
==== Front
iScience
iScience
iScience
2589-0042
Elsevier

S2589-0042(24)02028-5
10.1016/j.isci.2024.110803
110803
Article
Spatiotemporal motion features resulting from tactile interface layouts influence tactile speed perception
Ujitoko Yusuke yusuke.ujitoko@gmail.com
14∗
Takenaka Yuko 2
Hirota Koichi 3
1 NTT Communication Science Laboratories, Nippon Telegraph and Telephone Corporation 3-1 Morinosato Wakamiya, Atsugi, Kanagawa 243-0198, Japan
2 Graduate School of Information and Engineering, The University of Electro-Communications, Chofu 182-8585, Japan
3 Graduate School of Information and Engineering, The University of Electro-Communications, Chofu 182-8585, Japan
∗ Corresponding author yusuke.ujitoko@gmail.com
4 Lead contact

30 8 2024
20 9 2024
30 8 2024
27 9 11080327 3 2024
28 7 2024
20 8 2024
© 2024 The Author(s)
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Summary

Representing tangential motion between objects and the skin using tactile displays enables humans to manipulate virtual objects and recognize their surface properties. To design effective tactile stimuli that accurately represent motion, it is important to understand how humans perceive tactile motion based on spatiotemporal features, an area that remains relatively unexplored. This study elucidates the spatiotemporal features that influence the perceived speed of tactile motion represented by a tactile display with discrete stimulation points. The findings show that the average spatial spacing between stimulation points affects the perceived speed, even though the average spatial spacing does not vary with the speed itself, but rather varies with the stimulation point layout of the tactile display. No significant effects from other features were observed on the perceived speed. The results suggest that perceived speed can be controlled by considering the average spatial spacing during tactile stimulus design.

Graphical abstract

Highlights

• Tactile interfaces form unique spatiotemporal features when presenting motion

• We manipulated the spatiotemporal features using a pin-array display in experiments

• Average spacing between stimulation points influences the perceived speed of motion

• Perceived speed can be controlled by varying the stimulus spacing in tactile design

Behavioral neuroscience; Cognitive neuroscience

Subject areas

Behavioral neuroscience
Cognitive neuroscience
Published: August 30, 2024
==== Body
pmcIntroduction

Haptic technology has immense potential for use in various areas, such as communication, training, and entertainment.1 A critical feature of such displays is their ability to represent the tangential motion between virtual objects and the skin of the user during touch interactions. This tangential motion enables users to perceive the haptic properties of the objects, including their shapes, materials, and textures.2,3,4 Moreover, the perception of motion plays a crucial role in everyday interactions with the environment, as evidenced by our ability to grasp and manipulate objects effectively.5

Tactile motion cues, such as spatiotemporal features or tangential deformations, must be provided to ensure that users can perceive motion through tactile displays.6 The present study focuses on spatiotemporal features because displays capable of presenting spatiotemporal features have been actively developed, in contrast to those capable of presenting other motion cues.6 Common displays that provide spatiotemporal features on the skin have discrete stimulation points in a matrix pattern (hereafter referred to as “stimulation matrix”) (see Figure 1A).7,8,9,10,11 The current spatial resolution of the stimulation matrix is on the millimeter scale, given practical requirements such as wearability. This discrete skin stimulation is different from continuous skin stimulation in the real world and thus forms unique spatiotemporal features on the skin. Understanding tactile motion perception based on unique spatiotemporal features is essential to design stimuli that make humans perceive motion as intended.Figure 1 Tactile display with discrete stimulation matrix and five spatiotemporal features that vary with the arrangement of the stimulation matrix

(A) Assume a tactile display with a discrete stimulation matrix in the future.

(B–F) The green box represents a virtual object that is moving straight down at a constant speed on the skin. The white and blue dots represent the discrete points on the matrix of the tactile display. The blue dots represent the points at which skin is stimulated, as these points intersect with the trajectory of the virtual object. The spatiotemporal features of the stimuli are defined by the locations of the blue dots and the timing of actuation at each of the dots. The features vary with the translation, rotation, and scaling of the matrix relative to skin. (B) Example of variation in spatial frequency owing to rotation of the stimulation matrix. (C) Example of variation in spatial width owing to translation of stimulation matrix. (D) Example of variation in the transverse component of motion owing to rotation and scaling of the stimulation matrix. (E) Example of variation in the equality of spatial spacing owing to translation of the stimulation matrix. (F) Example of variation in periodicity or non-periodicity owing to rotation of the stimulation matrix.

(G) Time at which the stimulus is presented. The stimulus is presented when the centers of the stimulation points overlap with any part of the virtual moving object.

Despite its necessity, the perception of tactile motion based solely on spatiotemporal features remains relatively unexplored. Most previous studies have investigated the perceptual characteristics of tactile motion speed in scenarios wherein a real object was sliding across the skin.12,13,14,15,16,17 In these scenarios, multiple motion cues (e.g., spatiotemporal features, slip-induced vibrations, and tangential skin deformations) were integrated into the perception of tactile motion. However, as the motion cues were not isolated in the experiments, the role of spatiotemporal features in speed perception remains unclear in scenarios wherein humans perceive tactile motion speed based solely on spatiotemporal features.

Moreover, scant consideration has been given to situations wherein spatiotemporal features are represented by a discrete stimulation matrix. Under certain stimulus conditions in such situations, humans can perceive tactile apparent motion,18 which is the sensation of continuous motion created by sequentially activating discrete points on the skin. This phenomenon is similar to how a sequence of still images can create the illusion of motion in visual perception.19,20 However, also in the context of the tactile apparent motion, which spatiotemporal features affect the perceived speed is unknown.

