Effect of gazing at the camera during a video link on recall

Abstract

The impact of looking into the camera during a presentation over a video link (resulting in the perception of mutual gaze) on information recall was investigated. In a face-to-face context mutual gaze has been shown to facilitate the encoding and subsequent recall of information [Fry, R., Smith, G.F., 1975. The effects of feedback and eye contact on performance of a digit-coding task. J. Soc. Psychol. 96, 145–146; Otteson, J.D., Otteson, C.R., 1980. Effect of teacher's gaze on children's story recall. Percept. Motor Skill. 50, 35–42; Sherwood, J.V., 1988. Facilitative effects of gaze upon learning. Percept. Motor Skill. 64 (3 Part 2), 1275–1278]. One explanation for these findings is that gaze acts as an arousal stimulus, which increases attentional focus and therefore enhances memory [Kelley, D.H., Gorham, J., 1988. Effects of immediacy on recall of information. Commun. Edu. 37(3), 198–207]. Two studies were conducted in order to test whether gazing at the camera during video-mediated presentations resulted in similar benefits as mutual gaze in a face-to-face context. In study 1 a confederate presented information about two fictitious soap products. In one condition, the confederate gazed at the camera for 30% of the presentation, therefore giving the participants the impression that he was gazing in their direction. In the other condition the confederate did not gaze at the camera. Participants viewed the sales presentations from both conditions. In the condition where gaze was directed at the camera, participants recalled significantly more information about the sales presentation. Study 2 employed the same pre-recorded sales presentations used in study 1, however they were delivered to the participants under audio-only conditions (therefore, the image was switched off). Results from study 2 indicated no recall differences between the two conditions. Findings from these studies would seem to indicate that the perception of gaze aversion over a video link (a consequence of the salesman not looking into the camera) has a negative impact on information recall. This has practical implications for video-mediated presentations. In a distance learning environment lecturers could be advised to look into the camera in order to promote more efficient learning in students.

Introduction

Video-mediated communication (VMC) is regarded as a valuable tool for applications such as remote collaboration, conferencing, and distance learning (Finn, 1997). According to Campbell (1998), videoconferencing systems duplicate the experience of face-to-face meetings as closely as possible without the burden of travel. Unlike audio-conferencing, VMC allows participants access to visual information, and therefore there is a likelihood that many of the advantages associated with co-present face-to-face interaction can be replicated (Sellen, 1997). Currently, corporate and academic sectors appear to be making the most use of videoconferencing technology. In a business environment, VMC is used to serve a host of functions, for example interviews, meetings, product announcements and training (Videotalk, 1999). In the academic sector, “distance learning” is a relatively new application, which incorporates audio and video technologies for educational purposes, so that widely dispersed students can attend training seminars and courses without travelling to where the course is being presented (Videotalk, 1999).

The implementation of videoconferencing systems is largely based upon the assumption that visual signals improve human interaction. If visual signals were unimportant, then communication over the telephone would surely suffice? Rutter (1987) however has argued that visual signals are less effective during a videoconference compared to face-to-face interactions. Heath and Luff (1991) also noted that non-verbal behaviours have less of an influence on communication over video: in other words they are either ignored, or do not serve any communicative benefit. Although it is clear that VMC systems allow users access to non-verbal signals, one problem is that VMC results in an attenuation of visual cues (Doherty-Sneddon et al., 1997). One aspect of this is that the complexities of human gazing behaviour are not replicated in most VMC systems. During a videoconference information from the eyes is limited. Due to the manner in which normal video systems are set up, mutual eye contact is impossible. The camera is usually placed above the monitor and not inside of it, and therefore if one participant looks at the eyes of another person, it will appear to the other user that he/she is looking in a downward direction.

As well as natural eye contact being compromised, Monk and Gale (2002) indicate that full gaze awareness is also difficult to achieve in a normal videoconferencing set-up and is dependent upon the scope of the image provided. Traditionally, VMC is set-up in such a manner that only an image of another person's head and shoulders is available. The problem with this is that individuals in a videoconference will be unaware as to where and at what their partner is looking at. Monk and Gale (2002) indicate that providing wider coverage to expand to the environment around participants may therefore be beneficial. In doing so, images of the participants’ faces will become less clear, however, Daly Jones et al. (1998) indicate that benefits of facial expressions have been over stated in most task contexts.

Various novel attempts have been made by researchers to provide full gaze awareness in VMC. Velichkovsky (1995) and Vertegaal (1999) for example used eye-tracking devices to help detect where individuals were looking. Gemmel and Zhu (2002) suggest using a software solution to correct gaze in videoconferencing. The developers have designed a system which provides eye contact and full gaze awareness by modifying the head and eye position to a desired head and eye position. Monk and Gale (2002) report the benefits of full gaze awareness in VMC. Using apparatus that supported gaze awareness (GA Display), a number of effects were found. In comparison to two conditions (VMC with eye contact and audio-only), the gaze awareness set-up resulted in a lower number of turns and words in order to complete the task. The researchers go on to suggest that this finding can be explained in terms of understanding. Essentially, full gaze awareness provided an alternative non-linguistic method for checking one's own and another individual's understanding. In conditions where gaze awareness is not possible, individuals must signal understanding verbally, which is less efficient and therefore takes a longer period of time.

