Detection of single vs multiple feature types Having come to the view above that a range of feature types might be viably detected in iMars imagery by citizen scientists, we carried out an optimisation experiment to assess whether it might be better to ask individual participants to search for a range of different features (a “General” strategy) or whether it might be more optimal to partition participants so that they looked only for a specific feature (a “Specific” strategy). Our analysis of the literature, discussed below, demonstrated that there were arguments on both sides generated through research interest primarily in industrial inspection. Thus, we had a fundamental human science issue to resolve. In this experiment non-planetary imagery was used (photographs of cork tiles some of which contained ‘features’ in the form of defects – scratches, notches and so on) in order to rapidly generate a high volume of appropriate imagery. The most relevant research base to the iMars citizen science task of inspecting planetary imagery for changed features is that which concerns industrial visual inspection, particularly in manufacturing quality control, but also in security and medical contexts (i.e., the inspection of X-rays images). Although the industrial (and consequently, research) emphasis on rapid visual inspection has generally declined in the manufacturing arena as computer-controlled manufacturing made visual inspection of finished goods less important as a locus of quality control, there remains a significant literature that examined the best ways of organising this activity. Two main strategies of improving performance for inspection have been identified in literature. One approach (the “General strategy”) is to ask inspectors to search for a range of defects. This increases the amount of targets that can be found and makes the task more stimulating and avoids the vigilance decrement, wherein performance falls in a repetitive task where valid targets only rarely appear. Alternatively, another approach (the “Specific strategy”) is to ask inspectors to only look for a specific type of defect. Reducing the number of things to “look out for” and allowing the viewer to specialise can improve performance (Xxxxxx & Xxxxxx, 1969, Megaw et al., 1979). While both these claims can be seen as intuitively valid, on consideration one might also notice that in any specific situation these would be contradictory approaches. In a real world task, a Specialised approach means that in any...
