THE UN I VE RS IT Y OF MI CHI GAN COLLEGE OF LITERATURE, -SCIENCE, ADRTS Department of Psychology Final Report INFORMATION-HANDLING AND DECISION-MAKING BY INDIVIDUALS AND SMALL TEAMS:! - M' O'RA.: Proj~Jet 02814 under contract with' AIR FORCE OFFICE OF SCIENTIFIC RESEARCH OFFICE OF AEROSPACE RESEARCH CONTRACT NO. AF 49(638)-449 WASHINGTON, D C. administered througho OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR February 1964

f.

OBJECTIVES OF -THE RESEA.RCH PROGRAM The purpose of the research.program under Contract No. AF 49(638)-449, which:was initiated in September, 1958, was to analyze: the performance of individuals working singly, and in small teams in a variety of informationprocessing tasks, including not.only information-transmission tasks but also tasks in which they would have to learn to deal with probabilistic information, to categorize.and filter information, to utilize feedback from previous responses, and to apply rules or concepts in dynamic choice or decision situations. An information-theoretic approach wass followed in all of this work, but emphasis was upon aspects of behavior in which the individual could adapt in- various ways to the special demands of the task. Emphasis was. also focused on rate of performance in all of the research. Team performance was considered as an extension of individual,performance, with varying information-exchange processes between team: members and a well-defined output- criterion.

SOME HIGHLIGHTS OF THE RESEA.RCH FINDINGS Research findings from this programhave been published in professional journals, theses, and special technical reports, and rkported at various scientific meetings. Therefore, no attempt will be made here to review or summarize all of the results of the research. However, a few highlights can be singled out for emphasis and by way of summary of the results. AN INFORMATIONAL APPROACH TO THINKING Nearly all previous studies of human capacity in information-processing tasks have employed information-conserving tasks, in which all of the stimulus (input) information must be preserved in the responses (output). Under the present research program considerable progress was made in the.analysis of performance in tasks ordinarily considered to involve thinking or higher-level cognitive processes. Information reduction is one broad class of such tasks. The individual must select relevant information (filtering) or must categorize information into fewer classes than there are stimuli after consideration of all stimulus characteristics (condensing)..Some outstanding contributions to the understanding of information-reduction processes were made by Dr. Michael Posner in his Ph.D. thesis, "A.n Informational Approach t6 Thinking" (ORA. No. 02814-9-T), which received the Marquis Award of the Department of Psychology at The University of Michigan for the best thesis in Psychology in 1961, and also won first place in nationwide competition for the American Institute of Research Creative Talent Award for the best thesis in experimental psychology. One of the principal findings from this research was that given some specified information source, the greater the degree of information condensing required (the smaller the output information) the more time required for the information processing by human subjects. In terms of response information, this is just the opposite relation to that which holds for information-conserving tasks. COGNITIVE ASPECTS OF INFORMATION PROCESSING A series of numbered reports in the Journal of Experimental Psychology bearing this main title have been published, and a paper with this title was presented and is being published as part of the proceedings of the XVII International Congress of Psychology, which was held in Washington in August, 1963. These reports have analyzed the important role of processes such as cognitive set (-familiarity with sets of alternatives, adjustments for speed vs. accuracy, etc.) in human-information processing. Thus an individual is viewed as using many adaptive processes which permit him to handle different informationprocessing tasks in different ways-and the research contributes to an understanding of man's versatility. 2

SPEED-ACCURA.CY RELATIONS It has long been known that man has great capacity for achieving accuracy (reliability, precision, etc.) at the expense of speed and vice versa. A. decision-theoretic model has now been proposed to account for how this trade-off is accomplished. It treats perception, discrimination, and information processing as involving rapid sequential stimulus sampling. Results of studies using this approach have been reported and show considerable promise especially in accounting for man ts capacities in dealing with redundant sequences. These findings emphasize that-errorless performance i; a questionable concept.; some risk of error must be accepted in any task accomplished under time stress, although it may be possible to make error rate arbitrarily small. Such a theory has great significance for application to military reliability work. HIERARCHICA.L AND: TEMPORA.L ASPECTS OF INFORMA.TION-PROCESSING: SKILLS The view has long been held that skills are organized hierarchicaliy, both in terms of sequences of actions, and levels of Organization.' However, little recent evidence has been advanced on thiS issue, especially as regards information-processing skills. In several recent studies accomplished under the. present research program, including Dr. Richard Pew's Ph. D. thesis, "Temporal Organization in Skilled Performance" (ORA. No. 02814-llT):, analytic data have been obtained which shed new light on the role':tf hierarchical processes in skills. In terms of temporal organization, individuals were shown to organize short patterns of motor behavior, lasting for up to one half second or more, in advance of initiating the response sequence. Dr. Pew was able to demonstrate two processes in.a control task, a continuous rapid control process (up to six or eight responses per second), which was modolated by a slower, higher-level regulatory process. In other studies it has been demonstrated that perceptual.or' decision time, and movement execution time, are quite independent processes. Various stages in skill learning have been proposed and detailed expositions published as.chapters in two separate books on learning and training. In essence it i.s hypothesized that individuals develop semiautomatic habits, or subroutines, that run off essentially:as "open-loop" processes, but which in turn are governed by slower and intermittent cognitive processes that can be viewed as analogous to executive programs. USE OF HEURISTIC PROCEDURES IN PROBLEM SOLVING Several studies were completed that extend information-theoretic concepts to tasks in which subjects are asked to find solutions to problems for which no algorithm (procedure for guaranteeing an optimum solution) is available. Typical tasks included the transportation-assignhent and the personnel-assignment problems. Individual performance was found to be surprisingly good on such problems, from the outset of training, in comparison with solutions produced by a random process, by reliance on simple heuristic rules, and by 53

