel. That is, activating the muscles of one arm/hand does not lead to a movement of the\\nother. For instance, when playing the guitar one hand frets the chords while the other hand\\nPLOS ONE\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n1 / 24\\na1111111111\\na1111111111\\na1111111111\\na1111111111\\na1111111111\\nOPEN ACCESS\\nCitation: Glowania C, Plaisier MA, Ernst MO, Van\\nDam LCJ (2020) No need to touch this: Bimanual\\nhaptic slant adaptation does not require touch.\\nPLoS ONE 15(7): e0236824. https://doi.org/\\n10.1371/journal.pone.0236824\\nEditor: Matthew Longo, Birkbeck University of\\nLondon, UNITED KINGDOM\\nReceived: February 25, 2020\\nAccepted: July 14, 2020\\nPublished: July 31, 2020\\nCopyright: © 2020 Glowania et al. This is an open\\naccess article distributed under the terms of the\\nCreative Commons Attribution License, which\\npermits unrestricted use, distribution, and\\nreproduction in any medium, provided the original\\nauthor and source are credited.\\nData Availability Statement: All raw data files are\\navailable from the figshare database (Experiment 1:\\n10.6084/m9.figshare.11889564; Experiment 2: 10.\\n6084/m9.figshare.11889606; Experiment 3: 10.\\n6084/m9.figshare.11889609).\\nFunding: Author CG was supported by the Cluster\\nof Excellence Cognitive Interaction Technology\\n’CITEC’ (EXC 277) at Bielefeld University, which is\\nfunded by the German Research Foundation (DFG).\\nWe acknowledge support for the Article Processing\\nCharge by the Deutsche Forschungsgemeinschaft\\nand the Open Access Publication Fund of Bielefeld\\n\\n\\nplucks the guitar strings without the one task interfering mechanically with the other because\\neach hand is controlled by a separate set of muscles. However, for performing such bimanual\\ntasks the two hands do of course still need to be coordinated by the Central Nervous System\\n(CNS) leading to the question to what extent and at what stages sensory information is com-\\nbined. Even when haptically exploring objects we often use both of our hands in a coordinated\\nfashion. [1] investigated object exploration with both one and two hands and showed that the\\nmodes of exploration used to obtain information about the object properties are very special-\\nized and coordinated across the hands. That is, the exploratory actions we make are very spe-\\ncific to the object property we want to explore. For instance, we dynamically slide with the\\nfingers over a surface for texture information but we statically hold an object in our hands to\\nestimate its weight; and when exploring the shape of an object, we often hold the object with\\none hand and move with the other over its surface. However, object shape information can be\\nobtained in multiple ways: we can do so by statically touching the object with a large portion\\nof our hand(s) (static exploration) or by dynamically moving with our finger(s) over its surface\\n(dynamic exploration). Moreover, we can explore object shape using either one or both hands.\\nIt is important to note however, that research on haptic shape perception has often involved\\nparadigms that use only one hand instead of two. This is particularly the case for haptic shape\\nadaptation studies in which participants are exposed to a curved or slanted surface for a pro-\\nlonged period of time. Afterwards a flat/level surface is perceived as curved or slanted in the\\nopposite direction (the haptic adaptation aftereffect). So far, haptic shape adaptation studies\\nfocused on conditions in which only a single hand was adapted, be it by sliding over a surface\\nwith one finger [2, 3], touching the surface with the whole hand [4, 5] or multiple fingers [3],\\ntouching a small part of a surface with the fingertip [6] or rubbing thumb and fingers along the\\nsides of a bar [7]. In the present study, we will instead investigate bimanual haptic adaptation\\nby using the index fingers of both hands simultaneously to make a perceptual judgment, and\\nthe potential transfer to other exploration modes.\\nNote that in the mentioned examples, often one hand or even one finger was sufficient to\\nobtain the required information to estimate the surface shape. Using two hands instead of one\\nin these cases would mean that each hand provides a separate estimate of object shape. That is,\\nthe two hands would provide redundant information. However, for large curvatures or slanted\\nsurfaces one finger, if used in a static fashion, does not provide very meaningful information\\nof such global shapes. In such cases, one finger alone samples too small a portion of the surface\\nto provide a very reliable estimate of the curvature or slant [8, 9]. This means that for global\\nshape estimation by static touch, at least one additional finger is needed, be it from the same or\\nopposite hand. In this case, the information provided by the additional finger is no longer\\nredundant; instead, this information is necessary to estimate the shape. The difference in posi-\\ntion between the fingers when touching the object (e.g. due to the difference in height at which\\nthe object is touched) would be informative about the object’s shape [10].\\nPrevious studies have focused on shape perception using multiple fingers from one hand\\n(e.g. [2, 3]) and found that adaptation largely depends on the posture of the hand. However,\\nwhereas two fingers from the same hand are mechanically coupled to some extent (i.e. they\\npartially use the same set of muscles), the fingers from the opposite hands share no mechanical\\ncoupling, in e.g. muscles and skin, and thus do not directly share any low-level receptors at\\nwhich adaptation can occur. Therefore, any bilateral control or coupling of sensory informa-\\ntion between the hands has to take place in the CNS, e.g. through bilateral tactile receptive\\nfields in the primary somatosensory cortex [11–14] which is another potential stage at which\\nadaptation may occur. However, it is unclear which of these stages would contribute to percep-\\ntual shape adaptation aftereffects in the case of static bimanual exploration. In order to investi-\\ngate whether shape adaptation aftereffects still occur in this case, the present study will\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n2 / 24\\nUniversity. The funders had no role in study\\ndesign, data collection and analysis, decision to\\npublish, or preparation of the manuscript.\\nCompeting interests: The authors have declared\\nthat no competing interests exist.\\n\\n\\nparticularly focus on the situation when two fingers from our two separate hands are used for\\nadaptation (we will use both the index fingers of the left and right hand). In order to perceive\\nthe global shape by using the left and right index finger, the two hands need to share their posi-\\ntion information to create a combined percept. If we find adaptation aftereffects for this mode\\nof exploration, the intuitive conclusion seems to be that adaptation occurs at this bimanual\\nposition sharing stage. However, as will become evident our results rather point towards static\\nbimanual adaptation still being posture based and at the level of the individual hands.\\nWe conducted three experiments. Experiment 1 and 2 tested contrasting predictions of non-\\nredundant bimanual slant adaptation being posture based or occurring at the level of the bimanual\\nsurface representation. Experiment 1 tested whether non-redundant bimanual adaptation transfers\\nto conditions that include a dynamic exploration component and Experiment 2 investigated\\nwhether or not a surface is needed to be felt for haptic slant adaptation to occur. As will become\\nclear the results of both these experiments indicated that haptic adaptation was driven by posture,\\nrather than adaptation occurring at the processing level at which the surface is represented. This\\nwould mean that bimanual adaptation aftereffects are based on the comparison of two individually\\nadapted hands by the brain [15, 16], and thus, adapting only one hand might be sufficient to show\\nadaptation aftereffects. This was confirmed in a third and last experiment in which only one hand\\nwas adapted to a position in space and clear aftereffects of adaptation were found.\\nExperiment 1\\nIn Experiment 1 we tested whether static bimanual slant adaptation occurs when the information\\nof the two hands is non-redundant (i.e. the slant estimate cannot be obtained using one hand\\nalone). If so, it would seem intuitive that such adaptation occurs at the level at which the informa-\\ntion of the two hands is shared. Evidence for information sharing between the hands for shape\\nperception was previously found for dynamic unimanual exploration by studies that investigated\\ntransfer of haptic adaptation between the hands. In a study by Van der Horst et al. [2] participants\\nadapted dynamically to haptic curvature (i.e. they moved a single finger back and forth over the\\nsurface) and showed transfer of the aftereffects to the fingers of the opposite hand, which were\\nnever directly involved in the adaptation process. Van der Horst and colleagues concluded that\\nthe adaptation occurred at a level at which the dynamic information of the two hands is shared.\\nThe same was found for virtual surfaces for which adaptation to curvature using a dynamic explo-\\nration mode also transferred intermanually [17]. However, for static contact of the surface the\\nintermanual transfer effects were much reduced [6] or even absent [5], suggesting that static\\ntouch adaptation might be more specific to the hand used during adaptation. In other words, for\\nstatic unimanual exploration the literature points towards a more receptor-based adaptation. This\\nsuggests that information sharing between the hands may depend on the mode of exploration.\\nThe present case of non-redundant bimanual static adaptation to shape however naturally\\nrequires the sharing of information across the hands and therefore may be occurring at a level\\nthat generally couples the information from the two hands regardless of exploration. A previ-\\nous study by Dupin et al. [18], for instance, showed that the kinaesthetic information coming\\nfrom one hand and tactile information coming from the other hand can be combined in the\\nbrain to form a single percept of object shape. If indeed the adaptation occurs at such a general\\nbimanual coupling level at which information of the two hands is available, one could expect\\nadaptation to transfer to conditions with a dynamic component (see e.g. [2, 6]). However, in\\nline with adaptation transfer studies finding different results in static and dynamic conditions,\\na recent study found that when using the same hand, aftereffects do not transfer between static\\nand dynamic exploration modes [3]. This suggests very distinctive processing pathways for\\nthese separate modes of exploration. Furthermore, it is known that the primary and secondary\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n3 / 24\\n\\n\\nnerve endings in the muscle spindles respond to either position as well as movement or to\\nposition alone, respectively. Therefore, it is also possible that any static bimanual adaptation is\\nexploration mode specific and thus does not occur at a higher level at which bimanual\\ndynamic information is represented. Thus, if bimanual adaptation is exploration mode spe-\\ncific, this would point to adaptation occurring at a less general and thus likely a more pre-CNS\\nstage involving skin and muscle receptors or the very early processing thereof in the CNS.