The Impact of Imagination on Computer Mediated Telepresence

Emil Coman (coman@uconn.edu)

Christian Rauh (chris@rauh.net)

University of Connecticut

Storrs, CT, USA

Paper presented in the Presence panel of the Information Systems Division

International Communication Association

May 2003 Annual Conference, San Diego, CA

ABSTRACT

Imagined realities, interactions, and actions seem to be a recurrent theme in CMC contexts that induce a strong sense of tele-presence. This study is designed to assess the impact that mental imagery has on the sensation of presence experienced in CMC. An experiment was conducted to check whether visual mental imagery causes tele-presence. 178 participants were divided into high, medium, low imaginers and asked to chat in dyads of same group, then reported on presence and emotional and rational involvement. Individual ability to generate visual images is not a causal factor for self-reported presence. Chatting via the computer is a highly involving experience, and is perceived to be more rational than emotional, and the sense of tele-presence in chatting experiences does not show a reliable departure dimension, giving support only to the arrival dimension of presence.

 FULL TEXT

This study looked into the possible causal relationship between metal imagery and the sensation of presence in computer-mediated communication. The main argument asserts that since the concept of presence involves a particular kind a perceptual experience, considered less ”real” than an un-mediated experiences, we should evaluate it in relation to individual ability to mentally escape and “create” reality, which is mental imagery generation. We should expect then people with higher imagining abilities to report a higher sense of presence. First a short introduction of the concepts of presence, mental imagery and qualia will be presented, then the affect-reason approach on involvement will be presented, our hypotheses will follow, and finally the methods, results and conclusions of our experiment will be detailed here.

Presence

Presence has become a reputed construct in social scientific research lately, and recent reviews have been devoted entirely to the explication of the very concept (e.g. Lombard, 1997). The International Society for Presence Research defines tele-presence as a psychological state or subjective perception partly or totally generated by and/or filtered through human-made technology, in which part or all of the individual’s perception fails to accurately acknowledge the role of technology in the experience. Different dimensions of presence have been documented and studied, among the major ones spatial and sensory presence, social realism, engagement, and social presence. We chose to investigate in this study the social presence aspect of it, with some emphasis on psychological immersion (involvement, engagement). Involvement here is viewed as the directing of one’s perception towards the world created by the technology, and away from the surrounding “real” sensory world.

All the dimensions of presence specified by Lombard (1997) seem to contain a perceptual component, most of them pointing toward a distancing from actual reality. The perception of sociable medium, the feeling of being there (or of being here, or of being together), the perception of mediated identities as social actors, all involve a certain degree of imaginative engagement on the part of the individual who communicates through computer (we "imagine" us being there, somebody being there, etc, based on limited information). Several psychological characteristics of the actual consumer (user) of a medium have been explored also, among them some relevant indicators of susceptibility to immersion, like absorption, creative imagination, empathy, or cognitive style (Sas & O’Hare, 2003). Sas & O”Hare reported (2003) significant correlation between presence and creative imagination and empathy, as well as a positive correlation with cognitive type (favoring feeling type rather than thinking), and positive but not significant correlation with absorption.

Mental Imagery

The French anthropologist Gilbert Durand argued that “imaginary is the ‘implicate’ order through which all understanding necessarily passes, and even all explanations of individual or collective human behavior as well” (Durand, 1993, p. 17). In the same tone, Feshbach stated that "an impressive amount of daily cognitive activity is fantasy in nature" (1976, p. 71). As for the overall role of imagination in cognitive activities, Kant (1933) for instance even argued that imagination is a necessary ingredient of perception process itself. He distinguished imagination (the power of synthesis) from both sensibility and understanding, and treated it as a separate faculty or `subjective source of knowledge' (in Pendlebury, 1996). The representation of objects in the form of mental images is based upon the same cognitive mechanisms that are involved in the perception of those objects. Finke (1989) characterized this idea as the principle of functional (or perceptual) equivalence.

Mental imagery (sometimes colloquially called visualization, or "seeing in the mind’s eye") is experience that resembles perceptual experience, but which occurs in the absence of the appropriate stimuli for the relevant perception (cf. Finke, 1989; McKellar, 1957). Very often these experiences are understood by their subjects as echoes or reconstructions of actual perceptual experiences from their past; at other times they may seem to anticipate possible, often desired or feared, future experiences. In general mental imagery has been conceptualized according to Richardson (1999) in four different ways. Firstly, imagery is a personal or phenomenal experience essentially, secondly a "private" or "subjective" experience, similar to what Buck (1999) calls emotion III – subjective experience of feelings and desires (affects), then, imagery was seen as a stimulus attribute, and the focus can be set upon imagery also as a process that is under strategic control.

