Frans Verstraten¹, Renèll Rodrigues¹, Guang Yang¹, David Yu¹, Reuben Rideaux¹, David Alais¹; ¹The University of Sydney, School of Psychology

In the mental health domain, like substance abuse research, questionnaires are prone to be contaminated by (social desirability) bias. For that reason, there is a trend to replace traditional questionnaires by free-viewing paradigms where for example eye movements and pupil size are registered when looking at naturalistic stimuli. In the realm of (binge)drinking research several stimuli databases of bottles of alcoholic and non-alcoholic drinks have been developed and claim to have reliability and (construct) validity. However, often the saliency of the stimuli is not considered, other than that the products are photographed under the same conditions (e.g. light source, camera location, etc.). We try to find out if differences in the individual stimuli content can account for choice preferences/biases. We calculated the rms-contrast of the stimuli, where the background was not taken into consideration, for luminance (the standard deviation of luminance fluctuations around the mean) as well as the rms-contrast in each color channel (RGB). We ordered the stimuli in increasing rms-contrast over these 4 dimensions separately. In experiment 1, in each trial observers were presented with two stimuli from the database and asked to indicate the most salient one. This resulted in an individual preference order. A correlation approach shows that the rms contrasts have no or very limited predictive value for stimuli preference. However, if we take brightness and saturation into account by using the HSV coordinates (hue, saturation, value), there is a strong correlation between stimuli and choice preference. We next used a noise paradigm where one of 2 noise patterns had a hidden object and became visible when the noise decreased as a function of time. In that case the observer was strongly relying on the object’s contours. The results show that stimuli in addiction research need a careful look at possible intervening stimuli characteristics.