A. Karakolidis, D. Scully, M. O'Leary

Centre for Assessment Research, Policy and Practice in Education (CARPE), Dublin City University (IRELAND)
Technology has had an important impact on our lives. It has changed the way we communicate, relax, work and learn. Its impact on the field of assessment is no exception: many tests have moved from paper-and-pencil to computer-based testing, with the aim of improving the efficiency of test delivery and scoring. Although this has been a major advance, it is only the first of several, as computers can be much more than simply an alternative means of delivering paper-based tests.
Computerisation can greatly increase the quantity and quality of the stimulus materials that students are presented with, from simple text to basic audio, still images, high quality animated videos and interactive virtual environments. Computer simulations that either involve animated videos that mimic situations or provide interactive virtual environments are becoming increasingly popular, as test-takers are presented with more realistic contexts in which to demonstrate knowledge and skills.
Given the fact that the development of animations and computer simulations is a complex and expensive process, test developers need to consider carefully important aspects for developing a quality assessment tool. One of the most important features of a simulation is its fidelity, or the degree to which a simulation accurately presents a situation. Intuitively, it is often assumed that higher fidelity leads to better simulations. However, there are some instances in which this may not be the case, for example, when designing animated characters.
In 1970, Masahiro Mori hypothesised that humanlike objects (e.g. robots and animated characters) that are designed to be very realistic may evoke a feeling of eeriness in the perceivers. He argued that trying to make robots or animated characters more humanlike, tended to increase perceivers’ affinity with them up to a point. Once that point of humanlikeness was exceeded, animated characters become too realistic and viewers experienced an eerie sensation. Mori coined the term Uncanny Valley to describe this phenomenon. Although Mori formulated his theory more than four decades ago, the Uncanny Valley still seems a relevant and valid concept with great importance today.
Although it has been many years since Mori published his theory, empirical research around this topic is in the early stages of development. This paper discusses the advantages of using computer simulation in educational assessment, presents an up-to-date review of Mori’s Uncanny Valley and provides practical implications for the development of computer-generated simulations, with a specific focus on the features of animated characters.
It is recommended that poorly produced multimedia can negatively impact test-takers’ attitudes not only towards the assessment procedure, but towards the body/organisation that set the assessment as well. Moreover, although animated characters with universally attractive characteristics can mediate the effects of the Uncanny Valley, very realistic characters should be avoided. It is argued that 3D or 2D caricatured animations could be a safe alternative, providing satisfactory authenticity without falling into the valley. Finally, it is recommended that simulation designers put emphasis on characters’ facial expressions, voice and movement so as to mitigate the negative impact of the Uncanny Valley.