Simoland is a Virtual Social Environment (VSU) which is inhabited by several Simos. These Simos are autonomous agents, pursuing their own goals in a (more or less) intelligent way. Their "psychological apparatus" is modeled after the Zurich Model of Social Motivation, a system theoretic model of social motives and distance regulation.

One of the Simos can be controlled by the participant. The research question now is: How does he or she act and react to events in the virtual environment?

Watch the video (it's a shortened montage of the separation scene)

Investigating close relationships

Having a close relationship is amongst the highest rated goals in our society (Deutsche Shell Holding, 2006), and intimate relationships have numerous benefits for health and well-being (Prager, 1995). For these and many other reasons, the investigation of close relationships is a fruitful and exciting field in psychological science. However, research in this area has some specific challenges. The American Psychological Association’s "Decade of Behavior" comes to an end, and several scholars emphasized the importance of behavioral observations, in contrast to relying exclusively on self-report measures (Baumeister, Vohs, & Funder, 2007; Furr, 2009). Especially in the field of relationships and social interactions, researchers have a strong distrust in self-report measures and argue that the observation of interactional processes is the key for understanding relationship outcomes (Gottman, 1998). As behavioral observations are costly, labor-intensive, and frustrating, many researchers are reluctant to undertake this effort (Gottman, 1998). Beyond that, it is hard or impossible to create certain situations with real couples. For example, most studies on jealousy have been conducted with written vignettes or questionnaires. These vignettes, however, lack the social significance and social consequences of real situations, and the validity of self-reported hypothetical reactions to these hypothetical situations can be questioned (Furr, 2009). Especially with such an emotionally laden topic, observations of actual behavior would be particularly informative. However, it is hard to create a situation of serious jealousy with real couples in the laboratory - neither romantic partners, nor ethic committees will approve such an approach. In this and other situations, virtual environments might be useful tools for the generation and observation of behavior.

A New Approach For the Study of Close Relationships: Virtual Social Environments

Virtual environments (VEs) are advocated as a powerful tool to bridge the gap between increased realism and dynamics of social interactions, and the increased controllability of laboratory experiments (Asendorpf, 2004; Blascovich et al., 2002). VEs have the potential to overcome many of the restrictions expressed above, as they are supposed to have several advantages in contrast to conventional methods like self-report measures, laboratory studies, or interviews. First, both data collection and data analyses can be more easily accomplished: Testing can be done over the internet, and due to automatic coding, reliable behavioral indices can be obtained without the effort of coding hours of video material. Second, the researcher has full control over the actions and reactions of the virtual interaction partners, which opens up new possibilities of experimentally varying the partner's behavior. Tying in with the jealousy example from above, special scenarios can be created that are hard or impossible to create in the laboratory with real persons. Third, in contrast to situations described in vignettes, interpersonal situations in VSEs are anchored in the experiential system (McClelland, Koestner, & Weinberger, 1989; Schultheiss, 2001). For example, situation-contingent facial expressions of the agents, background music, or triggered events can be used to increase the feeling of immersion of the participant and to elicit spontaneous and automatic reactions. And in contrast to hypothetical self-reported reactions, in VSEs spontaneous behaviors as well as the dynamics of behavior over time can be investigated (see Chapter 3).
These features of VEs look very promising. Concerning the social interactivity with other inhabitants of these environments, however, the majority of existing research in VEs so far only has covered relatively simple scenarios. Examples are the measurement of interpersonal physical distance towards a static agent that does not interact at all (Bailenson, Blascovich, Beall, & Loomis, 2003; Dotsch & Wigboldus, 2008), or giving a talk in front of some agents, which either show friendly or hostile reactions, regardless of participants' actual performance. Only one study is known to the authors, which investigated close relationships in virtual worlds. Frey, Blunk, and Banse (2006) sent couples into the same virtual world (each partner sat in another room at a PC and their avatars met in the virtual world). In this study, actually human-human interaction of a couple was studied, mediated through the VE.
To my knowledge, no study so far has been conducted where ongoing human behavior towards an interactive agent is embedded in a rich social context. Hence, for the purpose of this dissertation, the following framework is proposed: A virtual social environment1 (VSE) is a virtual environment that is populated with autonomous agents. These agents show a sufficiently believable behavior, such that relatively natural interactions can take place. In the implementation used in this dissertation, participants can control one of these agents (called the protagonist) that has a virtual romantic relationship to another agent in the VSE (the virtual spouse). Participants can instruct the protagonist to perform more than 30 different actions with the spouse (for details, see Chapter 3), and the spouse reacts to them according to an underlying psychological model (for details, see Chapter 2). Several indices derived from gaming behavior serve as dependent variables: behavioral indices (i.e., which actions were chosen to be performed?), the physical distance between the protagonist and other agents in the VE, as well as short experience sampling questionnaires displayed during the game.
Skeptics may state that behavior in such VEs is completely arbitrary and has nothing to do with "real life". In the remainder of this dissertation it will be theoretically argued and empirically shown that, quite contrary to that apprehension, under appropriate conditions virtual behavior indeed reflects qualities of real life. Nonetheless, if close relationships are investigated in VSEs, a key assumption is that characteristics of the real life relationship or generalized internal working models of close relationships are transferred to the virtual world.


