Self-Benefit and Others’ Benefit in Cooperative Behavior in Shared Space

Abstract

Objective

The objective is to clarify the nature of cooperative moving behavior that realizes smooth traffic with others from the viewpoint of the trade-off between self-benefit and others’ benefit in the shared space.

Background

The shared space is not constrained by formal rules or behavioral norms, and is a potentially ambiguous situation where it is not clear who has priority. Therefore, the nature of cooperative behavior in the shared space is unclear.

Method

An experimental task was conducted to compare cooperative and nonurgent moving behavior regarding completion time (self-benefit), the amount of interruption (others’ benefit), and the amount of operation (cognitive effort).

Results

First, cooperative behavior benefits others. Second, although cooperative behavior decreases self-benefit compared to the baseline without any instructions, it can obtain relatively more self-benefit than nonurgent behavior without considering self-benefit. Third, cooperative behavior requires cognitive effort.

Conclusion

Cooperative behavior provides benefit to both oneself and others by spending cognitive effort in not interrupting others.

Application

If the nature of the cooperative behavior can be clarified, a cooperative module can be implemented into the algorithms of various mobilities.

Keywords

cooperative behavior nonurgent behavior self-benefit others’ benefit cognitive effort

Cooperative Behavior in Traffic

The importance of cooperation with others in traffic has been discussed in many previous studies. Cooperative behavior improves the efficiency and safety in traffic (Fiosins et al., 2016; Fujii et al., 2010), and generates positive emotions in surrounding traffic participants (Mcknight, et al., 2011; Zimmermann et al., 2015). In contrast, the effects of uncooperative behavior have also been described; for example, unreasonable lane changing can cause serious collisions and delays (Tang et al., 2018; Matsubayashi et al., 2020), and drivers may experience stress and anger if asked to merge when there is only a small gap between vehicles (Riener et al., 2013).

Merging and lane changing are traffic situations in which cooperative behavior is especially likely to appear. Accelerating or decelerating a vehicle so that other vehicles can easily change lanes is considered as cooperative behavior (Hidas, 2005; Stoll et al., 2019, 2020). Such acceleration or deceleration in the main lane benefits other drivers attempting to change lanes by relieving stress and arousing positive emotions. In addition, it improves traffic safety by avoiding accidents.

Meanwhile, the acceleration or deceleration behavior requires cooperative cost. Operating gas, brake pedal, and steering are necessary to accelerate or decelerate a vehicle, and the cost of these operations becomes too large to accept another vehicle changing lanes. This tendency is even stronger under time pressure (Lütteken et al., 2016). Therefore, if the cost is perceived to be high, the driver may not accept another vehicle. In such cases, other drivers will not be able to obtain such a benefit.

Trade-Off Between Self and Others

In many traffic situations, one of the two participants must always yield. For example, on a road where two lanes merge into one, it is essential to decide whether one should go first (i.e., giving priority to one’s own benefit) or the other should go first (i.e., giving priority to the other’s benefit) because it is physically impossible to obtain both benefits at the same time (Shimojo et al., 2020). In this study, a situation in which two participants cannot obtain benefit at the same time is considered a trade-off between self-benefit and the others’ benefit. One’s benefit is how quickly one can reach their destination, and the other’s benefit is how quickly the other can reach their destination; in other words, how much the other is not interrupted by one.

The trade-off between oneself and others has a significant impact on decision-making (review: Bekkers & Wiepking, 2011), but this trade-off is expected to be more explicit in traffic situations. However, such a trade-off may appear only when moving behavior is rigidly constrained (e.g., merging). When the constraint is small, the trade-off between self-benefit and others’ benefit is unclear.

Shared Space

In recent years, shared space has appeared as an alternative to the traditional separated space (Moody & Melia, 2014). While each traffic participant is provided with dedicated space in the separated space (e.g., sidewalk for pedestrians or motorway for vehicles), various types of traffic participants such as pedestrians, bicycles, and vehicles share the same traffic space in the shared space. The separated space imposes strong constraints on traffic participants, especially for vehicles, which can move only in one dimension on the road. Meanwhile, in the shared space, such constraints are small, and all traffic participants can move bidimensionally (Weifeng et al., 2003). Although the shared space looks dangerous, it has been reported to improve safety compared to the separated space (Hamilton-Baillie, 2008; Kaparias et al., 2012; Kaparias & Wang, 2020; Yamamoto, 2020). Additionally, the implementation of the shared space reduces vehicle travel time and delays (Frosch et al., 2019). The shared space can also take various forms depending on the stage of integration (Barr et al., 2021). In some cases, a section of a city is shared by traffic participants over a wide area, while in other cases, a limited space, such as a parking lot, is shared by pedestrians and vehicles. Furthermore, some mobilities, such as transport carts in airport or guide robots in museums, enter a space that was originally occupied only by pedestrians, which accordingly becomes a shared space.

Shared space in traffic is not based on formal rules or behavioral norms, and is a potentially ambiguous situation where it is not clear who has priority (Uttley et al., 2020). In such ambiguous situations, explicit communication such as eye contact and signal (De Ceunynck et al., 2013) has been used to achieve cooperative interaction (Imbsweiler et al., 2018; Uttley et al., 2020).

However, it is imperative to verify what kind of implicit communication via moving behavior (e.g., acceleration, deceleration, or stop; De Ceunynck et al., 2013; Markkula et al., 2020) is perceived to be cooperative without relying on explicit communication. Some previous studies have shown that pedestrians tend to use vehicle-based implicit behavior rather than explicit communication cues from drivers (Lee et al., 2021).

