The hyperpersonal model of CMC (Walther, 1996) proposes a set of concurrent theoretically based processes to explain how CMC may facilitate impressions and relationships online that exceed the desirability and intimacy that occur in parallel off-line interactions. The model follows four common components of the communication process to address how CMC may affect cognitive and communication processes relating to message construction and reception: (1) effects due to receiver processes, (2) effects among message senders, (3) attributes of the channel, and (4) feedback effects. The model has received a great deal of attention in the literature. At the same time, extensions and revisions to the model have been proposed on the basis of both conceptual and empirical contributions. Certain aspects of the model remain underresearched – such as the holistic integrity of its subcomponents as well as the reciprocal effects of feedback?although some progress has been made with respect to these issues.

Receivers. When receiving messages from others in CMC, an individual may tend to exaggerate perceptions of the message sender. In the absence of the physical and other cues that face-to-face encounters provide, rather than fail to form an impression, receivers fill in the blanks with regard to missing information. This often takes the form of idealization if the initial clues about another person are favorable. The original articulation of the model drew explicitly on SIDE theory (Lea & Spears, 1992) in formulating receiver dynamics. The SIDE model also describes how CMC users make overattributions of similarity when communicating under conditions of visual anonymity if contextual cues suggest that a conversational partner shares some salient social identity with the receiver. It further proposes that communicators experience heightened attraction in these circumstances. The SIDE model argues that the specific form of attraction is focused on one's attachment to the group identity rather than to the individual person.

Recent rearticulations of the hyperpersonal model, however, have attempted to broaden the concepts related to receiver dynamics (see Walther, 2006). The hyperpersonal approach now suggests that an initial impression may be activated not only by group identifications but through individual stereotypes, such as personality characteristics, or due to the vague resemblance of an online partner to a previously known individual (see Jacobson, 1999). Analysis of online impressions using social relations analysis (Kenny, 1994), which assesses how uniform or differentiated one's impressions of other group members are, offers a promising approach to the question of group- or interpersonally based impressions in CMC (see Markey & Wells, 2002).

Senders. Text-based CMC facilitates selective selfpresentation. Online, one may transmit only cues that an individual desires others to have. It need not be apparent to others what one's physical characteristics are (unless one discloses them verbally), nor do individuals generally transmit unconscious undesirable interaction behaviors such as interruptions, mismanaged eye contact, or nonverbal disfluencies of the kind that detract from desired impressions face-to-face. Instead, CMC senders may construct messages that portray themselves in preferential ways, emphasizing desirable characteristics and communicating in a manner that invites preferential reactions. Selfdisclosure quite naturally plays a role in this process, by which individuals not only disclose what content they wish to be known but also, through disclosure, breed intimacy. Research has found that disclosure and personal questions constitute greater proportions of utterances in online discussions among strangers than they do in comparable face-to-face discussion (Joinson, 2001; Tidwell & Walther, 2002). This may be a simple adaptation to the lack of nonverbal expressive behavior, which would normally provide uncertainty-reducing information. Yet CMC users' disclosures are more intimate than those of face-to-face counterparts, suggesting a strategic aspect to this difference as well.

Apart from explicit disclosures, much of what senders selectively self-present is conveyed through the content of the exchanges in terms of how communicators express their evaluations of various subjects, their agreement with partners, word choice, and any number of ordinary expressions of affinity. A recent study (Walther, Van Der Heide, Tong, Carr, & Atkin, 2010) asked one member of an online dyad, who was about to discuss the topic of hamburgers with an online partner, to behave online in a way that prompted the other person to like or to dislike the individual. The significant differences in liking for the actor following the CMC conversation were associated with the actor's level of agreements versus disagreements and concurrence versus divergence in statements about the other partner's favorite hamburger. Online (and perhaps elsewhere), we manipulate our desirability to others not so much by overt statements of interpersonal affect but through the way we complement or contest others' views of things in the world. In other research, systematic differences among individuals' construction of stories about themselves online led to changes in their self-perceptions. Gonzales and Hancock (2008) asked participants to write about their experiences in a manner that would lead others to perceive them as either extraverted or introverted. Half of the participants in the experiment posted these responses in a blog, presumably accessible to other CMC users, whereas the other half of the participants recorded their answers in a private document for ostensible analysis at a later time, anonymously. The blog writers generated significantly different self-perceived extraversion/introversion scores following the experience, in accordance with the characteristic they had been assigned. Gonzales and Hancock concluded that selective selfpresentation online provides a potent influence not only on others but also on the transformation of an individual's self, a phenomenon they called “identity shift.”

