Fig. 2. Results from the `Kamin control' experiment with patterns and colours as elements of the compound. (A) Pooled compound training data for all flies used in this experiment. The final 2 min test period (i.e. minutes 14-16) of each subgroup is depicted in B—E (N=103). (B) Compound control (N=25). (C) Colours alone. T-patterns were replaced by four vertical bars (N=28). (D) Patterns alone. Colour filters were removed (N=25). (E) Nonsense compound. The contingency between patterns and colours was reversed such that positive scores would indicate a correct colour choice and negative scores a correct pattern choice (N=25). Statistical analyses are the results of a Wilcoxon matched-pairs test against zero: ^***significant at P<0.001; ^**significant at P<0.01; NS, not significant. Cross-hatched columns, training; open columns, test period; shaded areas, compound stimulus; unshaded areas, single stimulus. Values are means ± S.E.M.