\\nIn short, the purpose of Experiment 1 was twofold: First we investigated whether static\\nbimanual slant adaptation occurs when the information of the two hands is non-redundant. In\\norder to do so participants adapted to a slanted surface by touching the surface with their two\\nindex fingers statically. The adaptation aftereffect was measured using this same static biman-\\nual exploration mode in the test phase. Second, to test whether static bimanual adaptation is\\nexploration mode specific as well as to gain insights into the level at which bimanual static\\nadaptation may occur, Experiment 1 included transfer conditions that had a dynamic explora-\\ntion component (either moving one finger over the surface or moving one finger and keeping\\nstatic contact with the other).\\nMaterial and methods experiment 1\\nParticipants.\\nInformed consent was acquired prior to participation and participants were\\ntreated in accordance with the Declaration of Helsinki. Ethical approval was obtained from the\\nBielefeld University ethics committee. Thirteen people (including the authors CG and LD) vol-\\nunteered to participate in the experiment (11 female, all participants were right-handed upon\\nself-report, age range: 19–38). Note that this number of participants is generally sufficient for\\nhaptic adaptation studies, since effect sizes of haptic adaptation aftereffects tend to be relatively\\nlarge (e.g. [2, 4, 5, 8] used participant numbers ranging between 2 and 8 for separate experi-\\nments). The students received financial compensation (6€/h) for their participation. None of\\nthe participants reported any somatosensory deficits.\\nSetup.\\nThe participants were seated behind a haptic workbench on which two PHANToM\\nforce-feedback devices (PHANToM premium 1.5, SensAble Technologies, Inc. Woburn, MA)\\nwere mounted–with their body midline aligned with the centre of the bench. On each side of\\nthe workbench one PHANToM force-feedback device was placed. Participants placed their\\nright and left index fingers into thimble-like holders, attached to each PHANToM (see Fig 1A).\\nFig 1. Experimental and virtual setup. A: Experimental Setup. The participant was seated in front of a visuo-haptic workbench consisting of a CRT-monitor,\\nan opaque mirror and two PHANToM force feedback devices which were attached to the participants left and right index fingers; B: Virtual Setup. workspace\\nbox that contains the virtual surface (depth 26 mm) as well as response zones at the top left and right of the box; The red dashed line indicates the threshold that\\nparticipants had to cross with both index fingers in order to start the trial.\\nhttps://doi.org/10.1371/journal.pone.0236824.g001\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n4 / 24\\n\\n\\nThe PHANToMs were used to render virtual slanted surfaces and the haptic rendering could be\\nswitched on and off independently for each finger. Thus, haptic information could be displayed\\nto both fingers simultaneously or to only one of the fingers individually. Furthermore, the\\nPHANToMs were used to record the participant’s movement trajectories during exploration to\\nverify adherence to the task. For the current experiment, the system was setup to record the fin-\\nger positions with a sampling rate of 47Hz. To inform the participants about the next trial, a\\nCRT monitor (Sony CPD G500/G500J, Sony Europe Limited, Weybridge, UK; 140 Hz) was\\nused.\\nStimuli & procedure.\\nFor adaptation, we always used a static bimanual exploration mode\\nwhereas for the test trials there were three exploration modes: Static Bimanual (adapted condi-\\ntion), Dynamic Unimanual (transfer condition 1) and Mixed Bimanual (transfer condition 2).\\nIn the Static Bimanual mode, participants kept static contact with the surface using the index\\nfingers of the left and right hands. In the Dynamic Unimanual condition, the participants\\nmoved their right index finger across the surface in an area spanning 140 mm left to right, cen-\\ntred at body midline, in order to explore the slanted plane. In this condition, the haptic render-\\ning for the left index finger was switched off and thus no haptic information was provided to\\nthat finger. In the Mixed Bimanual condition, the surface was again rendered for both the\\nright and left index fingers. In this case, however, participants kept static contact with the left\\nindex finger on the left side of the slanted surface and moved across the surface with the right\\nindex finger. The Mixed Bimanual condition tested the influence of the bimanual adaptation\\non an exploration mode that contains both a static and a dynamic component. To avoid the\\ndynamic finger from making contact with the static finger as much as possible, the participants\\nwere told to place the static left index finger close to the left end of the surface and to make\\nmovements that do not interfere with the static finger. In order to prevent the participants\\nfrom moving diagonally over the surface in the Dynamic Unimanual and Mixed Bimanual\\nconditions and thus creating the impression of a less slanted surface, we limited the space in\\nthe z-direction (depth) by flanking each side of the slant with hard vertical surfaces. The so\\nrestricted area for exploration was limited to 26mm in depth, while keeping the entire width of\\n140 mm.\\nBefore the trial started, participants were informed about which exploration mode to use\\nfor the upcoming trial. For this purpose, colour cues were used (red, green and blue), which\\ncovered the full range of the screen. A red screen indicated that participants should use the\\nStatic Bimanual exploration mode; A green screen was used for the Dynamic Unimanual\\nmode and a blue screen was used for the Mixed Bimanual mode. To make sure the participants\\nused the colour cues adequately, each participant practiced using the correct exploration\\nmodes corresponding to the colour cues before the start of the experiment. Moreover, during\\nthe experiment, the participant’s finger positions were recorded using the PHANToMs to be\\nable to verify whether the participants adhered to the cues.\\nIn order to start a trial, participants first lifted their fingers above a programmed threshold\\nof 75 mm above the height at which the surface would be rendered. The moment they passed\\nthis threshold the colour cue disappeared, and no visual information was provided. Next par-\\nticipants lowered their fingers until they reached the surface and explored the surface for 1s\\nusing the exploration mode indicated by the colour cue. The exploration time started as soon\\nas one finger touched the surface and after 1s the surface disappeared. The participants’ task\\nwas to indicate the slant of the surface by judging which side of the surface felt higher: left or\\nright. Participants provided their response by moving their index finger into the correspond-\\ning “response zone” located at the top left and right of the programmed PHANToM workspace\\n(see Fig 1B). Note that also while responding the participants could not see anything on the\\nscreen or their finger positions to prevent any interaction from visual cues. The left response\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n5 / 24\\n\\n\\nzone indicated that the left side was perceived to be higher and vice versa for the right response\\nzone. After providing their response, the exploration mode colour cue for the next trial was\\nshown.\\nIn order to determine the Point of Subjective Equality (PSE)–the point at which the partici-\\npant perceived the surface as horizontal–we used an adaptive 1-up/1-down staircase procedure\\n(for further information see [19] or [20]). The step size between trials started with 8deg. After\\ntwo reversals in the responses, the step size was decreased to 4deg. and after another two rever-\\nsals to 2deg. After 12 reversals, the staircase was terminated.\\nTo measure the effect of slant adaptation we used a pre- versus post-test procedure. In the\\npre-test as well as in the post-test phases, there were two staircases for each exploration mode.\\nTo control for possible hysteresis effects within the staircase procedure one staircase started\\nwith a positive angle (+20 deg, right side higher) and the other with a negative angle (-20 deg,\\nleft side higher). Hence, 6 staircases were used for each phase (3 exploration modes x 2 stair-\\ncases) and the trials for these staircases were presented in a randomly interleaved fashion.\\nAfter all staircases for the pre-test were finished, a message on the screen told the participant\\nto take a break to prevent fatigue from influencing the results. After the break, participants\\nwere presented with the adaptation stimulus (surface slant of ±10 deg) for 30s. The direction\\nof adaptation surface slant (to the left or right) was counterbalanced across participants. A col-\\nour cue on the screen, like the ones used for test-trials, informed the participant about the\\nexploration mode to use during adaptation. For adaptation, it was always the cue for Static\\nBimanual exploration. During adaptation participants were not asked to decide which side felt\\nhigher. After adaptation, the post-test started. Again, the trials for the 6 staircases were ran-\\ndomly intermixed. However, in the post-test phase, each trial was preceded by 4s top-up adap-\\ntation. This means that before the actual trial, the adaptation stimulus was presented for 4s to\\nprevent de-adaptation over time. The top-up adaptation interval was again preceded by the\\nred colour cue, instructing the participant to use the Static Bimanual exploration mode. After\\nthe top-up adaptation interval, a second colour cue indicated which exploration mode to use\\non the upcoming test-trial.\\nAnalysis.\\nTo calculate the PSEs for each condition we pooled the data from the two stair-\\ncases (i.e. the staircase starting with a negative slant and the one starting with a positive slant)\\nfor each condition in the pre/post-test stage and fitted psychometric curves (cumulative Gauss-\\nian). The 50% cut-off point of the psychometric curve (i.e. the point at which there are equal\\namounts of left-side-higher and right-side higher responses for a given condition) was taken as\\nthe PSE. We then subtracted the pre-test PSEs from the post-test PSEs of each condition to\\nobtain the size of the adaptation after-effect (taking the direction of the adaptation slant into\\naccount).