The design criteria, acceptance, issues and implications of computer-based intimate friends are explored through discussions and interviews and a review of the psychological literature on interpersonal relationships. The importance of close friendships to mental and physical health and personal development is discussed along with a look at the state of such relationships in our society. The ability of computer friends to address some contemporary interpersonal problems is analyzed.

Intimate friends are more important now than ever. We need to have close friends that we can talk to about our feelings and problems in order to get empathetic, non-judgmental, hearings as a means of stabilizing and guiding our frantic lives. Psychologists have demonstrated the crucial role that friends play in everything from our development of self identity to self esteem, and the important role they play in mitigating the increasing stress in our lives (Perlman and Fehr, 1987). Not only does having a close confidant make people better able to cope with marital disruption, money problems, failure, illness, or death, but children also reap the benefits when their parents have close, dependable friends to call on in a crisis (Pogrebin, 1987). At least one survey has indicated that close personal relationships are what give meaning to most people's lives (Klinger, 1977).

Unfortunately, the state of friendship in our society is in trouble. Many people, and men in particular, would say they are too busy for friends, given the increasing demands of work, commuting, consumerism, child care, second jobs, and compulsive commitments to television and physical fitness (Forbes, 1985). I personally have spent the last five years deliberately without friends because I felt that I could not afford the time commitment. Apparently I am not alone, as documented in the McGill Report on Male Intimacy (McGill, 1985):

Technology may provide a solution. In the last several years "virtual pets" such as the Tamagotchis, DogZ, CatZ, and Furbies have been introduced into the marketplace. These computational artifacts provide increasingly sophisticated simulations of animal pets which require some level of care-taking and are designed explicitly to provoke an emotional response from their users. ELIZA and its descendants (Turkle, 1997) are simulated conversationalists which respond to English sentences in a startlingly realistic manner. A natural convergence of these two technologies could provide "virtual friends" which are designed to interact with you in the same way an intimate human friend would, listening to you empathetically, providing encouraging feedback, and flattering you when necessary. Products in this niche are already starting to appear.

Regardless of the moral or ethical implications, such virtual friends will be introduced into the market as soon as their construction is technically feasible. This paper explores the notion of simulated intimacy and its design criteria, as well as some of the issues and ramifications of the use of such artifacts.

Intimate Friendship

I want to clearly differentiate between casual friendships, intimate friendships, and intimate romantic relationships. Psychologists distinguish between social loneliness stemming from a lack of friendship ties and emotional loneliness stemming from a lack of intimate relationships, and conclude that emotional loneliness is the more severe of the two forms (Weiss, 1973). Intimate romantic relationships involve additional behavioral systems that I will discuss in a later section. Thus, my initial focus will be on intimate friendships.

Most people I asked about friendship said that it had something to do with reliability (being able to depend on friends), mutual interests or a shared history or understanding of the other person. Intimate friendships, in comparison, have one additional quality that was mentioned by almost everyone I talked to--self-disclosure--the ability to talk openly about one's inner thoughts and feelings. As Hao, a graduate student with a background in Linguistics and Computer Science, put it:

This focus on self-disclosure and partner responsiveness to it (empathy) is echoed in the literature on the psychology of intimate relationships (Laurenceau, Barrett and, Pietromonaco, 1998). In fact, the word intimacy derives from intimus, the Latin term for "inner" or "inmost," and points to knowledge of the core of something, an understanding of the inmost parts (Perlman and Fehr, 1987).

Why do we need intimates?

There are both developmental and mental and physical health reasons for having friends in general and intimates in particular. Children (3 - 10) with close friends have significantly higher self-esteem than those without, and grow up to enjoy more intimate friendships later in life (Pogrebin, 1987). Intimacy problems account for the primary reason people seek out therapists for work on interpersonal relationships (Horowitz, 1979), and lack of a confidant has been shown to lead to an order-of-magnitude increase in likelihood of depression following severe life events (Brown and Harris, 1978). There is a considerable body of research which shows that married individuals enjoy better health and well-being than non-married individuals (Fehr and Perlman, 1985). In addition, some psychologists argue that the absence of close male relationships is strongly related to the significantly higher rate of suicide among men (Goldberg, 1976).

