Auditory distraction compromises random generation. Falling back into old habits?

Auditory distraction of random generation--a quintessentially executive control task--was explored in three experiments. Random number generation was impaired by the mere presence of irrelevant auditory sequences that comprise digits, but not letters, and then only if the digits were heard in a canonical order (1, 2, 3 … or 3, 2, 1 …), not in random order (Experiments 1 and 2). Random letter generation was impaired by irrelevant letters heard in alphabetical order (a, b, c …) and reversed alphabetical order (i, h, g …), but not by numbers in canonical order or letters in random order (Experiment 3). Attempting to ignore canonical sequences--with items that are members of the same category as the to-be-generated items--reduced the randomness of the generated sequence, by decreasing the tendency to change the direction of the produced sequence for random number generation, and by increasing resampling of responses for random letter generation. Like other selective attention tasks, the cost of distraction to random generation appears to stem from preventing habitual responses assuming the control of action.