\\nExclusion of participants from the analysis.\\nWe removed all participants who needed\\nmore than 40 trials to finish at least one of the staircases in the design, since this is indicative of\\nthe staircases not converging. This resulted in the removal of 2 female participants. This\\nmeans that 11 participants (9 female, age range: 19–38 years) remained for the analysis.\\nResults experiment 1\\nAfter the Static Bimanual adaptation to a 10.0 deg surface slant, there was a significant afteref-\\nfect (Fig 2) when using the Static Bimanual exploration mode also in the test phases (two-tailed\\nOne sample t-test against 0, t(10) = 6.00, p<0.001; Bonferroni corrected using an alpha of\\n0.0167; Cohen’s d = 1.81), though adaptation was not complete (6.9 deg ± 1.1 deg instead of\\nthe 10.0 deg adaptation angle). This means that the angle at which the surface was perceived as\\nlevel had significantly changed between pre- and post-test.\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n6 / 24\\n\\n\\nHowever, there was no significant transfer of adaptation to the Dynamic Unimanual explo-\\nration mode, (One sample t-test against 0, t(10) = 1.22, p = 0.25; Bonferroni corrected using an\\nalpha of 0.0167; Cohen’s d = 0.37). There was also no significant transfer to the Mixed Biman-\\nual condition, in which a mixture of the static and the dynamic exploration was used (One\\nsample t-test against 0, t(10) = 2.14, p = 0.06; Bonferroni corrected using an alpha of 0.0167;\\nCohen’s d = 0.64). Using an one-way ANOVA we tested for differences between the conditions\\nand found a significant effect (F(2,30) = 5.14, p = 0.01; partial η2 = 0.26). Post-hoc paired-sam-\\nples t-tests revealed—after Bonferroni correction using an alpha of 0.0167—that the size of the\\naftereffect in the Static Bimanual condition differed significantly from the Mixed Bimanual\\ncondition (Paired t-test, t(10) = 3.20, p<0.01; Cohen’s d = 0.96) as well as the effect for the\\nDynamic Unimanual condition (Paired t-test, t(10) = 3.18, p<0.01; Cohen’s d = 0.96). The\\naftereffects for the Mixed Bimanual condition and the Dynamic Unimanual condition, how-\\never, were not significantly different from each other (Paired t-test, t(10) = 0.57, p = 0.58;\\nCohen’s d = 0.17). Together these results indicate that bimanual haptic slant adaptation is pos-\\nsible if the information of the two hands is non-redundant and furthermore, that this adapta-\\ntion is condition specific.\\nDiscussion experiment 1\\nIn Experiment 1, we tested if bimanual adaptation is possible and if this adaptation transfers to\\na dynamic movement condition when using only one hand. Our results show a significant\\naftereffect when the two index fingers statically touch the adaptation surface (Static Bimanual\\nFig 2. Adaptation aftereffect and transfer of bimanual static adaptation. On the x-axis the different movement conditions are shown: Static Bimanual (left), the\\nDynamic Unimanual (middle) and the Mixed Exploration condition (right). The y-axis shows the aftereffects as calculated by subtracting the PSE of the pre-test from\\nthe PSE of the post-test. The dashed line indicates the point at which full adaptation would occur. Error bars represent the standard error.\\nhttps://doi.org/10.1371/journal.pone.0236824.g002\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n7 / 24\\n\\n\\ncondition). This shows that also with slant input derived from two hands adaptation is possible\\n(Bimanual Adaptation).\\nSince in our experiment a slant-estimate for Static Bimanual exploration was only possible\\nwhen the information of both index fingers is combined, it seems that the interaction between\\nthe hands is adaptable. However, it has to be noted that this adaptation cannot occur at the\\nsame level at which intermanual transfer was previously observed for dynamic exploration [2,\\n6], since in the present Experiment 1 the adaptation did not transfer to exploration modes that\\ninvolved a dynamic component. This is in line with a study by Van Dam et al. [3], which\\nshowed that information from unimanual static and dynamic exploration modes do not trans-\\nfer between modes even when using the same hand. Van Dam et al., concluded that static hap-\\ntic adaptation is largely a low-level, i.e. posture based adaptation, which is dependent on the\\nexploration mode. Our results of Experiment 1 are consistent with this conclusion. They show\\nthat it is enough to include a dynamic component in the mode of surface exploration to\\ndecrease adaptational transfer effects. This can be seen most clearly in the Mixed Bimanual\\ncondition in which the position estimates of the two hands are both available and informative\\nabout the slant, yet no transfer to this condition was observed. One explanation for this might\\nbe an independent adaptation of static and dynamic exploration, as found by Van Dam et al.\\n[3], even in the case of bimanual exploration. Since the exploration mode used during adapta-\\ntion was the Bimanual Static mode, the neurons/receptors coding for static exploration\\nadapted, but the neurons coding for dynamic exploration did not adapt. Thus, the dynamic\\nexploration is unaffected by static adaptation aftereffects.\\nThis, however, raises the question whether a distal stimulus, i.e. a haptic slant, is needed to\\nadapt to slant. From the study by Van Dam et al. [3] it is known that static unimanual haptic\\nadaptation to slant is heavily dependent on the hand posture. If this is also the case for biman-\\nual adaptation a distal stimulus should not be necessary for adaptation to occur. Thus, we con-\\nducted a second experiment in which in one condition participants adapted to a haptically\\nrendered surface and in a second condition to just the finger positions by holding the index\\nfingers at fixed points in the air. For pure adaptation of posture, touching an actual object and\\nthus receiving haptic feedback from the object should not be necessary. In other words, remov-\\ning the object and adapting purely proprioceptively by holding the fingers in mid-air should\\nelicit the same effect as adapting by touching an actual surface.\\nExperiment 2\\nThe results of Experiment 1 showed that bimanual slant adaptation is exploration mode spe-\\ncific and no transfer was found to exploration modes that included a dynamic component.\\nThis suggests that even static bimanual adaptation may be heavily posture based. If so, this\\nraises the question whether an object is really needed for haptic slant adaptation to occur. To\\ninvestigate this, we conducted a second experiment in the present study. This second experi-\\nment included two conditions: In the first condition, we adapted participants in a static\\nbimanual fashion (i.e. keeping static contact with the surface using both index fingers) to a sur-\\nface slant that was rendered haptically (surface present). That is, like in the first experiment the\\nsurface could be felt and haptic feedback was provided when touching it. In the second condi-\\ntion, participants adapted–also in a static bimanual fashion–to just the corresponding position\\nin space. That is, in the second condition participants held their fingers in mid-air at the posi-\\ntions where the slant was programmed, just that now there was no surface that could be felt\\n(surface absent). Should aftereffects be present in the condition without any haptic feedback\\nand furthermore, should those effects transfer to the condition in which haptic feedback is\\navailable and vice versa, this would be clear evidence that the static bimanual adaptation is\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n8 / 24\\n\\n\\nposture based. However, if there are no aftereffects in the condition without haptic feedback,\\nor should the aftereffects not transfer, this would point towards adaptation needing the inter-\\naction with a physical surface rather than being purely posture based. Several studies showed\\nthat for instance Area 2 of the primary somatosensory cortex is particularly sensitive to the\\nspecific combinations of proprioceptive (posture) and tactile (haptic feedback) information\\n(e.g. [21–23]). This would suggest that also the combination of posture and haptic force feed-\\nback (and thus the presence of a surface) could play an important role in haptic shape percep-\\ntion in general and adaptation in particular.\\nMaterial and methods experiment 2\\nParticipants.\\nA total of 14 people volunteered to participate in the experiment (9 female,\\nage range: 20–32 years). They were all self-reported right-handed and received 6€/h as com-\\npensation for participation. They gave informed consent prior to the experiment.\\nSetup & conditions.\\nBecause we were interested in the object dependence of slant adapta-\\ntion, we had two conditions: adaptation to slant when a surface provided haptic feedback (Sur-\\nface Present condition) and adaptation to “slant” by holding the fingers in mid-air without\\ntouching a surface (Surface Absent condition). The setup was the same as in Experiment 1. In\\nExperiment 2, however, we used only the Static Bimanual exploration mode for both adapta-\\ntion as well as testing. The experiment was divided into two sessions, which for each partici-\\npant were performed on two different days. In one session, the participants adapted in the\\nSurface Present condition and in the other they adapted to posture alone in the Surface Absent\\ncondition. The order of the sessions was counterbalanced across participants. In both sessions,\\nthe test conditions were the Surface Present and the Surface Absent conditions, to test for con-\\ndition specific adaptation as well as transfer.\\nProcedure.\\nThe same adaptation procedure as in Experiment 1 was used. This time, how-\\never, no information about the upcoming trial was given. Instead the screen gave information\\nabout the finger position relative to the surface (see Fig 3). This was particularly important for\\nFig 3. Presenting information about the vertical finger distance relative to the surface. The computer screen was split in half. The left side corresponded to\\nthe left finger, the right side to the right finger. The solid line represents the surface, i.e. a touchable surface in the Surface Present condition and in the Surface\\nAbsent condition an imaginary surface. Participants initially moved their hand downward, i.e. along the gravitational axis, to reach the correct position for a\\ngiven trial. The colour of each screen half depended on how close the participant’s fingers were to the surface: the corresponding screen half turned from red to\\nyellow 15 mm above and below the surface and when the participant (would) touch the surface the corresponding screen half turned green (2.5 mm above the\\nsurface for the surface present condition, 5 mm above and below the surface for the surface absent condition).