Theoretical Frameworks

There are several theoretical frameworks in psychology which deal with different aspects of friendship and intimacy. Attachment Theory, while initially developed to describe infants' feelings of attachment and loss with respect to their primary caregivers, has more recently been applied to adult relationships (Feeney and Noller, 1996). In this theory, the attachment figure serves as a "secure base" from which the individual feels safe to explore and master their environment. Individuals are categorized according to their experiences with intimacy as infants: secure individuals had parents which were available, responsive and warm, and are sociable and engage in high levels of exploration; avoidant individuals had parents who were rejecting, rigid, hostile, and averse to contact, and respond with defensiveness and avoidance of close contact; and anxious-ambivalent individuals had parents who were insensitive, intrusive, and inconsistent and respond with anxious behaviors such as crying and clinging. These categories can be useful in identifying the types of individuals who may be especially receptive to using virtual friends.

Various frameworks have also been developed to describe different aspects of intimacy (Perlman and Fehr, 1987). Some theorists maintain that the establishment of an intimate relationship is a crucial stage in the life development of an individual, and can only be achieved once a reasonable sense of self identity has been established. Other theorists approach intimacy as an enduring motivation to experience closeness, warmth and communication. Equilibrium models of intimacy account for the fact that individuals seek to achieve a fixed level of intimacy in a given interaction, and if this equilibrium is disturbed (e.g., by topic change in the conversation) they will seek to restore the overall balance of intimacy through other channels (e.g., eye contact, smiling, and physical proximity). Finally, equity theory states that an individual tries to maintain a balance between their own input (cost, investment) / outcome (perceived benefits) ratio and their partner's in an intimate relationship.

Other researchers have concluded that equity theory is more descriptive of new acquaintances (grouped under the more general category of exchange theory), while interactions between intimates is better characterized by communal relationships, in which people give in response to the other's needs, regardless of whether they are "paid back" (Aronson, Wilson, and Akert, 1994).

Finally, Maslow's Needs Hierarchy has been used to describe "addiction" to graphical chat rooms such as the Palace (Suler, 1996). The model describes human needs according to a hierarchy ranging from very fundamental biolgoical needs to higher order ones of an aesthetic and self-actualizing nature. When a person is able to satisfy needs at one level, he is then prepared to move upward to the next. In Suler's characterization, cybersex fills part of the individual's basic biological needs (the first level of the hierarchy), chat fills the need for interpersonal contact, social recognition and a sense of belonging (second level), learning how to use the environment fills the need for learning, accomplishment, and mastery (which help improve self-esteem; the third level), and the openness afforded by anonymity helps provide a channel for self-actualization (the top of the hierarchy). Many of these needs (the second and third levels in particular) may also be satisfied through interaction with a virtual friend, allowing users to spend more time on spiritual and self-actualization activities.

Friendship and Intimacy Today

As mentioned in the introduction, intimate friendship in our society seems to be in a sorry state. As Pogrebin (Pogrebin, 1987) puts it:

Even people with many acquaintances can feel lonely. As one study found (Fischer, 1982), "People with the richest, most supportive personal networks are as likely or more likely to say they 'wished they knew more people' as those who had marginal or sparse friendships."

Although the Internet has the potential to bring more people closer together than ever before, a recent study has shown a direct correlation between Internet use, social isolation and depression (Schindler, 1998). Thus it would seem, at least with current technologies and typical use, that the onslaught of the digital world is in fact leading to a decrease in our sociability. What is missing from the world-wide web and casual email exchanges may be a sense of intimacy, which virtual friends could provide in cyberspace.

The Market for Intimacy

Given our consumer-oriented society, it is no wonder that many industries have already sprung up to meet this demand for intimacy, for a price of course. Psychotherapists, phone-in psychics, and phone sex lines all provide intimacy for hire.

Therapy has become such a common avenue for intimacy, especially among women, that "Have you thought of seeing a therapist?" is a standard response whenever one admits to despair, depression, or any number of other negative or distressing emotions (Russell, 1996). "Many therapists act like a perfect friend or lover--interested, understanding, empathetic, concerned, supportive, sometimes even warm, loving and seductive." Many women remain in therapy despite major financial difficulties in meeting their basic expenses. My own experience with therapists is that I could have gained as much benefit from talking to a close friend as seeing a therapist for many of the sessions that I attended by myself (couples therapy is another matter).