Objective

The objective of this study is to clarify the nature of cooperative moving behavior that realizes smooth traffic with others from the viewpoint of trade-off between self-benefit and others’ benefit in the shared space where traffic participants can move freely in a two-dimensional space. The reciprocal of the time before one reaches their destination is defined as self-benefit, and the reciprocal of the total time when the other is interrupted by one is defined as others’ benefit.

A condition in which participants are directly required to behave cooperatively is set up (hereafter, cooperative condition). In the cooperative condition, participants are required to reach their destination with the utmost consideration for other people. Others’ benefit is expected to be greater in the cooperative condition than in the baseline condition without any instructions because the participant in the cooperative condition deliberately attempts not to interrupt others.

The main question in this study is what the nature of cooperative behavior in the shared space is. Specifically, this study clarifies what benefits cooperative behavior provided to oneself and others, and the relationship between self-benefit and others’ benefit. Therefore, the nonurgent condition is set up as contrastive to the cooperative condition. Participants in the nonurgent condition are told that they have no work to do and can reach their destination slowly. The main question consists of the following three research questions:

Our first research question is whether a trade-off between self-benefit and others’ benefit exists even in the shared space, which is ambiguous based on informal rules. There is a strong trade-off between self-benefit and others’ benefit in conventional traffic situations, such as merging and lane changing. However, it is questionable whether such an expectation is also true in the shared space.

RQ1: Are the trade-offs between self-benefit and others’ benefit observed in each of the cooperative and nonurgent conditions?

The assumption of a trade-off also involves a direct comparison of the benefits between cooperative and nonurgent conditions. When one is required to consider others, others’ benefit is expected to be larger. However, simultaneously, self-benefit may be smaller than when one is required not to consider oneself, given a strong trade-off between self-benefit and others’ benefit. This direct comparison reveals the major nature of cooperative behavior.

RQ2: Is there a difference in the benefits gained in cooperative and nonurgent conditions? Especially, does self-benefit decrease in the cooperative condition?

Providing benefit to others requires a certain amount of cognitive effort. Acceleration and deceleration, which make it easier for others to change lanes, is a cooperative cost (Stoll et al., 2019, 2020) and can be considered a cognitive effort. If a certain amount of cognitive effort is necessary to provide benefit to others, a more cognitive effort will be paid in the cooperative condition than the nonurgent condition.

RQ3: How does cognitive effort differ between the cooperative and nonurgent conditions?

The purpose of this study is to clarify the nature of cooperative behavior in terms of the trade-off between self-benefit and others’ benefit in the shared space. An experiment is conducted to answer the above three RQs.

Method

Participants

Forty-seven participants participated in the experiment. Powers were calculated with G*Power version 3.1 (Faul et al., 2009). As a result, all powers were high enough for each analysis (1 – β > .91, given α = 0.05 and effect size f = 0.10). The details of the calculation of powers are described in the Appendix. This research complied with the American Psychological Association Code of Ethics and was approved by the Institutional Review Board at Institutes of Innovation for Future Society (MIRAI), Nagoya University. Informed consent was obtained from each participant.

Design

The experiment was designed using two within-subject independent variables and three dependent variables. The first independent variable was the three-level Instruction factor to compare cooperative behavior with the other two behaviors. The second independent variable was the three-level Number of others factor to control crowdedness. The three dependent variables were the behavioral data of the participants in a task, which corresponded to self-benefit, others’ benefit, and cognitive effort (details in Procedure).

Stimulus

Although explicit communication, such as gestures, are available in everyday situations, this study focused only on moving behavior as implicit communication, and an experimental task in which all other factors were controlled was developed.

In this task, an agent controlled by the participants (hereafter, own agent) and other autonomous agents (hereafter, other agents) moved in a simulated shared space (Figure 1), which the participants had a bird’s-eye view of. The screen size was 900 pixel × 900 pixel and the size of agents was 30 pixel in diameter. The own agent and other agents were displayed in blue and gray, respectively. The own agent became red when in contact with other agents. The own destination was displayed as an 80 pixel × 80 pixel square, and the participants were required to direct the own agent to its destination for each trial with a Microsoft Xbox controller. The frame rate was 50 fps, and all agents could move up to 4 pixels per frame in each X/Y direction. At the beginning of each trial, the own agent was placed in one of four corners, and its destination was always placed on the diagonal, that is, the shortest distance to the destination was the same for all trials.

Figure 1.

Overview of the Experimental Task. Note. Three screenshots indicate the conditions where Number of others is 5 (left), 10 (middle), and 20 (right), respectively.

Number of others in the torus-shaped space was 5, 10, or 20 and always constant during a trial. The other agents were set to move randomly by sometimes changing their speed and direction. Furthermore, other agents slowed down or stopped based on the following algorithm when different agents were in their direction to simulate a realistic crossing interaction: Other agents slowed down if the distance from different agents was within 100 pixel. Additionally, they stopped if the distance was 50 pixel. However, in this algorithm alone, once two other agents stopped, neither would be able to move again. Therefore, when two other agents were likely to stop together, one of them would pass slowly.

Number of others was three levels (5, 10, and 20) as a within-subject independent variable, and the starting point of own agent was one of the four corners. These 12 trials were repeated twice, and eventually one set consisted of 24 trials. Since the starting point of own agent was collapsed, eight trials were assigned for each level in Number of other factor.

Procedure

The following instructions were given to the participants to help them understand that the situation of the task corresponded to an actual shared space: “Imagine a crowded lobby of a hospital, airport, shopping center, etc., and reach your goal.” Afterward, the participants practiced operating the controller. In this practice phase, other agents and the destination were not presented. After this, five main sets were conducted. In Sets 1 and 2, the participants were instructed to reach their own goals. Set 1 intended to help the participants become familiar with the other agents. The data in Set 2 were used as a baseline without any instruction.