The imagining process is not usually a conscious one, as it has been considered by researchers distinguishing between knowledge by acquaintance versus knowledge by description, or experiential versus rational cognitive systems (Epstein & Pacini, 1999, Buck 1999). In psychological terminology, such subjective non-conscious or pre-conscious experiences are also called qualia (Buck, 1993). Michael Persinger e.g. (2001) has studied such subjective experiences in the absence of artificial technology, dominated by a sensed presence (not an actual, real one) that appears to involve the acquisition of information from distances beyond those normally obtained by classical senses, and imply distortions in physical time (para-normal experiences, p. 515). Mental images have an impact on one’s information processing because it resembles at times regular perceptual processes, and moreover it is involved in memory storage and retrieval, as well as in emotional experiences (Lang, 1979). Lang actually considered the image an active response process itself, an emotional experience altogether, rather than a stimulus in the head to which we respond.

Some elements that determine presence have been studied in the context of communication-evoked mental imagery. In general, induced imagery (or imagination, or fantasy, or daydreaming) gets combined with the actual mediated message in producing the final mental representation or triggering behaviors. An old model of communication actually builds in daydreaming in the perception phase of the communicative behavior, inasmuch as imagination becomes part of the perceiving process itself (Palmgreen, 1971).

Our main assumption takes into account some of the features of computer-mediated experiences, contrasted usually against the “real” un-mediated ones, like lack of nonverbal cues and reduced richness of communication. The immediate implication is that what may contribute to keeping up appearances or reality in a mediated interaction is the mind-generated perceptual-like representations, and hence individual differences in ability to imagine should translate into different accounts of the degree of presence. We proposed therefore specifically that:

H1: Higher visual imagery ability causes a higher sense of tele-presence in computer-mediated interactions.

            The second area of investigation in our study was the emotional and rational components of the involvement (engagement) process experienced in computer-mediated interactions. We drew most of the theoretical reasoning from Buck’s developmental-interactionist theory (1999). The theory provides a foundation for the conceptualization of emotion in general and for subjectively experienced affect in particular. Buck distinguished three main types of emotion, Emotion I (arousal), serving the functions of adaptation and homeostasis, Emotion II (expression) serving the function of social coordination through expressive displays, and Emotion III (experience) serving the function of self-regulation through subjectively experienced feelings and desires, or affects. In this conceptualization, affect constitutes a kind of knowledge, syncretic cognition corresponding to the concept of knowledge by acquaintance, proposed by  Betrand Russel’s (1912, “presentational immediacy of experience”) and William James’ (1890).

            Both affect and reason are held to be cognitions, reason being more sequential and analytic, while affect more syncretic and holistic. Chaudhuri & Buck (1995) defined hence the level of involvement as “depth and quality of both syncretic and analytic cognitive responding”. The ARI measure of involvement includes therefore both an affective and a rational component, and is defined as the average of the two: ARI= (A+)R/2. The relative influences of affect and reason are captured by the A/R ration, which together with the ARI level define the ARI three dimensional solid. Based on this conceptualization and on the engagement characteristic of tele-presence, we hypothesized that:

H2: The ARI involvement measure will correlate with self-report of presence.

and asked the following Research Question:

RQ1: Which of the two types of cognitions are likely to dominate in the computer mediated chat interaction?

            We decided also to check the dimensionality of Kim & Biocca (1997) presence scale, the relation between the "not being here" and "being there" (departure and arrival) dimensions, and explore their relationship with the imagery variable. If these dimensions tap into different aspects of the sense of presence, we would expect to find stronger correlations between the “being there” part and visual imagery, because the act of generating visual images is an emotional experience rather than a response to a stimulus generated by the mind (imagining a scene resembles more actually “being” in that scene, than departing from this real one). Therefore we hypothesize that:

H3: Visual imagining ability will correlate better with the arrival dimension of tele-presence, than with the departure one.