Phone sex lines are a multi-billion dollar a year industry. In 1994, the four international exchange carriers collected $900 million in profits from international 900 number calls alone (Flowers, 1998). According to Flowers, "What callers are really buying is a few minutes of human contact within a society where personal contact is fraught with ambiguity." Although most men who call these lines are looking for sex talk, a significant portion call just to talk about the events and problems of their day.

Intimate Virtual friends

In addition to virtual pets and ELIZA-like conversational engines, synthetic characters and affective computing technologies may play a crucial role in the design of virtual friends. The field of synthetic computer characters has as its goal the construction of interactive, believable, lifelike computer animated characters, with identifiable personalities and emotions. Initial development and testing in this field has demonstrated that users socially and emotionally react to synthetic characters as if they were real people (Rousseau and Hayes-Roth, 1997). Compared to other kinds of simulations, synthetic characters rely on users not knowing what is in the "black box" of their design (Starr, 1994); their believability relies on users "suspending their disbelief" during interactions.

The field of affective computing (Picard, 1995) seeks to develop computing that "relates to, arises from or deliberately influences emotions." An initial project in this area is the CASPER system (Klein, 1999) which deals with user's frustrations through empathy. Although the system is not conversational (users simply selected from multiple-choice lists of responses), it significantly reduced their level of frustration in performing computer-based tasks.

What would a virtual friend look like, and how would you interact with it? I described the general concept of a synthetic character with an ELIZA-like conversational capability that also prompted the user to self-disclose and was able to empathize to several people to get their reactions. One of the first things that I asked was what it should look like. Most people thought that it should be human in form, although a few thought an animal or even disembodied speech or text would be OK. Toby said, "I don't think it should be a cartoon character, because I wouldn't take it seriously. I don't think it should be an animal either, because I don't think you could relate your intimate feelings to a little dog. It should look like a person, perhaps a relative or movie star."

Most initially thought that the gender of a virtual friend should be the same as the user. Jennifer, a graduate student with a background in Psychology and Philosophy said that she would want it to be, "female, just because that's who my best friends tend to be. I have friends who are guys, but it's not the same kind of process as with other women. I think it has a lot to do with having a better understanding about what is going on with other women."

However, upon further reflection most of the men said that for an intimate friend, they would prefer a female character. As Hao put it, " My best friends are men, but I also have some female friends that are not that close. But it's kind of strange, if you really have something to say you need to find a female listener. I don't know why. I've talked to several people and they have the same feeling." Toby also said, "Probably female. I find it easier to communicate with females, and they are usually better communicators than men in general. In my experience, its been easier to talk to a female than to a male."

The age, cultural background, and opinions of the character should also match that of the user. Research on liking and attraction has shown that we tend to like others who are most like us (Aronson, Wilson, and Akert, 1994), because they provide a source of validation for ourselves. My own feeling is that in order to understand me and my problems from my point of view, a close friend must have a similar cultural background to my generation or they just wouldn't "get it."

When asked how they would like to interact with a virtual friend most said they would actually prefer to type. As Toby said, "I think I can type my feelings easier on email than talking about it. It's like the 'Dear John' letter. Sometimes it's easier to express my feelings by typing than by talking about it." I would personally prefer text input as well, not only because I find it easier to type about difficult problems, but because with text you don't have the time pressure of a verbal conversation and because you can edit the text to get it just right.

Affect

Some recent research has been done on the detection of mood and emotional state in users, both using physical sensing (Picard, 1995) and cognitive attribution theory (Elliott, 1992). When asked if they would want their virtual friend to be able to determine their affective state, most said that they thought it would be essential. As Toby said, "If you are really upset and it doesn't read the inflection and comes back with a real positive response, that would shut you off from it."

On the other hand, most interviewees felt that while some emotional variability was necessary for believability, they would not want their character to have mood swings, especially negative ones. Jennifer said:

Experiments have demonstrated that we tend to like people better if they evaluate us positively than if they evaluate us negatively (Aronson, Wilson, and Akert, 1994). A more recent study (Reeves & Nass, 1996) was conducted in which computers flattered users, with the result that the subjects reported "more positive affect, better performance, more positive evaluations of the interaction, and more positive regard for the computer, even though subjects knew that the flattery from the computer was simply non-contingent feedback." It would seem that flattery could be used, at least initially, by a virtual friend to get users to like it.

