CHI '95 ProceedingsTopIndexes
Short PapersTOC

To influence Time Perception


Erik Geelhoed, Peter Toft, Suzanne Roberts & Patrick Hyland


Hewlett Packard Labs.,
Bristol, UK, BS12 6QZ
eg@hplb.hpl.hp.com
+44 272 228038

© ACM

Abstract

Within IT there are no studies on how to reduce waiting time perception. We demonstrate the relevance of this issue within the context of synchronous voice and data communication and conclude that waiting time perception can be reduced.

Keywords:

CSCW, time perception, user interface.

Introduction

IT research has been carried out on the acceptability of different computer response times[1,9], learning curves[10] and time distortions during computer games[5]. However there are no studies regarding perception of waiting times or the ways in which this can be influenced. Time issues become more pertinent with the increase of multiuser communication systems where lag between action and effect can adversely affect communication[2].

Time perception has long been of interest to psychology. Priestly[8] suggests that being busy reduces time judgements whereas Ornstein[7] relates the amount of information stored to the perception of that interval (being busy could increase time judgement). More recent research[3] has attempted to reconcile the two views.

The DeskSlate prototype is a pen based shared workspace, featuring simultaneous voice and data, on a single telephone line[6]. The system allows a user to share documents for joint annotation by sending a copy to her/his telephone partner. The interface displays a bar-gauge filling up during transmission. We developed an alternative method, using the metaphor of an incoming fax which enables discussion and annotation during transmission thereby enhancing productivity and possibly reducing time judgement. We report four experiments that evaluate people's ability to distinguish between different waiting times and compare different displays of incoming electronic faxes.

System

We used PCs equipped with flat screen LCD display tablets and pen input. A software application was written to simulate the arrival of fax pages from a remote system. Two modes of indicating transmission progress were used: the traditional bar-gauge, and progressive display. The latter showed the page arriving from top to bottom, allowing discussion and annotation of page sections as soon as they arrived.

The software could simulate different data arrival rates and used actual fax formats for the data. Pages were 'received' in one of two formats: Group 3 fax format[4][fig.1], and a non-standard format denoted as G3K. G3K[fig.2] was a proprietary variation on one of the fax formats which allowed the page image to be displayed in two passes per page. The first pass generated a coarse image quickly, the second pass filled in the full-detail image. Four grey levels were used to compensate for the lower resolution of the displays compared with paper faxes.

Fig 1 : G3 one pass progressive display.

Fig 2 :G3K two pass progressive display

EXPERIMENTS & RESULTS

The experiments used a within subjects design and were all balanced for order.The first three experiments each used eight subjects and the last studied eight pairs of subjects.

In the first two experiments, employing the one pass (G3) display, subjects carried out a shadowing task, i.e. they read the text out loud as it was progressively displayed. First we evaluated time judgements of transmission rates of 6 and 7.2 kbits/sec with a duration of 77 and 67.7 sec respectively, a 9.3 sec difference. Only three subjects distinguished adequately which of the two faxes was quicker to arrive. Irrespective of the actual transmission speed, most subjects judged the second one to be faster.

We then compared 4.8 and 6 kbits/sec with a corresponding duration of 95.6 and 77 sec, the differences being 18.6 sec. Here six subjects estimated correctly that the 6 kbits/sec transmission was faster. This did not result in a significant c2. To a lesser extent than the first experiment there still was a tendency for subjects to estimate the second transmission to be faster irrespective of actual durations.

The third experiment compared single pass and two pass displays, and text with graphics using a 4.7 kbits/sec transmission rate. The two pass transmission took about 70 seconds and was faster than single pass by 60 seconds. Subjects were instructed to identify and mark either a word or a symbol as the document appeared on the screen, using the pen. Although two pass was considerably faster, five participants were only able to detect a small time difference. Seven subjects (c2=4.5, df 1, p< .05) expressed a preference for the one pass display and said that this format made the task easiest. There were no significant differences between the text and graphic symbols tasks, but subjects comments indicated that for a text task the single pass format was desirable whereas for the graphic symbols task the need was not so great.

And finally we compared single pass with a bar-gauge. This study examined co-operation between two subjects, in voice and data communication performing a marking task, individually judging time duration following a one pass transmission and following a wait whilst a bar-gauge filled. In both cases transmission took 74 sec. Although there was no time difference between the two transmission times nine (out of 16) subjects thought that there was, stating that the progressive mode was faster. Subjects time estimates for the progressive transmission was significantly faster (F1,12= 8.29, p<.05). Fourteen preferred the progressive mode (c2=9, df 1, p <.01) and 12 said it made the task easiest. (c2=4, df 1, p <.05). Subjects commented:

 
'I could start work straight away' 
''The transmission time was similar to my reading speed which was a help' 
' I prefer to see something happen on the screen' 
''The bar was irritating and slow to watch' 
Across the experiments people's time estimates were considerably lower than the actual durations with the exception of the bar-gauge condition where subjects were on average closer to the actual time.

CONCLUSIONS

Accurate time judgements are difficult to make . One implication for simultaneous voice and data communication systems (when bandwidth is limited and trade-off decisions about voice compression and data transmission rates need to be made) is: a few seconds may not be crucial. However time perceptions in long term use may well be different from "one off" experiments.

Faster is not always better. Although the two pass display was a whole minute faster, it was not perceived as being that much faster. In addition it was not preferred, possibly because of the confusing nature of the two pass display.

Progressive display is a more satisfactory method of displaying an incoming electronic fax than waiting for a bar-gauge to fill. It seems that being busy reduces the time experience confirming Priestly's view[8]. When designing interfaces there may be some real value in finding an alternative to the traditional bar- gauge or the hour glass.

References

[1]Dannenbring G. L. 1983. The effect of computer response time on user performance and satisfaction: A preliminary investigation. Behaviour Research Methods and Instrumentation, 15 (2), 213-216.
[2]Dix,A. Pace and Interaction. in Proceedings of HCI'92: People and Computers VII, A. Monk, D. Diaper and M. Harrisons(eds.), Cambridge University Press, 1992.
[3]Hogan H. W. 1978b. A theoretical reconciliation of competing view of time perception. American Journal of Psychology, September 1978, Vol. 91, No 3, 417-428.
[4] McConnell K., D. Bodson & R. Schaphorst. 1992. FAX: Digital Facsimile Technology & Applications, Artech House Inc.
[5]Myers D. 1992. Time, Symbol Transformation, and Computer Games. Play and Culture, 5, 441-457. [6]O'Conaill, B., E. Geelhoed & P.Toft. 1994.DeskSlate: A shared Workspace for Telephone Partners. Proceedings of CHI '94 Conference Companion, 303-304, ACM New York.
[7]Ornstein R. E. 1969. On the Experience of Time. Penguin Books.
[8]Priestly J.B. 1964. Man and time. Aldus Books Ltd. London
[9]Shaefer F. 1990. The effect of system response times on temporal predictability of work flow in Human Computer Interaction. Human Performance, 3 (3), 173-186.
[10]Shneiderman B. 1986. Designing the User Interface: Strategies for effective Human Computer interaction. Addison-Wesley Publishing Company.