Transference towards humans is a concept with a long-standing tradition in psychotherapy. It describes the phenomenon that characteristics of past relationships to significant others are transferred to unknown people. While in the psychoanalytic tradition transference describes a pathological process, the concept has been re-conceptualized in social-cognitive terms and has been described as a normal phenomenon that happens every day (Saribay & Andersen, 2007). Transference is a well documented phenomenon. For example, Andersen and colleagues found that features of unknown target persons are inferred from features of a significant other, and that evaluations of the target are influenced by transference, both in explicit evaluations and in facial expressions (Andersen & Thorpe, 2009). Transference effects also influence expectancies for acceptance or rejection in a new relationship, triggers specific interpersonal behavior of the participants (Berk & Andersen, 2000), and activates specific motivations and goals (Berk & Andersen, 2008).
Comparably, attachment theory emphasizes the idea of transferring internal working models acquired in past relationships, especially in early childhood, to new relationships (Ainsworth, Blehar, Waters, & Wall, 1978; Bowlby, 1980; Fraley, 2007). While there is some debate about the specificity of these working models (Baldwin & Fehr, 1995; Fraley, 2007; La Guardia, Ryan, Couchman, & Deci, 2000; Pierce & Lydon, 2001; Sibley & Overall, 2008), empirical results generally support the effect of transference (e.g., Brumbaugh & Fraley, 2006, 2007; Mallinckrodt & Chen, 2004).
In this dissertation I propose two extensions to the framework of transference. First, just like towards other humans, transference can also take place towards a virtual agent. If the aim is to investigate internal working models of a person through transference, these agents might even be preferable as targets of transference, as they present a blank slate which is not distorted by unique characteristics of a real target person. Second, not only characteristics of past relationships are transferred to the virtual spouse, but also own emotions and behavioral tendencies are transferred to the protagonist, who performs these behaviors as a proxy person of the participant.
If behavior assessed in virtual environments should be of diagnostic value for real life behavior, it has to be shown that indeed behavioral programs, reaction norms, internal working models, motives, and other psychological constructs are transferred to relationships and interactions of the virtual world. Two manuscripts of this dissertation (Chapter 3 and 4) are the first studies that demonstrate that transference in fact happens in VSEs.

The Construction of a Virtual World

Requirements for a VSE

Due to the high effort of testing participants in the laboratory, studies in immersive virtual environments typically have samples sizes between 10 and 30 for each experimental condition, or even overall (e.g., Bailenson et al., 2003; Dotsch & Wigboldus, 2008; Klinger et al., 2005; Slater et al., 2006; Tichon & Banks, 2006). For the study of interindividual differences, however, sample sizes of 100 and more participants are desirable to have sufficient statistical power. To achieve such sample sizes, the first requirement for Simoland was that the studies could be run over the internet. Technical hurdles should be kept as low as possible, thus a secondary goal was that the game should run in any browser without the requirement to install additional plug-ins or software, which would repel many participants. Second, as I sought for a broad sample including older participants, the controls and handling of the game should be as easy as possible. Finally, I also ruled out the use of collaborators who control the other agents, as this would make studies both more expensive, and less controllable. Instead of that, I wanted to work with autonomous agents with whom the protagonist interacts.