When I asked people if they would like their virtual friend to flatter them, I received a fairly universal initial negative response. However, after thinking about it for a few minutes, many of them qualified their answers and said they wouldn't mind as long as the statements appeared to be non-repetitive. As Kenny, a graduate student in Electrical Engineering and Computer Science, said, "I don't think it's something I'd want it to do, but I wouldn't mind it doing it. I wouldn't want to know that it's doing that."

Experience

Many interviewees said that their virtual friend should have experiences of its own to talk about, and relating those would be an important part of a two-way conversation. As Jan said, "It would be important that it relate its own similar experiences, as people do when they have conversations. I don't know how you would give it experiences and background, but I think that would be important."

One way that virtual friends could be programmed with experiences is using a database of advice in the form of stories, and indexing the stories based on the user's problem statement. A system that uses this approach to construct storytelling systems for children is the SAGE project at the MIT Media Lab (Umaschi, 1997).

Other Characteristics

Everyone I talked to took it for granted that the virtual friend would be secure (i.e., no one else could find out about your conversations with it) and reliable, probably the two most-cited advantages of a virtual friend over a human one. As Jennifer put it, "Of course, it would have to be secure…a good friend doesn't give away your secrets. Unless of course, you tell it that you were going to kill yourself, and then it should tell your doctor." Another desirable characteristic mentioned was the ability to turn it off. As Kimiko, a graduate student at MIT said, "It's important that I be able to turn it off. I wouldn't want it with me all the time. Even intimate human friends have privacy and time away from each other."

Acceptance of Intimate Virtual Friends

Would people really use virtual friends? Would they be able to establish intimate relationships with them? Based on people's reactions to existing technology, the answer would seem to be yes, if the implementation was "believable." The wild success of virtual pets such as Tamagotchis (20,000 sold on the first day in the U.S., Duckett, 1997), user's acceptance of ELIZA (Turkle, 1997), and the treatment of computers as "social actors" in a general sense (Reeves and Nass, 1996) all indicate that the general population is primed for the introduction of virtual friends. A deeper reason for this acceptance can be found in Dennett's book on the "Intentional Stance," (Dennett, 1997) in which he states that humans readily ascribe rational agency to any sufficiently complex system which cannot be explained by other, simpler models.

Would users disclose personal information to a virtual friend? A study was performed in which pairs of subjects (strangers to each other) were asked to disclose information of an increasingly personal nature to each other (Aron, et al, 1997). As expected, the couples were found to have greater interpersonal closeness at the end of the exercise (as measured by standard instruments for measuring degree of intimacy) than those in a control group whose members only engaged in social small-talk. In another study, low-disclosing subjects disclosed significantly more when paired with high-disclosers (Jourard, 1971), indicating that even users who normally do not disclose much about themselves may be motivated to do so when interacting with a virtual friend which offers disclosures of its own.

After discussing the characteristics that they would want their virtual friend to have, I asked people if they thought they would really use it and if they thought they could establish an intimate relationship. Only two of the interviewees gave an immediate yes, although most said they would at least try it (as Jennifer said "I think I would use it as much as if it were a human friend, if it were realistic."). Personally, I would use a virtual friend if I was very frustrated about something and needed to vent, and I didn't feel like there was another friend available and I didn't have the time to find one. I take as a conceptual starting point the keeping of a journal or diary, or writing a letter that I never intend to send describing my problems. If I could add to that process a believable character that could determine the emotional tone of my letter and empathize, it would be a big help. Just the process of writing about one's troubles has been shown to reduce stress and improve health (Psychology Today, 1997).

The objections to using a virtual friend seemed to fall primarily into two categories--technical unfeasibility and inability to relate to an artifact. The technical issues arose almost exclusively from those with computer science backgrounds, who simply thought that it would be impossible to build any kind of program that could hold even a superficial conversation. Some of those who felt that they could not relate to an artifact seemed to think that they would actually use the system if there were a human controlling the virtual friend (e.g., Kenny said, "If you told me it was a person sitting on the other side typing, it would be OK, as long as I didn't think it was just code.").

Most people did not seem to have a problem with disclosing personal information to a virtual friend. As Toby said, "I wouldn't have any problem being intimate, knowing that it wouldn't go anywhere, knowing that no one could find out about it… I think it could have it's place."

Advantages of Intimate Virtual friends

Most of the people I discussed this idea with felt that, if it could be made to work, virtual friends could be very advantageous to society. Reliability and security were mentioned repeatedly as the primary benefits.