\\nhttps://doi.org/10.1371/journal.pone.0236824.g003\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n9 / 24\\n\\n\\nthe Surface Absent condition because the participant could not feel the surface. Yet we needed\\nthem to take up the specific postures that relate to a given surface slant. For providing the par-\\nticipant with information about the distance of the finger to the surface, the screen was split in\\nhalf. The right half of the screen corresponded to the right finger and the left half of the screen\\nto the left finger. To inform the participant about the vertical position of the finger, a traffic\\nlight symbolism was used. If the screen-half was red the finger(s) were far away from the sur-\\nface. As soon as the finger was closer than 15 mm to the surface, the corresponding screen half\\nturned yellow and as soon as the finger was closer than 2.5 mm (Surface Present) or 5 mm\\n(Surface Absent) the corresponding screen half turned green. The two thresholds for the green\\nlight for the Surface Present and Surface Absent conditions were different because we observed\\nin pilot experiments that with a 5 mm threshold in the Surface Present condition the partici-\\npants sometimes did not touch the surface at all during a trial if their approach was too careful.\\nOn the other hand, for the Surface Absent condition the 2.5 mm threshold turned out to be\\ntoo difficult to maintain in mid-air for both fingers simultaneously. For this reason, we chose\\nto use two slightly different thresholds in the two conditions. Depending on the condition, the\\nparticipants could feel a surface (Surface Present) or not (Surface Absent). When both fingers\\nwere in the “green zone” the trial time started. After one second the screen turned black and\\nthe participant decided which side was higher using the response zones as in Experiment 1\\n(see Fig 1B). Then the next trial started.\\nThe same statistical analysis as for Experiment 1 was used and Bonferroni correction was\\napplied for the one- and paired-sample t-tests to correct for multiple comparisons (i.e. alpha\\nwas set to 0.0125).\\nResults experiment 2\\nWhen adapting using the Surface Present condition (Fig 4, bars with solid outline), the\\nadaptation after- and transfer effects for the test conditions Surface Present (5.8 deg ± 1.6\\ndeg) and Surface Absent (4.6 deg ± 1.4 deg) were both significantly different from zero (Sur-\\nface Present, One-sample t-test: t(13) = 3.66, p<0.01; Cohen’s d = 0.98; Surface Absent,\\nOne-sample t-test: t(13) = 3.18, p<0.01; Cohen’s d = 0.85) and not significantly different\\nfrom each other (Paired t-test: t(13) = 1.02, p = 0.33; Cohen’s d = 0.27). These results con-\\nfirm the finding from Experiment 1 that bimanual adaptation to surface slant using the two\\nindex fingers in a non-redundant static fashion, leads to adaptation aftereffects for test-con-\\nditions that have the same static exploration mode. Experiment 2 shows that this is true\\nregardless of the presence of the surface. The bars in Fig 4 with a dashed outline show the\\nresults when the participants adapted to the Surface Absent condition. In this case partici-\\npants held their fingers in mid-air at the indicated positions using the screen traffic light\\nsystem. Similar to the results for adapting with a rendered surface (solid outline bars), the\\nadaptation aftereffect of the Surface Absent test condition (4.6 deg ± 1.4 deg) is significantly\\ndifferent from zero (One-sample t-test: t(13) = 3.15, p<0.01; Cohen’s d = 0.84). Again this\\naftereffect fully transferred to the Surface Present test condition (5.1deg ± 1.3deg) which\\nwas also significantly different from zero (One-sample t-test: t(13) = 3.89, p<0.01; Cohen’s\\nd = 1.04). Again, there was no significant difference between the two test conditions (t(13)\\n= 0.42, p = 0.68; Cohen’s d = 0.11).\\nThe fact that the Surface Absent and Surface Present conditions led to similar aftereffects\\nand that these fully transferred between conditions, clearly demonstrates that posture and\\nnot object presence is a crucial factor in slant adaptation. However, this raises the question\\nof whether we are dealing with bimanual adaptation at all. That is, it is not clear whether it\\nis the relative static posture between the hands that adapts (i.e. the way the position of one\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n10 / 24\\n\\n\\nhand may in part be judged in relation to the other hand), or if the results of Experiment 1\\nand 2 can fully be explained by very low-level unimanual posture adaptation (each hand\\nadapting in isolation but to slightly different postures and in this way leading to the\\nobserved aftereffects). If it is the relative positions between the hands that adapts this rela-\\ntive difference, and thus the adaptation aftereffect, should fully transfer when testing at a\\ndifferent height compared to where adaptation occurred. Adapting one hand only by keep-\\ning it in a certain posture for a period of time should however in this case not lead to any\\n“slant” aftereffects, since no adaptation of relative hand positions should occur. In contrast,\\nin the case of pure unimanual posture adaptation, proprioceptors and muscles in each hand\\nand arm get adapted. This should then lead to slightly misperceived position estimates\\nwhen the hand is moved away from the adaptation position (e.g. through muscle condition-\\ning; for further information see e.g. [15, 16, 24–26]). This means that it should be possible\\nto find adaptation effects when adapting only a single hand to a certain height and then test-\\ning how this affects position estimates when the hand is next moved to a different height. If\\nboth hands adapt at the same time in this manner but to slightly different positions, this can\\naccount for the results in the previous experiments.\\nFig 4. Adaptation effects in the two main conditions. Solid outline: The adapted condition was the Surface Present condition; Dashed outline: The adapted condition\\nwas the Surface Absent condition. On the x-axis the two test conditions are shown. The y-axis shows the adaptation aftereffect. The dashed line marks the point at which\\nfull adaptation would occur. The error bars represent the standard error.\\nhttps://doi.org/10.1371/journal.pone.0236824.g004\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n11 / 24\\n\\n\\nMaterial and methods experiment 3\\nTo distinguish between an effect due to a relative static posture adaptation and an effect based\\non a low-level unimanual adaptation, we conducted a third experiment. Here the assumption\\nwas the following: if adaptation is based on the position of each hand (unimanual) rather than\\nthe relative position between the hands, a change in position, here height, after adaptation\\nshould lead to an overestimation of the change in height for the adapted hand(s) [16, 25, 26].\\nHowever, if the relative position between the hands gets adapted, i.e. the difference in positions\\nbetween the left hand and the right hand adapts over time rather than each hand adapting\\nindividually, a change in height should not show an overestimation of the height change when\\nadapting unimanually. Rather in this case, even after bimanual adaptation, aftereffects for the\\nrelative position between the hands should not depend on the test height at all and thus remain\\nequal at different testing heights. To test these different predictions Experiment 3 included\\nadaptation conditions that involved both hands set at a “slant” by placing the two hands at dif-\\nferent heights corresponding to that “slant”. Moreover, Experiment 3 included conditions in\\nwhich only one hand was adapted by placing it at a specific height for a period of time. For\\nboth types of adaptation, the test condition consisted of placing one hand at one of three pre-\\ndefined heights and setting the other hand such that it was perceived to be at the same height.\\nParticipants.\\nFor Experiment 3 ethical approval was obtained from the University of\\nEssex Ethics Committee. A total of 11 people, including the authors CG and LD volunteered to\\nparticipate in the experiment (10 female, age range: 20–40 years). They were all self-reported\\nright-handed and student volunteers received course credits as compensation for their partici-\\npation. They gave informed consent prior to taking part in the experiment.\\nGeneral setup.\\nThe findings that the observed “slant” aftereffects seem to be posture\\nbased, rather than requiring haptic force feedback about the object, allowed us to move away\\nfrom the PHANToM force feedback devices which have only a limited workspace. For Experi-\\nment 3 we instead used the Oculus Rift VR headset and touch controllers (Oculus Rift CV1\\nFacebook Technologies, LCC) to both guide the participants to the correct hand position for\\neach adaptation and test condition as well as measure the hand positions using the touch con-\\ntrollers. This furthermore allowed us to measure adaptation aftereffects at more extreme\\nheights compared to what would be possible with the PHANToM force feedback devices. To\\nbe able to verify that the participants followed the instructions, the hand positions during vari-\\nous stages of the trials were recorded with a sampling frequency of 90 Hz.\\nIn Experiment 3, in the pre- and post-test phases the participants were guided to place one\\nof their hands at a certain position in 3D space using a visual guidance system in the VR head-\\nset (see Fig 5). Once their hand was in the correct position, they then had the task to match the\\nheight of their “set hand” with their “free hand”. This way we obtained on each individual trial\\na measure of the height differences at which the participants perceived their two hands to be at\\nthe same level. During the adaptation phase, the same visual guidance system was used to have\\nparticipants place either one or both of their hands (depending on the condition) in such pre-\\ndefined 3D positions.\\nTo guide the participants to the correct position for the set hand(s), we gave visual feedback\\nas seen in Fig 5. The left cross corresponds to the left hand, the right cross to the right hand.