Method & Measures

The design of the study was an experiment with a manipulation of conditions based on the visual imagery ability. Participants were first administered the Vividness of Visual Imagery Questionnaire-VVIQ test (Mark, 1973), a 15 item self-report questionnaire asking to rate the vividness of the visual images from 0-perfectly vivid to 5-no image at all. The test has two versions, one asking participants to visualize with their eyes open, the other with eyes closed; no difference was found for the eyes condition in an extensive reviewed by McKelvie (1995), so we decided to use the closed eyes format. Based on 14 studies reporting means for poor and good visualizers, McKelvie (1995) computed confidence intervals which we used to discriminate between conditions (high visualizers 0 to 1.95, medium 1.96 to 2.49, poor 2.5 through 5). The scale has recorded Kronbach alphas between .870 and .906, and delayed test-retest reliabilities between .072 and .782 (McKelvie, 1995, p. 27-28).

            To assess self-reported tele-presence, we used the 9-item social presence scale built largely by Champness (1973), reported in Short, Williams, & Christie (1976, p. 66-72). The other presence scale used was Kim & Biocca’s (1997). I addition, we controlled the manipulation of visual imagery with a post-test of the visual part (5 items) of Betts’ Questionnaire upon Mental Imagery (QMI, 1909). To assess the emotional versus rational nature of the experience, we used two items asking directly about the degree to which participants engaged in feeling or thinking during the interaction, following suggestions taken from Buck et al (1995).

Design

172 participants (mean age=20), filled out the pretest on VVIQ. Participants were first divided into three groups, high, medium, and low imaginers, and then asked (after more than a week) to interact over the computer with another person. We assigned them to interaction dyads on same-level imagining ability, and asked them to chat about anything they want for 20 min., without revealing their identities. We decided to set up a chat environment online, with a lobby room where the researchers assigned participants to individual chat rooms based on their pre-test scores, instead of bringing participants to our lab, partly to better replicate the real chatting environment and avoid problems of external validity, and partly due to convenience. We instructed them to find a quiet and without distractions area with a PC connected to Internet and complete the experiment there. After interaction, they filled out the post-test questionnaires online.

Results

            Valid data on the initial pre-test on visual imagery (Mark’s VVIQ) has been collected online from 172 participants, and 104 of them completed the experiment and reported on their sense of presence, but 6 participants were discarded because they filled out the post-test without chatting (two of them chatted High-Low); based on an initial outlier analysis (see Figure 1) we could exclude 3 outliers, numbers 23, 44, and 86, but we decided not to. Of the 98 participants, 42 were males and 56 females (mean age 20.24), and the breakdown in terms of visual imagery levels was: high imaginers 39, low imaginers 26, and middle values 33. The manipulation was successful, the three groups differed on the visual part of QMI (the manipulation control test) significantly, F=10.681, p=0.000, df=97, and also significant differences were found between all 3 conditions in terms of the manipulation check measure, QMI (all p=0.000, the QMI-visual part and VVIQ are very similar).

            Due to the experimental nature of our study, we could rely only on a correlational analysis to test the influence of visual imagining on presence. The correlation between visual imagery and social presence was not significant overall, r=0.159, p=0.118 (effect size Hedges-corrected d=0.32), but it was significant for the medium and high imagining groups merged, r=0.327 , p<0.01 (effect size Hedges-corrected d=.685). A one-way ANOVA showed that hypothesis 1 was not supported, F=1.783 and p=0.174, df=97, proving that there is not an overall significant difference in self-reports of presence between the three groups of imaginers. A post-hoc multiple comparison Bonferroni test revealed that the only difference that approached somewhat significance was actually between the medium and high imaginer groups, p=0.097.

            The departure dimension of Kim & Biocca’s presence scale was different between the three imagining groups, F=3.509, p=0.034, which is due to the middle-high groups difference (post-hoc p=0.028). A direct t test for the difference between the medium and high imaginers in social presence showed almost a significant (p= 0.063, corresponding to an effect size Hedges-corrected d=0.443) difference opposite the expected direction, t=-1.894. When looking separately at females and males, only for females there are some variables that are different by imagining group, Kim & Biocca’s departure dimension, p=0.057 (for presence p= -0.130), which comes mostly from the difference between the high and medium groups, p=0.054. A quadratic regression model has been fitted for the regression of social presence on imagery, and the fit almost reached significance, F=2.51, p=0.086. When considering the medium and high groups together, the same fit became significant, F=4.32, p=0.0171 (see Fig. 7).