I discussed the idea of intimate computers with Doug, who is a family therapist in Sacramento. He felt that it was great idea for providing a sense of stability, especially for those who had problems communicating with others:

Virtual friends may also fill a useful role in the workplace where it can often be difficult to find an appropriate channel to express one's anger or frustration. As Toby said:

There is a lot of concern over the increasing isolation of individuals brought about by technology, and this was reflected in the discussions that I had with people about virtual friends. However, Doug thought that while virtual friends might reinforce an existing degree of isolation it would not cause it, and "I think it would make them less isolated in some ways."

The other frequently-mentioned problem with virtual friends was over-reliance. As Kenny put it, "People can take it too seriously. Somehow this program is going to fail, and when it breaks down it would have a big effect on the person who was using it. If they had too much emotional dependence on it and it crashed, their world would crash."

I was also interested in how people felt virtual friends might impact their human relationships. Almost everyone I talked to felt that virtual friends would not cause them to give up their human friends. However, there is some experimental evidence that intimacy is a zero-sum game, in which an increase in intimacy with a virtual friend could lead to decreased intimacy with existing confidants. In an experiment on the amount of self-disclosure before and after marriage, it was found that the total amount an individual disclosed to all acquaintances remained constant, but following marriage it was redistributed with more disclosures to the spouse (Jourard, 1971).

Some people that I talked to felt that intimate virtual friends could impact their human relationships in other ways. When asked if she could ever feel jealous about her boyfriend using a virtual friend, Jennifer responded:

What kind of people might be especially well-suited to using virtual friends? Attachment theory indicates that anxious-ambivalent individuals would be good customers, since they tend to experience more distress than others based on their low levels of self-esteem and heightened level of anxiety, and are inclined to be overly solicitous given their high need for other's approval (Feeney and Noller, 1996).

Ability to self-disclose in general would seem to be a trait indicative of likely users. A series of studies on self-disclosure (Jourard, 1971), showed that it increases with age (between 17 and 40 for opposite-sex friends), and varies with occupation (police officers disclosed significantly less than college students) and cultural background (Americans disclosed more than British or Puerto Rican subjects). These studies and more like them could be used to form a profile of likely users.

Virtual friends might also be used to ease people into therapy who are frightened of it for some reason. As Doug said, "If they are struggling with intimacy in their life and they are not able to reach out, and the thought of going to therapy seems overpowering or overwhelming. But if they could go to a computer and pull up a character, that might eventually lead to them feeling that it's OK to seek out therapy. " Virtual friends brought to you by the APA.

Crossing the Line

The discussion up to this point has focused on intimate friendships. The other common form of intimate relationship is romantic intimacy. According to one model, romance is the integration of three behavioral systems--attachment, care-giving, and sexuality (Feeney and Noller, 1996).

Care-Giving

The commercial success of the Tamagotchi toys is at least partly attributed to the care-giving relationship it fosters in its owners; they are required to take care of their virtual pet or it dies. I was interested in whether a more subtle form of care-giving could be used with virtual friends. For example, the user might be required to self-disclose (or at least interact) with the friend once a day or it becomes unhappy, or perhaps the user could plant flowers in the friend's environment to improve its mood (which die over time and must be replenished).

Most people I talked to about this did not like the idea of having to take care of a virtual friend in the same way as a Tamagotchi. Hao said it would be alright as long as it could detect if he is busy and "not jump out and interact with me" during those times. Toby felt that it would be good to start with, to draw the user into interacting with the system, and then tapering off once the relationship was established.

Doug also exercised caution, especially with troubled adults:

Aside from the obvious pornographic possibilities, there are more subtle, sensual design dimensions that could be explored with virtual friends. One possibility is to realize the friend not as a desktop PC application, but as a tangible artifact which responds to voice and touch for its interface, designed along the lines described in (Hofmeester, Kemp and Blankendaal, 1996). In addition, a tangible object could be not only portable, but potentially wearable (MIT Wearable), allowing a user to carry their virtual friend with them wherever they go, and providing immediate comfort anytime, anywhere.

Conclusions and Implications

The creation of virtual friends appears to be technically feasible, and based on the current market for virtual pets will likely be accepted and used by at least some portion of society. The real open question is if the depth of the intimate relationships that are formed will be enough to fulfill users' needs for intimacy to the degree that they will reap the mental and physical benefits afforded by such relationships with other people.

