Features of Two Embodied Processes in Spatial Perspective-Taking Across the Lifespan

1 Introduction

Spatial perspective-taking plays a fundamental and essential role in supporting smooth movements and spatial orientation in everyday life. It is the ability to “imagine how a scene looks like from different viewpoints” (Hegarty & Waller, 2004, p. 175). It is developmentally described as Level 1 (judging if an object can be seen or not by another person) and Level 2 (judging if an object is to the left/right or above/below another person; Flavell et al., 1981). Level 2 encompasses two qualitatively different processes of imagining the viewer’s perspective, and tracing a line of his or her sight. The former is enabled by placing oneself in a different position (perspective operation) and the latter means adding the cognitive processing required for task situations (information processing) to it (Michelon & Zacks, 2006). Moreover, Watanabe (2022) predicted that the perspective operation process could be subdivided into the generation of a representational self from the physical self (detachment) and the subsequent movement operation of representational self (self-representation movement). Self-representation is a few characteristics of one’s own body detached from inherent experiences and behaviors (Wallon, 1934). Healthy people, who can control their spatial perspective-taking intentionally, will feel difficulty imagining both self-representation and their self-body schema simultaneously in daily life, while those with hallucinatory symptoms unintentionally have an experience as if their own body schema were viewed by their self-representation (Blanke & Arzy, 2005). In this sense, the detachment process could be defined as a process of generating self-representation, which does not necessarily require a body image, from one’s physical self through self-body schema, which includes body image. The detachment will increase perspective objectivity leading to being free from one’s own physical self, and decrease egocentrism (Piaget & Inhelder, 1956). Therefore, decentering in spatial perspective-taking will proceed with the development of detachment. Egocentrism is observed until early childhood and attenuates considerably after adolescence (Rubin, 1973). Therefore, children execute the detachment process with greater difficulty than adults.

Self-representation does not necessarily have a real body shape, and its movement means imagining the self-perspective moving around in real space and being sent to another point. Nonetheless, the ability to operate self-representation is often influenced by real body movements (Surtees et al., 2013). Ruby and Decety (2001) measured brain activity using positron emission tomography when participants were imagining mimicking an act. Their results revealed a strong association between participants’ own perspectives and regions involved with somatic sensation (e.g., the inferior parietal lobe, precuneus, or postcentral gyrus). Wraga et al. (2005) reported that a physical constraint arising in simulating real-world physical motions in the supplementary motor cortex can manifest in the movement of the representational self. These findings imply that the operation of the representational self and actual body movement are homologous.

Children and older adults have weaker physical control than younger adults. Gouleme et al. (2014) studied the development of postural control in healthy children (aged 4–16 years) and healthy adults. They suggested that the growing of sensorial inputs and cerebellar integration in the development process could promote more efficient postural control. Molina et al. (2008) studied the development of postural control in 5-year-olds and 7-year-olds using a mental chronometry paradigm. They reported that the correlation between actual and virtual walking durations was observed only in 7-year-old children. These results lead to the anticipation that younger children’s poor performance in controlling their own bodies should also lead to poor performance in operating their representational selves. In older adults, more direct evidence has been presented. Concerning the development of motor imagery, Schott (2012) measured motor images in older adult men in three age groups (i.e., 60–69, 70–79, and ≥ 80 years) and young males (aged younger than 20 years). Schott reported that the ability of men older than 70 years was inferior to that of young men. This may lead to a gradual slowdown in the self-representation movement process of spatial perspective-taking in older adults (Watanabe & Takamatsu, 2014).

However, few cases exist in which these processes of detachment and self-representation movement in perspective operation across a lifespan were examined operationally. Thus, confirming this prediction will give us important cues to the core of the perspective-taking system by revealing that the body representational operation in spatial perspective-taking means imagining the self-image moving around in real space. By examining the differences among children, university students, and older adults on either the detachment or the self-representation movement processes, this study demonstrates the developmental features of these two processes in perspective operations.

For this purpose, an experimental task of spatial perspective-taking should be one that is able to separate the detachment process from the self-representation movement. Zacks and Michelon (2005) suggested that the speed of mental self-rotation depends upon the degree of angular disparity between one’s own and a target’s frame of reference. Watanabe and Takamatsu (2014) designed a task to elicit mental self-rotation separately from other cognitive processing and separated the perspective operation and the information processing as the slope and the intercept, respectively, using the regression line of the reaction times corresponding to each perspective. Their tasks will also be used in this study and the perspective operation time measured in it will be further divided into two parts here. First, the reaction latency (RL) from the stimulus presentation to the beginning of the gaze movement can be used as the time index for the detachment because detachment is assumed to precede self-representation movement in the perspective operation process. Inevitably, the time for the self-representation movement will be the remaining time (RT) calculated from the perspective operation time minus RL. In summary, the mental mechanism from stimulus presentation to the end of the response consists of three processes for spatial perspective-taking, which are the detachment, self-representation movement, and information processing to generate an appropriate response, which respectively correspond to the RL, RT between the perspective operation times and RL, and times that the intercept of regression line implies. These three kinds of indicators will clarify the developmental changes of spatial perspective-taking across a lifespan.