The Implementation of Simoland

Most studies in virtual environments drew upon existing technical frameworks, for example sophisticated VEs with head-mounted displays (e.g., McCall, Blascovich, Young, & Persky, 2009), or existing computer games like Quake III (Frey et al., 2006; Frey, Hartig, Ketzel, Zinkernagel, & Moosbrugger, 2007). While both hard- and software of the first solution are very expensive (in sum 30'000 € - 50'000 €), the second solution can be used for free on standard PCs. Both solutions, however, do not meet my requirements stated above. Hence, both for economical reasons and for design considerations, I decided to implement my own VSE - an online computer game called "Simoland" which was populated with autonomous agents called "Simos". After several pre-test studies and having written more than 28'000 lines of code, Simoland was stable and ready to run (for a screen shot, see Figure 1 in Chapter 3).
Simoland was implemented as a two-dimensional game using the Adobe Flash technology (version 9). With that approach the game runs in any internet browser, the only requirement being an installed Flash Player plug-in (which applies to over 98% of internet users in Europe and the US, Adobe Systems Inc., 2009). The file size of the final program was 2.7 Mb, which means a loading time of some seconds with a regular broadband connection. After several pre-tests, a graphical user interface was developed that was sufficiently easy to handle. In a pilot study with 241 participants, over 84% agreed or strongly agreed that the handling of the game was easy, 12% were undecided, and only 4% disagreed. These percentages were comparable in the subsample of older participants (>40 years).

Autonomous Agents: The Behavior of the Simos

All agents in Simoland are autonomous agents. That means, they have several needs and they have a representation of the virtual world along with the behavioral programs to satisfy these needs. The Simos have five motivational systems: hunger, thirst, sleep, affiliation, and security. The first three of these are basic needs that have to be satisfied on a regular basis. The need for affiliation is implemented as the need to do leisure activities with moderately familiar others, and the need for security is implemented as the need to have contact with highly familiar others. Beyond these motivational systems, the agents' behavior generally followed a tit-for-tat rule, that is, they reciprocated the positivity of each interaction that was initiated.
All motivational systems are implemented as feedback control systems which continuously compare an actual value with an internal set point (Bischof, 1993). Any deviation from this set point leads to an activation to reduce this discrepancy. In the implementation of the psychological model of motivation, several challenges had to be overcome, like the problem of action selection (i.e., if multiple, incompatible motivations are present, which gains control over the behavioral system?), or a sensible balancing of the parameters of these control systems. Details on these issues can be found in Chapter 2.
The protagonist is autonomous as well, although its behavioral program is restricted compared to that of the other agents. That means, when the participant does not give any commands to the protagonist for some time, it will show basic behaviors like searching for food and water, taking a nap, or listening to a MP3 player. In contrast to the other agents, however, the protagonist does not initiate social interactions (although it responds to interactions initiated by other agents). The security system of the virtual spouse is adjusted such that it seeks the closeness of the protagonist every 2-3 minutes. In the case of the security appetence, the mere closeness to a familiar other usually is sufficient to reduce the activation (Bischof, 2001). In cases of a very high activation of the security system, however, the spouse initiates an interaction to the protagonist (e.g., wants to talk about what happened today, or wants to kiss the protagonist).

The Virtual Spouse: An Avatar of the Real Life Partner?

