An experiment is reported that is an extension of Payne (1995) and of Vu and Proctor (2003). These authors used various light/key arrangements to determine the ability of naive subjects to rate the usability of interface designs and found that naive judgements were not accurate, apart from selecting a best design. In this experiment, there were one, two, four and eight arrangements of lights and response keys with varying levels of compatibility between them. Response time is shown to be determined by two main factors: the level of response uncertainty (the number of light/key combinations); the correlation between stimulus-light and response key location, accounting for up to 93% of the experimental variance. Subjective response time (SRT), or judged response time, was responsive to the level of response uncertainty and also the correlation between light/key combinations, showing good correspondence to actual reaction times. It was found that SRT showed a stronger relationship to actual response time when subjects were presented with the full set of situations that they were to respond to, prior to judgements being made, rather than just individual sets for comparison. In interface design it is essential that, in order to reduce reaction times and error rates, there is a high level of user expectancy between locations of the stimulus and corresponding response. This research indicates how designs might be evaluated, based on the geometrical layout of stimuli and response arrangements.