2 Materials and Methods

2.2 Procedure

The experiment was conducted individually and was initiated after establishing communication with the participants to ensure they could approach the task calmly. Participants sat on the chair, which was approximately 50 cm from the display that was set on a table about their waist height. They were requested to wear a head cap to measure eye movements, and it was ensured verbally that it was not unduly uncomfortable, nor would it interfere with performing the experimental task. This study adopts Watanabe and Takamatsu’s (2014) video game task (Figure 1). The game controller detected the player’s hand movement, and a virtual palm was projected on a monitor. The controller’s position was adjusted so that the virtual palm could be moved to every corner of the display screen when participants moved their own palms within a 30 (W) × 20 (D) cm (11.8 × 7.9 in.) quadrilateral around their chests. The game progressed automatically by placing the virtual palm for 2,000 ms on the icon on the screen. Before the measurement began, the aim and procedures of the game and the functions of the game controller were explained verbally, and it was confirmed that sufficient understanding had been reached. A practice session was introduced before the experimental session to familiarize participants with the procedure. Participants were judged to have a sufficient understanding of the task procedure when they could correctly answer three times in a row during the practice session. In both practice and experimental sessions, a child character hid behind either of the two windows of the house. Then, a countdown of “3, 2, 1” appeared in the center of the screen for 3,000 ms, followed by displaying the “Start” signal for 1,000 ms accompanied by a woman’s voice. Participants were requested to move the virtual palm onto the target window to find the hiding child. While all questions were presented at the 0° position (in which the house was not rotated) in the practice session, at the beginning of the experimental session, two questions were presented at the 0° position, followed by seven questions in other positions, when the house was rotated. The angle of rotation in the picture plane (45°, 90°, 135°, 180°, 225°, 270°, and 315° counterclockwise) was randomized by the computer program. This rotation took a maximum of 300 ms in the case of turning the house 180°. Before the “Start” signal appeared, participants had to keep the virtual palm at the standby region under the house. In this session, participants played the video game twice; the results of the second trial were adopted unless they were insufficient, and the data from the first trial were substituted if the second trial failed. The RLs were intervals between the presentation of the rotated stimulus on the screen and the beginning of the gazing point movement. The RT calculated from the perspective operation time minus RL was defined as the time calculated by subtracting RLs from the elapsed time between the “Start” signal and the starting moment of a player putting their virtual palm on any window.

Figure 1

Schematic of an experimental session in the video game task. An example of the house rotated to 180°.

3 Results

In the first step of data cleaning, a linear formula using the least squares method was calculated for each participant with the degrees of rotation between 45° and 180° as an independent variable and response times as a dependent variable following Watanabe and Takamatsu (2014). In doing so, data corresponding to rotation angles greater than 180° were included in ones corresponding to angles less than 180° as the average response times for the seven locations from 45° to 315° displayed a hill-like curve that peaked at around 180° in each age group. Gradient values over two and a half standard deviations from the mean were treated as outliers to reduce the likelihood that participants were temporarily distracted from the task. The data of five children, two students, and seven older adults with an outlier were excluded from the analysis. Six data with negative gradient values (three children, two students, and one older adult) were also excluded from the analysis. Additionally, data from three children who scored under five out of nine points were excluded from the analysis to increase the possibility that all participants understood the task procedures. The final sample comprised 25 participants in each age group.

The average value of RL at seven rotation angles was subtracted from the response time at each rotation angle to create values of the response times of mental rotation to each position after the detachment of the representational self (RTs). A two-way mixed-design analysis of variance (ANOVA) was conducted on the RTs with age group as an independent variable and the degrees of rotation as the repeated measure (Figure 2). The results revealed significant main effects of both the age group (F _{(2, 72)} = 12.47, p < 0.01, partial η ² = 0.257) and the degrees of rotation (F _{(6, 432)} = 19.72, p < 0.01, partial η ² = 0.215). Holm’s multiple comparisons test on the age group revealed that the child group (M = 1113.10 ms) was slower than the other age groups (students, M = 317.09 ms; older adults, M = 645.71 ms). The older adults were significantly slower than the students. Holm’s multiple comparisons test on the degrees of rotation also revealed some significant differences. The interaction between the age group and the degrees of rotation was also significant (F _{(12, 432)} = 2.78, p < 0.015, partial η ² = 0.072), which meant that the slopes of curves in each age group were different.