The objective of this study was to elucidate the spatiotemporal features that contribute to speed perception when humans perceive speed based solely on these features, with a specific focus on the discrete stimulation matrix. We focused on the basic spatiotemporal features that vary according to the overall layout (i.e., translation, rotation, and scaling) of the discrete stimulation matrix relative to the skin surface: spatial frequency (Figure 1B), spatial width (Figure 1C), transverse component of local motion (Figure 1D), equality of spatial spacing (Figure 1E), and spatial periodicity (Figure 1F). For example, in the scenario shown in Figure 1B, the spatial frequency of the stimuli changes owing to the rotation of the stimulation matrix. We assumed a case in which a virtual object moved at a constant speed toward the matrix. A stimulus was presented at a point (colored red in Figure 1G) that intersected the object trajectory. The speed in the principal direction of travel is herein referred to as “speed.” To enhance the clarity of the interpretation of the results, we chose not to consider conditions under which multiple points were simultaneously stimulated. In the experiments, we manipulated each spatiotemporal feature using a pneumatically driven pin-array display.9

Results

Experiment 1: Spatial frequency significantly influences perceived speed

Experiment 1 was aimed at clarifying whether the spatial frequency and duration of a periodic stimulus affected the perceived speed. As the spatial frequency may change based on the arrangement of the matrix of the tactile display, as shown in Figure 1B, we focused on the spatial frequency as a factor. In addition to the spatial frequency, we included the stimulus duration because it is known to affect the subjective quality of motion (e.g., impressiveness or continuity) in the context of tactile apparent motion research.18 If the underlying perceptual mechanism between the speed and quality of motion is related, the perceived speed may also be affected by the stimulus duration.

The participants placed their right hand on the pin-array display and received a pressure stimulus from the display (Figures 2A–2C). The display can control the pressure on the order of milliseconds (Figure S1). The display simulated a virtual object moving at a constant speed from the index fingertip to the wrist (Figure 2D). Pressure was applied to raise any pin whose center overlapped with any part of the virtual moving object (Figure 1G). The participants judged the speed of motion of the virtual object. The start and end positions of the stimulation were randomized to prevent the participants from judging the speed based on the global motion distance.Figure 2 Apparatus

(A) Components of experimental system and data flow.

(B) The right hand of the participant was located on the pin-array display. A cardboard barrier was positioned to prevent the participant from seeing his or her right hand.

(C) Pin-array display.

(D) Virtual moving object (colored in green). The moving object was represented by the pin-array display in the experiment. The start and endpoints were randomly chosen.

We manipulated the stimulus spacing (3 and 6 mm), virtual object length (1.5 and 3 mm) (see Figure 3A), and virtual object speed (0.04 and 0.08 m/s) to determine the effects of the spatial frequency and stimulus duration on the perceived speed. A total of eight possible combinations of conditions existed. We herein refer to the stimulus corresponding to each of these eight combination conditions as the “reference stimulus”.Figure 3 Experimental conditions for and results of experiment 1, in which the effects of the spatial frequency and stimulus duration were investigated

(A) Combinations of conditions of stimulus spacing and virtual object length. We manipulated the stimulus spacing, virtual object length, and virtual object speed (not illustrated) of the reference stimulus. The spacing of the comparison stimulus was 3 mm, and its virtual object length was 3 mm.

(B) Timeline of a single trial.

(C) Mean PSE (speed of the comparison stimulus that is perceptually equal to the speed of the reference stimulus). Error bars denote 95% CI.

To quantify the perceived speeds of the eight reference stimuli, we asked the participants to compare them with the speed of the “comparison stimulus” (stimulus spacing: 3 mm and object length: 3 mm). Prior to the experiment, the pressure values of all stimuli were configured such that the perceptual intensities matched because the intensity had a minor but significant effect on the perceived speed.21 The participants were sequentially presented with two stimuli (a reference stimulus and a comparison stimulus). The participants judged whether the second stimulus (the comparison stimulus) was faster than the first stimulus (the reference stimulus), and they were required to make a two-alternative forced choice (Figure 3B). We employed a randomly interleaved staircase method with two staircases. For each participant and condition, we fit a psychometric function to the proportion of times the comparison stimulus was judged to be faster than the reference stimulus as a function of speed. We obtained the point of subjective equality (PSE) using the psychometric function.

Figure 3C shows the PSE, or speed of the comparison stimulus subjectively perceived to be equal to the speed of the reference stimulus for each condition. We conducted a three-way repeated-measures analysis of variance (ANOVA) on the PSE with the factors of stimulus spacing, virtual object length, and virtual object speed. Significant main effects of the stimulus spacing (df = 1, F = 85.4, p < 0.001, ηp2 = 0.49) and speed (df = 1, F = 401.5, p < 0.001, ηp2 = 0.82) were found. No significant main effect of object length was observed (df = 1, F = 0.1, p = 0.75, ηp2 = 0.001). Significant two-way interaction effects were observed between the stimulus spacing and object speed (df = 1, F = 8.2, p = 0.005, ηp2 = 0.09). No other significant two- or three-way interaction effects were found (p > 0.05). Owing to a lack of a significant main effect of the object length and its minor effect size, the duration can be concluded not to play a significant role.

One may suspect that the perceived speed can be explained by the stimulus onset asynchrony (SOA), which refers to the temporal interval between the onset of two stimuli. The reason for this suspicion is that significant main effects of both the spatial spacing and speed exist. However, in a Welch’s t test, a significant difference in PSEs was observed between the condition with a speed of 0.08 m/s and spacing of 6 mm and the condition with a speed of 0.04 m/s and spacing of 3 mm (t[28.4] = 6.75, p < 0.001, Cohen’s d = 1.95), although the SOA was the same under both conditions. This finding suggests that the SOA cannot explain the perceived speed.

In summary, the spatial spacing (spatial frequency) and speed contribute to the perceived speed. The spatial spacing (spatial frequency) is a critical determinant when engineers intend to make users perceive a certain speed. When the spatial frequency was doubled, the perceived speed was approximately 1.3 times the original speed.

Experiment 2: No significant effect of spatial width

The aim of experiment 2 was to clarify whether the spatial width of the stimulus affected the perceived speed. We focused on the spatial width as a factor because the spatial width can change owing to the arrangement of the matrix of the tactile display, as shown in Figure 1C. We manipulated the spatial width (small, medium, and large) (Figure 4A) and virtual object speed (0.04 m/s and 0.08 m/s). A total of six possible combinations existed. We herein refer to the stimulus corresponding to each of these six combination conditions as the “reference stimulus.” The perceived intensities of each stimulus were aligned before the experiment. In a manner similar to the previous experiment, the perceived speed of the reference stimulus was quantified by comparing it with the “comparison stimulus.” The spatial width of the comparison stimulus was set to small.Figure 4 Experimental conditions for and results of experiment 2, in which the effects of the spatial width were investigated

(A) Three conditions of spatial width. We manipulated the spatial width and virtual object speed (not illustrated) of the reference stimulus. The stimulus with small spatial width was used as a comparison stimulus.

