Introduction

The quality of social interaction depends on various non-social factors just as the quality of verbal communication depends on many non-verbal parameters [1]. In social robotics, physical human–robot interaction typically improves the social interaction between humans and robots [2]. One of the important issues in the design of communication robots is the selection of the covering material with appropriate touch sensations to improve the personality impressions of robots. The appearance of robots affects their personality impressions [3,4,5,6]. However, for humans, personality impressions may be influenced by not only visual perception but also some other modalities. Volume and tone are major components that inform the listener about the emotional state of the speaker [7]. Tactile sensations produced by touching products affect the quality impressions and attractiveness of the products. Therefore, it seems reasonable to consider that the touch of robots also influences their personality impressions.

Fig. 1
figure 1

Research overview of a potential causal relationship between the touch sensations of a robot part and the personality impressions of the entire robot (modified from [8]). Factor analysis was carried out to identify abstract factors from evaluation scores obtained by assessing questionnaire results. Path analysis was carried out, which revealed significant causal relationships between these factors. Then, an analysis of variance of the identified factors was carried out to investigate the differences in the evaluations among the different forearms

Several studies have attempted to realize touch-based interactions between humans and robots [9,10,11,12], and designers have selected comfortable covering materials for robots to influence the mental states of humans through the sensation of touch [13,14,15,16,17]. However, the design of the touch sensation of robots to improve their intended personality impressions is a challenge because the types of touch sensations that affect the personality impressions and the manner in which these sensations produce these effects have not been systematically investigated.

The touch sensations of robot skins [18], personality impressions of hugging dolls [19], and both the touch sensations and personality impressions of robot hands [20] were evaluated under different conditions, but the manner in which touch sensations affect personality impressions has not been reported so far. Therefore, touch sensations of robots have been determined intuitively and empirically and tuned in a try-and-error manner by designers. This inefficient situation could be improved if the causal relationship from touch sensations of robots to personality impressions of them is revealed. In other words, the designers can determine appropriate touch sensations of robots that provide their intended personality impressions systematically and efficiently.

To systematically understand the causal relationships between touch sensations and personality impressions, we have examined the perceived impressions of a robot by asking human participants to answer questionnaires containing a large number of questions [8]. In our previous study, we found significant causal relationships between tactile impression factors and personality impression factors with four types of silicone rubber forearms attached to a stationary robot. In this study, we analyzed the causal relationships in more detail to investigate how each forearm was evaluated by the participants. Figure 1 shows an overview of our study. Participants were instructed to touch and grab the silicone rubber forearm of a stationary robot using their right hand and then answer the evaluation questionnaires on touch sensations (19 items) and personality impressions (46 items) under each of the four conditions associated with different forearms that provide different sensations of hardness upon touching. A factor analysis of the evaluation scores was carried out to identify abstract impression factors for touch sensations and personality impressions [8]. A path analysis was then carried out, which revealed significant causal relationships between tactile impression factors and personality impression factors [8]. Finally, an analysis of variance (ANOVA) was carried out to investigate the difference in the evaluations by the participants among the different forearms.

Method

Participants

The participants were 20 healthy Japanese adults, including 10 males (mean age = 21.9, SD = 2.3) and 10 females (mean age = 22.9, SD = 1.2). Of these, 17 participants had no experience of contact with humanoid robots until the experiment, and 11 participants had no knowledge of humanoid robots. Two participants had contact with infants in the past 5 years. All participants provided written informed consent approved by the Ethics Committee of Graduate School of Engineering, Osaka University.

Robot

A child-like android robot named Affetto [21] was set on a desk in front of a chair, as shown in Fig. 2. The robot had a head and upper body, which were covered with a cloth or gloves; only its face and left forearm were exposed. The joints of the robot were physically fixed to maintain a particular posture and its left hand was held out to the participants so that the joint movements did not affect its impressions on the participants. A partition was placed between the robot and participants so that the robot could be hidden during forearm exchanges and could be shown to the participants only during the experiment.

Fig. 2
figure 2

Reproduced with permission from Yamashita et al. [8]

Appearance of the robot with exposed left forearm

Fig. 3
figure 3

Modified from Yamashita et al. [8]

Structural overview of the left forearm

Four different types of left forearms—A, B, C, and D—with identical sizes and appearances were prepared, and they were set (one at a time) between the left elbow and left hand of the robot under each experimental condition. Figure 3 shows the structural overview of the left forearm. Overall, the forearms were cylindrical with a diameter of 40 mm and a length of 105 mm. The main material of the forearm was cured platinum silicone rubber (Dragon Skin FX-Pro, Smooth-On Inc.), and its center was supported by a metal rod with a diameter of 8 mm. The outer surface of the silicone rubber was wrapped with a thin (7 \(\upmu \)m) polyurethane film (Airwall UV, Kyowa Ltd.) to ensure constant surface friction along the forearm.