One major decision had to be taken on how to visually design the virtual spouse. Generally, two approaches can be taken, each with its own appeal. On the one hand, one could try to model the real life partner as realistic as possible. This promises to instigate partner-specific scripts and reaction tendencies that reflect the current relationship. On the other hand, one could aim for a decrease of ego-involvement, explicit attitudes, and self-enhancement, to enable a measurement of rather implicit working models. In this case, one would take advantage of the mechanism of transference where internal working models automatically are applied onto unknown persons.
In a first attempt, I pursued the first alternative and tried to maximize the resemblance of the virtual spouse to the real life partner. For this purpose, a facial avatar editor was created with which participants could model an avatar (i.e., representation) of their real life partner. The avatar had the same first name as the real life partner and participants could adjust the hair color, hair-do, the shape, position, and color of mouth, nose, ears, eyes, etc. (see Figure 1). This avatar then was used as the virtual spouse in a second step of the study.

Pasted Graphic

Although participants had some fun modeling their real life partner, a pretest revealed that they expressed enduring concerns about the accurateness of the virtual avatar and got stuck in ruminations about inconsistent details. This finding is consistent with the hypothesis of the "uncanny valley" (Mori, 1982), which states that in some region short of 100% realism, users are jolted by some minor inconsistencies, which completely destroy the illusion of realism. While in several scenarios increased realism indeed might have an impact on believability and immersion, theoretical considerations and the experiences from my pretest led me to the conclusion that in the special case of a virtual spouse, “less is more”.
In addition, I did not define the virtual spouse as an avatar (i.e., a representation) of the real life partner, just as I did not define the protagonist as a representation of the participant. The protagonist was rather introduced as "the Simo you can give commands to", and the virtual spouse was introduced as "the romantic partner of the protagonist". Hence, participants were not instructed to act out their current real life relationship - at least not explicitly. Support for this approach can be drawn from the assessment of implicit motives by means of picture story exercises (Schultheiss & Pang, 2007) like the Thematic Apperception Test (TAT). Although stories written to these pictures usually are written about other persons from a third person’s view, valid inferences about the implicit motives of the writer can be drawn. In some coding systems, stories written from the first person’s perspective even are discarded (Fivush, 2006; Waters & Waters, 2006).

Besides the resemblance to the real partner, another decision has to be taken concerning the visual fidelity of the virtual spouse. While many other applications of VEs strive for photo realistic visual displays, I argue for the contrary: a symbolic depiction. As already discussed, too many details can lead to disturbances and prevent a successful immersion. A symbolic depiction of the virtual partner, in contrast, decreases feelings of strangeness and inconsistency, and rather provides a blank slate where behavioral and emotional habits can be transferred to (for a detailed discussion, see Chapters 3 and 4). Empirical results give support for this approach. For example, in an extension of Heider and Simmel's (1944) classic study, Barrett, Todd, Miller, and Blythe (2005) could show that motion trajectories from simple geometrical shapes could be correctly classified as activities like chasing, mating, or fighting; and a ball tossing game with very crude stick men elicited strong feelings of exclusion ("cyber-ostracism", Williams, Cheung, & Choi, 2000). Likewise, Sanchez-Vives and Slater (2005) concluded in a review of virtual reality research that "[s]urprisingly, however, there is strong evidence that people respond to relatively crude virtual humans as if they were real people." (p. 335).
To summarize, symbolic virtual agents can be seen as a unique combination of social realism and the necessary indefiniteness which fosters the mechanism of transference. On the one hand, they provide the necessary behavioral reactions and social consequences which are lacking in pure vignettes. On the other hand, their symbolic depiction provides a rather blank slate onto which mental representations can be applied, in contrast to real persons who might distort the process due to their own characteristics.


