Flash behaviour is widespread in the animal kingdom and describes the exposure of a hidden conspicuous signal by an otherwise cryptic individual as it is fleeing from predators. Recent studies have demonstrated that the signal can enhance survivorship by leading pursuing predators into believing the flasher will also be conspicuous at rest.  Naturally, this illusion will work best if potential predators are ignorant of the flasher’s resting appearance. One way in which this could be achieved is by the prey fleeing when the predator is far away. To test this hypothesis, we compared the survival of flashing and non-flashing computer-generated prey with different flight initiation distances (FIDs) using humans as model predators. This experiment found that flash displays confer a significant survivorship advantage only to those prey with a sufficiently long FID that the predator has little opportunity to view them at rest. A complementary phylogenetic analysis of Australian bird species supports these results: after controlling for body size, species with putative flashing signals had longer FIDs than those without. Species with putative flashing signals also tended to be larger, as demonstrated in other taxa. The anti-predation benefit of flash displays is therefore related to the nature of escape behaviour.

We provide two data sets. 

The data set in FlashGameData represents the responses of 30 human participants playing our flash display game.  Each participant was given a unique identifier (ID).  Eight virtual prey were used in training followed by four virtual prey as test treatments. The prey were presented sequentially and in a random order (Order). The colour of the prey when at rest (Stationary) and when fleeing (Mobile) is given. Human subjects “disturbed” the virtual prey when their mouse cursor moved within a certain reactive distance (React, measured in Twips) of the prey, at which point the prey fled off the screen through a directed random walk. Our subjects were then invited to follow the fleeing prey into a new screenshot where the prey may have settled.  Three of the training prey were invisible in the settling stage (Visible = 0) making them impossible to find. Subjects searching for the settled prey could then either click on the prey if they saw it (Attack = 1), or if they could not find it, press the "give up" button (Attack = 0).  The time in seconds taken for each human subject to either attack the prey item or press "give up" is given by TimetoAct.  

The data set in BirdClassificationData represents the classification of Australian bird species as to whether (1) or not (0) they were judged by human observers (n = 30) to have hidden colours that are only exposed in flight.  Observers were interviewed separately in a randomized block format (3 blocks of 10 subjects) with each block having a unique presentation sequence.  Here we give the bird species, their mean FID in metres (from Blumstein DT. 2003 J. Wildl. Manag. 67, 852 – 857), and their mean mass in grams.  The final columns refer to the % of subjects that agreed on the most popular classification and show the modal classification of a subset of bird species where there is 100%, 90%, 80% and 70% agreement.