In Sets 3, 4, and 5, one of the following three instructions was given as Instruction factor, as a within-subject independent variable: “Reach your destination while considering others” (cooperative condition), “You have enough time and can go to your destination slowly” (nonurgent condition), and “You do not have enough time and should reach your destination as fast as you can” (urgent condition). The urgent condition was set as a contrastive to the cooperative and nonurgent conditions, that is, the cooperative and nonurgent conditions were expected to decrease self-benefit and increase others’ benefit compared to the baseline, while the urgent condition was expected to increase self-benefit and decrease others’ benefit. Specifically, the urgent condition was set to confirm the trade-off between self-benefit and others’ benefit.

The participants answered a questionnaire on the reflection on moving behavior in the previous set, which was conducted immediately after Sets 2, 3, 4, and 5 on a 7-point scale. This questionnaire consisted of four items based on the four features of prosocial behavior in social psychology (Kikuchi, 1988, 2014). The four items were “Your behavior included assisting others,” “You expected the reward from others,” “Your behavior resulted in some loss for you,” and “Your behavior was spontaneous.”

The three conditions in Instruction factor (cooperative, nonurgent, and urgent) corresponded to Sets 3, 4, and 5. Furthermore, a Set consisted of 24 trials, where eight trials were averaged for each of three levels in Number of others factor (5, 10, and 20). The orders of three conditions in Instruction and three levels in Number of others were counterbalanced among the participants, respectively.

The following behavioral data were obtained for each trial as dependent variables to answer the three RQs: the completion time corresponding to self-benefit, the amount of interruption corresponding to others’ benefit, and the amount of operation corresponding to self-cognitive effort. The completion time was the total time it took for the own agent to reach the destination. The amount of interruption was the total amount of time that the own agent slowed or stopped the other agents. If the own agent interrupted more than two other agents simultaneously, the number of interrupted agents was counted. The amount of the interruption was divided by the completion time as an index per unit time because it tended to increase as the completion time increased. Finally, the amount of operation was calculated as the total amount of controllers operated by the participants. In other words, it was the sum of the vector differences between previous and current time points. It was calculated to be zero while the own agent was in constant speed or straight movement, and it increased when the speed or direction changed.

The differences calculated from the baseline in Set 2 were used as data for all statistical analyses.

Results

Questionnaire

The responses to the questionnaire were used to confirm whether the manipulation of the instruction was appropriate. The differences were calculated from the baseline for each of the three conditions and then compared. Table 1 shows the results of a one-factor ANOVA for the four items. First, the main effect of Instruction factor was significant for “assistance for others” (F (2, 92) = 114.21, p < .001, η² = .36), and the result of multiple comparisons showed that there were significant differences among all conditions (cooperative-nonurgent: t (92) = 2.12, p = .036, r = .22; nonurgent-urgent: t (92) = 11.89, p < .001, r = .78; urgent-cooperative: t (92) = 14.02, p < .001, r = .83). The main effect was also significant for “expectation of reward” (F (2, 92) = 11.04, p < .001, η² = .09), and the value in the urgent condition was significantly lower than those in the cooperative and nonurgent conditions (cooperative-nonurgent: t (92) = 0.30, p = .764, r = .03; nonurgent-urgent: t (92) = 3.91, p < .001, r = .38; urgent-cooperative: t (92) = 4.21, p < .001, r = .40). Furthermore, the main effect was also significant for “self-loss” (F (2, 92) = 5.39, p = .006, η² = .03), and the value in the nonurgent condition was significantly lower than those in the cooperative and urgent conditions (cooperative-nonurgent: t (92) = 2.91, p = .004, r = .29; nonurgent-urgent: t (92) = 2.76, p = .006, r = .28; urgent-cooperative: t (92) = 0.15, p = .878, r = .02). Meanwhile, the main effect was not significant for “spontaneity” (F(2, 92) = 1.05, p = .353, η² = .01). These results indicated that the instruction makes significant differences in the reflection on moving behavior, confirming that the conditional manipulations were appropriately conducted.

Table 1:

Means, Standard Errors, and One-Way ANOVA Statistics for Questionnaire

Item	Cooperative		Nonurgent		Urgent		F (2, 92)	p	η ²
Item	M	SE	M	SE	M	SE	F (2, 92)	p	η ²
Assistance for others	2.09	0.29	1.60	0.28	−1.15	0.25	114.21	<.001^***	.36
Expectation of reward	0.53	0.22	0.47	0.18	−0.36	0.15	11.04	<.001^***	.09
Self-loss	0.32	0.29	−0.49	0.32	0.28	0.25	5.39	.006^**	.03
Spontaneity	−0.34	0.24	−0.09	0.24	0.00	0.18	1.05	.353	.01

Note. Each value indicates the distraction from the baseline. ^**p < .01. ^***p < .001.

Comparison with Baseline

As RQ1, it was examined whether a trade-off between self-benefit and others’ benefit was observed in the cooperative and nonurgent conditions. In this analysis, the completion time and the amount of interruption were adopted as indices for self-benefit and others’ benefit, respectively. These two indices for each of the three conditions were compared with the baseline to confirm the trade-off.

Completion Time (Self-Benefit)

Table 2 shows the results of the one-sample t-test for the completion time in each condition. The results indicated that all the differences between the baseline and the cooperative, nonurgent, and urgent conditions were significant at 5, 10, and 20 other agents, respectively. In sum, cooperative and nonurgent behaviors decreased self-benefit compared to the baseline without any instructions, while the urgent behavior increased self-benefit.