Although video-mediated technologies constrain gazing behaviour, some perceived degree of gazing behaviour can be replicated in videoconferencing technologies through looking directly into the camera: this gives the viewer the impression that the other participant is gazing in the direction of their eyes (and therefore results in a perception of mutual gaze). An example of this is the strategy employed by television presenters to give the impression that they are talking to the audience. Users of VMC technology, of course, focus attention on the monitor (displaying the image of the other conversational participant) and not at the camera. The result of this practice is that users appear to be looking away from the person(s) with whom they are communicating, which in turn means that they do not look like they have conviction in what they are saying (Tiffin and Rajasingham, 1995). Technological solutions to this problem include the ‘videotunnel’ (Smith et al., 1991), which replicates natural eye contact through the use of strategically placed ‘half-silvered mirrors.’Doherty-Sneddon et al. (1997) compared VMC using videotunnels and normal videoconferencing with no eye contact, finding that when eye contact was possible users tended to over-gaze. Participants in the videotunnel condition gazed on average 239 times at their conversational partner (more than double that recorded in face-to-face dialogues), compared to participants in the normal videoconferencing condition who gazed on average 144 times. Consequently, users became distracted by their partner's face and took significantly longer and used significantly more words to achieve a comparable level of task success. Indeed, Beattie (1981) has indicated that over-gazing interferes with cognitive processing and speech planning. Doherty-Sneddon et al. (1997) go on to argue that this over-gazing effect may be a direct result of the novelty of using such equipment, and are interested to note whether this would change over a period of time. Over-gazing may also be explained in terms of equilibrium theory. According to Argyle and Dean (1965) there are many cues to intimacy (for example proximity, touch and eye contact), and in cases where such cues are restricted individuals may compensate with other available cues. In this case participants may have compensated for a lack of proximity by over-gazing.

Although there are a number of VMC systems which can replicate the complexities of gazing behaviour, for example Gazemaster (Gemmel and Zhu, 2002), companies still make use of systems which do not allow natural eye contact to take place. Considering the wealth of research which suggests that eye gaze plays an important role in human communication, it would be expected that the inability to use such cues will have an adverse effect on communication. In a face-to-face context gaze has been shown to have a number of communicative benefits, for example helping to regulate speaker exchanges (Kendon, 1967), and as an indicator of interpersonal information, for example signalling levels of attentiveness (e.g. Kleinke et al., 1975). In addition to its social impact gazing behaviour has also been shown to have a number of cognitive effects. Beattie (1981) for example, indicated that excessive levels of inappropriate gazing result in high levels of physiological arousal, which in turn may interfere with cognitive processing. Such interference may have an effect on an individual's ability to perform cognitive tasks, for example memory tasks. This theory is also borne out by Glenberg et al. (1998) cognitive load hypothesis. Glenberg et al. found that people averted their gaze when attempting to answer questions that were deemed moderately difficult. The authors go on to propose that such behaviour is beneficial as it allows the individual to disengage from environmental stimuli (for example, the other individual's face), and therefore enhances performance directed by non-distracting stimuli.

Whereas research indicates that excessive amounts of gazing can interfere with cognitive processing, experimental evidence also suggests that gazing behaviour can improve memory for verbal information. For example, Fry and Smith (1975) showed that students remembered more instructions from a teacher who gazed at them more frequently. Similarly, primary school students remembered more of a story when their teacher gazed more frequently (Otteson and Otteson, 1980). Sherwood (1988) also found positive effects of gaze upon recall: verbal presentations with gaze improved memory for information compared to presentations without gaze. Titsworth (2000) found students retained more information in the long-term when their teachers employed immediacy behaviours (for example eye contact). Such behaviours are said to produce a greater perception of closeness between individuals. This effect is not limited to verbal information. Using a forced-choice recognition task, Hood et al. (2003) found that faces displaying direct gaze were encoded and retrieved by adults and children more successfully than faces with deviated gaze. The findings from these studies can be explained in terms of arousal. Kelley and Gorham (1988) suggest that gaze acts as an arousal stimulus, which increases attentional focus and therefore enhances memory. It may also be the case, however, that when accompanying verbal information, gazing behaviour serves as a non-verbal indicator to important information that requires attention, much in the same way that eyebrow movements are used to reinforce important aspects of speech (Ekman, 1979; Whittaker and O’Conaill, 1997).

Considering the positive effects that gaze have on human memory, can it be used to serve this same function in video-mediated contexts? This is an important question: if gaze can be used effectively to facilitate processing and recall of information even when it is mediated, this supports the importance of gaze in human cognition and also has design implications for VMC. Although it has been argued that non-verbal signals have less of an impact during VMC, some perceived level of mutual gaze can be achieved if the user looks directly into the camera. In order to test the effects of gaze across a video link on information recall, participants were instructed to watch a salesman presenting information about fictitious soap products. In one condition the salesman looked at the camera (video-mediated gazing), and in the other condition did not look at the camera. It is expected that looking directly into the camera will improve information recall.