Statistical tests.

Extra sum of squares F tests were used to determine if the fits to the data describing thresholds for detection and identification of the patterns of each test Pairs 1(a)-(f) were statistically different for the group of observers (see Figure 2). The thresholds (at 75% correct performance) are assumed to be the same unless p<0.05. Three asterisks highlight p<0.001, two p<0.01 and one p<0.05. χ2 tests were performed as per Watson & Robson [15] to determine if the patterns were perfectly discriminated. Adopting the criteria that p>0.95 indicates perfect discrimination and p<0.05 not the tests show that only Pairs 1(a) and 1(c) are can be considered perfectly discriminated and the other pairs are demonstrably not perfectly discriminated.