Channel. The third dimension of the hyperpersonal model involves characteristics of the channel and how CMC as a medium contributes to the deliberate construction of favorable online messages. One part of the channel factor focuses on the mechanics of the CMC interface, suggesting that users exploit the ability to take time to contemplate and construct messages mindfully. In many CMC applications (especially asynchronous systems), users may take some time to create optimally desirable messages without interfering with conversational flow, very much unlike the effects of face-to-face response latencies. The hyperpersonal model further suggests that CMC users capitalize on the ability to edit, delete, and rewrite messages to make them reflect intended effects before sending them. The introduction of the model further suggested that CMC users may redirect cognitive resources into enhancing one's messages, without the need to pay attention to the physical behaviors of one's conversational partner or oneself, or to the ambient elements where one is physically located when communicating (in contrast to these demands on attention in face-to-face conversations). CMC users can focus their attention on message construction to a greater extent than they would in face-to-face conversations.

Recent research supported a number of these suggestions (Walther, 2007). A study led college student participants to believe that they were joining an asynchronous discussion with a prestigious professor, who was described in much detail; with a relatively undesirable high school student in another state, also described in detail; or with another college student, about whom no details were provided except for the student's name. Participants' message composition was recorded in real time and later coded and rated, and a different group of participants provided ratings of how desirable each type of target would be as an interaction partner. Results of the study revealed that the more desirable the partner was, the more editing (deletions, backspaces, and insertions) the participants exercised in composing their messages to that partner. The degree of editing corresponded to the degree of relational affection ascribed to the messages by raters. Participants self-reported their level of mindfulness during message production, which had been expected to differ based on the attractiveness of the ostensible message target. It did not, and neither did the time they spent composing their messages differ as a result of the different types of targets. However, those who were more mindful spent more of their time editing the messages they had written, whereas those who were lower in mindfulness spent more time choosing what to write. These results add a level of verification to the model's contention that CMC users exploit the unique mechanical features of the medium to enhance relational qualities of their messages.

Another facet of the channel component of the hyperpersonal model has been more difficult to interpret, and research results have challenged the model's original assertions about asynchronous versus synchronous CMC. The model originally posited that asynchronous CMC allowed users to avoid the problems of entrainment associated with face-to-face meetings. Entrainment, in the small group communication literature (Kelly & McGrath, 1985), refers to the ability to synchronize attention and interaction with collaborators. It is proposed to be difficult to accomplish when participants have competing demands on their time and attention. Time pressures work against entrainment in face-to-face meetings, leading communicators to neglect group maintenance behaviors in favor of impersonal, task-related discussions. Since CMC users working asynchronously can interact with others at times that are convenient and available to them, the model suggested that CMC should not suffer from a lack of maintenance behavior. CMC users would be more likely to engage in off-task, interpersonal discussions than in face-to-face meetings since, without meeting in real time, there is no time pressure constraining such exchanges.

This aspect of the model was challenged very quickly. Roberts, Smith, and Pollock's (1996) ethnographic observations and interviews reflected that individuals who enter real-time, multiplayer online games and chat systems (as opposed to asynchronous discussions) very rapidly exhibit sociable exchanges. Likewise, Pena and Hancock (2006) demonstrated that the conversations in a real-time multiparty sword-fighting game reflected more socio-emotional utterances than game-related statements even during online duels. The sociability benefits originally ascribed to asynchronous CMC in the introduction of the model are fairly clearly an aspect of many synchronous systems as well, at least those in which socializing is a goal that users bring to the system. A recent review of communication that takes place in certain online, real-time, role-playing games describes a great proportion and a wide variety of interpersonal communication behaviors among associates and fellow “clan” members (Klimmt & Hartmann, 2008). Although these findings suggest greater scope for the development of hyperpersonal dynamics, the entrainment explanation has not been tested since the model was developed, and the conceptual and empirical status of this aspect of the channel component of the model is unclear.

Feedback. The hyperpersonal model of CMC suggested that the enhancements provided by idealization, selective self-presentation, and channel effects reciprocally influenced matters, forming a feedback system by which the CMC intensified and magnified the dynamics that each component of the model contributes. That is, when a receiver gets a selectively self-presented message and idealizes its source, that individual may respond in a way that reciprocates and reinforces the partially modified personae, reproducing, enhancing, and potentially exaggerating them. The manner in which the dynamics of these reciprocated expectations may modify participants' character was suggested to reflect the process of behavioral confirmation.