There are many cultural implications of virtual friends. Such artifacts could very easily be used to control people in explicit or subtle ways, by governments or commercial institutions. What better way to influence someone than to have their best friend constantly whisper suggestions to them? Governments have been felled this way.

There is a continuing concern over such artifacts leading to further isolation of the individuals who use them. However, virtual friends could actually aid in bringing people together by functioning as social "transitional objects" (Winnicott, 1971) which serve to acclimate individuals to intimacy who have avoided it thus far in their lives. On the other hand, isolation may not be such a bad thing. As mentioned earlier, virtual friends could free some women ("relationship junkies") from their need for constant approval from others. They may even free humankind to spend more time in reflective solitude, free from the demands and unpredictability that other people can place on us (Koch, 1994) and help relieve the stress that some tend to put on their marriages as the only intimate relationship that they have.

The use of virtual friends may even change our notions of religion and spirituality. As Hannes, a graduate student at the Media Lab, put it, "The idea of having an intimate friend that is always with you reminds me of the role that religious or spiritual figures play in some people's lives. They have an ongoing, internal dialog with these figures over issues in their lives." This is also supported by research into attachment theory accounts of religious behavior (Feeney and Noler, 1996):

As Pogrebin says in a discussion of the role of friendship in an individual's development (Pogrebin, 1987), "It's not too glib to say that as we travel across the life span, we make friends and friends make us." When our friends are objects of our own construction, we find this sentiment echoed in the words of Csikszentmihalyi (Csikszentmihalyi, 1981), "Man is…the maker and user of objects, his self to a large extent a reflection of things with which he interacts." It would seem that the construction of our future intimates is an enterprise that should be undertaken with great care.