Different amounts of two types of additives were mixed with the silicone rubber to provide different touch sensations to the forearms. One of the additives was a plasticizer (Silicone Thinner, Smooth-On Inc.), which reduces hardness, whereas the other was a thickener (Slacker, Smooth-On Inc.), which increases viscosity while reducing hardness. The additive contents were adjusted so that the hardness decreased from forearm A to forearm DFootnote 1, and so that forearms C and D had higher viscosity than forearms A and B. The different percentages of the plasticizer and thickener used for preparing the forearms are listed in Table 1.

Table 1 Amounts of two types of additives used for preparing the four forearms

Questionnaires

The semantic differential (SD) method [22] was used to measure the touch sensations and personality impressions. Two sets of a touch sensation questionnaire (TSQ) and personality impression questionnaire (PIQ), each with lists of several pairs of opposite Japanese adjectives, were provided to the participants. They were instructed to choose their responses on a seven-point scale between opposite adjectives, e.g., “Soft or Hard” for the TSQ and “Active or Passive” for the PIQ.

Table 2 summarizes the 19 adjective pairs used in the TSQ. Most of these pairs were selected from previous studies on touch sensations of artificial skins with eight adjective pairs [18] and touch sensations of robot hands with 10 adjective pairs [20]. Table 3 summarizes the 46 adjective pairs used in the PIQ. Most of these pairs were derived or selected from previous studies on personality impressions of robot hands with 12 pairs [20], impressions of hug dolls with 12 adjective pairs [19], quantification of impressions of humanoid robots with 33 adjective pairs, and meta-analysis of SD adjective pairs for personality impressions [23]. Thus, our adjective pairs were prepared such that we could investigate touch sensations and personality impressions thoroughly. These adjective pairs were translated into English by a professional translator for use in this manuscript.

Table 2 Adjective pairs used in the touch sensation questionnaire (TSQ)
Table 3 Adjective pairs used in the personality impression questionnaire (PIQ)

Procedure

The experimental procedures were divided into three sessions. In the first one, the participants were instructed on the manner in which the robot was to be evaluated. In the second one, they practiced evaluating their own forearms in the instructed manner. In the third one, they evaluated the robot by touching its forearm and then answering the questionnaires.

In the first instruction session, a movie describing the manner of touching the robot and the two questionnaires (TSQ and PIQ) were shown to the participants. The movie showed a demonstrator pinching one of the forearms of the robot with his thumb and forefinger, touching the forearm with the pads of his fingers, and holding the forearm with his hand several times. The participants were told to look at the forearm and touch it with their dominant hand as shown in the movie. In the second session, for training purposes, the participants were told to touch their own forearm with their dominant hand and answer the TSQ and PIQ. This session was conducted to check whether the participants had understood the instructions regarding the manner of touching the forearm and to allow the participants to get acquainted with the questionnaires. The third evaluation session was divided into four subsessions in which the participants evaluated each of the four forearms attached to the robot and the entire robot. In each subsession, the participants were instructed to touch the forearm of the robot for arbitrary time durations and then answer the TSQ. After answering it, they were told to touch the forearm again and then answer the PIQ. This subsession took approximately 10 min. Between subsessions, the robot was hidden from the participants by the partition, and the robot’s attached forearm was changed; then, the robot was shown to the participants again. The forearm replacement was completed in 1 min. The order in which each forearm was showed and the order of the adjective pairs in the questionnaires were shuffled for each participant. The entire series of sessions was completed in an hour.

Table 4 Factor matrix for the touch sensation factors
Table 5 Factor matrix for the personality impression factors

Data Processing

An exploratory factor analysis was carried out to identify several underlying factors, each of which was statistically reflected by several observed variables or evaluation scores of the adjective pairs. The maximum-likelihood estimation and varimax rotation were chosen as the factor extraction and rotation methods, respectively. The number of factors chosen according to the scree test was investigated by the Bayesian information criterion (BIC) and the root mean square error of approximation (RMSEA).

Path analysis was carried out to determine significant causal relationships between the found factors. Here, we assumed a multivariate multiple regression model whose independent variables were the touch sensation factors and the dependent variables were the personality impression factors. The maximum-likelihood estimation was used for estimating the model parameters. The R language (version 3.2.2) [24] was used for the above analyses.

