"Inhibition of return" (IOR) was originally described in the field of spatial attention, but it has also been observed in random number generation tasks. Subjects showed a tendency of "repetition avoidance," which can be considered as equivalent to IOR in another cognitive domain. As temporal factors have been suspected to play an important role in random number generation, we aimed to examine how such factors might influence regularities such as repetition avoidance in random number generation tasks. Participants were instructed to verbally generate a sequence of numbers at a certain pace, that is, with either 0.5, 1.5, 3 or 4 s between each response. Each number in the sequence should have the same probability of appearance and should be independent from the others. However, it was observed that the human-generated sequences differed drastically from computer-simulated pseudo-random sequences. The distribution of the repetition gap, which indicates how many different numbers are reported between two identical numbers in the generated sequences, showed a "three-phase" characteristic: a phase of avoidance of the same number, an oscillatory component for coming back to the same number, and finally an exponential decay of number selection frequencies. This three-phase characteristic was independent of the time interval between responses. These observations indicate an item-based process in random number generation, making a time-based control in this task rather unlikely as has been hypothetically assumed.
Read full abstract