Bibliography

1. Aron, A., Melinat, E., Aron, E., Vallone, R., and Bator, R., "The experimental generation of interpersonal closeness: a procedure and some preliminary findings," Personality and Social Psychology Bulletin v. 2,3 April 1997, pp. 363-377.
2. Aronson, E., Wilson, T., and Akert, R., Social Psychology: The Heart and the Mind, Harper Collins, New York, 1994, pp. 370-415.
3. Brown, G.W., and Harris, T., Social origins of depression: A study of psychiatric disorder in women, Free Press, New York, 1978 as quoted in Perlman, D., and Fehr, B., "The Development of Intimate Relationships," in Perlman, D., and Duck, S., eds., Intimate Relationships, Sage Publications, Newbury Park, CA, 1987, pp 13-42.
4. Csikszentmihalyi, M., The Meaning of Things, Cambridge University Press, 1981.
5. Dennet, D.C., The Intentional Stance, MIT Press, 1987.
6. Duckett, J., "Toy Consumers Getting Latest Tickle Out of 'Virtual Pets'," Morning Call, June 23^rd, 1997.
7. Elliott, C.D., The Affective Reasoner: A process model of emotions in a multi-agent system, Ph.D. Dissertation, Northwestern University, June, 1992.
8. Feeney, J., and Noller, P., Adult Attachment, SAGE Publications, Thousand Oaks, CA, 1996.
9. Fehr, B., and Perlman, D., "The family as a social network and support system," Handbook of family psychology and therapy, Vol. 1, Dow-Jones Irwin, Homewood, IL, 1985, pp. 323-356.
10. Fischer, C.S., To Dwell Among Friends: Personal Networks in Town and City, University of Chicago Press, 1982, as quoted in Pogrebin, L.C., Among Friends, McGraw-Hill, New York, 1987.
11. Forbes, January 14, 1985, p.12.
12. Flowers, A., The Fantasy Factory: An Insider's View of the Phone Sex Industry, University of Pennsylvania Press, Philadelphia, 1998.
13. Goldberg, H., The Hazards of Being Male: Surviving the myth of masculine privilege, Nash, New York, 1976.
14. Hofmeester, G., Kemp, J., and Blankendaal, A., "Sensuality in Product Design: a Structured Approach," in Proceedings of CHI96.
15. Horowitz, L., "On the cognitive structure of interpersonal problems treated in psychotherapy," J. of Consulting and Clinical Psychology, 47, 1979, pp. 5-15 as quoted in Perlman, D., and Fehr, B., "The Development of Intimate Relationships," in Perlman, D., and Duck, S., eds., Intimate Relationships, Sage Publications, Newbury Park, CA, 1987, pp 13-42.
16. Jourard, S.M., Self-Disclosure: An Experimental Analysis of the Transparent Self, Wiley-Interscience, New York, 1971.
17. Klein, J. T., Computer Response to User Frustration, Master's Thesis, MIT, February, 1999.
18. Klinger, E., Meaning and void: Inner experience and the incentives in people's lives, Minneapolis, University of Minnesota Press, 1977.
19. Koch, P., Solitude: A Philosophical Encounter, Open Court, Chicago, 1994.
20. Laurenceau JP, Barrett LF, Pietromonaco PR, "Intimacy as an interpersonal process: The importance of self-disclosure, partner disclosure, and perceived partner responsiveness in interpersonal exchanges," JOURNAL OF PERSONALITY AND SOCIAL PSYCHOLOGY, 74: (5) 1238-1251 MAY 1998.
21. McGill, M.E. , The McGill Report on Male Intimacy, Holt, Rinehart & Winston, 1985, as quoted in Pogrebin, L.C., Among Friends, McGraw-Hill, New York, 1987.
22. MIT Wearable Computing Web Page (http://wearables.www.media.mit.edu/projects/wearables/).
23. New York Times, "Psychology Revising Its View of Women," March 20, 1984.
24. Perlman, D., and Fehr, B., "The Development of Intimate Relationships," in Perlman, D., and Duck, S., eds., Intimate Relationships, Sage Publications, Newbury Park, CA, 1987, pp 13-42.
25. Piorkowski, G. K., "Back Off: Our overemphasis on intimacy," Psychology Today, v. 28 (Jan/Feb), 1995, pp. 50-53.
26. Picard, R., Affective Computing, Media Laboratory VisMod Technical Report 321, 1995.
27. Pogrebin, L.C., Among Friends, McGraw-Hill, New York, 1987.
28. "Can You Write Your Way to Good Health?," Psychology Today v. 30, Jan./Feb. 1997, pg. 20.
29. Reeves, B. and Nass, C., The Media Equation, Cambridge University Press, Cambridge, 1996.
30. Rousseau, D., and Hayes-Roth, B., Interacting with Personality-Rich Characters, Stanford University Knowledge Systems Laboratory Technical Report KSL-97-06, September 1997.
31. Russell, D.E., "Intimacy for Sale," WOMEN & THERAPY, 19: (4) 15-25 1996.
32. Schindler, P.E., "Study: Home Internet Use Breeds Depression," TechWeb, August 31^st, 1998 (http://www.techweb.com:3030/internet/story/NIN19980831S0001).
33. Starr, P., "Seductions of Sim," American Prospect, no. 17, 1994, pp. 19-29.
34. Suler, J., Why is This Thing Eating My Life: Computer and Cyberspace Addiction at the "Palace", (http://www1.rider.edu/~suler/psycyber/eatlife.html).
35. Turkle, S., The Second Self, Simon and Schuster, 1984.
36. Turkle, S., Life on the Screen, Simon and Schuster, 1995.
37. Umaschi, M. "Soft Toys with Computer Hearts: Building Personal Storytelling Environments" Published in CHI'97 Proceedings, ACM, 1997, pp. 20-21.
38. Weiss, R.S., Loneliness: The experience of emotional and social isolation, MIT Press, Cambridge, 1973 as quoted in Pogrebin, L.C., Among Friends, McGraw-Hill, New York, 1987.
39. Winnicott, D.W., "Transitional Objects and Transitional Phenomena," Playing and Reality, Basic Books, 1971, pp. 1-20.
40. Yankelovich, D., New Rules: Searching for Self-Fulfillment in a World Turned Upside Down, Random House, 1981 as quoted in Pogrebin, L.C., Among Friends, McGraw-Hill, New York, 1987.

Interview with Ron, November 26^th, 1998, 15 minutes.

Interview with Hao, November 29^th, 20 minutes.

Interview with Petra, November 29^th, 15 minutes.

Interview with Jennifer, November 29^th, 25 minutes.

Interview with Doug, December 2^nd, 20 minutes.

Interview with Toby, December 3^rd, 20 minutes.

Interview with Kenny, December 3^rd, 20 minutes.

Interview with Kimiko, December 4^th, 20 minutes.

Interview with Jan, December 4^th, 20 minutes.

Interview with Hannes, December 4^th, 15 minutes.