Table 2:

Means, Standard Errors, and t-test Statistics for Completion Time

Condition	M	SE	t (46)	p	r
Cooperative
5	0.63	0.11	5.72	<.001^***	.64
10	0.93	0.13	6.85	<.001^***	.71
20	1.30	0.17	7.58	<.001^***	.75
Nonurgent
5	1.09	0.12	4.87	<.001^***	.58
10	1.40	0.24	5.68	<.001^***	.64
20	1.81	0.22	8.21	<.001^***	.77
Urgent
5	−0.47	0.07	−6.46	<.001^***	.69
10	−0.68	0.12	−5.72	<.001^***	.64
20	−0.94	0.11	−8.44	<.001^***	.78

Note. Each value indicates the distraction from the baseline. ^***p < .001.

Interruption (Others’ Benefit)

Table 3 shows the results of the same one-sample t-test for the interruption. Different tendencies were found among the three conditions. Cooperative behavior caused less interruption and significantly increased others’ benefit compared to the baseline. Meanwhile, there were no significant differences between the nonurgent condition and the baseline. Additionally, urgent behavior caused more interruption and significantly decreased others’ benefit.

Table 3:

Means, Standard Errors, and t-test Statistics for Interruption

Condition	M	SE	t (46)	p	r
Cooperative
5	−0.010	0.004	−2.62	.011^*	.36
10	−0.012	0.006	−1.96	.055	.28
20	−0.034	0.011	−3.18	.002^**	.42
Nonurgent
5	−0.008	0.004	−1.97	.054	.28
10	−0.003	0.007	−0.48	.633	.07
20	−0.010	0.011	−0.85	.397	.12
Urgent
5	0.016	0.005	3.14	.002^**	.42
10	0.024	0.008	3.07	.003^**	.41
20	0.030	0.011	2.70	.004^**	.37

Note. Each value indicates the distraction from the baseline. ^*p < .05. ^**p < .01.

Results for RQ1

A trade-off between self-benefit and others’ benefit was confirmed when the three conditions were compared with the baseline in RQ1. The results showed that different types of trade-offs occurred. A trade-off was confirmed in the cooperative condition, since cooperative behavior decreased self-benefit and increased others’ benefit. Conversely, nonurgent behavior also decreased self-benefit; however, did not increase others’ benefit, which indicated no trade-off. Urgent behavior increased self-benefit and decreased others’ benefit. Notably, cooperative behavior significantly increased others’ benefit.

Differences in Benefits Between Cooperative and Nonurgent Behaviors

The results of RQ1 showed a significant difference in others’ benefit between the cooperative and nonurgent conditions. In RQ2, the cooperative condition was directly compared to the nonurgent condition using the differences from the baseline, which had the same values as in Tables 2 and 3. A 2 (Instruction factor: cooperative and nonurgent) × 3 (Number of others factor: 5, 10, and 20) ANOVA within-subject analysis was conducted.

Completion Time (Self-Benefit)

A 2 × 3 ANOVA was conducted for completion time (Figure 2). The results showed that the main effect of Instruction was significant (F (1, 46) = 9.05, p = .004, η² = .03). While the main effect of Number of others was also significant (F (2, 92) = 17.11, p < .956, η² = .04), the interaction was not significant (F (2, 92) = 0.04, p = .956, η² = .01). Thus, self-benefit in the cooperative condition was greater than in the nonurgent condition.

Figure 2.

Means of Completion Time. Note. Each value indicates the distraction from the baseline. Error bars indicate the standard errors.

Interruption (Others’ Benefit)

The same ANOVA was conducted for interruption (Figure 3). The results showed that the main effect of Instruction was significant (F (1, 46) = 7.18, p = .010, η² = .01) while the main effect of Number of others and the interaction were not significant (F (2, 92) = 1.48, p = .231, η² = .01; F (2, 92) = 2.85, p = .062, η² = .01). Thus, others’ benefit was greater in the cooperative condition than in the nonurgent condition.

Figure 3.

Means of Interruption. Note. Each value indicates the distraction from the baseline. Error bars indicate the standard errors.

Results for RQ2

The results for RQ1 showed that both cooperative and nonurgent behaviors decreased self-benefit. A direct comparison in RQ2 showed that the decrease in self-benefit was smaller in cooperative behavior than in nonurgent behavior. Furthermore, it was directly confirmed that the cooperative behavior tended to bring more benefit to others than nonurgent behavior. In other words, cooperative behavior brought relatively more benefits to both oneself and others than nonurgent behavior did.

Differences in Cognitive Effort Between Cooperative and Nonurgent Behaviors

Cooperative behavior could bring relatively more benefit both to oneself and others, which seemed difficult to achieve. The difference in the cognitive effort required in the cooperative and nonurgent conditions was examined as RQ3. Here, the total amount of controllers operated by the participants was used as the index of cognitive effort.

A 2 × 3 ANOVA was conducted for the amount of operation (Figure 4). The results showed that the main effect of Instruction was significant (F (1, 46) = 11.49, p = .001, η² = .02) while the main effect of Number of others and the interaction was not significant (F (2, 92) = 0.97, p = .381, η² = .01; F (2, 92) = 1.98, p = .143, η² = .01).

Figure 4.

Means of Amount of Operation. Note. Each value indicates the distraction from the baseline. Error bars indicate the standard errors.

Therefore, the answer for RQ3 was that cooperative behavior included a greater amount of operation than nonurgent behavior. This indicated that the participants in the cooperative condition changed their speed and direction frequently or significantly.