            Only one significant gender difference was found in terms of the pre- and post-experiment measures, for AR ratio (emotional/rational), amazingly significantly higher for men (.96) than women (.78), based on t-test for difference of means, t=-2.413, p=0.018, none other approaching significance (effect size Hedges-corrected d=0.489). Males’ AR ratio was 0.95, while females’ 0.80, therefore even though as an emotion/ration ratio the chat experience was perceived as more rational, males have seen it as more emotional than females. Age had no impact on any of the measures (but the sample was range restricted to the 19 to 21 age bracket).

            Hypothesis 2 was supported, the ARI involvement measure correlated with the social presence scale (r=0.419, p=0.000, see Table 1, effect size Hedges-corrected d=0.915), the Kim & Biocca scale (r=0.212, p=0.039, effect size Hedges-corrected d=0.43), and the arrival dimension of Kim&Biocca’s (r=0.279, p=0.006, effect size Hedges-corrected d=0.576), but not with the departure dimension (r=-0.1, p=0.335, effect size Hedges-corrected d=0.2). The two dimensions actually correlated negatively, so whether they measure the same concept is questionable. It is likely that the departure dimension does not tap into characteristics of the tele-presence measure, and could be dropped entirely as a separate dimension of it. If we look at how “arrival” scores relate to the social presence scale (r=0.835 correlation), we could argue that the arrival part of Kim&Biocca’s scale can be used to assess presence without the departure one.

            To respond to our research question we looked at the mean of AR ratio, which is 0.86, showing that overall the chat experience mediated by computer is viewed more as a rational activity, than emotional one (values over 1 would indicate a stronger emotional component). The last hypothesis seems to be weakly confirmed, sine the correlation of VVIQ with the arrival dimension is -0.003, while that with the departure one -0.162 (both are non-significant, however, a test of significance for the difference would be appropriate).

            An unexpected result has been obtained for the difference in the ratings of presence between the High and Medium groups of visual imaginers, an almost significant difference (means of 3.94 for High and 3.39 for Medium, p=0.063, t=-1.984, effect size Hedges-corrected d=0.443). A significant difference of the opposite direction has been noted with respect to the departure dimension of the Kim&Biocca scale (p=0.009, t=2.683, effect size Hedges-corrected d=0.63).

Discussion and implications

            Based on the results recorded thus far, individual ability to generate visual images is a  causal factor for self-reported presence, but their relationship seem to be curviliniar. Chatting via the computer is a highly involving experience, and is perceived to be more rational than emotional, and the sense of tele-presence in chatting experiences does not show a reliable departure dimension, giving support only to the arrival dimension posited by Kim&Biocca (1997).

            The discriminating power of the VVIQ test as a measure of individual visual imaging ability needs to be reassessed, in our opinion, because the expected differences between the more extreme groups (high and low) have not been found, but significant differences have been seen between the medium and high groups. It is possible also that the imaginal ability may exert a nonlinear evolution from no image to vivid portrayal, such that one people start forming images the effects of these mental images may be increasingly important the closest we get to the higher extreme of the spectrum. The interaction task may have contributed also to the pattern of results, so that for instance a task solving design instead of a chat interaction may help disentangle the interpersonal dynamics from the tele-presence realm.

            More investigations are required to account for other measures of susceptibility to immersion in technology-mediated experiences. Broader and more reliable measures of individual imaginal abilities should be employed, as it was proposed for instance by Kwekkeboom (2000), who devised the Imaginal Ability Questionnaire by combining items from the Absorption, VVIQ and QMI scales. Other designs are also suggested here, with combinations of conditions (in addition to same condition dyads), other computer-mediated experiences (say browsing highly involving Internet content).

References

Betts, G.H. (1909). The distribution and functions of mental imagery (Contributions to Education, No. 26). New York: Columbia University, Teachers College.

Buck, R. (1993). What is this thing called subjective experience? Reflections on the neuropsychology of qualia. Neuropsychology, 7 (4), pp. 490-499.

Buck, R. (1999).  The biological affects: A typology.  Psychological Review, 106(2), 301-328.

Buck, R., Chaudhuri, A., Georgson, M. & Kowta, S. (1995). Conceptualizing and operationalizing affect, reason, and involvement in persuasion: The ARI model and the CASC scale. Advances in Consumer Research, 22, 440-447.