Results

Factor Analysis

To detect the model structure, the number of dimensions should be reduced. We carried out factor analyses of each questionnaire by using the number of factors determined by inspecting the scree plot of eigenvalues. Four touch feeling factors (BIC \(=-\,292.2\) , RMSEA = 0.094) and three personality impression factors (BIC \(=-\,2774.5\) , RMSEA = 0.009) were extracted from the evaluation scores of each questionnaire. Tables 4 and 5 list the factor matrices for the touch sensation factors and personality impression factors, respectively. Based on the adjectives with high loadings for each factor, we named factor 1 as “Preference,” factor 2 as “Resilience,” factor 3 as “Smoothness,” and factor 4 as “Naturalness” for touch sensations. We named factor 1 as “Likability,” factor 2 as “Capability,” and factor 3 as “Vitality” for personality impressions.

Fig. 4
figure 4

Path diagram in which standardized coefficients represent the relationships between the touch sensation factors and personality impression factors [8]. The insignificant paths were removed. **p < 0.01, ***p < 0.001

Path Analysis

We carried out path analysis to investigate how tactile feelings affect the impressions of a humanoid robot. The seven variables considered were the factor scores calculated from the factor analyses. A satisfactory goodness-of-fit index (RMSEA < 0.001) for a full model including all of the possible paths was achieved.

Figure 4 shows the path diagram in which standardized partial regression coefficients represent the relationships between touch sensation factors and personality impression factors. The path thickness represents the magnitude of the coefficient. Solid and dotted lines represent positive and negative relationships, respectively.

We found that the Likability personality impression was highly and positively affected by the Preference touch sensation (\(\beta = 0.877\)); the Capability personality impression was positively influenced by the Resilience touch sensation (\(\beta = 0.665\)); and the Vitality personality impression was positively affected by the Preference and Resilience touch sensations (\(\beta = 0.385\) and 0.311). Additionally, the Smoothness touch sensation affected the Capability personality impression negatively but weakly (\(\beta =-0.200\)), and the Naturalness touch sensation affected all three personality factors weakly (\(\beta = 0.084, -0.276\), and 0.191).

ANOVA of Factor Scores

To evaluate the differences among the four types of forearms, we carried out seven ANOVAs for the three touch sensation factors and four personality impression factors. For each analysis, we used a one-way repeated-measures ANOVA and a multiple comparisons test using Shaffer’s modified Bonferroni procedure. All statistical thresholds were set at p < 0.05.

Fig. 5
figure 5

Comparisons of factor scores related to each touch sensation factor among the forearms. Each error bar shows the standard error of the mean. \(\dagger \)p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001. a Preference factor, b resilience factor, c smoothness factor, d naturalness factor

Figure 5 shows the comparisons of the factor scores related to each touch sensation factor among the forearms. The main effect of the Preference factor was significant [F(3, 57) = 6.720, p < 0.001]. Forearms B and C had higher scores than forearms A and D, which means that forearms B and C were evaluated to be more preferable (good-feeling and pleasant; Fig. 5a). The main effect of the Resilience factor was significant [F(3, 57) = 41.87, p < 0.001]. Forearm A had the highest score related to the Resilience factor, whereas forearm D had the lowest score, which means that forearm A was evaluated to be the most resilient (supple and tense; Fig. 5b). The main effect of the Smoothness factor was significant [F(3, 57) = 5.112, p = 0.003]. Forearm A had the lowest score related to the Smoothness factor, which means that forearm A was evaluated to be the smoothest (rough and coarse; Fig. 5c). The main effect of the Naturalness factor was significant [F(3, 57) = 5.235, p = 0.003]. Forearm D had the highest score related to the Naturalness factor, whereas forearm A had the lowest score, which means that forearm D was evaluated to be the most natural (rounded and moist; Fig. 5d).

Figure 6 shows the comparisons of the factor scores related to each personality impression factor among the forearms. The main effect of the Likability factor was significant [F(3, 57) = 5.469, p = 0.002]. Forearm C had the highest scores related to the Likability factor whereas forearm A had the lowest scores, which means that forearm C was evaluated to be more likable (desirable and agreeable; Fig. 6a). The main effect of the Capability factor was significant [F(3, 57) = 31.10, p < 0.001]. Forearm A had the highest score related to the Capability factor whereas forearm D had the lowest score, which means that forearm A was evaluated to be the most capable (robust and strong; Fig. 6b). However, the main effect of the Vitality factor was not significant [F(3, 57) = 1.841, p = 0.150]. No significant difference was found in the scores related to the Vitality factor as shown in Fig. 6c.