Behavioral confirmation (Snyder, Tanke, & Berscheid, 1977) describes how one interaction partner's impression about a target partner leads the first partner to behave and how that behavior alters the responses of the target partner in return. The original behavioral confirmation study involved male participants who were shown photos priming them to believe that their upcoming female telephone interaction partners were physically attractive or unattractive (even though the actual partners were not really those depicted in the photos but were randomly selected female participants). Not only did this expectation affect the males' involvement, it affected the females' personality-related responses as well, as revealed in outside raters' evaluations of the females' personalities based on audio recordings of their conversations. The hyperpersonal model appropriated this construct, suggesting that one's idealized impressions of an online partner may lead a CMC user to reciprocate based on that impression, transmitting messages that, in turn, may shape the partner's responses, shifting the target's personality in the direction of the communicators' mutually constructed and enacted impression. In this way, feedback may intensify the hyperpersonal effects of idealization, selective self-presentation, and channel exploitation.

The feedback component of the hyperpersonal model has received little formal research attention until recently. One study (Walther, Liang, et al., 2011) examined whether feedback to a CMC communicator enhanced the identity shift phenomenon described by Gonzales and Hancock (2008; see above). As Gonzales and Hancock had done, this experiment called on half the participants to answer several questions as if they were extraverted and the other half, as if introverted. Participants posted their responses to a blog or pasted them into a Web-based form. Departing from Gonzales and Hancock, in each condition, participants either did or did not receive feedback confirming their (extraverted or introverted) personality performances. When participants subsequently completed self-report measures of their extraversion/introversion, those who received feedback expressed more extreme scores in the direction of the initial prompting. This study also helps establish a link between two components of the hyperpersonal model?selective selfpresentation and feedback?showing that the activation of these components jointly produces stronger effects than in isolation.

Several CMC studies have generated findings consistent with a behavioral disconfirmation effect (see Ickes, Patterson, Rajecki, & Tanford, 1982; Burgoon & Le Poire, 1993). Behavioral disconfirmation takes place when one individual anticipates an unpleasant interaction with a target person and, to avert the unpleasantness, overaccommodates in order to improve the person's demeanor. One was the Walther (2007) study described above, in which participants anticipated online communication with a high school? age loner, a college student, or a professor. Despite pretest indications that the high schoolers were the least desired communication partners, male participants who believed that they were communicating with a male high schooler expressed greater editing and affection than with a male peer or professor. No voice-based or faceto-face comparisons were done in that study.

As discussed earlier, two recent studies explored the effects of preinteraction expectancies on subsequent impressions following CMC or voice-based communication (Epley & Kruger, 2005; Walther, DeAndrea, & Tong, 2010). Manipulations in both studies instilled preinteraction expectancies among interviewers regarding their partners' high or low intelligence. Manipulations in both studies involved the bogus presentation of one of two sets of a partner's ostensible photograph, grade point average, major, and self-reported greatest high school achievement. In Epley and Kruger's (2005) research, half the interviewers used a phonelike system to speak to a real interviewee, and half the interviewers used CMC to obtain responses that were transcribed from a person other than the actual interviewee. The results superficially appear to reflect greater behavioral confirmation in CMC than on the phone: Interviewers' posttest assessments of interviewees' intelligence were different in CMC but not in voice conditions. The methodology in that study, however, was such that the CMC interviewer could not actually have influenced his or her partner's behavior. Walther, DeAndrea, and Tong's (2010) replication involved actual interviewees in both voice and CMC conditions. The post-CMC ratings indicated relatively greater intelligence assessments than did those following the voice-based interviews, reflecting behavioral disconfirmation in CMC relative to voice. Further research is exploring the reasons for these voice versus CMC differences in confirmation and disconfirmation.

Extensions. In addition to research that has added, supported, or challenged the hyperpersonal model's claims, a variety of extensions to the model have been made, and it has been applied to new social technologies as well.

Research exploring the dynamics of online date-finding systems has applied aspects of the hyperpersonal model in several ways. Many of these systems require users to create profiles that feature photos and self-descriptions. Ellison, Heino, and Gibbs's (2006) interviews with online daters revealed that users make overattributions from minimal cues that prospective dates exhibit. These include gross inferences based on spelling errors and projections about individuals' character on the basis of what time of day or night he or she initiates a date request. Gibbs, Ellison, and Heino (2006) also drew on selective self-presentation principles in their documentation of the dilemmas faced by daters when honest self-presentations produce fewer dates than do self-aggrandizing or deceptive self-presentations (see also Whitty, 2008).

Research on deceptive self-presentation in online dating profiles has made particular use of the hyperpersonal model. Innovatively acquired data demonstrate that most online daters misrepresent their age, weight, and/or height online (Toma et al., 2008; see also Hall, Park, Song, & Cody, 2010). In several cases, these findings have been attributed to CMC's facility for selective self-presentation and editing under asynchronous communication conditions (Toma et al., 2008). This hyperpersonal perspective has most recently been applied to the manner in which dating system users select or retouch the photographs they post to their electronic profiles (Hancock & Toma, 2009).

