Fig. 4. Discovery time (probing time between feeding attempts) for four different treatments. (A) Full lobe, (B) full lobe with transparency film, (C) large disk and (D) large disk with transparency film. Data points are medians, whiskers represent first and third quartiles. Statistical values refer to goodness-of-fit to an exponential decline function (one factor), a classical `learning curve'. Moths exploiting full lobe flowers with no transparency film show exponentially decreasing discovery times. When exploiting large disks, or either shape with transparency film, moths show larger variances in their responses, which do not fit an exponential decline function.