(B) Mean PSE (speed of comparison stimulus that is perceptually equal to the speed of the reference stimulus). Error bars denote 95% CI.

Figure 4B shows the PSE, or speed of the comparison stimulus that was subjectively perceived to be equal to the speed of the reference stimulus for each condition. We conducted a two-way repeated measures ANOVA on the PSE, with the stimulus width and speed as within-participant factors. A significant main effect of the speed was found (df = 1, F = 198.7, p < 0.001, ηp2 = 0.75), although no significant main effect of the stimulus width was observed (df = 2, F = 1.6, p = 0.21, ηp2 = 0.05). A significant interaction between the spatial width and object speed was observed (df = 2, F = 4.5, p = 0.015, ηp2 = 0.12). We conducted a post-hoc test regarding the interaction effect to find a simple main effect of spatial width at each object speed. We conducted a one-way repeated measures ANOVA with the spatial width as a factor on PSEs under the speed conditions of 0.04 and 0.08 m/s. We did not find a significant main effect of the stimulus width either under the speed condition of 0.04 m/s (df = 2, F = 2.3, p = 0.12, ηp2 = 0.12) or under the speed condition of 0.08 m/s (df = 2, F = 2.8, p = 0.07, ηp2 = 0.15). Owing to the lack of a significant main effect and a simple main effect of the spatial width, the spatial width can be concluded not to play a significant role.

Experiment 3: No significant effect of transverse component of local motion

Experiment 3 aimed to clarify whether the transverse component of local motion in a periodic stimulus affects the perceived speed. As the magnitude of the transverse component can change owing to the arrangement of the matrix of the tactile display, as shown in Figure 1D, we focused on the magnitude of the transverse component as a factor. We manipulated the magnitudes of the transverse components (zero, small, and large values) (Figure 5A), as well as the virtual object speed (0.04 and 0.08 m/s). In total, six combinations of conditions were possible. We herein refer to the stimulus corresponding to each of the six combination conditions as the “reference stimulus.” The magnitude of the transverse component of the comparison stimulus was set to zero. The perceived intensities of the stimuli were aligned before the experiment. Similar to the approaches used in the previous experiments, the perceived speed of the reference stimulus was quantified based on a comparison with the comparison stimulus.Figure 5 Experimental conditions for and results of experiment 3, in which the effects of the transverse component were investigated

(A) Three conditions of the magnitude of the transverse component. We manipulated the magnitude of the transverse component of local motion and the speed of the virtual object (not illustrated) of the reference stimulus. The stimulus with zero transverse component was also used as a comparison stimulus.

(B) Mean PSE (speed of comparison stimulus that is perceptually equal to the speed of the reference stimulus). Error bars denote 95% CI.

Figure 5B shows the PSE, or speed of the comparison stimulus subjectively perceived to be equal to the speed of the reference stimulus for each condition. We conducted a two-way repeated measures ANOVA on the PSE with the magnitude of the transverse component and speed as within-participant factors. A significant main effect of the speed was observed (df = 1, F = 199.9, p < 0.001, ηp2 = 0.75), whereas no significant main effect of the magnitude of the transverse component was found (df = 2, F = 0.6, p = 0.55, ηp2 = 0.02). No significant interaction was observed between these variables (df = 2, F = 1.1, p = 0.35, ηp2 = 0.03). Owing to a lack of significant main effect of transverse component and its minor effect size, the transverse component can be concluded not to play a significant role.

Experiment 4: No significant effect of equality of spatial spacing of periodic stimuli

Experiment 4 aimed to clarify whether the equality of spatial spacing in a periodic stimulus affects the perceived speed. We focused on the equality of spatial spacing as a factor because the equality or inequality of the spatial spacing may change owing to the arrangement of the matrix of the tactile display, as shown in Figure 1E. We manipulated the equality of spatial spacing (see Figure 6A) and virtual object speed (0.04 and 0.08 m/s) of the reference stimulus while maintaining a constant average spatial spacing. The comparison stimulus used equal spatial spacing. The perceived intensities of the stimuli were aligned prior to the experiment. Similar to the methods used in the previous experiments, the perceived speed of the reference stimulus was quantified based on a comparison with the comparison stimulus.Figure 6 Experimental conditions for and results of experiment 4, in which the effects of the spatial spacing were investigated

(A) Two conditions of spatial spacing (equal or unequal). We manipulated the equality of spatial spacing and speed of the virtual object (not illustrated) of the reference stimulus. The stimulus with equal spatial spacing was also used as a comparison stimulus.

(B) Mean PSE (speed of the comparison stimulus that is perceptually equal to the speed of the reference stimulus). Error bars denote 95% CI.

Figure 6B shows the PSE, or speed of the comparison stimulus subjectively perceived to be equal to the speed of the reference stimulus for each condition. We conducted a two-way repeated measures ANOVA on the PSE, with the equality of spatial spacing and speed as within-participant factors. A significant main effect of the speed was found (df = 1, F = 104.9, p < 0.001, ηp2 = 0.70), but no significant main effect of the equality of spatial spacing was observed (df = 1, F = 0.03, p = 0.86, ηp2 = 0.001). No significant interaction was found between them (df = 1, F < 0.001, p = 1.00, ηp2 < 0.001). Owing to the lack of a significant main effect of the equality of spatial spacing and its minor effect size, the equality of spatial spacing can be concluded not to play a significant role.

Experiment 5: No significant effect of spatial periodicity

Experiment 5 had three main objectives. First, we aimed to clarify whether the presence of spatial periodicity of the stimulus affected the perceived speed. Because the presence or absence of spatial periodicity can change owing to the arrangement of the matrix of the tactile display, as shown in Figure 1F, we focused on this aspect. In experiment 4, we did not observe a significant difference in the perceived speed between the stimuli with equal and unequal spatial spacings when the average stimulus spacings were the same. Accordingly, we investigated whether the perceived speed did not differ between stimuli with or without spatial periodicity as long as the average spatial spacing remained the same.

Second, we investigated whether the magnitude of the transverse component of local motion affected the perceived speed for spatially non-periodic stimuli. The results of experiment 3 show that the magnitude of the transverse component does not influence the perceived speed in the case of spatially periodic stimuli. We examined whether this was also true for spatially non-periodic stimuli.