Discussion

The purpose of this study is to clarify the nature of cooperative behavior in terms of the trade-off between self-benefit and others’ benefit in the shared space. Pursuant to this purpose, the cooperative condition, in which they are required to behave cooperatively, was compared to the nonurgent condition, in which they have no work to do and can reach their destination slowly. The three RQs and their validation results are as follows.

RQ1 whether a trade-off between self-benefit and others’ benefit was observed. As a result, cooperative behavior decreases self-benefit and increases others’ benefit, indicating a trade-off. However, the trade-off does not appear in the nonurgent condition because it does not increase others’ benefit while it decreases self-benefit.

RQ2 directly answered whether there is a difference in benefits between the cooperative and nonurgent conditions. The result showed that cooperative behavior obtained significantly more self-benefit and others’ benefit than nonurgent behavior did.

RQ3 examined the process by which cooperative behavior provides benefit both to oneself and others. The result showed that cooperative behavior required more cognitive effort to provide benefit than nonurgent behavior.

Based on the above answers to the RQs, the following three points are the nature of cooperative moving behavior. First, cooperative behavior provides benefit to others. Second, although cooperative behavior decreases self-benefit compared to the baseline without any instructions, it can obtain relatively more self-benefit than nonurgent behavior without considering self-benefit. Third, cooperative behavior requires much cognitive effort.

The Nature of Cooperative Behavior

Providing Benefit to Others

Among the three conditions in the experiment, only the cooperative condition increased others’ benefit. It was greatly decreased in the urgent condition and did not change significantly in the nonurgent condition.

In this study, others’ benefit was calculated as the reciprocal of the total time when other agent is interrupted by the own agent. Specifically, situations where the own agent approached the other agent were considered an interruption. This means that when the own agent interrupted the other agent, the own agent was always ahead of and passed in front of the other agent. The situation where the other agent went first was not considered an interruption, which provided benefit to the other agent.

The result that giving way provides benefit to others is consistent with previous findings that the person giving way contributes more to collision avoidance than the person crossing first (Knorr et al., 2016; Olivier et al., 2013). However, in these studies, the path passed by the participants was roughly determined, while the agents in this study could move freely in a two-dimensional space. Additionally, it was impossible for the other agent to go ahead of the own agent because of the settings of the other agent. Although there are some slight differences, the situation in this task is similar to those in previous studies in that cooperative behavior is realized through giving way.

Providing Benefit to Oneself

Surprisingly, cooperative behavior obtained more self-benefit than nonurgent behavior. This behavior can be interpreted as selfish altruism (Yamagishi, 1990), in which the decision one makes for others provides self-benefit eventually. In this experiment, the participants required to behave cooperatively could inhibit interruption to others, and as a result, could reach their destination earlier.

Although selfish altruism has been confirmed in prisoner’s dilemma situations, it is surprising that it was found in the traffic situation. Yamagishi (1990) states two requirements for selfish altruism as follows: (1) the choice of cooperation or noncooperation must be made between two persons, not among a group, and (2) the relationship between the two persons must be long-lasting, not temporal. However, neither of requirements were satisfied in this task. There were multiple other agents, and the moving behavior would affect many other agents. In addition, most of the relationships among the agents must have been temporal because they had only one chance to face each other. It is curious that selfish altruism appeared when cooperative behavior was required. This is because the participants may have been aware that they could obtain self-benefit by providing benefit to others. According to the questionnaire, the participants required to be cooperative had an explicit intention to assist others, and might have had an implicit intention to assist themselves. This should be verified in future studies.

Selfish altruism in cooperation might be closely related to the minimization of decision entropy (Kawaguchi et al., 2021). This solution urges drivers to accelerate or decelerate, and reduces their mental load eventually, that is, minimizing a decision entropy can be considered cooperative behavior which provides benefit both to themselves and others.

Spending Self-Effort

Cooperative behavior required a lot of cognitive effort. The questionnaire also revealed that cooperative behavior entailed subjective self-loss compared to nonurgent behavior. Although this may be a reason for the unavailability of explicit communication, it is likely that the participant could not provide benefit both to oneself and others without such a large amount of effort. Meanwhile, lack of effort may be a reason for nonurgent behavior not increasing others’ benefit.

Previous studies have proposed that pedestrians have a common principle of collision avoidance with minimal path adjustment (Basili et al., 2013). Conversely, the participants in the cooperative condition in the present study took the initiative to change speed and direction more frequently than those in the nonurgent condition. Although path adjustment in the cooperative condition was not minimal or grounded in the principle, it was less likely for the own agent to interrupt or collide with others.

The cooperative behavior in this study provides benefit to both oneself and others. Specifically, cognitive effort in cooperative behavior provides self-benefit via others’ benefit (Hulst et al., 1998). This indicates that the own agent allowed others to go first by changing its own speed and direction, and as a result, could reach its destination earlier. Cooperative behavior also contributed significantly to collision avoidance alongside a person giving way as reported in previous studies (Knorr et al., 2016). It is likely that the agent giving way spends more effort in the actual crossing situations. The detail of allocation of effort may be clarified by comparing the subjective cognitive effort between the person going first and the person giving way.

Cooperative Behavior in Shared Space

Previous studies in social psychology have discussed why humans are cooperative and attempt to provide benefit to others, even if self-benefit may be decreased (e.g., Rubaltelli et al., 2020). Specifically, nonmonetary motivations, such as warm glow (e.g., Andreoni, 1990) or psychological benefits (Bekkers & Wiepking, 2011; Meier & Stutzer, 2008), have been suggested as factors that encourage one to assist others, even if it decreases their own benefit in charity or donation.