Fig. 6
figure 6

Comparisons of factor scores related to each personality impression factor among the forearms. Each error bar shows the standard error of the mean. \(\dagger \)p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001. a Likability factor, b capability factor, c vitality factor

Discussion

Several significant causal relationships between the touch sensation factors and personality impression factors were identified by the path analysis in accordance with our hypothesis. In other words, even when a robot had a fixed appearance, its personality impression could be modified by touch sensations. Thus, this study systematically showed how different touch sensation factors affected different personality impression factors.

First, we found that the Likability personality impression was improved if the robot provided preferable and natural touch sensations to the participants. This supports the conventional idea of covering communication robots with good-feeling and lifelike materials such as fur [14,15,16] and soft silicone rubber [20, 25]. In particular, preferable touch sensations strongly affect the Likability personality impression, and this emphasizes the importance of designing preferable touch feelings to build likable robots. We consider that the strong causal relationship between preferable touch sensation and likable personality is the result of a type of halo effect, known as human cognitive bias, and is defined as the influence of a global evaluation on the evaluations of individual attributes of a person [26]. The Likability personality impression has been considered as one of the important properties for communication robots, and therefore, several studies have attempted to improve the likability of robots by modifying their appearances and behavior [27, 28]. The application of the halo effect by touching robots will be another effective design technique to improve the likability of robots.

Second, the Capability personality impression was mainly enhanced by imparting resilient touch sensation to the participants. However, the Capability personality impression was reduced by smooth and natural touch impressions. This suggests that supple, tight, rigid, and hard covering materials with rough, coarse, and dry surfaces and sharp and squared shapes are desirable to impart the Capability personality impression to robots. From this perspective, hard covering materials such as metal, plastic, and hard rubber are considered to be desirable to enhance the Capability personality impression of robots.

Third, the Vitality personality impression was enhanced by three of the four touch sensation factors: Preference, Resilience, and Naturalness. These results are unexpected but valuable findings especially for social robot designers.

Here, we encounter a new research question: whether both Likability and Capability personality impressions can be improved using the same covering materials. When we cover robots with soft and comfortable materials, the Likability personality impression will be improved, whereas the Capability personality impression will be degraded. On the other hand, when we cover robots with hard and unnatural materials, the Capability personality impression will be improved, whereas the Likability personality impression will be degraded. To improve both the personality impressions, covering materials that can provide both preferable and resilient touch sensations simultaneously to humans should be identified.

We can identify several issues for future consideration from the ANOVA results. First, the relationship between a robot’s appearance and a human’s preference for forearms should be investigated. In this study, forearms A and D, which were evaluated to be the most or least resilient, respectively (Fig. 5b), failed to help the robot to provide the Likability personality impression to the participants (Fig. 6a). The possible reason for this is that these forearms gave the participants a sense of incongruence in terms of the forearms of the child-type android robot because their surface elasticities measured by the durometer were stronger than those of the co-contracted forearms of males or weaker than those of the relaxed forearms of females. However, we could not conclusively determine whether this result was universally applicable to different types of robots. We should investigate how the factor scores related to the Preferable factor change when these forearms are attached to different types of robots such as a non-android robot.

Second, additional experiments are necessary to confirm whether the viscosity of the forearms affect the touch sensations and therefore the personality impressions. We prepared four types of forearms having different amounts of two types of additives that changed the elasticity and viscosity of these forearms. However, in this experiment, the order of scores in terms of the Resilience factor (Fig. 5b) was similar to the order of surface elasticities reduced by using both the additives, and therefore, we could not conclude that the difference in viscosity affected the evaluation by the participants. To make such a conclusion, we should use forearms with different viscosities and the same elasticity in a future experiment.

Third, we should be careful while considering the mutual relationships among touch sensation factors. Although the four types of forearms had cylinders of the same size and were covered with the same thin film, their Smoothness and Naturalness touch sensations were different: Forearm A, which had the highest elasticity, was evaluated to be rougher and coarser (Fig. 5c), whereas Forearm D, which had the lowest elasticity, was evaluated to be the moistest and most rounded. Robot designers should have sufficient knowledge about such human touch sensation properties; otherwise, it would be difficult to design touch sensations for effective personality impression design of robots.

Conclusion

In this study, factor analysis, path analysis, and ANOVA were carried out on the evaluation scores of a robot’s touch sensations and personality impressions by using SD questionnaires to show how the touch sensations affect the personality impressions (e.g., in the case of halo effect). Several significant causal relationships between touch sensations and personality impressions were found. The study results are expected to help in the design of the personality impressions of robots by facilitating a more systematic design of touch sensations. Further studies are required to reveal the types of covering materials that will provide preferable touch sensations to humans when using several types of robots, including non-android robots.