\\nThe goal for the participant was to get all squares yellow. As soon as the controller left the goal\\narea in a certain direction, the corresponding square(s) turned red indicating to the participant\\nthey had to place their hand more in the opposite direction. The goal area was defined as a\\n3-dimensional box spanning 2.0 cm in the horizontal and vertical directions and 4.0 cm in\\ndepth. The goal area along the depth direction was double the size since it was harder to main-\\ntain compared with the other two dimensions. Furthermore, the depth direction was not of\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n12 / 24\\n\\n\\nmain interest in this experiment and therefore did not require the same level of precision. To\\ncontrol for the right position in depth, we used vibration. As soon as the participant moved\\nout of the goal area to the front or back the controller(s) started to vibrate, telling the partici-\\npant to correct for depth. It is important to note that the visual placement of the crosses was\\nfixed for the whole course of the experiment and thus its position in virtual space did not cor-\\nrespond in any meaningful way to the position of the hand in real space. Therefore, this guid-\\nance system only provided feedback to correct the hand position if necessary and did not\\nprovide visual feedback as to the precise 3D coordinates of the hand(s) in space. Note that the\\ncross(es) for the “set hand” in the visual display remained visible throughout the experiment\\n(i.e., also during adaptation and test phases) in order to allow readjustments in case partici-\\npants unintentionally left the goal area with their hand.\\nFor bimanual adaptation both crosses of the visual guidance system were shown. The goal\\nareas for the hands were 7.0 cm to the left of the body midline for the left hand (using the posi-\\ntion of the VR-headset as a reference) and 7.0 cm to the right for the right hand, with a height\\ndifference between the hands of 10.0 cm centred around the shoulder area (20.0 cm below the\\nVR-headset). The hands furthermore needed to be placed at a distance in depth of 30.0 cm.\\nNote that the height difference roughly corresponds to a slant of 36 deg instead of 10 deg as\\nused in the previous experiments. This was done since we had to allow for the range of goal\\nareas in which participants placed their hands as well as for the idea that we were working with\\nhand position rather than fingertip positions. A “slant” of 10 deg would have easily been lost in\\nthe possible variable placement of the hands within the respective goal areas.\\nFor unimanual adaptation only the cross corresponding to the adapted hand was shown\\nusing the colour representations described above. The adapting position would again be placed\\n7.0 cm to the left or right, depending on whether the left or right hand was adapted, at roughly\\nshoulder height (i.e. 20.0 cm below the position of the VR headset) and 30.0 cm in depth from\\nthe VR headset. The squares making up the cross corresponding to the non-adapting hand\\nwere visible but black. The non-adapting hand was held down in a relaxed fashion.\\nGeneral procedure.\\nThe experiment started with a short training block in which the par-\\nticipants were familiarized with the setup and how to interpret the colour coding and vibra-\\ntional feedback. After the training session the experiment started. The experiment was done in\\na blocked design, i.e. each adaptation condition was done in a separate block of trials. After\\nFig 5. Visual feedback the participants received to get to the correct positions with their hands. Shown is an example for a bimanual\\nadaptation phase. In this example the participant holds the right hand in the correct x- and y-coordinates (+/- 1.0 cm). The left hand is held at\\nthe correct y-coordinates (+/- 1.0 cm) but more than 1.0 cm to the right of the goal coordinates. Therefore, the right square of the left cross is\\nshown red.\\nhttps://doi.org/10.1371/journal.pone.0236824.g005\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n13 / 24\\n\\n\\neach block there was a break of 10 minutes in which the participants were allowed to rest their\\narms, take off the VR headset and were encouraged to do things with their hands to help the\\nde-adaptation (e.g. drink, eat a snack, using the smartphone etc.). After the break, the next\\nblock with the next adaptation condition started.\\nEach block consisted of a pre-test phase, the adaptation phase and a post-test phase, as in\\nthe previous experiments.\\nBimanual adaptation condition.\\nTo be able to compare our results of Experiment 3 to\\nthe previous experiments we had a bimanual adaptation condition in which both hands had a\\ngoal area during the adaptation phases. Each participant performed two blocks of trials for the\\nbimanual adaptation condition. In one block the right hand was held higher during the adap-\\ntation phases (positive slant), in the other block the left hand was held higher during adapta-\\ntion (negative slant). Fig 6 shows sketches of the different adaptation conditions and the\\ndifferent testing heights. In Fig 6A the controller positions (for a positive slant) as well as the\\nvisual feedback given by the VR glasses are shown. For the main adaptation phase participants\\nheld their hands in the indicated goal area for 30 seconds. In the pre- and post-test phases, we\\nused the testing conditions as explained above: one hand (the set hand) was guided to one of\\nthe three testing heights (see Fig 6C) using the visual guidance system (the other cross was\\nblack) and participants next had to match it with the other hand (the free hand) without any\\nvisual feedback. Once satisfied that their hands were at the same height, participants pressed\\neither “X” or “A” on one of the controllers to start the next trial. Which hand was used as the\\nset hand and which as the free hand was counterbalanced across trials. Per set hand each test-\\ning height was repeated three times. This led to a total number of 36 test trials for each block (2\\nhands x 3 heights x 3 repetitions = 18 test trials for each of the pre and post-test phases). The\\norder of the conditions was randomized in each test-phase.\\nAs in the previous experiments, the post-test differed from the pre-test, i.e. that each test-\\ntrial was preceded by a 4 second top-up adaptation interval in which participants were guided\\nto take up the same hand positions as during the main adaptation phase. Participants were\\nnotified what they needed to do at each stage through messages displayed in the virtual envi-\\nronment (e.g. keep hands in the same position for adaptation intervals, or move the “free”\\nhand to the same height as the “set” hand in the test-phases).\\nUnimanual adaptation condition.\\nIn the unimanual adaptation condition, only one\\nhand was adapted at shoulder height. There were two blocks of trials for the unimanual condi-\\ntion. In one block the left hand was adapted, in the other block the right hand was the adapted\\nhand. For the adaptation phases the hand to be adapted was guided to the correct adaptation\\nheight using the visual guidance system explained above. Participants were instructed to hold\\nthe other arm down in a resting position during the main adaptation phase (30 seconds) as\\nwell as during the top-up adaptation intervals (4 seconds) of the post-test phase. Fig 6 shows\\nthe controller positions and the visual feedback for a right-hand adaptation condition. Note\\nthat in this case one cross, namely the cross of the unadapted hand, was shown in black, i.e. no\\nvisual feedback was provided for the non-adapting hand. For test trials the adapting hand for\\nthat block was guided to one of the three testing heights as seen in Fig 6C and participants next\\nhad to try and match the felt height with their non-adapted hand. Each test-height was\\nrepeated 3 times in each of the pre- and post-test phases. Therefore, the number of trials in the\\nunimanual adaptation conditions was 18 trials per block (9 trials in the pre-test + 9 trials in the\\npost-test).\\nAs indicated above, we used three different testing heights in the pre-test phase as well as in\\nthe different post-test phases to which one hand (the “set hand”) of the participant was guided\\nto. One testing height was at eye level (called “Head”), as determined by the location of the VR\\nheadset in space. The second testing height was at 20.0 cm below the centre of the VR headset\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n14 / 24\\n\\n\\nFig 6. Sketches of the conditions and testing heights in the third experiment. A: Controller positions during bimanual adaptation. Both hands are raised to\\nkeep the two crosses in the visual feedback yellow; B: Controller positions during unimanual adaptation (right hand). The adapted hand is raised (in this\\nexample the right hand) whereas the left hand is held in a relaxed position. In the unimanual conditions the cross corresponding to the unadapted hand was\\nshown black. Note that in the picture the visual feedback shown is the one the participant sees in the VR glasses (i.e. mirrored to the observer); C: Testing\\nheights of the experiment. The red lines show the testing heights in relation to the participant’s body. Note that we used the coordinates of the VR headset as\\nthe reference for the correct placement of the set hand. Thus, the testing positions relative to the body differed slightly between participants, depending how tall\\nthe participant was. The dashed line marks the adaptation height.\\nhttps://doi.org/10.1371/journal.pone.0236824.g006\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n15 / 24\\n\\n\\nwhich roughly corresponded to shoulder height (called “Shoulder”). The third height for test-\\ning was at 40.0 cm below the centre of the VR headset, which roughly corresponds to chest\\nheight (called “Chest”, see Fig 6C).\\nAnalysis.\\nTo analyse the effects of adaptation, we analysed the height hand settings for pre-\\nand post-test trials. This is the height at which participants felt their hands to be at the same\\nheight. To determine these heights, we took the y-coordinate of the hands at the moment the\\nparticipant pressed the “X” or “A” button on the Oculus TouchTM controllers to indicate that the\\nmatching of the hands was complete. We then subtracted the coordinate of the left and right\\nhand to calculate the relative height difference for each trial. Furthermore, we pooled the data\\nacross the two blocks for each of the bimanual and unimanual adaptation conditions (mirroring\\nthe data where necessary), as the effects were symmetric for the two hands. Here handedness did\\nnot play a role. For the statistical analysis we compared the mean results in terms of the relative\\nheight differences in the settings for each adaptation condition and each testing height to zero\\nwith a one-sample t-test and we used paired-sample t-tests for comparisons between the different\\ntesting heights for each adaptation condition. Bonferroni correction was applied for the one-\\nand paired-sample t-tests to correct for multiple comparisons (i.e. alpha was set to 0.0167).\\nResults experiment 3\\nFig 7 shows the results for the Bimanual Adaptation condition of Experiment 3 (Fig 7A)\\ntogether with the Unimanual Adaptation condition (Fig 7B). The x-axis shows the height at\\nwhich the test was performed relative to the height that was used for adaptation and the y-axis\\nthe size of the aftereffect in cm.