In contrast, as for traffic in the shared space, the results of this experiment indicate that the trade-off between self-benefit and others’ benefit is not strong. Nonurgent behavior decreased self-benefit but did not increase others’ benefit, which means no trade-off. Conversely, cooperative behavior also decreased self-benefit but did not neglect others’ benefit by spending more cognitive effort. Although perceived cost of cooperation has been shown to influence decision-making (Stoll et al., 2019, 2020), it may be different when the constraints on movement are small as in the shared space. The results of the questionnaire also show that the subjective intention to assist others was higher in the cooperative condition, and that the cost of cooperation may have been estimated to be smaller. The subjective differences between cooperative and nonurgent behaviors need to be examined in future studies.

Various behaviors toward others are important to realize cooperation in ambiguous traffic situations. For example, in the T-intersection or narrow path, a driver is likely to accept the partner’s offer and go first when perceiving the partner’s willingness to cooperate (Imbsweiler et al., 2018). Defensive intentions (intention to give way) with deceleration or flashing light and offensive intentions (intention to go first) with acceleration contributes to cooperation. In parking lots, nonverbal communication reduces ambiguity, and realizes cooperative interaction between vehicles and pedestrians (Uttley et al., 2020).

Meanwhile, no such explicit communication was available in this experimental task. The nature of cooperative behavior was clarified in terms of the implicit communication such as acceleration or deceleration and changing direction. These findings will be useful in the future when automated vehicles interact with pedestrians, bicycles, and other vulnerable road users (Stoll et al., 2019), or the types of mobilities vary.

Application

In the shared space, various types of small mobilities participate in the traffic. The forms of mobilities will become increasingly diverse in the near future, which includes wheelchairs and single-seater personal mobility vehicles. If the nature of cooperative behavior can be clarified, a cooperative module can be implemented into the algorithms of various mobilities. Traffic participants in the shared space can show various behaviors due to its small constraints, but a cooperative module may realize cooperative decision-making for others with the indices adopted in this study regarding self-benefit and others’ benefit. This study suggests that one will eventually obtain self-benefit if they behave cooperatively toward other traffic participants.

Future Work

The Number of Others

Although Number of others factor was manipulated to control crowdedness in this experiment, most of its effects were not observed. However, as Number of others increased, the nature of cooperative behavior tended to become more salient. This implied that cooperative complied with the presence of others. Therefore, its nature may disappear when the Number of others is extremely large, as shown in Figure 4. Perhaps, its effect may not always be linear, and there may be a limit to how many people can be considered simultaneously.

The effect of the Number of others did not influence the general conclusions of this study since there was no interaction between the instruction and the Number of others. However, the constraints under which cooperative behavior can work must be verified.

Intention Perception

It is imperative to infer others’ intention to cooperate in traffic. Social interaction is based on the understanding and prediction of others’ mental state that led them to take a certain action (Baker et al., 2009). Additionally, in traffic, the intentions of all relevant traffic participants should be transparent (Stoll et al., 2019). In fact, cooperation often includes the process of inferring the intentions of others; for example, behaviors of changing speed or direction have been shown to help the partner to infer their intention (Sebanz & Knoblich, 2009).

It is easy to infer the intention of cooperative behavior in this experiment. Although other agents were not actual persons, the participant may have been trying to show their intention explicitly to others. However, an internal dilemma between selfish and prosocial decisions may have arisen (Mellers et al., 2010). Despite such dilemma, cooperative behavior provided benefit both to oneself and others. The kinds of intentions in cooperative behavior should be examined. In addition, it should be verified whether such intentions are easily discernable by others, and whether they are perceived as cooperative or not.

Actual Shared Space

It is essential to apply this experimental paradigm to an actual shared space. In this case, various factors are assumed to affect cooperative behavior (Amini et al., 2019). For example, the difference in mobility between oneself and others may affect the cooperative behavior (Chauvin & Lardjane, 2008). If one has high mobility and can move quickly or agilely, they may be regarded as cooperative when they do not give way to others and go first in the crossing situation. Additionally, the results of this study indicate no notable effect of the number of other agents, but a nonlinear pattern with it was found in some cases. In an actual shared space, the effect of other agents is likely to be more pronounced because of the first-person perspective. Future studies should verify this.

Key Points

This study clarified the nature of cooperative moving behavior in the shared space in terms of three indices, namely others’ benefit, self-benefit, and cognitive effort.

First, cooperative behavior provides others’ benefit by decreasing interruption.

Second, although cooperative behavior decreases self-benefit compared to the baseline behavior without any instructions, it can obtain relatively more self-benefit than nonurgent behavior without considering self-benefit by reaching own destination earlier.

Third, cooperative behavior requires cognitive effort by way of changing own speed or direction frequently or significantly.

Footnotes

Acknowledgments

Support for this work was given by JSPS KAKENHI Grant Number 18H05320 and Toyota Motor Corporation (TMC). However, note that this article solely reflects the opinions and conclusions of its authors and not TMC or any other Toyota entity.

Author’s Contributions

Shota Matsubayashi: Review of previous studies, Theory construction, Development of the experimental task, Analysis of the experimental data, and Discussion

Kazuhisa Miwa: Theory construction and Discussion

Hitoshi Terai: Theory construction and Discussion

Asaya Shimojo: Review of previous studies and Analysis of the experimental data

Yuki Ninomiya: Review of previous studies and Analysis of the experimental data

Ethical Approval

This research complied with the American Psychological Association Code of Ethics and was approved by the Institutes of Innovation for Future Society (MIRAI), Nagoya University (Approval Number: 2020-25). Informed consent was obtained from each participant.

