Percepfual ar~diMolor Skills, 1999, 88, 138-146. O Perceptual and Motor Skills 1999 A REVERSED TURING TEST OF HUMAN RANDOM NUMBER GENERATION ' ALBERT0 MONTARE Hztrnan Learnlrzg and Cogt~itiorz Luboratoy Williarn Patersorz Utziverszty Strrnmary.--Based upon distincuons benveen true randomness, probabilistic ran- domness, and stochastic randomness, a reversed Turing Test was performed ro com- pare human-generated ro computer-generated random numbers. Tests oE three hypoth- eses showed that humans more often fail to behave randomly when assessed by a "probabilistic interrogator" (based on distribution-free nonparametric tests) than by a "stochastic interrogator" (based on parametric testing), that computer-generated num- bers displayed both probabilistic and stochastic randomness, and that human Failure to pass the reversed Turing Test may be attributed to a nonrandom response pattern embedded in the group data corresponding to the highly automatized human counting skill. In addition to supporrlng the ubiq~~itous observation that humans do not behave randomly, these find~ngh supl;esr that humans, unlike computers, may spontaneously interpose relative arnounrh of order that preclude successful random generation re- quiring relative amounts of disorder. The original Turing Test of artificial intehgence (Turing, 1950) called for a human interrogator to compare machine performance against the stan- dard of human performance. If the interrogator failed to distinguish ma- chine in tehgence from human intelligence, the computer passed the test. The original intent of Turing's test is widely reflected today in the work of those wishing to assess the qualitative capacity of machines to act k e hu- mans. For example, Spector, Arnold, and Wilson (1996) used the original Turing Test to study the capacity of computers to act like human teachers. Ginsburg and Karpiak (1995) and Spatt (19961, while not explicitly referring to the original Turing Test, have recently used computers to simulate human performance on random digit generacion tasks. Turing (1950) recognized the inherent reversibhty of his "imitation game" when he cautioned against an unfair comparison in the quantztative realm because the machine was designed to outperform humans in speedy arithmetic calculations. The present study ut~lized the inherent reversibhty of the Turing Test and reversed the original figure-to-ground relationships in order to make human-to-machine comparisons that assessed a human's capacity to act hke a computer in the cognitive task of random number gen- 'The author thanks Cianni Migliaccio, one of our undergraduates, €or his enthusiastic assis- tance in the data analysis. Please send correspondence and reprint requests to Abetto Montare, Department of Psychology, W~Lliarn Paterson University, Wayne, New Jersey 07470 or e-mail (montarea@frontier.~vilpaterson.educ).