Third, we investigated whether the average spatial spacing affects the perceived speed in the case of spatially non-periodic stimuli. The results of experiment 1 reveal that the spatial spacing (spatial frequency) influences the perceived speed in the case of spatially periodic stimuli. We examined whether this was also true for spatially non-periodic stimuli.

To achieve these objectives, we prepared five reference stimuli, as shown in Figure 7A (the actual stimulation layout is depicted in Figure S2). These stimuli included (1) a spatially periodic stimulus, (2) spatially non-periodic stimuli without transverse components, and (3) a non-periodic stimulus with transverse components. Moreover, we manipulated the average spacing of spatially non-periodic stimuli without transverse components (4.5, 6, and 9 mm). The spatially periodic stimulus was also used as a comparison stimulus. The speed of the reference stimulus was fixed at 0.08 m/s. The perceived intensities of the stimuli were aligned before the experiment. Similar to the approaches used in the previous experiments, we quantified the perceived speed of the reference stimulus based on a comparison with the comparison stimulus.Figure 7 Experimental conditions for and results of experiment 5, which focused on the periodicity of the stimuli

(A) The reference stimuli included a periodic stimulus, non-periodic stimulus without transverse components, and non-periodic stimulus with transverse components. Additionally, we manipulated the average spacing of non-periodic stimuli without the transverse component. The periodic stimulus was also used as a comparison stimulus.

(B) Mean PSE (matched speed of comparison stimulus) with respect to the speed of the reference stimulus. Error bars denote 95% CI.

Figure 7B shows the PSE, or speed of the comparison stimulus subjectively perceived to be equal to the speed of the reference stimulus under each condition. A Welch’s t test revealed no significant difference in PSEs between the periodic and non-periodic stimuli without a transverse component at an average spacing of 6 mm (t[19.8] = 0.55, p = 0.59, Cohen’s d = 0.23). The Welch’s t test results do not show any significant difference in PSEs among the non-periodic stimuli with and without a transverse component at an average spacing of 6 mm (t[21.9] = 0.24, p = 0.81, Cohen’s d = 0.10). Owing to such minor effect sizes, the presence of periodicity or transverse component can be concluded not to play a significant role. The one-way repeated measure ANOVA with average spacing as a within-participant factor shows that the average spacing has a significant effect on the perceived speed in the case of non-periodic stimuli (df = 2, F = 10.4, p = 0.0002, ηp2 = 0.32).

Discussion

Interpretation of main results

In this study, we investigated the contributions of spatiotemporal features in the context of perceiving tactile motion speed when humans can only use features represented by the discrete stimulation matrix of a tactile display. In the experiments, we systematically manipulated the spatiotemporal features that may vary based on the layout of the stimulation matrix. The experimental results contribute to understanding how humans perceive tactile motion based on spatiotemporal features and have important implications for engineering aimed at presenting desired speeds.

We found that the average stimulus spacing in the traveling direction is a critical determinant of speed perception. The observed changes in the perceived speed owing to the average stimulus spacing were larger than those in previous studies wherein the changes in the perceived speed of tactile motion were investigated.12,21 For instance, in experiment 1, when the average interval between stimuli on the skin was halved, the perceived speed increased by approximately 1.3 times. Furthermore, the effect of the average stimulus spacing on the perceived speed was consistently observed irrespective of the stimulus duration, stimulus speed, or spatial periodicity of the stimulus. These findings suggest that the average stimulus spacing is a fundamental feature effective for human speed perception.

It may appear counterintuitive for the average spatial spacing to affect speed perception because the average spatial spacing does not depend on the virtual object speed, but rather on the layout of the stimulation matrix of the tactile display. To delve into the seemingly counterintuitive effect of average spatial spacing on speed perception, we assume that the motion presented in this study is that of a physical object with an uneven surface rotating on the hand without friction. Under this assumption, the human perceptual characteristic of the inability to separate tactile signals from surface roughness and speed when perceiving speed12 may be related to this phenomenon. Dépeault et al. demonstrated that in situations wherein real objects slide on the skin surface, the perceived magnitude of speed is influenced by both the sliding speed of the object and the surface roughness.12 In this situation, the speed and surface roughness signals from mechanoreceptive afferents may be confounded and humans may not be able to distinguish between them precisely to perceive speed. In our experimental scenario, the presented motion can be regarded as a dotted object surface rolling on the skin without friction, with the participants perceiving surface roughness through spatially discrete stimuli. We speculate that when humans decode the motion speed of objects with surface roughness, the confounding effect between the surface roughness of the object and the speed of movement may affect the perception of speed, even in our situation.

The neuronal coding behind speed perception remains unclear, even when this speculation is true. Given that both speed and roughness directly affect the perception of speed, Dépeault et al. assumed simple intensive coding based on the firing rates of peripheral mechanoreceptive afferents, which are sensitive to both roughness and speed.12 Although intensive coding is known to contribute to speed perception,21 the effect of average spatial spacing on speed perception in our experiment cannot be explained solely by intensive coding for the following reasons. First, we equalized the perceived intensities of all stimuli before each experiment to eliminate the potential effect of the perceived intensity on the perceived speed. Second, even if the intensity of the stimulus is doubled, the perceived speed increases by 1.14 times at most, as demonstrated previously.21 Thus, the observed increase of approximately 1.3 times in experiment 1 cannot be explained solely by the differences in stimulus intensity or simple intensive coding.

An interaction may exist between time and space in touch perception. The dimensions of time and space are well known to interact constantly.22,23,24 For example, the kappa effect refers to the illusion wherein the perceived duration of a time interval is affected by the spatial distance between two stimuli.25 When the spatial spacing is smaller, the perceived time may be shorter than the actual time, and the speed may be accordingly perceived as faster. Note that the relationship between these time-space interactions and speed perception remains unclear.

We did not find significant contributions from other features (i.e., spatial width, stimulus duration, transverse component, equality of spatial spacing, and periodicity). Although we did not extensively investigate these results in this study, the implications for the characteristics of human speed perception may be inferred. For instance, the size perception of an object may not be related to its speed perception. When the spatial width of the stimulus points changes two or three times, the perception of the size of a moving virtual object changes. However, the non-contribution of spatial width to speed perception indicates that humans can accurately and robustly perceive speed without being influenced by the perceived size of an object when perceiving speed.