ORCID iDs

Shota Matsubayashi

Hitoshi Terai

Appendix

Author Biographies

Shota Matsubayashi is a designated assistant professor in the Institutes of Innovation for Future Society (MIRAI) at Nagoya University in Nagoya, Aichi, Japan. He received his Ph.D. in information and science from Nagoya University in 2020.

Kazuhisa Miwa is a professor in the Graduate School of Informatics at Nagoya University in Nagoya, Aichi, Japan. He received his Ph.D. in engineering from Nagoya University in 1989.

Hitoshi Terai is a professor in the Faculty of Humanity-Oriented Science and Engineering at Kindai University in Iizuka, Fukuoka, Japan. He received his Ph.D. in information and science from Nagoya University in 2006.

Asaya Shimojo is a Ph.D. Student in the Graduate School of Informatics at Nagoya University in Nagoya, Aichi, Japan.

Yuki Ninomiya is a Ph.D. Student in the Graduate School of Informatics at Nagoya University in Nagoya, Aichi, Japan.

References

Amini

R. E.

Katrakazas

Antoniou

(2019). Negotiation and decision-making for a pedestrian roadway crossing: A literature review. Sustainability (Switzerland), 11(23), 1–24.

Andreoni

(1990). Impure altruism and donations to public goods: A theory of warm-glow giving. The Economic Journal, 100(401), 464–477. https://doi.org/10.2307/2234133

Baker

C. L.

Saxe

Tenenbaum

J. B.

(2009). Action understanding as inverse planning. Cognition, 113(3), 329–349. https://doi.org/10.1016/j.cognition.2009.07.005

Barr

Lampkin

Dawkins

Williamson

(2021). Shared space: Negotiating sites of (un)sustainable mobility. Geoforum, 127(1), 283–292. https://doi.org/10.1016/j.geoforum.2021.11.012

Basili

Saĝlam

Kruse

Huber

Kirsch

Glasauer

(2013). Strategies of locomotor collision avoidance. Gait and Posture, 37(3), 385–390. https://doi.org/10.1016/j.gaitpost.2012.08.003

Bekkers

Wiepking

(2011). A literature review of empirical studies of philanthropy: Eight mechanisms that drive charitable giving. Nonprofit and Voluntary Sector Quarterly, 40(5), 924–973. https://doi.org/10.1177/0899764010380927

Chauvin

Lardjane

(2008). Decision making and strategies in an interaction situation: Collision avoidance at sea. Transportation Research Part F: Traffic Psychology and Behaviour, 11(4), 259–269. https://doi.org/10.1016/j.trf.2008.01.001

De Ceunynck

Polders

Daniels

Hermans

Brijs

Wets

(2013). Road safety differences between priority-controlled intersections and right-hand priority intersections. Transportation Research Record, 2365(1), 39–48. https://doi.org/10.3141/2365-06

Faul

Erdfelder

Buchner

Lang

A. G.

(2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–1160. https://doi.org/10.3758/BRM.41.4.1149

10.

Fiosins

Friedrich

Görmer

Mattfeld

Müller

J. P.

Tchouankem

(2016). A multiagent approach to modeling autonomic road transport support systems. Autonomic Road Transport Support Systems, 67–85. https://doi.org/10.1007/978-3-319-25808-9_5

11.

Frosch

Martinelli

Unnikrishnan

(2019). Evaluation of shared space to reduce traffic congestion. Journal of Advanced Transportation, 2019, 1–10. https://doi.org/10.1155/2019/6510396

12.

Fujii

Yoshimura

Seki

(2010). Multi-agent based traffic simulation at merging section using coordinative behavior model. CMES - Computer Modeling in Engineering and Sciences, 63(3), 265–282.

13.

Hamilton-Baillie

(2008). Towards shared space. Urban Design International, 13(2), 130–138. https://doi.org/10.1057/udi.2008.13

14.

Hidas

(2005). Modelling vehicle interactions in microscopic simulation of merging and weaving. Transportation Research Part C: Emerging Technologies, 13(1), 37–62.

15.

Hulst

M. Van Der

Rothengatter

Meijman

(1998). Strategic adaptations to lack of preview in driving. Transportation Research Part F: Traffic Psychology and Behaviour, 1(1), 59–75. https://doi.org/10.1016/s1369-8478(98)00004-7

16.

Imbsweiler

Ruesch

Weinreuter

Puente León

Deml

(2018). Cooperation behaviour of road users in t-intersections during deadlock situations. Transportation Research Part F: Traffic Psychology and Behaviour, 58, 665–677. https://doi.org/10.1016/j.trf.2018.07.006

17.

Kaparias

Bell

M. G. H.

Miri

Chan

Mount

(2012). Analysing the perceptions of pedestrians and drivers to shared space. Transportation Research Part F: Traffic Psychology and Behaviour, 15(3), 297–310. https://doi.org/10.1016/j.trf.2012.02.001

18.

Kaparias

Wang

(2020). Vehicle and pedestrian level of service in street designs with elements of shared space. Transportation Research Record, 2674(9), 1084–1096. https://doi.org/10.1177/0361198120933627

19.

Kawaguchi

Tran

A. T.

Okuda

Suzuki

(2021). Development and evaluation of driving speed controller for lane merging considering surrounding driver’s intention toward stress-free driving. SICE Journal of Control, Measurement, and System Integration, 14(1), 128–139. https://doi.org/10.1080/18824889.2021.1939627

20.

Kikuchi

Akio

(1988). Omoiyari wo kagaku suru: koushakaiteki koudou no shinri to sukiru (in Japanese). Tokyo: Kawashima Shoten.