\\nUsing these results, we first verified whether the same effects of bimanual adaptation also\\nappear with the VR setup, i.e. in 3D virtual space without force feedback. To do so here in\\nFig 7. Results of the bimanual and unimanual adaptation. A: Results of the Bimanual Adaptation conditions; B: Results of the Unimanual Adaptation conditions. The\\nx-axes show the different testing heights relative to the adaptation height. The y-axis in A shows the bimanual “slant” aftereffect and in the unimanual adaptation the\\nheight of the “free” hand relative to the set hand. “Shoulder” is the adaptation height, “Chest” is the testing height 20 cm below the adaptation height and “Head” is the\\ntesting height 20 cm above the adaptation height. The errorbars represent standard errors.\\nhttps://doi.org/10.1371/journal.pone.0236824.g007\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n16 / 24\\n\\n\\nExp. 3 we used the one bimanual adaptation condition that was the most similar to the static\\nbimanual adaptation conditions of the previous Experiments 1 and 2, except for using the\\nOculus Rift with the Touch controllers instead of the PHANToM force-feedback devices. This\\nwas the bimanual adaptation condition for which both test and adaptation occurred at the\\nsame “Shoulder” level height (see Fig 7A, the middle bar). It can be seen that an aftereffect\\noccurs also in this case (one-sample t-test against zero for “Shoulder” level: t(10) = 5.06,\\np < 0.01; Cohen’s d = 1.53) despite the fact that there were no boundaries and thus no force or\\nother kind of external haptic feedback was present.\\nNext, we tested for effects of adaptation transfer at “Chest” and “Head” level also in the\\nBimanual Adaptation condition (Fig 7A, “Chest” level: left bar and “Head” level: right bar). It\\ncan be seen that such a transfer effect occurred at least to some extent for the “Chest” level\\n(one sample t-test: t(10) = 4.22, p<0.01; Cohen’s d = 1.27) but not for the “Head” level (one\\nsample t-test: t(10) = 0.01, p = 0.99; Cohen’s d<0.01). However, both the results for the “Shoul-\\nder” and “Chest” level are significantly different to the “Head” level (paired samples t-test\\nChest-Head: t(10) = 5.36; p<0.001; Cohen’s d = 1.62; Shoulder-Head: t(10) = 4.47; p<0.01;\\nCohen’s d = 1.35) but not significantly different to each other (Chest-Shoulder: t(10) = 1.46;\\np = 0.18; Cohen’s d = 0.44). Thus, the transfer effects, when testing at different heights than the\\nadapted height, were significantly reduced only in one condition (“Head” level) whereas a sig-\\nnificant transfer effect was observed for the second transfer condition (“Chest” level). Since\\nthese results are mixed, it is difficult to make any strong conclusions. However, the above\\nshown results—together with the results of the previous experiments (which point towards\\nreceptor based adaptation)—hint towards the assumption that it may not be the relative posi-\\ntion between the hands at a bimanual stage that gets adapted, in which case we would have\\nexpected the adaptation aftereffect to more or less fully transfer to both the different testing\\nheights. Since this is not the case, adaptation may perhaps actually be occurring at the uniman-\\nual level.\\nWe used the Unimanual Adaptation condition to verify this suggestion. If bimanual adapta-\\ntion occurs at the unimanual level, adapting only one hand to a certain height and then mov-\\ning it to another height, should lead to an overshoot in the position estimation of this hand.\\nThus, in the Unimanual Adaptation condition only one hand was adapted to the “Shoulder”\\nlevel and we then measured whether aftereffects, i.e. a misjudgement of the “set hands” posi-\\ntion, occurred at the same and different testing heights. The results are shown in Fig 7B. The\\nx-axis shows the testing height relative to the adaptation height (“Chest” = -20.0 cm, “Shoul-\\nder” = 0.0 cm, “Head” = +20.0 cm); the y-axis represents the height difference between the free\\nhand and the adapted “set” hand at which the hands are perceived to be at the same height.\\nWhen testing at the same height as the adaptation took place, no significant difference in\\nheight perception occurred (one-sample t-test t(10) = 1.85, p = 0.09; Cohen’s d = 0.56. The\\nresults show a significantly negative distance for the testing height at “Chest” level, indicating\\nthat the participants perceived the adapted hand to be lower than it actually was (t(10) = 5.34,\\np<0.001; Cohen’s d = 1.61). For the testing height “Head” however, the distance is significantly\\npositive, indicating that the participants perceived the adapted hand to be held at a higher posi-\\ntion than it actually was (t(10) = 7.14, p<0.001; Cohen’s d = 2.15). This means that for both\\nthe “Chest” testing level and the “Head” testing level the participants overestimated the dis-\\ntance that the hand had moved from the adaptation level, which is consistent with adaptation\\neffects in perception. Furthermore, the results of the three conditions are significantly different\\nfrom each other (paired-sample t-test Chest-Shoulder: t(10) = 3.32, p<0.01; Cohen’s d = 1.00;\\nChest-Head: t(10) = 7.86, p<0.001; Cohen’s d = 2.37; Shoulder-Head: t(10) = 7.00, p<0.001;\\nCohen’s d = 2.11). These results confirm that haptic adaptation can occur for a single hand\\nposition individually.\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n17 / 24\\n\\n\\nDiscussion experiment 3\\nThe results again confirm that bimanual adaptation in 3D space is possible without needing to\\ntouch any surface. This means, that even when the participant is simply holding their hands in\\na certain position in 3D space without external force feedback, adaptation aftereffects occur.\\nThe results of the unimanual adaptation show that the participants significantly misjudge the\\nposition of the adapted hand when this hand is moved. That is, the adapted hand is perceived\\nsignificantly lower when moved downwards and significantly higher when moved upwards.\\nThis effect was already described by Gregory et al. [26] and was confirmed here. Furthermore,\\nthis shows that adaptation to height is possible with a single hand and thus points towards\\nadaptation at the level of the individual hands (e.g. through adaptation of the muscle spindles)\\nrather than an adaptation of the two hands in relation to each other. Though the results for the\\nbimanual condition are not entirely conclusive, the finding that the Bimanual Adaptation\\ntransfer effect is significantly reduced when tested at “Head” level is in line with this interpreta-\\ntion. Adaptation of relative hand positions instead of adaptation of each individual hand\\nshould be independent of the location/posture at which adaptation and testing occurs, and we\\nwould expect aftereffects to fully transfer to any other location. In the present experiment this\\nwould mean that for bimanual adaptation the results at non-adapted locations (“Chest” and\\n“Head” levels) should have been the same as at the adapted height (“Shoulder” level). This is\\nevidently not the case in the present results when testing at “Head” level. This absence of trans-\\nfer of the aftereffect to “Head” level cannot simply be due to biomechanical constraints because\\nwe did find strong unimanual aftereffects at this height. Therefore, our results show that at the\\nvery least such an adaptation is again posture dependent and does not necessarily transfer to\\nall non-adapted postures. It has to be noted however, that since we did not observe a signifi-\\ncant reduction of adaptation transfer when testing at the “Chest” level, it would be premature\\nto completely rule out a role of adaptation of relative hand positions.\\nTaken together, the results from all three experiments confirm that the posture at which\\nadaptation occurs is the most important factor. This indicates at the very least a very important\\nrole for unimanual adaptation processes for generating such aftereffects. Moreover, the unim-\\nanual condition in Experiment 3 highlights that bimanual aftereffects could potentially even be\\nfully explained by unimanual adaptation.\\nLastly, it is of interest to note that, across the three experiments we observed very similar\\nadaptation aftereffects for the bimanual adaptation conditions. Yet, in Experiment 3 we used\\ncontrollers, which had to be grasped by the participants while in the other experiments we\\nused the PHANToM robot arms in which only the fingertips were used. Combined, the pres-\\nent results therefore suggest that the haptic slant adaptation is likely related to the position of\\nthe arms and shoulders and not solely on the finger positions per se.\\nGeneral discussion\\nIn the first part of the present study, we investigated if bimanual adaptation to slant is possible\\nin conditions in which it is essential that the information from both hands is used (non-redun-\\ndant information). The results of Experiment 1 showed that Static Bimanual slant adaptation\\ndoes occur. Furthermore, the Static Bimanual adaptation aftereffect transferred neither to the\\nDynamic Unimanual condition nor to the Mixed Bimanual condition in which dynamic and\\nstatic exploration were mixed and position information for both fingers was available (Mixed\\nBimanual). These results extend the findings by Van Dam and colleagues [3], who found that\\nstatic and dynamic exploration adapt independently when tested within one hand, to the\\nbimanual case. In Experiment 2 we tested whether a distal stimulus is needed for adaptation\\nand showed that a physical object is not necessary to elicit haptic adaptation aftereffects. This\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n18 / 24\\n\\n\\nsuggests that also bimanual adaptation is posture based. Finally, Experiment 3 provides evi-\\ndence that this adaptation is most likely linked to adaptation at the level of the individual\\nhands rather than at a level at which the relative position differences between the hands is\\nrecalibrated.\\nBimanual adaptation to slant\\nIn the present study we showed, for the first time, that adaptation to a haptic feature, in this\\ncase slant, also works when the two hands are simultaneously involved in the adaptation pro-\\ncess. In earlier studies adaptation to haptic features was already shown (size and volume: e.g.\\n[27]; curvature: e.g. [2, 4–6]; slant: [3]), but only within one hand. Our study extends these\\nfindings by showing that adaptation to slant also occurs when slant is estimated using two fin-\\ngers from different hands. Here it is important to note that in this study as well as in the study\\non slant adaptation by Van Dam et al. [3], one static finger was not enough to estimate the\\nslant of the surface. One needs a second finger to be able to make a judgment of the surface\\nslant by estimating the difference in position between the fingers. In the present study the two\\nfingers used were from the two different hands and thus the slant could only be estimated by\\ncombining information from the two hands. Our findings show that this nevertheless resulted\\nin adaptation aftereffects.