Adobe Systems Inc. (2009). Adobe Flash Player version penetration. Retrieved from
Ainsworth, M., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Hillsdale: Lawrence Erlbaum Associates.
Andersen, S. M., & Thorpe, J. S. (2009). An IF–THEN theory of personality: Significant others and the relational self. Journal of Research in Personality, 43, 163-170.
Asendorpf, J. B. (2007). Psychologie der Persönlichkeit [Psychology of personality]. 4. Auflage. Berlin: Springer.
Bailenson, J. N., Blascovich, J., Beall, A. C., & Loomis, J. M. (2003). Interpersonal distance in immersive virtual environments. Personality and Social Psychology Bulletin, 29, 819-833.
Baldwin, M. W., & Fehr, B. (1995). On the instability of attachment style ratings. Personal Relationships, 2, 247-261.
Barrett, H. C., Todd, P. M., Miller, G. F., & Blythe, P. W. (2005). Accurate judgments of intention from motion cues alone: A cross-cultural study. Evolution and Human Behavior, 26, 313-331. doi:10.1016/j.evolhumbehav.2004.08.015
Baumeister, R. F., Vohs, K. D., & Funder, D. C. (2007). Psychology as the science of self-reports and finger movements: Whatever happened to actual behavior. Perspectives on Psychological Science, 2, 396-403.
Berk, M., & Andersen, S. M. (2000). The impact of past relationships on interpersonal behavior: Behavioral confirmation in the social-cognitive process of transference. Journal of Personality and Social Psychology, 79, 546-562.
Berk, M., & Andersen, S. M. (2008). The sting of lack of affection: Chronic goal dissatisfaction in transference. Self and Identity, 7, 393-412.
Bischof, N. (1993). Untersuchungen zur Systemanalyse der sozialen Motivation I: Die Regulation der sozialen Distanz - Von der Feldtheorie zur Systemtheorie [On the regulation of social distance - from field theory to systems theory]. Zeitschrift für Psychologie, 201, 5-43.
Bischof, N. (2001). Das Rätsel Ödipus. Die biologischen Wurzeln des Urkonflikts von Intimität und Autonomie. [The riddle of Oedipus. The biological roots of the core conflict between intimacy and autonomy]. München: Piper.
Blascovich, J., Loomis, J., Beall, A. C., Swinth, K. R., Hoyt, C. L., & Bailenson, J. N. (2002). Immersive virtual environment technology as a methodological tool for social psychology. Psychological Inquiry, 13, 103-124.
Bowlby, J. (1980). Attachment and loss. New York: Basic Books.
Brumbaugh, C. C., & Fraley, R. C. (2006). Transference and attachment: How do attachment patterns get carried forward from one relationship to the next? Personality and Social Psychology Bulletin, 32, 552-560.
Brumbaugh, C. C., & Fraley, R. C. (2007). Transference of attachment patterns: How important relationships influence feelings toward novel people. Personal Relationships, 14, 513-530.
Deutsche Shell Holding. (2006). Jugend 2006. 15. Shell Jugendstudie. Frankfurt am Main: Fischer.
Dotsch, R., & Wigboldus, D. H. (2008). Virtual prejudice. Journal of Experimental Social Psychology, 44, 1194-1198. doi:10.1016/j.jesp.2008.03.003
Fivush, R. (2006). Scripting attachment: Generalized event representations and internal working models. Attachment & Human Development, 8, 283-289.
Fraley, R. C. (2007). A connectionist approach to the organization and continuity of working models of attachment. Journal of Personality, 75, 1157-1180.
Fraley, R. C., & Shaver, P. R. (2000). Adult romantic attachment: Theoretical developments, emerging controversies, and unanswered questions. Review of General Psychology, 4, 132-154.
Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysis of self-report measures of adult attachment. Journal of Personality and Social Psychology, 78, 350-365.
Frey, A., Blunk, H. A., & Banse, R. (2006). Psi-Land: Paarinteraktionsforschung in einer virtuellen Umgebung [Psi-Land: Investigating interactive behaviour in romantic couples in a virtual environment.]. Zeitschrift für Sozialpsychologie, 37, 151-159.
Frey, A., Hartig, J., Ketzel, A., Zinkernagel, A., & Moosbrugger, H. (2007). The use of virtual environments based on a modification of the computer game Quake III Arena (R) in psychological experimenting. Computers in Human Behavior, 23, 2026-2039.