Additional work has added new explanatory extensions to the model. Jiang, Bazarova, and Hancock (2011) developed a framework for understanding the exceptional impact of selfdisclosure on intimacy in CMC compared with face-to-face communication. Although individuals disclose proportionately more, and more intimately, in CMC than in face-to-face communication (Tidwell & Walther, 2002), questions remained over whether receivers (over) interpret disclosures in a way that increases intimacy in CMC more intensively than in off-line interactions. Jiang et al. (2011) hypothesized that the degree to which receiving disclosure from a conversational partner affects intimacy is shaped by the attributions a receiver makes for the partner's motivation to disclose. A 2 × 2 experiment included CMC chat versus face-to-face interactions between a naive participant and a confederate who offered several personal disclosures in one condition and no disclosures in a control condition. Posttest measures revealed that the CMC participants receiving disclosures experienced greater intimacy than did face-to-face participants. Among those who were exposed to a greater degree of disclosure, the CMC participants more frequently perceived that the discloser's behavior was motivated by some aspect of their relationship rather than by the medium or the discloser's disposition, compared with the face-to-face participants. The type of attribution fully mediated the relationship between the disclosure-bymedium interaction and intimacy. In addition to documenting a hyperpersonal effect of disclosure on intimacy, this study provided a new attributional mechanism to explain the effect, which is also affected by the medium.

A self-attribution dynamic may also be operating online that leads to exaggerated intimacy as a result of online self-disclosure, a hypothesis that has not appeared in the literature previously. Although it is commonly understood that when another person discloses to us, we experience intimacy with that person, Collins and Miller's (1994) meta-analysis of the relationship between disclosure and liking demonstrates an alternative connection as well: When we disclose to another person, our own disclosure increases our feelings of intimacy toward the recipient. Thus, when users naturally adapt to the absence of nonverbal cues in CMC by disclosing proportionately more than they do in face-to-face interaction (Joinson, 2001; Tidwell & Walther, 2002), it may be due to their own expression of relatively greater disclosure (in addition to or instead of the reception of others' disclosures) that they attribute greater intimacy to disclosive CMC conversations. Although this contention warrants empirical verification, it suggests an interesting contribution to the hyperpersonal cycle.

Another form of self-perception affecting intimacy can be hypothesized on the basis of findings that it takes several times longer to have a conversation online than exchanging the same amount of verbal content in a face-to-face meeting (see Tidwell & Walther, 2002). If CMC chatters have an online conversation that feels as though it should only have taken an hour but turns out to have taken four hours, and if the communication rate differential is not apparent to CMC interactants (as it is apparently unapparent to online game players; Rau, Peng, & Yang, 2006), this temporal distortion may also lead to exaggerated inferences about the desirability of the online partner. When time seems to pass more quickly than it actually does, people attribute enjoyment to the events that occurred during that time (Sackett, Nelson, Meyvis, Converse, & Sackett, 2009).

Other researchers have also examined the role of disclosures in the development of relatively more intimate relations online and their effects. Valkenburg and Peter (2009) identify three relationships among four specific processes that explain how CMC may be related to improvements in adolescents' well-being. For reasons that have appeared in the literature (see above; for a review Kim & Dindia, 2011; see also Schouten, Valkenburg, & Peter, 2007), the first important relationship in the model is the effect of CMC in promoting online self-disclosure. Drawing on extensive literature, Valkenburg and Peter (2009) proceed to connect self-disclosure with the development of higher quality relationships among people. Finally, the authors point out the connection between high-quality relationships and development of psychological well-being. The first two linkages in particular implicate CMC as a catalyst in the relationally-based development of adolescent adjustment.

In contrast to Valkenburg and Peter's depiction of the beneficial effects of CMC to wellbeing, another application of the hyperpersonal model is seen in Caplan's (2003) approach to the study of problematic Internet use. Caplan focuses on the usage and consequences of CMC by individuals who have social skill deficits in their face-to-face communication abilities and who experience disruptive communication-related anxieties. To such people, Caplan has shown that Internet interaction is especially appealing, particularly real-time discussion systems. Because CMC provides individuals greater control over their messages and their self-presentation, it reduces anxiety (see also Amichai-Hamburger, 2007). Under these conditions, individuals may develop what Caplan (2005) refers to as a preference for online social interaction, “characterized by beliefs that one is safer, more efficacious, more confident, and more comfortable with online interpersonal interactions and relationships than with traditional (face-to-face) social activities” (p. 723). This use of CMC is paradoxical and problematic, according to Caplan's research, because such individuals experience a decline in their off-line social skills in conjunction with their more socially rewarding online interactions.