Scholars may be interested in the connection between the perception of speed and motion quality (e.g., impressiveness or continuity). The motion presented to users in this study can be considered a form of tactile apparent motion that is achieved by sequentially presenting discrete stimuli. Prior research on apparent motion has demonstrated that both the stimulus spacing and duration affect the subjective quality of motion.18,26 In contrast to the results of previous studies, we did not find any contribution of the stimulus duration to speed perception. This finding suggests that the underlying perceptual mechanisms of the subjective quality of motion and perceived speed may not be the same.

Possible application to tactile stimulus design

The results of this study have engineering implications, particularly in the design of tactile stimuli for conveying the intended motion speed to users. When engineers naively represent relative motion based on the stimulation matrix of the tactile display, the average spatial spacing between the stimuli varies based on the layout of the stimulus matrix (Figure 1B). A problem exists with unintended variations in the perceived speed owing to variations in the average spatial spacing, which depends on the layout of the stimulation matrix. Unintended variations in the perceived motion speed may directly affect the perception of surface properties or object manipulation in virtual reality.

These unintended variations in perceived speed can be avoided using a tactile stimulus design. One method involves selecting stimulus points within the trajectory of the object motion such that the average stimulus spacing maintains a constant value (Figure 8). This approach enables humans to perceive a consistent speed regardless of the stimulation matrix layout, the manipulation of grasped objects, and the recognition of object attributes independent of the layout of the stimulation matrix.Figure 8 Possible application of our results in tactile stimulus design

Tactile stimuli enable users to perceive similar motion speeds irrespective of the stimulation matrix layout. If we select the stimulus points within the trajectory of the object in motion to maintain a constant average spacing, the perceived speed will not change based on our experiments.

We did not clarify speed perception under conditions beyond the stimulus range that we used (e.g., conditions involving the simultaneous activation of multiple pins, non-constant object motion speeds, or complex shapes of virtual objects). Investigating speed perception under these conditions will facilitate the extension of the aforementioned tactile stimulus design method in future research.

Limitations of the study

This study has certain limitations in terms of the participant demographics, stimulation methods, and spatiotemporal features. First, we only recruited male participants because the stimulation area of the pin array display was designed to fit males. They were all in their 20s, as they were recruited at a university. As tactile perception may vary based on age and gender,27,28 the perception of movement based on the tactile stimuli investigated in this study may have been influenced by these factors. In future research, developing personalized devices based on hand size could facilitate the investigation of this aspect.

It remains unclear whether the perceptual characteristics revealed using pin-array displays in this study are also observed (1) when humans interact with real objects, (2) when using tactile displays with different actuation mechanisms, or (3) when using tactile displays to stimulate different skin areas. The spatiotemporal features used for speed perception by the participants in this study can be replicated by smoothly rotating real objects with textured surfaces on the skin. The spatiotemporal features can be replicated by employing stimuli such as vibrations8 and electrical stimuli,29 even to the arms instead of the hands. Clarifying whether the findings from pin-array displays can be extended to other stimulation methods can deepen our understanding of the mechanisms of tactile motion perception. For example, if similar results are not obtained when applying electrical skin stimulation, the perceptual characteristics discovered may be attributed to the skin deformation.

Moreover, this study focused only on basic spatiotemporal features, but not more complex spatiotemporal features. For instance, stimuli involving acceleration or stimuli the size of which deform could be considered. Analyzing responses to more complex spatiotemporal features could lead to a deeper understanding of human perceptual characteristics.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Yusuke Ujitoko (yusuke.ujitoko@gmail.com).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• Data will be shared by the lead contact upon request.

• Code will be shared by the lead contact upon request.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

There is no funding source for this study.

Author contributions

Conceptualization: Y.U. methodology: Y.U., Y.T., K.H. experiment: Y.T. analysis: Y.U. and Y.T. visualization: Y.U. and Y.T. supervision: K.H. writing—original draft: Y.U. writing—review and editing: Y.U., Y.T., and K.H.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER	
Software and algorithms	
	
RStudio 2024.04.2	Posit, USA	https://posit.co/downloads	
R programming language 4.2.1	R Foundation, USA	https://www.r-project.org/	
Microsoft Visual Studio 2022 Community Edition	Microsoft, USA	https://visualstudio.microsoft.com/vs/community/	
AMD Vivado v2022.2.2 (64bit)	AMD, USA	https://japan.xilinx.com/products/design-tools/vivado.html	

Experimental model and study participant details

Participants and ethics statement

Twelve different Japanese males participated in five experiments. The mean ages of the participants in Experiments 1, 2, 3, 4, and 5 were 23.9 (SD: 0.9), 23.4 (SD: 1.3), 23.9 (SD:0.9), 23.4 (SD: 1.3), and 23.4 (SD: 1.2), respectively. We measured the hand sizes of the participants along a straight line following the pin layout and found that the averages in Experiments 1, 2, 3, 4, and 5 were 176.9 (SD: 5.9) mm, 176.1 (SD: 6.4) mm, 176.9 (SD: 5.9) mm, 176.1 (SD: 6.4) mm, and 175.3 (SD: 5.5) mm, respectively. The reason for exclusively choosing male participants was that the stimulation area of the developed pin-array display was tailored to fit the hand size of males. Because female hands tend to be smaller, the stimulation area of the device would not have fit properly. All participants were right-handed, which was assessed based on self-reporting by the participants. All participants were naive to the purpose of the study and reported no sensorimotor disorders. The participants were paid for their participation. Ethical approval was obtained from the ethics committee of the University of Electro-communications (approval number: 23007). The experiments were conducted in accordance with the principles that have their origin in the Helsinki Declaration.

Method details

Apparatus

We used the same apparatus that was employed in a previously reported study.21 Here, we briefly describe the apparatus specifications. An in-depth description of the experimental apparatus can be found in the previous report.21 The system comprised a pin-array display, an air pressure controller, and a PC (Figure 2A).