Furr, R. M. (2009). Personality psychology as a truely behavioral science. European Journal of Personality, 23, 369-401.
Gottman, J. M. (1998). Psychology and the study of marital processes. Annual Review of Psychology, 49, 169-197.
Gottman, J. M., & Levenson, R. W. (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health. Journal of Personality and Social Psychology, 63, 221-233.
Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. American Journal of Psychology, 57, 243-259.
Klinger, E., Bouchard, S., Legeron, P., Roy, S., Lauer, F., Chemin, I., & Nugues, P. (2005). Virtual reality therapy versus cognitive behavior therapy for social phobia: A preliminary controlled study. CyberPsychology & Behavior, 8, 76-88.
La Guardia, J. G., Ryan, R. M., Couchman, C. E., & Deci, E. L. (2000). Within-person variation in security of attachment: A self-determination theory perspective on attachment, need fulfillment, and well-being. Journal of Personality and Social Psychology, 79, 367-384.
Mallinckrodt, B., & Chen, E. C. (2004). Attachment and interpersonal impact perceptions of group members: A social relations model analysis of transference. Psychotherapy Research, 14, 210-230.
McCall, C., Blascovich, J., Young, A., & Persky, S. (2009). Proxemic behaviors as predictors of aggression towards Black (but not White) males in an immersive virtual environment. Social Influence, 4, 138-154.
McClelland, D. C., Koestner, R., & Weinberger, J. (1989). How do self-attributed and implicit motives differ? Psychological Review, 96, 690-702.
Mikulincer, M., & Shaver, P. R. (2005). Attachment theory and emotions in close relationships: Exploring the attachment-related dynamics of emotional reactions to relational events. Personal Relationships, 12, 149-168.
Mori, M. (1982). The Buddha in the Robot. Charles E. Tuttle Co.
Pierce, T., & Lydon, J. E. (2001). Global and specific relational models in the experience of social interactions. Journal of Personality and Social Psychology, 80, 613-631.
Prager, K. J. (1995). The psychology of intimacy. New York: Guilford Press.
Sanchez-Vives, M. V., & Slater, M. (2005). From presence to consciousness through virtual reality. Nature Reviews Neuroscience, 6, 332-339.
Saribay, S. A., & Andersen, S. M. (2007). Are past relationships at the heart of attachment dynamics? What love has to do with it. Psychological Inquiry, 18, 183-191.
Schneider, M. (2001). Systems theory of motivational development. In N. J. Smelser & P. B. Baltes (Eds.), International Encyclopedia of the Social & Behavioral Sciences (pp. 10120-10125). Oxford: Elsevier.
Schultheiss, O. C. (2001). An information processing account of implicit motive arousal. In M. L. Maehr & P. Pintrich (Eds.), Advances in motivation and achievement (Vol. 12: New directions in measures and methods) (pp. 1-41). Greenwich: JAI Press.
Schultheiss, O. C., & Pang, J. S. (2007). Handbook of research methods in personality psychology. In R. W. Robins, R. C. Fraley, & R. F. Krueger (Eds.), Measuring implicit motives (pp. 322-344). Guilford Press.
Sibley, C. G., & Overall, N. C. (2008). Modeling the hierarchical structure of attachment representations: A test of domain differentiation. Personality and Individual Differences, 44, 238-249.
Slater, M., Guger, C., Edlinger, G., Leeb, R., Pfurtscheller, G., Antley, A., Garau, M., et al. (2006). Analysis of physiological responses to a social situation in an immersive virtual environment. Presence, 15, 553-569.
Tichon, J., & Banks, J. (2006). Virtual reality exposure therapy: 150-Degree screen to desktop PC. CyberPsychology & Behavior, 9, 480-488.
Waters, H. S., & Waters, E. (2006). The attachment working models concept: Among other things, we build script-like representations of secure base experiences. Attachment and Human Development, 8, 185-197.
Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). Cyberostracism: Effects of being ignored over the internet. Journal of Personality and Social Psychology, 79, 748-762.
© 2018 Felix Schönbrodt | Impressum | Datenschutz | Google Analytics deaktivieren | Contact