EXPERIENCE WITH AN EYE TRACKER IN VISUAL COMMUNICATION EVALUATION
Jacques R. Charlier, Cathy Buquet · 1991
A consumer who purchases a product no longer deals with a local shop keeper, but makes a direct selection from the shelves of a supermarket or from the pages of a catalogue. When he buys gas, he uses a self service gas pump and pays his purchase with a credit card to an automated cashier. As a large part of today's service industry relies less and less upon direct human interaction, there is an increasing need for objective tools to assess the effectiveness of these new communication modalities. Among the numerous methods available, the analysis of eye movements should furnish the most objective evidence of the centers of interest within a picture (BUSWELL, 1935). The rationale behind this statement is that observers fixate mainly informative areas in pictures : a local variation in the spatial frequency content, a pattern irregularity or a cognitive irregularity (LOFTUS, 1983). The applications of eye movements analysis have been limited to research laboratories mainly for technical reasons. As the general pattern of eye movements when looking at a picture shows wide individual variations (BUSWELL, 1935), the identification of specific features cannot be based upon the recording from one individual but requires a large group of subjects. For exemple, if data is to be collected from a group of 60 subjects within one day, no more than 8 minutes are left for each subject's installation, explanations, adaptation, calibration and recording. This cannot be accomplished with most currently available eye movements recording instruments as they require the attachment of some device to the eye or to the head and necessitate time consuming procedures for their adjustment and calibration. Most existing systems are also limited to a qualitative analysis of eye movements scan paths. Specific tools are needed to allow users with a limited knowledge about computers and statistical analysis to develop their own procedures for a quantitative analysis of the results collected from groups of subjects . We present in this paper several solutions which have been implemented and tested in order to overcome these limitations.