Figure 2B shows the pin-array display. The participants were instructed to place their right hand on the base without applying any pressure to it other than the weight of the hand. A hand-shaped dent was present on the surface, and pin holes 2 mm in diameter (see Figure 2C)) existed on the base. Pins with a diameter of 1.9 mm were arranged in a grid pattern with a 3 mm space between the centers of the pins. This spacing was determined based on the spatial acuity of fingertip (two point discrimination: 2–3 mm).30 Only some pins were used in each experiment. The air pressure at each pin was controlled independently. Calibration was performed for each pin to ensure accurate delivery of the target pressure stimulation within a pressure range of up to 0.06 MPa.

Regarding temporal resolution, we confirmed that pin control was achievable on the order of milliseconds. To evaluate the temporal resolution of the display, we manipulated the stimulus onset asynchrony (SOA) of adjacent pins with 3 mm spacing, corresponding to the speed manipulation of motion. We manipulated the SOA with a minimum duration of 20 ms (150 m/s), a maximum of 300 ms (10 m/s), and increments of 10 ms. The SOA was measured based on the images captured at 960 fps by a high-speed camera (SONY, RX0M2). Figure S1A shows the measured SOA, indicating that pin control was achievable on the order of milliseconds. Figure S1B shows the relationship between the control and measured speed and is a replot of Figure S1A.

The experimental software running on the PC transmitted the control values to the field programmable gate array (Xilinx, XC7S50) in the air pressure controller through USB serial communication. The internal regulator of the air pressure controller (SMC, VY1B00) controlled the valves. Two air compressors (RYOBI, ACP-50, and ACP-60) were employed as the air sources for the regulator to efficiently generate the air required for the system. The regulator operated with a response time of 30 ms. The response time was from the onset of the command to the pressure regulator to the completion of pressure control pushing the pin.

The participants were seated comfortably in chairs. They wore earplugs and noise-cancelling headphones that played white noise to block external sounds. The participants rested their right arm on an armrest with their right hand placed on a pin-array display. A cardboard barrier was positioned over the right hand to prevent the participants from seeing the pin-array display (Figure 2C).

Stimuli

The pin-array display simulated a virtual moving object with a constant speed whose movement spanned a certain length and width from the fingertip to the wrist, as illustrated in Figure 2D. Pressure was applied to raise the pin whose center overlapped with any part of the virtual object in motion (Figure 1G). In each experiment, the pins to be enabled were predetermined, which helped manipulate the spatiotemporal features presented to the participants. For example, if only the middle column of the pins was enabled, the pins in that column were pressured to push out when their centers overlapped with the virtual object. The pins in the other columns were not activated.

To prevent the participants from judging the speed based on either the global movement distance or duration, we randomly selected one of the nine rows near the fingertip as the starting row for the movement of the object for each stimulus. Similarly, we randomly selected one of the nine rows near the wrist as the end row for the movement of the object. Consequently, the shortest and longest movement distances were 111 mm (38 pins) and 159 mm (54 pins), respectively. This implementation, involving the random assignment of starting and ending rows for the movement stimulus, is related to the configuration of the stimulation area adopted in the experiments. Unlike previous studies that focused solely on stimulating the fingertip area (e.g.,12), we conducted stimulations across both the finger and palm areas. By activating a series of pins over this extended area, we increased the pool of pins available for random selection as the starting and ending rows, thereby enhancing the randomization of the movement distance.

Adjustment of perceived intensity of stimuli in each experiment

Prior to the experiment, the pressure values of all stimuli were configured such that the perceptual intensity matched by three authors (YU, YT, and KH) because the intensity had a minor effect on the perceived speed.21 We did not let participants in the main experiment perform the adjustment for two reasons: 1) to prevent the experiment duration from becoming excessively long, which could have imposed a significant burden on the subjects and potentially affected the experimental results, and 2) the intensity tendencies after adjustment were generally common among the authors.

In the adjustment for Experiment 1, the perceived intensities of eight reference stimuli and one comparison stimulus at speeds of 0.01, 0.02, 0.04, 0.06, 0.08, 0.10, and 0.11 m/s were adjusted according to the perceived intensity of a specific reference stimulus (stimulus spacing: 3 mm, speed: 0.04 m/s, pressure: 0.02 MPa). Three authors adjusted the pressure value by using a foot button without a time limit. They performed two trials for each stimulus (eight reference stimuli and one comparison stimulus). The results are listed in Table S1. In the main experiments, the mean intensity value was used. Regarding the comparison stimulus, the pressure value used was interpolated based on this adjustment experiment.

In the adjustment for Experiment 2, the perceived intensities of six reference stimuli and one comparison stimulus at speeds of 0.01, 0.02, 0.04, 0.06, 0.08, 0.10, and 0.11 m/s were adjusted to the perceived intensity of a specific reference stimulus (spatial width: small, pressure: 0.02 MPa). The methods were the same as those mentioned above. The results are listed in Table S2. Regarding the comparison stimulus, the pressure value used was interpolated based on this adjustment experiment.

In the adjustment for Experiment 3, the perceived intensities of six reference stimuli and one comparison stimulus at speeds of 0.01, 0.02, 0.04, 0.06, 0.08, 0.10, and 0.12 m/s were adjusted to the perceived intensity of a specific reference stimulus (no transverse component, pressure: 0.02 MPa). The methods were the same as those mentioned above. The results are listed in Table S3. Regarding the comparison stimulus, the pressure value used was interpolated based on this adjustment experiment.

In the adjustment for Experiment 4, the perceived intensities of four reference stimuli and one comparison stimulus at speeds of 0.01, 0.02, 0.04, 0.06, 0.08, 0.10, and 0.11 m/s were adjusted to the perceived intensity of specific reference stimulus (no transverse component, pressure: 0.025 MPa). The methods were the same as those mentioned above. The results are listed in Table S4. Regarding the comparison stimulus, the pressure value used was interpolated according to this adjustment experiment.

In the adjustment for Experiment 5, the perceived intensities of five reference stimuli and one comparison stimulus at speeds of 0.04, 0.06, 0.08, 0.10, and 0.11 m/s were adjusted to the perceived intensity of specific reference stimulus (periodic stimulus, pressure: 0.025 MPa, speed: 0.04 m/s) in Experiment 4. The adjusted pressure value for the comparison stimulus was extracted from the values in Experiment 4. The methods were the same as those mentioned above. The results are listed in Table S5. Regarding the comparison stimulus, the pressure value to present was interpolated according to this adjustment experiment.

