Pergamon 0028-3932(95)0006f~6 Neuropsychologia, Vol. 34, No. 2, pp. 97-103, 1996 Copyright © 1996 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0028-3932/96 $15.00 + 0.00 Random number generation in dementia of the Alzheimer type: A test of frontal executive functions P. BRUGGER,*t A. U. MONSCH,]- D. P. SALMONS and N. BUTTERS]" tDepartment of Veterans Affairs Medical Center and SDepartment of Psychiatry and Neurosciences, University of California, San Diego, California, U.S.A. (Received 7 January 1995; accepted 3 May 1995) Abstract--The quality of attempts at generating a random sequence of the numbers 1~5 was studied in 30 patients with dementia of the Alzheimer type (DAT) and 30 elderly normal control (NC) subjects. Three main findings emerged: (1) DAT patients' subjective random sequences were more stereotyped (contained fewer digit combinations) than those of NC subjects. (2) This difference in response stereotypy was due to patients' enhanced tendency to arrange consecutive numbers in an ascending series ('counting bias'). (3) In the patient group, degree of sequential nonrandomness was positively correlated with overall severity of dementia and with the extent to which performance on neuropsychological tests specifically assessing executive functions (fluency, naming, error monitoring) was impaired. These results illustrate a loss of behavioral complexity in the course of dementia and are interpreted as reflecting a frontal dysexecutive syndrome in DAT. Key Words: random generation; Alzheimer type dementia; frontal lobes; neuropsychological tests; review. Introduction When human subjects attempt to generate a random series of numbers they quite consistently fail [3, 27]. While in true random sequences consecutive numbers are independent from one another, there are character- istic sequential dependencies in subject generated number series. In particular, a human random generator produces too few direct repetitions of the same number and too many ordinal subseries. Although "cognitive psychology has engendered few examples of so much support for and agreement among researchers about the prevalence of a cognitive bias" ([29], p. 352), random generation has found relatively little application in experimental and clinical neuropsychology. Bearing in mind the early recognition that randomization tasks have received in clinical psychiatry [e.g., 28, 41, 42], especially in German literature [e.g., 2, 14, 23], this omission within neuropsychology may seem puzzling. Table 1 summarizes all published studies, of which we are aware, that examine randomization perfor- mance in neuropsychological patient groups. Despite the considerable variability in design and procedures, *Correspondence to be addressed to Peter Brugger, Depart- ment of Neurology, University Hospital Zfirich, CH-8091 Zfirich, Switzerland; e-mail: pbrugger@npsy.unizh.ch. the main findings of these studies are strikingly similar: without exception, sequences generated by brain damaged patients were found to be less random than those generated by their respective control group. However, most of the studies listed in Table 1 employed non-specific measures of sequential random- ness which reflect the degree of a subject's response stereotypy by contrasting observed and expected frequencies of consecutive numbers. From these strictly quantitative measures it is not possible to determine whether an observed deviation from ideal randomness is due to perseverative responding, to a complete avoidance of repetitions, to excessive use of a counting strategy, or to some other form of response patterning. Thus, most of these studies provide little information about the neuropsycholo- gical processes and specific brain structures that are important for ,random generation. The few studies that have examined the processes underlying the impaired random number generation performance of brain damaged patients provide evi- dence that different neurological or neurobehavioral deficits are associated with qualitatively different sequential biases. For instance, excessive ordinal sequencing is particularly pronounced in patients with frontal lobe dysfunction [34], most likely due to a failure to inhibit interference from a highly overlearned 97