21.

Kikuchi

Akio

(2014). Sarani omoiyari wo kagaku suru: koushakaiteki koudou to shakaiteki sukiru (in Japanese). Tokyo: Kawashima Shoten.

22.

Knorr

A. G.

Willacker

Hermsdörfer

Glasauer

Krüger

(2016). Influence of person- and situation-specific characteristics on collision avoidance behavior in human locomotion. Journal of Experimental Psychology: Human Perception and Performance, 42(9), 1332–1343. https://doi.org/10.1037/xhp0000223

23.

Lee

Y. M.

Madigan

Giles

Garach-Morcillo

Markkula

Fox

Camara

Rothmueller

Vendelbo-Larsen

S. A.

Rasmussen

P. H.

Dietrich

Nathanael

Portouli

Schieben

Merat

(2021). Road users rarely use explicit communication when interacting in today’s traffic: Implications for automated vehicles. Cognition, Technology and Work, 23(2), 367–380. https://doi.org/10.1007/s10111-020-00635-y

24.

Lütteken

Zimmermann

Bengler

K. J.

(2016). Using gamification to motivate human cooperation in a lane-change scenario. IEEE Conference on Intelligent Transportation Systems, Proceedings, Rio de Janeiro, 01-04 November 2016.

25.

Markkula

Madigan

Nathanael

Portouli

Lee

Y. M.

Dietrich

Billington

Schieben

Merat

(2020). Defining interactions: A conceptual framework for understanding interactive behaviour in human and automated road traffic. Theoretical Issues in Ergonomics Science, 21(6), 728–752. https://doi.org/10.1080/1463922x.2020.1736686

26.

Matsubayashi

Terai

Miwa

(2020). Development of a driving model that understands other drivers’ characteristics HCII. 2020(2), 29–39.

27.

Mcknight

D. H.

Carter

Thatcher

J. B.

Clay

P. F.

(2011). Trust in a specific technology: An investigation of its components and measures. ACM Transactions on Management Information Systems, 2(2), 1–25. https://doi.org/10.1145/1985347.1985353

28.

Meier

Stutzer

(2008). Is volunteering a reward in itself? Economica, 75(297), 39–59.

29.

Mellers

B. A.

Haselhuhn

M. P.

Tetlock

P. E.

Silva

J. C.

Isen

A. M.

(2010). Predicting behavior in economic games by looking through the eyes of the players. J Exp Psychol Gen, 139(4), 743-55. https://doi.org/10.1037/a0020280

20853991

30.

Moody

Melia

(2014). Shared space – research, policy and problems. Proceedings of the Institution of Civil Engineers - Transport, 167(6), 384–392.

31.

Olivier

A. H.

Marin

Crétual

Berthoz

Pettré

(2013). Collision avoidance between two walkers: Role-dependent strategies. Gait and Posture, 38(4), 751–756. https://doi.org/10.1680/tran.12.00047

32.

Riener

Zia

Ferscha

Ruiz Beltran

Minguez Rubio

J. J.

(2013). Traffic flow harmonization in expressway merging. Personal and Ubiquitous Computing, 17(3), 519–532. https://doi.org/10.1007/s00779-012-0505-6

33.

Rubaltelli

Hysenbelli

Dickert

Mayorga

Slovic

(2020). Asymmetric cost and benefit perceptions in willingness-to-donate decisions. Journal of Behavioral Decision Making, 33(3), 304–322.

34.

Sebanz

Knoblich

(2009). Prediction in joint action: What, when, and where. Topics in Cognitive Science, 1(2), 353–367. https://doi.org/10.1111/j.1756-8765.2009.01024.x

35.

Shimojo

Ninomiya

Matsubayashi

Miwa

Terai

Okuda

(2020). Decision-making in interactions between two vehicles at a highway junction HCII. 2, 104–113

36.

Stoll

Lanzer

Baumann

(2020). Situational influencing factors on understanding cooperative actions in automated driving. Transportation Research Part F: Traffic Psychology and Behaviour, 70, 223–234. https://doi.org/10.1016/j.trf.2020.03.006

37.

Stoll

Müller

Baumann

(2019). When cooperation is needed: The effect of spatial and time distance and criticality on willingness to cooperate. Cognition, Technology and Work, 21(1), 21–31.

38.

Tang

Liu

Zhang

Zou

(2018). Lane-changes prediction based on adaptive fuzzy neural network. Expert Systems with Applications, 91, 452–463. https://doi.org/10.1016/j.eswa.2017.09.025

39.

Uttley

Lee

Y. M.

Madigan

Merat

(2020). Road user interactions in a shared space setting: Priority and communication in a UK car park. Transportation Research Part F: Traffic Psychology and Behaviour, 72, 32–46. https://doi.org/10.1016/j.trf.2020.05.004

40.

Weifeng

Lizhong

Weicheng

(2003). Simulation of bi-direction pedestrian movement using a cellular automata model. Physica A: Statistical Mechanics and Its Applications, 321(3–4), 633–640.

41.

Yamagishi

(1990). Shakai teki jirenma no shikumi (in Japanese ). Tokyo: Saiensu-sha.

42.

Yamamoto

(2020). Road environmental design for last one mile in the age of population decline: Possibility and adaptation of shared space. Journal of the Sceience of Design, 28(2), 88–93. https://doi.org/10.1016/s0378-4371(02)01732-6

43.

Zimmermann

Fahrmeier

Bengler

K. J.

(2015). A roland for an oliver? Subjective perception of cooperation during conditionally automated driving. International Conference on Collaboration Technologies and Systems, Atlanta, GA, 01-05 June 2015