General procedure

At the beginning of each experiment, the participants were provided with written instructions that explained the details of the experiment and were asked to provide informed consent in writing. Subsequently, they practiced the procedure before moving to the main sessions.

In each trial, during the practice and main sessions, the participants were sequentially presented with two stimuli—a reference stimulus and a comparison stimulus. The participants were asked to judge whether the second stimulus (the comparison stimulus) moved slower or faster than the first stimulus (the reference stimulus) and were required to make a two-alternative forced choice by pressing a foot button. We informed the participants in advance that the global distance of the object movement was randomized and instructed them not to base their judgment of motion speed on the global movement distance or duration. Once they had made their choices, they proceeded to the next trial.

We employed a randomly interleaved staircase method with two staircases. The presentation order of the two staircases was randomized: in one staircase, the initial speed of the comparison stimulus was faster than that of the reference stimulus by 0.025 m/s, whereas in the other staircase, the initial speed was slower than that of the reference stimulus by 0.025 m/s. We increased or decreased the speed of the comparison stimulus by 0.005 m/s in the subsequent step of the staircase in response to “slow” or “fast” responses of the participants, respectively. The experiment was terminated when the responses for both staircases were reversed six times.

Experiment 1: Spatial frequency and stimulus duration

In Experiment 1, we manipulated the stimulus spacing (3 and 6 mm), virtual object length (1.5 mm and 3 mm), and virtual object speed (0.04 and 0.08 m/s) (see Figure 3A). These two speeds were selected after considering the range of speeds that could be accurately presented based on the temporal resolution of the display (Figure S1). Eight combinations of conditions were possible. We refer to the stimulus corresponding to each of these eight combination conditions as the reference stimulus. We quantified the perceived speed of eight reference stimulus by comparing the speed of the comparison stimulus (stimulus spacing: 3 mm and object length: 3 mm).

The main session comprised eight blocks, each corresponding to one of the reference stimuli. The order of the eight blocks was randomized. Before the first block started, a practice session was conducted during which the participants performed only 10 trials. The reference stimulus in the practice session was the same as that in subsequent blocks. The procedure was the same in the practice session and in the main session.

Experiment 2: Spatial width

In Experiment 2, we manipulated the spatial width (small, medium, and large) (Figure 4A) and speed of the virtual object (0.04 and 0.08 m/s) of the stimuli. Six combinations were possible in total. We refer to the stimulus corresponding to each of these six combination conditions as the reference stimulus. The spatial width of the comparison stimulus was set to small. The other methods used in Experiment 2 were the same as those in Experiment 1.

Experiment 3: Transverse component of local motion

In Experiment 3, we manipulated the magnitude of the transverse components (zero, small, and large values) (Figure 5A) and virtual object speed (0.04 and 0.08 m/s). A total of six combinations of conditions were possible. We refer to the stimulus corresponding to each of these six combination conditions as the reference stimulus. The magnitude of the transverse component of the comparison stimulus was set to zero. The other methods used in Experiment 3 were the same as those in Experiment 1.

Experiment 4: Equality of spatial spacing of periodic stimuli

In Experiment 4, we manipulated the equality of spatial spacing (see Figure 6A) and virtual object speed (0.04 m/s and 0.08 m/s) while maintaining a constant average spatial spacing. The comparison stimulus had equal spatial spacing. The other methods used in Experiment 4 were the same as those in Experiment 1.

Experiment 5: Periodicity of stimuli

Experiment 5 included five reference stimuli: (1) a spatially periodic stimulus, (2) spatially non-periodic stimuli without transverse components, and (3) a non-periodic stimulus with transverse components. Moreover, we manipulated the average spacing of spatially non-periodic stimuli without transverse components (4.5, 6, and 9 mm). The average spacings between the other stimuli were 6 mm. The layouts of the five stimuli are shown in Figure S2. The stimulation points for spatially non-periodic stimuli were randomly chosen, with the only constraint being that the start and end points must match those of the spatially periodic stimulus. A spatially periodic stimulus was used for comparison. The speed of the reference stimulus was fixed at 0.08 m/s to reduce the entire duration of the experiment. The other methods used in Experiment 5 were the same as those in Experiment 1.

Quantification and statistical analysis

Data analysis

For each block and participant, we fit the responses to psychometric functions of the following form:Φ−1[P(Y=1)]=β0+β1vcomp,

where Φ−1 represents the probit link function and P indicates the probability of Y=1. The response variable Y is 1 if the participant reports that the object is moving faster in the comparison stimulus than in the reference stimulus, and 0 otherwise. On the right side of the equation, vcomp indicates the speed of the comparison stimulus, and β0 and β1 indicate the intercept and slope of the linearized equation, respectively. We analyzed the data of each participant using a generalized linear model. We computed the point of subjective equality (PSE=−β0/ β1) corresponding to the stimulus value yielding a response probability of 0.5.

In Experiments 1–4, we conducted repeated-measures ANOVA tests to determine whether the PSE changed according to different factors. If a violation of normality occurred, as determined by the Shapiro–Wilk test (with Bonferroni correction), we conducted an aligned rank transform (ART)31 on the data, followed by an ANOVA on the aligned ranks. As other conventional nonparametric statistical tests (e.g. Kruskal–Wallis test and Mann–Whitney U test) cannot elucidate the effects of multiple factors and their interactions, we adopted the ART procedure (see details in31). See the power analyses in Tables S6–S9.

In Experiment 5, we conducted three statistical tests. First, we conducted a Welch’s t-test to determine the difference in PSEs between periodic and non-periodic stimuli without a transverse component at a 6 mm average spatial spacing to determine whether the presence of spatial periodicity of the stimulus influences the perceived speed. Second, we conducted a Welch’s t-test to determine the difference in PSEs between non-periodic stimuli with and without a transverse component at an average spacing of 6 mm to determine whether the average spatial spacing affects the perceived speed in the case of spatially non-periodic stimuli. Third, we conducted a one-way repeated-measures ANOVA with average spacing as a within-participant factor among non-periodic stimuli to determine whether the average spatial spacing influences the perceived speed in the case of spatially non-periodic stimuli. See the power analyses in Tables S10–S12.