\\nNo transfer of aftereffects between exploration modes\\nIn this study we furthermore showed that the bimanual slant adaptation is exploration mode\\nspecific and does not transfer to conditions with a dynamic exploration component. Estimat-\\ning slant is also possible by using a single finger and moving it in a dynamic fashion to sample\\nthe height differences over time by sliding over the surface. Thus, there are two ways to obtain\\ninformation about slant (statically and dynamically) that intuitively might share common neu-\\nral pathways since they serve the same purpose. In this case, the adaptation should be indepen-\\ndent of the exploration mode and transfer between them. The Static Bimanual adaptation\\nfound in this study, however, did not transfer to conditions that had any form of dynamic\\ncomponent, even with two hands present on the surface and thus relative position estimates\\nbetween the hands still being available (Mixed Bimanual Condition of Experiment 1). An\\nexplanation for the lack of transfer is that Static Bimanual adaptation is dependent on the\\nexploration mode–i.e. based on the postures of the individual hands (for a review see [15,\\n16])–rather than at a stage at which both hands are represented. At first glance, this seems to\\ncontradict the findings of Van der Horst et al. [2, 6], who showed that adaptation to curvature\\ntransfers from the adapted hand to the non-adapted hand. Intermanual transfer was particu-\\nlarly found for dynamic information gathering, which points towards a bimanual processing\\nstage [2]. However, van der Horst and colleagues [6] also found that intermanual transfer was\\nmuch reduced or absent when using static contact with the curvature, showing that the biman-\\nual processing stage may be very particular to dynamic exploration only. This is in line with an\\nindependence between static and dynamic exploration modes and, rather than Static Bimanual\\nadaptation occurring at a bimanual level, suggested an alternative explanation for the present\\nresults of Experiment 1. In the present case, the slant percept is likely derived by estimating the\\ndistances between the fingers along the horizontal and vertical dimensions. If the perceived\\npositions of the individual fingers adapt (rather than the slant), this would lead to changes in\\nslant perception after adaptation, despite the adaptation not specifically occurring at a biman-\\nual processing stage that estimates the slant. This would also explain why we did not find trans-\\nfer to the Mixed Bimanual condition, since in that case one finger is not providing a stable\\nposition estimate. Yet, moving the fingers can provide a, perhaps more accurate, estimate of\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n19 / 24\\n\\n\\nthe slant based on its dynamic exploration that has remained unadapted. This is consistent\\nwith the results by Van Dam et al. [3] who showed that adaptation does not transfer between\\ndynamic and static exploration with the same hand.\\nAll in all, our results strongly suggest that Static Bimanual exploration is processed differ-\\nently compared to the bimanual stage for the dynamic exploration mode that Van der Horst\\nand colleagues [2] proposed. Furthermore, the results of both the current experiment 1 and 2\\nsuggest the strong posture dependence found by Van Dam and colleagues [3] is also true for\\nbimanual static adaptation to slant. This is in line with a study by Vogels et al. [5], that showed\\nthat for unimanual adaptation posture has an effect on the adaptation aftereffect. In their study\\nparticipants had to make either a fist, hold the hand passive in mid-air or bend and stretch the\\nfingers after adaptation and before testing. They then tested how fast the curvature adaptation\\ndecays in the different conditions. They found that when a fist was made before testing, the\\ndecay time is significantly shorter than when holding the hand passive in mid-air. That is, the\\nfist posture of the hand interfered with the adaptation aftereffect. This showed that posture is a\\nfactor in haptic adaptation, which is in line with our findings. However, the study by Vogels\\nand colleagues [5] did not investigate the bimanual case nor whether there is a difference\\nbetween adapting to posture alone and posture plus haptic feedback from the touched object\\n(or own hand).\\nInfluence of cutaneous cues\\nDue to the fact that we used force-feedback devices to present the slanted surface, there were\\nno direct cutaneous cues present for the slant of the surface. Instead, the cues available in the\\npresent study were the force-feedback from the surface (Experiments 1 and 2) and propriocep-\\ntive cues about the hand/finger postures (all three experiments). This is different from most\\nprevious studies in which real objects were presented and for which thus both proprioceptive\\nand cutaneous cues were available. From previous research it is known that such cutaneous\\ncues also adapt when available (e.g. [28]). However, even despite the difference in the presence\\nof cutaneous cues the results from this study are very consistent with the work from Vogels\\net al. [4, 5, 29] and Van der Horst et al. [2, 6], which are all studies involving real objects and\\nthus included both proprioception and cutaneous cues. Hence, it is likely–at least for adapta-\\ntion to global shape–that cutaneous cues play only a minor role. This may however be very dif-\\nferent for adaptation to predominantly tactile stimuli, such as the texture of a surface or other\\nstimuli that fit within the area of a single fingertip, for which adaptive interactions between the\\nhands have been observed in the CNS to at least some degree [13, 14].\\nPosture-based haptic slant adaptation\\nExperiment 2 addressed whether haptic adaptation is a purely proprioceptive adaptation. If\\nstatic adaptation is indeed posture based, after-effects should be found even in the absence of a\\nphysical surface during adaptation. In Experiment 2, we therefore removed the haptic surface\\nduring adaptation in one condition and the results show that Static Bimanual adaptation\\nindeed also occurs when adapting to posture alone (i.e. with the fingers held in mid-air). Fur-\\nthermore, there are no differences in magnitude of adaptation between the Surface Present\\nand Surface Absent conditions and adaptation fully transfers between these two conditions.\\nThis indicates that haptic feedback, i.e. the increased force when touching the surface and the\\ndifferences in muscle tension induced by this, makes no difference for haptic slant adaptation.\\nThis strongly supports the idea by Van Dam et al. [3], that static haptic adaptation to slant is\\nmainly posture based. In the study by Van Dam and colleagues [3] hand posture was a crucial\\nfactor for finding aftereffects in adaptation when testing using static contact with the object.\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n20 / 24\\n\\n\\nThey found that the average hand posture during the dynamic adaptation phases had a strong\\nimpact on the transfer effect to a static testing condition and a testing condition in which pos-\\nture and dynamic components were combined. This leads to the assumption that static haptic\\nslant adaptation is rather a proprioceptive adaptation that does not rely on haptic feedback\\nfrom an object, at least not to any measurable extent. Our results are consistent with the idea\\nthat each finger adapts individually to its own posture based on the proprioceptive sensory\\ninput from, for instance, muscle spindles and skin stretch (for reviews see e.g. [15, 16]). For\\nadapting one hand, posture adaptation makes sense, given that it can be linked to one group of\\nmuscles and joints. Interestingly, for adapting to slant using two hands, where the fingers of\\nthe separate hands act independently without a mechanical link, we still found similar results,\\ndespite the hands needing to share information to estimate slant.\\nComparison between possible explanations for the site of adaptation\\nBased on the finding that proprioceptive posture is a key factor for bimanual adaptation it\\nseems plausible that the proprioceptors of the individual hands are involved in the adaptation\\nprocess. In theory adaptation at this level could fully explain the present findings. However, it\\nis important to note that there is an alternative explanation for the present results, which is\\nthat adaptation occurs at a higher level at which the position of one hand is compared to the\\nposition of the other hand. When estimating the “slant”, or as in the Surface Absent condition\\nof our second experiment the relative positions of the two fingers, this requires the information\\nof the two hands to be shared. This means that this comparison necessarily has to take place at\\na processing stage at which both hands are represented. Adaptation at such a stage, rather than\\nadaptation at the level of the individual hands, would for instance explain why the adaptation\\nsurface itself tends to feel more level as time progresses. In other words, each hand may adapt\\nto the position of the opposite hand which then would lead to the stable percept of a level sur-\\nface over time. This is in line with the idea that symmetry is preferred by the body (e.g. for\\nvision: [30, 31]; for locomotion: [32]; for hand movements: [33, 34]; for joint information pro-\\ncessing: [35]). In the case of adaptation to slant one hand or finger is higher than the other,\\npossibly driving the adaptation to a point at which both hands/fingers feel level. Thinking of\\nnatural statistics this makes sense. If the two arms are passively hanging down from our shoul-\\nders, the fingers, hands and arms are roughly in symmetry. This raises the idea that during\\nadaptation the brain is adjusting what symmetry between the limbs feels like. In other words: a\\nreference for the position of one hand could in fact be the other hand, i.e. the right hand’s posi-\\ntion is the reference for the left hand’s position and vice versa. This way one would adapt in a\\nway that the perceived distance between the two hands decreases. This would also lead to the\\nalignment aftereffects found in the present study.