Champness, B. G. (1973). The assessment of the user reactions to confravision: Analysis and conclusions. Unpublished Communications Studies Group paper no. E/73250/CH.

Durand Gilbert. (1993). The Implication of the Imaginary And Societies, Sociologie-Contemporaine, 41(2), p. 17-32.

Epstein, S. Pacini, R. (1999). Some basic issues regarding dual process theories from the perspective of cognitive-experiential self-theory. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology. New York: Guilford. 1999.

Feshbach, Seymour. (1976). The Role of Fantasy in the Response to Television. The Journal of Social Issues,1976, 32, 4, fall, 71-85. (yes KSU)

Finke, R.A. (1989). Principles of Mental Imagery. Cambridge, MA: MIT Press.

Kant, I. (1933). Critique of pure reason (trans. N. Kemp Smith). London: Macmillan.

Kim, T. & Biocca, F. (1997). Telepresence via Television: Two Dimensions of Telepresence May Have Different Connections to Memory and Persuasion. Journal of Computer Mediated Communication, 3 (2).

Kwekkeboom, K. L. (2000). Measuring imaging ability: Psychometric testing of the Imaging Ability Questionnaire. Research in Nursing and Health, 23 (4), 01-309.

International Society for Presence Research (2000). The Concept of Presence: Explication Statement. Retrieved April 1, 2003 from: http://www.temple.edu/ispr/explicat.htm

James, W. (1890). The principles of psychology. New York: Henry Holt.

Lang, P. J. (1979). A bio-informational theory of emotional imagery. Psychophysiology, 16 (6), 495-512.

Lombard, M. (1997). At the heart of all: The concept of presence. Journal of Computer Mediated Communication, 3 (2).

Marks, D. F. (1973). Visual imagery differences in the recall of pictures. British Journal of Phylosophy, 64, 17-24.

McKellar, P. (1957). Imagination and Thinking. London: Cohen & West.

McKelvie, S.J. (1995). The VVIQ as a psychometric test of individual differences in visual imagery vividness: A critical quantitative review and plea for direction. Journal of Mental Imagery, 19(3 & 4), 1-106.

Palmgreen, P. (1971). A daydream model of communication. Association for Education in Journalism, Lexington, KY.

Pendlebury, M. (1996). The role of imagination in perception. South African Journal of Philosophy, 15, (4), 133-139.

Persinger, M. A. (2001). The neuropsychiatry of paranormal experiences. Journal of Neuropsychiatry and Clinical Neuroscience, 13, pp. 515-524.

Richardson, John T.E. (1999). Imagery. Psychology Press Ltd, Brunel University, UK

Russel, B. (1912). Problems of Philosophy. New York: Simon & Shuster.

Sas, C. & O’Hare, G.M.P. (2003). Impact of Cognitive Style upon Sense of Presence, Proceedings of 10th International Conference on Human Computer Interaction HCII’2003. August 22nd-27th 2003, Crete, Greece, Lawrence Erlbaum Associates, Publishers.

Sheehan, P.W. (1967). A shortened form of the Betts' Questionnaire Upon Mental Imagery. Journal of Clinical Psychology, 23, pp. 386-389.

Short, J., Williams, E., & Christie, B. (1976). The Social Psychology of Telecommunications. London: John Wiley & Sons, Ltd.  

Appendix A

Table 1. Bivariate correlations between measures

- * -Correlations significant at the 0.05 level, or ** - at the 0.01 level  (2-tailed).

- Correlations for clustered and individual groups: MH=Medium+High, LM=Low+Medium, L=Low, H=High

- Correlations approaching significance:, a - p=.079 , & b -  p=.064, c – p=.052

- On the diagonal – Kronbach’s alpha reliabilities / N, where N is the number of valid cases.

Table 2. Mean values per gender and level of visual imagery

- * = Significant gender difference, p<.05;

- # = Gender difference  for p=.069.

- @ = Significant difference between visual imagining groups, p=0.009.

- $ = Group difference approaching significance, p=0.55. ^$

Figure 1: Box plot for imagery-presence values per imagery groups and genders.

Figures 2-6: Mean values for the relevant variables, for each imagery level, and overall.

Fig. 7: The Quadratic fit of the imagery-presence relationship