Supplemental information

Document S1. Figures S1 and S2 and Tables S1–S12

Supplemental information can be found online at https://doi.org/10.1016/j.isci.2024.110803.
==== Refs
References

1 Culbertson H. Schorr S.B. Okamura A.M. Haptics: The present and future of artificial touch sensation Annu. Rev. Control Robot. Auton. Syst. 1 2018 385 409 10.1146/annurev-control-060117-105043
2 Gibson J.J. Observations on active touch Psychol. Rev. 69 1962 477 491 10.1037/h0046962 13947730
3 Katz D. Krueger L.E. The World of Touch 2013 Psychology Press
4 Lederman S.J. Tactual roughness perception: Spatial and temporal determinants Can. J. Psychol. 37 1983 498 511 10.1037/h0080750
5 Johansson R.S. Sensory input and control of grip Bock G.R. Goode J.A. Novartis Foundation Symposia 1998 Wiley 45 59 10.1002/9780470515563.ch4 discussion 59
6 Seizova-Cajić T. Fuchs X. Brooks J. Creating tactile motion Holmes N.P. Somatosensory Research Methods Neuromethods 2023 Springer 71 94 10.1007/978-1-0716-3068-6_4
7 Szabo Z. Enikov E.T. Development of wearable micro-actuator array for 3-D virtual tactile displays J. Electromagn. Anal. Appl. 04 2012 219 229 10.4236/jemaa.2012.46031
8 Summers I.R. Chanter C.M. Southall A.L. Brady A.C. Results from a tactile array on the fingertip Proceedings of Eurohaptics 2001
9 Ujitoko Y. Taniguchi T. Sakurai S. Hirota K. Development of finger-mounted high-density pin-array haptic display IEEE Access 8 2020 145107 145114 10.1109/ACCESS.2020.3015058
10 Velazquez R. Pissaloux E.E. Wiertlewski M. A compact tactile display for the blind with shape memory alloys Proceedings of the IEEE International Conference on Robotics and Automation. ICRA 2006 2006 2006 IEEE 3905 3910 10.1109/ROBOT.2006.1642300
11 Kawai Y. Tomita F. Interactive tactile display system: A support system for the visually disabled to recognize 3D objects Proceedings of the Second Annual ACM Conference on Assistive Technologies – Assets ’96 1996 ACM Press 45 50 10.1145/228347.228356
12 Dépeault A. Meftah E.-M. Chapman C.E. Tactile speed scaling: Contributions of time and space J. Neurophysiol. 99 2008 1422 1434 10.1152/jn.01209.2007 18199814
13 Ryan C.P. Ciotti S. Cosentino L. Ernst M.O. Lacquaniti F. Moscatelli A. Masking vibrations and contact force affect the discrimination of slip motion speed in touch IEEE Trans. Haptics 15 2022 693 704 10.1109/TOH.2022.3209072 36149999
14 Yang J. Kitada R. Kochiyama T. Yu Y. Makita K. Araki Y. Wu J. Sadato N. Brain networks involved in tactile speed classification of moving dot patterns: The effects of speed and dot periodicity Sci. Rep. 7 2017 40931 10.1038/srep40931
15 Delhaye B.P. O’Donnell M.K. Lieber J.D. McLellan K.R. Bensmaia S.J. Feeling fooled: Texture contaminates the neural code for tactile speed PLoS Biol. 17 2019 e3000431 10.1371/journal.pbio.3000431
16 Dallmann C.J. Ernst M.O. Moscatelli A. The role of vibration in tactile speed perception J. Neurophysiol. 114 2015 3131 3139 10.1152/jn.00621.2015 26424580
17 Moscatelli A. Scotto C.R. Ernst M.O. Illusory changes in the perceived speed of motion derived from proprioception and touch J. Neurophysiol. 122 2019 1555 1565 10.1152/jn.00719.2018 31314634
18 Kirman J.H. Tactile apparent movement: The effects of number of stimulators J. Exp. Psychol. 103 1974 1175 1180 10.1037/h0037355 4457592
19 Wertheimer M. Experimentelle studien uber das sehen von bewegung. In Zeitschrift fur psychologie 61 1912 161 165
20 Kenkel F. Untersuchungen über den Zusammenhang zwischen Erscheinungsgröße und Erscheinungsbewegung bei einigen sogenannten optischen Täuschungen 1913 Doctoral dissertation
21 Ujitoko Y. Takenaka Y. Hirota K. Effect of normal force intensity on tactile motion speed perception based on spatiotemporal cue IEEE Trans. Haptics 2024 10.1109/TOH.2024.3352042
22 Cholewiak R.W. The Perception of Tactile Distance: Influences of Body Site, Space, and Time Perception 28 1999 851 875 10.1068/p2873 10664778
23 Goldreich D. A Bayesian Perceptual Model Replicates the Cutaneous Rabbit and Other Tactile Spatiotemporal Illusions PLoS One 2 2007 1 11
24 Hidaka S. Tamè L. Zafarana A. Longo M.R. Anisotropy in tactile time perception Cortex 128 2020 124 131 32335326
25 Suto Y. The effect of space on time estimation (S-effect) in tactual space Japan. J. Psychol. 22 1952 189 200
26 Kirman J.H. Tactile apparent movement: The effects of interstimulus onset interval and stimulus duration Percept. Psychophys. 15 1974 1 6 10.3758/BF03205819
27 McIntyre S. Nagi S.S. McGlone F. Olausson H. The effects of ageing on tactile function in humans Neuroscience 464 2021 53 58 10.1016/j.neuroscience.2021.02.015 33607227
28 Peters R.M. Hackeman E. Goldreich D. Diminutive digits discern delicate details: Fingertip size and the sex difference in tactile spatial acuity J. Neurosci. 29 2009 15756 15761 10.1523/JNEUROSCI.3684-09.2009 20016091
29 Kajimoto H. Design of cylindrical whole-hand haptic interface using electrocutaneous display Isokoski P. Springare J. Haptics: Perception, Devices, Mobility, and Communication Lecture Notes in Computer Science 2012 Springer 67 72 10.1007/978-3-642-31404-9_12
30 Weinstein S. Intensive and extensive aspects of tactile sensitivity as a function of body part, sex and laterality In the First Int’l symp. on the Skin Senses 1968
31 Wobbrock J.O. Findlater L. Gergle D. Higgins J.J. The aligned rank transform for nonparametric factorial analyses using only ANOVA procedures Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (ACM) 2011 143 146 10.1145/1978942.1978963