\\nSince the two theories are in conflict with each other, we conducted a third experiment in\\nwhich we tested whether unimanual adaptation to a certain height leads to adaptation afteref-\\nfects. If the adaptation from the previous experiments was based on muscle spindle and skin\\nstretch adaptation, it should be possible to find adaptation effects when adapting only one\\nhand. If the previously found effects were based on adaptation of the relative hand positions at\\na bimanual stage, unimanual adaptation should not show any effects. The results show that\\nwhen adapting one hand to a certain position and then moving the hand up or down, leads to\\nthe impression that the hand moved further than it actually did. This is in line with the find-\\nings of Gregory et al. [26] who found that when flexing or stretching the elbow flexors the per-\\nceived limb position changes. The reason for this is that the firing rate of the involved\\nreceptors in the muscles and joints decrease their background discharge rates over time when\\nheld static in a certain position. Thus, when moving again the firing rate of the receptors in\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n21 / 24\\n\\n\\nrelation to the background discharge rate is higher, leading to the impression that a larger dis-\\ntance was moved [16, 36]. The findings of the third experiment match these previous findings,\\ntherefore suggesting that each arm or hand adapts individually. Since the task for the partici-\\npants was to match the height of the adapted hand with the unadapted hand, the brain still\\nneeds to compare the position of the two hands. However, since adaptation leads to the mis-\\njudgement of the position of the adapted hand [25], also the height difference at which the\\nhands are perceived as level is misjudged. As shown in Experiment 3, the effects of unimanual\\nadaptation were quite strong and therefore likely dominated also when adapting bimanually to\\nslant. This in part, if not completely, can also explain the findings for the bimanual adaptation\\nconditions in this experiment if the shift in perceived position depends on the distance moved.\\nIt has to be noted though that the conditions in Experiment 3 did not allow us to work out the\\nextent to which unimanual adaptation alone can account for all the adaptation effects in this\\nstudy. Therefore, a role of adaptation at a bimanual comparison stage, though unlikely, cannot\\nyet be completely ruled out. However, based on the present findings it can be safely assumed\\nthat if such adaptation at a bimanual comparison stage exists its role is likely relatively minor.\\nConclusion\\nOur results show that it is possible to adapt bimanually to slant using static touch and that this\\nadaptation does not transfer to conditions that involve a dynamic exploration component,\\neven if the relative positions of both hands are still informative about the slant. Furthermore,\\nwe demonstrated that for haptic adaptation the presence of an object is not necessary to elicit\\nadaptation aftereffects and that the observed aftereffects are based on the adaptation of posture\\nfor each hand and arm individually. Hence, taken together we conclude that although slant\\nestimation needs the input of both hands, Static Bimanual adaptation is largely of propriocep-\\ntive nature at the level of the individual hands. That is, the posture information of the individ-\\nual hands is already biased before it arrives at the stage in the CNS at which the hand positions\\nare compared.\\nSupporting information\\nS1 File. Video conditions experiment 3. This video shows the different conditions in experi-\\nment 3 as seen by the participant through the VR glasses.\\n(MP4)\\nAcknowledgments\\nWe gratefully thank Sarah Hanke for helping to conduct experiment 2.\\nAuthor Contributions\\nConceptualization: Catharina Glowania, Myrthe A. Plaisier, Loes C. J. Van Dam.\\nData curation: Catharina Glowania.\\nFormal analysis: Catharina Glowania, Loes C. J. Van Dam.\\nInvestigation: Catharina Glowania.\\nMethodology: Loes C. J. Van Dam.\\nProject administration: Catharina Glowania, Loes C. J. Van Dam.\\nSoftware: Catharina Glowania, Loes C. J. Van Dam.\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n22 / 24\\n\\n\\nSupervision: Myrthe A. Plaisier, Marc O. Ernst, Loes C. J. Van Dam.\\nVisualization: Catharina Glowania.\\nWriting – original draft: Catharina Glowania.\\nWriting – review & editing: Myrthe A. Plaisier, Marc O. Ernst, Loes C. J. Van Dam.\\nReferences\\n1.\\nKlatzky R, Lederman S. Hand Movements: A Window into Haptic Object Recognition. Cogn Psychol\\n1987; 19: 342–368. https://doi.org/10.1016/0010-0285(87)90008-9 PMID: 3608405\\n2.\\nVan der Horst B, Willebrands W, Kappers A. Transfer of the curvature aftereffect in dynamic touch.\\nNeuropsychologia 2008b; 46: 2966–2972.\\n3.\\nVan Dam LC, Plaisier MA, Glowania C, Ernst MO. Haptic adaptation to slant: No transfer between\\nexploration modes. Sci Rep 2016; 6: 34412. https://doi.org/10.1038/srep34412 PMID: 27698392\\n4.\\nVogels I, Kappers A, Koenderink J. Haptic aftereffect of curved surfaces. Perception 1996; 25(1): 109–\\n119. https://doi.org/10.1068/p250109 PMID: 8861174\\n5.\\nVogels I, Kappers A, Koenderink J. Investigation into the origin of the haptic aftereffect of curved sur-\\nfaces. Perception 1997; 26(1): 101–117. https://doi.org/10.1068/p260101 PMID: 9196695\\n6.\\nVan der Horst B, Duijndam M, Ketels M, Wilbers M, Zwijsen S, Kappers A. Intramanual and intermanual\\ntransfer of the curvature aftereffect. Exp Brain Res 2008a; 187: 491–496.\\n7.\\nWalker J, Shea K. Tactual size aftereffect contingent on hand position. J Exp Psychol 1974; 103(4):\\n668–674. https://doi.org/10.1037/h0037136 PMID: 4448966\\n8.\\nPont S, Kappers A, Koenderink J. Similar mechanisms underlie curvature comparison by static and\\ndynamic touch. Percept Psychophys 1999; 61(5): 874–894. https://doi.org/10.3758/bf03206903 PMID:\\n10499001\\n9.\\nGoodwin A, John K, Marceglia A. Tactile discrimination of curvature by humans using only cutaneous\\ninformation from the fingerpads. Exp Brain Res 1991; 86: 663–672. https://doi.org/10.1007/\\nBF00230540 PMID: 1761098\\n10.\\nPont S, Kappers A, Koenderink J. Haptic curvature discrimination at several regions of the hand Percept\\nPsychophys 1997; 59: 1225–1240. https://doi.org/10.3758/bf03214210 PMID: 9401457\\n11.\\nIwamura Y. Bilateral receptive field neurons and callosal connections in the somatosensory cortex. Phi-\\nlos Trans R Soc Lond B Biol Sci, 2000; 355: 267–273. https://doi.org/10.1098/rstb.2000.0563 PMID:\\n10724460\\n12.\\nIwamura Y, Taoka M, Iriki A. Bilateral Activity and Callosal Connections in the Somatosensory Cortex.\\nNeuroscientist 2001; 7(5): 419–429. https://doi.org/10.1177/107385840100700511 PMID: 11597101\\n13.\\nTamè L, Pavani F, Papadelis C, Farnè, A, Braun C. Early integration of bilateral touch in the primary\\nsomatosensory cortex, Hum Brain Mapp 2015; 36: 1506–1523. https://doi.org/10.1002/hbm.22719\\nPMID: 25514844\\n14.\\nTamè L, Braun C, Holmes NP, Farnè A, Pavani F. Bilateral representations of touch in the primary\\nsomatosensory cortex, Cogn Neuropsychol 2016; 33: 48–66. https://doi.org/10.1080/02643294.2016.\\n1159547 PMID: 27314449\\n15.\\nProske U, Gandevia S. The kinaesthetic senses. J Physiol 2009; 587(17): 4139–4146.\\n16.\\nProske U, Gandevia S. The proprioceptive senses: their roles in signaling body shape, body position\\nand movement, and muscle force. Physiol Rev 2012; 92: 1651–1697. https://doi.org/10.1152/physrev.\\n00048.2011 PMID: 23073629\\n17.\\nDenisova K, Kibbe M, Cholewiak S, Kim SH. Intra- and intermanual curvature aftereffect can be\\nobtained via tool-touch. IEEE Trans Haptics 2014; 7(1), 61–66. https://doi.org/10.1109/TOH.2013.63\\nPMID: 24845746\\n18.\\nDupin L, Hayward V, Wexler M. Direct coupling between hands. Proc Natl Acad Sci U S A 2014; 112,\\nNo. 2, pp. 619–624. https://doi.org/10.1073/pnas.1419539112 PMID: 25548179\\n19.\\nGescheider G. Psychophysics: Method and theory. Hillsdale, NJ: Lawrence Erlbaum Associates; 1976\\n20.\\nLevitt H. Transformed Up-Down Methods in Psychoacoustics. J Acoust Soc Am 1971; 49(2), Suppl. 2:\\n467+.\\n21.\\nGardner E, Costanzo R. Properties of kinesthetic neurons in somatosensory cortex of awake monkeys.\\nBrain Res 1981; 214: 301–319. https://doi.org/10.1016/0006-8993(81)91196-3 PMID: 7237173\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n23 / 24\\n\\n\\n22.\\nPrud’homme M, Kalaska J. Proprioceptive activity in primate primary somatosensory cortex during\\nactive arm reaching movements. J Neurophysiol 1994; 72: 2280–2301. https://doi.org/10.1152/jn.\\n1994.72.5.2280 PMID: 7884459\\n23.\\nTillery S, Soechting JF, Ebner TJ. Somatosensory cortical activity in relation to arm posture: nonuniform\\nspatial tuning. J Neurophysiol 1996; 76: 2423–2438. https://doi.org/10.1152/jn.1996.76.4.2423 PMID:\\n8899615\\n24.\\nProske U. Kinesthesia: The role of muscle receptors. Muscle Nerve 2006; 34: 545–558. https://doi.org/\\n10.1002/mus.20627 PMID: 16897766\\n25.\\nWhite O, Proske U. Illusions of forearm displacement during vibration of elbow muscles in humans. Exp\\nBrain Res 2008; 192: 113–120. https://doi.org/10.1007/s00221-008-1561-z PMID: 18787812\\n26.\\nGregory JE, Morgan DL & Proske U. Aftereffects in the Responses of Cat Muscle Spindles and Errors\\nof Limb Position Sense in Man. J Neurophysiol 1988; 59 (4): 1220–1230. https://doi.org/10.1152/jn.\\n1988.59.4.1220 PMID: 3373276\\n27.\\nMaravita A. Implicit processing of somatosensory stimuli disclosed by a perceptual after-effect. Neu-\\nroreport 1997; 8(7): 1671–4. https://doi.org/10.1097/00001756-199705060-00022 PMID: 9189912\\n28.\\nCrook M, Crook H. Adaptation to Cutaneous Pressure. Am J Psychol 1935; 47: 301–308.\\n29.\\nVogels I, Kappers A, Koenderink J. Haptic Surface Aftereffect is of Central, not Peripheral Origin. Stud-\\nies in Perception and Action III 1995: 319–322.\\n30.\\nRhodes G, Proffitt F, Grady JM, Sumich A. Facial symmetry and the perception of beauty. Psychon.\\nBull. Rev 5 1998, pp. 659–669.\\n31.\\nMachilsen B, Pauwels M, Wagemans J. The role of vertical mirror symmetry in visual shape detection.\\nJ. Vis 2009; 9, pp. 1–11.\\n32.\\nHannah R, Morrison J, Chapman A. Kinematic symmetry of the lower limbs. Arch. Phys. Med. Rehabil.\\n1984; 65, pp. 155–158. PMID: 6712430\\n33.\\nHaken H, Kelso J, Bunz H. A Theorethical Model of Phase Transitions in Human Hand Movements.\\nBiol. Cybern. 1985; 51, pp. 347–356.\\n34.\\nKelso J. On the oscillatory basis of movement. Bull Psychon Soc 1981a; 18: 63.\\n35.\\nHan J, Anson J, Waddington G, Adams R. Proprioceptive performance of bilateral upper and lower limb\\njoints: side-general and side-specific effects. Exp Brain Res 2013; 266: 313–323.\\n36.\\nGregory JE, Morgan DL, Proske U. Aftereffects in the responses of cat muscle spindles. J Neurophysiol\\n1986; 56: 451–461. https://doi.org/10.1152/jn.1986.56.2.451 PMID: 3760930\\nPLOS ONE\\nBimanual haptic slant adaptation does not require touch\\nPLOS ONE | https://doi.org/10.1371/journal.pone.0236824\\nJuly 31, 2020\\n24 / 24\",\"difficulty\":\"easy\",\"domain\":\"Single-Document QA\",\"length\":\"short\",\"question\":\"Why can VR Headset be used to help understand haptic slant adaptation in this article?\",\"sub_domain\":\"Academic\"}","display_format":"text","language":"","answer_status":"published","assets":[],"source_url":"https://huggingface.co/datasets/zai-org/LongBench-v2","history":"initial import","indexing_mode":"noindex","subproblems":[],"grids":[]}