Figure 2

Average response times of mental rotation of representational self (RT; ms) for each rotation angle in the three age groups.

Next, a two-way mixed design ANOVA was conducted on the RLs with age group as an independent variable and the degrees of rotation as the repeated measure (Figure 3). The results indicated a significant main effect of the age group (F _{(2, 72)} = 33.64, p < 0.01, partial η ² = 0.438). Holm’s multiple comparisons test on the age group revealed that the children group (M = 752.94 ms) was significantly slower than the other two age groups (students, M = 307.83 ms; older adults, M = 292.71 ms). No significant main effect of the degrees of rotation was detected. In addition, the interaction between age group and condition was not significant.

Figure 3

Average response latencies (RL; ms) for each rotation angle in the three age groups.

4 Discussion

Children and older adults required more time to rotate their representational selves than students. This finding, which implies that the operation of the representational self and actual body movement are homologous, supports the existence of the self-representation movement in perspective-taking processes anticipated in the Introduction. Further, this finding has been consistent with the finding about the embodied perspective-taking and the neurological origin that a physical constraint arising in simulating real-world physical motions in the supplementary motor cortex has been reported to manifest in the movement of the representational self (Wraga et al., 2005). If the operation of the representational self and actual body movement are homologous, then the developmental level of both should correspond. In fact, the evidence from Guillot and Collet (2005) suggests that imagined and actual movements are temporally aligned for highly automatic and cyclical movements like reaching or walking and that when a movement requires more physical effort, imagining this movement takes longer.

In addition, children’s response latency was greater than that of students and older adults. However, this result should be interpreted cautiously as RL is not the exact time required for the detachment process. Based on reports that the typical saccadic latency for adults is less than 200 ms (Walker et al., 1995), and that of young children is about 50 to 100 ms longer (Matsuo et al., 2015), it is inferred that the RLs of about 300 to 700 ms found in this study include approximately 100 to 400 ms for the detachment. The much larger values in children (400 ms) than in adults (100 ms) could be owing to the strength of their egocentrism. Contrastingly, the fact that older adults’ values could be comparable to students is a key finding for understanding the robustness of perspective-taking systems in aging.

This assumes that the strategies used in the performance of the experimental task are nothing but perspective-taking. Consequently, the following three possibilities must be excluded. First, the mental rotation strategy is the most likely alternative, which means continuous mental manipulation of the stimulus. However, it is unlikely that this strategy will be adopted because the cognitive load from mentally rotating the object is extremely high for young children. Perceptual updating of the second possibility, which implies a continuous covariant relationship between internal information and external input (Rieser & Rider, 1991), means an optical flow produced by the rotating stimulus identifying the target. However, this will also be unlikely owing to the discontinuous presentation of the stimuli. The third possibility is the localization using some kinds of spatial cues, which implies that the target position is judged with salient features such as shape or color on the stimulus. This possibility, however, is also unlikely because the stimuli used in the task were symmetrical, and there were no cues of correct judgment. Consequently, the most likely strategy in the experiment task of this study will be the perspective-taking strategy.

On the assumption of using the perspective-taking strategy, a stricter separation of the time required for saccadic latency from the one required for detachment is needed to explore the detachment process in greater depth. Three approaches are required to establish the presence of the expected two processes in the perspective operation. First, the most informative approach means to use some variant of drift-diffusion modeling to separate response times pertaining to non-decision components and task-related processing. This will allow for non-task-related differences in response times that could be present between the age groups; that is, basic visual processing or developmental changes in natural viewing behavior as well as time to initiate and execute a motor response. For this, it is necessary to design experiments and collect data on the assumption that the model will be applied.

Second, this requires massive data to ensure sufficient statistical power. The accuracy of drift-diffusion modeling should be preserved even after data cleaning of many outliers, which will occur in young children and older adults.

Third, and most importantly, whether the results obtained here are indeed evidence of spatial perspective-taking requires reconsideration because success could be owing to mental rotation ability. This task has been demonstrated as highly likely to use the spatial viewpoint-taking strategy in Watanabe and Takamatsu (2014) and Watanabe (2016). However, confirmation is essential because there is no example of specifying a mental strategy using the point of gaze as an index. Solving these questions will undoubtedly raise the prospect of building a more versatile perspective-taking theory.