optimizing procedures; however, learning was very slow and subjects were not able to give exact descriptions of the methods they employed in finding solutions. The time required to find acceptable solutions was found to be highly correlated withthe total number of possible solutions and the average information reduction per step in the solution (thus agreeing with findings in studies of information condensing). INFORMATION SHARING FOR TEAM PERFOMAMNCE Studies relating td the display of information, and the organization of information exchange procedures for efficient team performance in information processing and decision making, hae suggested two main concepts. First, under conditions of high-informationn oad and time stress, increased information exchange (communication) processes between team members usually slows up team performance; thus in a sense communnication processes compete for limited human channel capacity. Second, team performance may be considerably enhanced by the use of visual information exchange devices, i.e., by the provision of information sharing via visual rather than via auditory channels One reason wzhch is believed to account for this effect is that visual information exchange can thus be scheduled on an individual basis, to take advantage of momentary light periods of work load, without interrupting the work of two people at inopportune times. 4~

P — tLICATIONS The following is a list of reports of the work supported by this contract, which have been published {h books and professional journals, or as Ph.D. theses, and which are now available to scientists or are in press and will be available during the current year. BOOKS AND JOURNAL ARTICLES Fitts, P. M. and Switzer, Gail. Cognitive Aspects of Information Processing I: The Familiarity of S-R Sets and Subsets. J. Exp. Psychol., 1962,63, 321-329. Fitts, P. M., Peterson, J. R., and Wolpe, G.. Cognitive -Aspects of Information Processing II: Adjustments to Stimulus Redundance. J. Exp. Psychol., 1963, 65, 507-514. Fitts, P. M. and Peterson, J. R. The Information Capacity of Discrete Motor Responses. J. Exp. Psychol., 1964 (in press). Fitts, P. M. and Biederman, I. S-R Compatibility and Information Reduction (submitted to Journal of Experimental Psychology). Fitts, P. M. Skill Learning. Chap. in A. W. Melton (Editor), Categories of Human Learning, Academic Press, 1964. Fitts, P. M. Contribution to the Consideration of Information Processing and the Nervous System. Proceedings, International Congress of Physiological Sciences, Lieden, Holland, Sept., 1962 (Excerptia Medica, in press). Fitts, P. M. Cognitive Aspects of information Processing. Proceedings of the XVII International Congress of Psychology, Washington, D.C., Aug., 1963, NorthHolland Publishing Co. (in press). Fitts, P. M. Factors in Complex Skill Training. Chap. 6 in R. Glaser (Editor) Training Research and Education, Pittsburgh, U.Pitt. Press, 1962. It is expected that several additional reports, based in part on work initiated under this contract will be published subsequently with joint credit to the present contract and to a subsequent AFOSR contract (No. AX 49(638)-1235).

PH.D. THESES The following Ph.D. theses were: completed during the period of this contract: Moore, H. G. The Effects of fLad and.ess/ibility of Information Upon Performance of Smalll Teams. Ph.D. Thesis and AFOSR Report No. 1656,'October:. 1961. Posner, Michael I. "An Informational Approach to Thinking," Ph.D. Thesis and AFOSR Report No. 2635, April, 1962. Stimmel, Theron. Problem Solving Requiring the Use of Heuristic Procedures. Ph.D. Thesis, September-. 1962....:; Pew, Richard. Temporal Organization in Skilled Performance. Ph.D. Thesis and Report No. 02814-1-T, May, 1965. 6

EDUCATIONAL TEIEVISION Two 30-minute Educational Television Films were completed d-.ring the period of this contract and have been shown widely on the' National Educational Television network. These films -were not supported by the contract, but-they do report:aspects of the work accomplished under this contract. The first film was produced (and supported financially) by The University of Michigan and deals exclusively with work supported by this contract; the second was produced by the American Psychological A.ssociation, with funds provided..by the National Science Foundation:, _and covers work'.at Wright A.ir Development Division and elsewhere.as well as some..of the findings of the present program. The titles of these educational film.a.re: "The Stitch in Time." University of Michigan Television Center, Series, Of Men and Motives, File No'. 11D. "Of Men and'Machines." National Educational Television and Radio Center, Ann Arbor, Michigan. (P-59-9; Series, Focus on Behavior.) 7

PERSONNEL:. The following persons served on the staff for some period of time during the work on this contract: PROFESSIONAL STAFF. Dr. Paul M. Fitts (Project Supervisor) Dr. Roy Davis (Visiting Scientist from Oxford University) *Dr. Robert Isaacson (Psychologist) Dro Wilfred Kincaid (Consultant in Mathematics) GRADUATE STUDENTS (ASSISTANTS IN RESEARCH) Biederman, Irving Pickett, Ronald Brown, Betty Smith, Edward Henry, G. Croft Switzer, Gail Kibler, Austin (Major, USAF) Smith, Robert L. Moore, Heber G. Stimmel, David T. Passini, Frank Tomlinson, Percival Pew, Richard Von Buseck, Calvin -Peterson, James R. Wolpe, Gerson RESEARCH ASSISTANTS OTHER THAN GRADUATE STUDENTS Levine, Peter Roberts, Kelyn Mylrea, Kenneth Rosthal, Marianne Radford, Barbara Story, Vern Tobin, Paula It is estimated that approximately 900 University of Michigan students served as subjects in experiments under this program.

UNIVERSITY OF MICHIGAN 3 9015 02826 1 256111111 3901 02826 1256