Simulation of Human Behavior: Possibilities for Public Health Education Adelaide A. W. Williams, M.P.H. Chief, Health Education Oakland County Department of Health
Pontiac, Michigan Three levels of computer simulation of individual and aggregate human behavior are examined and applications to health education are reviewed General implications of this use of computers for the training and future role of public health educators are suggested .
The intent of this discussion is to introduce an area of computer usage which may have immediate application to the practice of public health education, but which certainly will have professional, ethical, and practical implications for the future of public health education and public health educators. It is not the purpose of this discussion to review what we already know about computers: how we use them; what benefits they have occasioned; or what threats they
pose.7,28,34 BASIS FOR CONCERN WITH COMPUTER SIMULATIONS
A basic question which concerns behavioral scientists engaged in research or attempting to apply research findings is how people make decisions. Health education programs are often designed to encourage decisions that will lead to behavior that will enhance health, but too often they appear to produce no effect that is measurable or generalizable from one population to another. Several models which are creating interest in health education programming include the Health Belief Model, Diffusion models, and Locus of Control models.3,11>14>1’ Each of these begins with different, but sometimes overlapping, assumptions; and each provides a partial explanation of how people structure decisions. Yet none is sufficient and necessary for the prediction of behavior. Similar problems abound with other models which have not been cited. Thus, the scientist or practitioner must structure his own decision about which one to use from among the competing theories. Once a model or combination of models has been chosen, the problem shifts to one of testing; and with testing, the concern with separating the trivial influences from the non-trivial in evaluating results. It is simplistic to state that there is almost an infinite number 181
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
of variables which could or should be considered; but no matter how obvious, the real problem will not disappear. This leads to the two fundamental reasons for concern with computer models and simulation of human behavior: (1) simulation permits experimentation with partial human systems without the ethical quandary of unexpected, undesirable, or morally questionable effects on people; (2) models can be structured to embody one or more theoretical concepts, and in more easily controlled situations, test the assumptions and dynamics of the concepts more exhaustively than is possible in real human systems. DEFINITION AND EXAMPLES OF SIMULATION
To simulate
an
object,
person,
or
a
process,
according
to
Merrian-Webster, is &dquo;to effect the appearance of’ that object, person, or process. A simulation, then, is a representation of something else. A photograph could be considered a simulation; Einstein’s equation, E = me 2, is a simulation of the process through which matter and energy change form. The definition of simulation to be used here is based on McLeod, who differentiates between a computer model and the assumptions, theory, program, and data needed to implement the model on the computer.23 The simulation is the dynamic process of carrying out experiments with the model on the computer. Sophisticated computer models of some behavior already are widely used. The most familiar is the election model .26 First developed in 1960 to estimate the effect of religious beliefs on the outcome of the presidential election, the basic model has been refined to the point that it can make accurate predictions of election results from small samples of votes within minutes of their tabulation. A less familiar use of simulation devolves from queuing theory and applications to industrial processes. A high level computer language, General Purpose Simulation System (GPSS), could be used to study and plan medical care delivery systems.l7 For example, an outpatient clinic may be defined by the elements of its facilities, staff, tasks performed, time needed for each task, and the pattern of utilization by patients. Varying these elements in simulation will yield data for choosing among alternative staffing patterns, patient flow, physical layout, and other configurations in the organization of the clinic. Such computer use does not, however, address the decisions of people to attend the clinic; whether they will enter the system early in the course of a disease episode; or if they will appear for regular health maintenance. The remainder of this discussion will focus on several computer models and simulations which do attempt to address these issues. The choice of simulations is entirely personal. They are 182
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
Figure l. Memory structure of permission of the publisher. not the
they
HOMUNCULUS Source: Loehlin, 1968. Reprinted by
only models which could be examined, nor is it claimed the most representative or the best models.
that
are
THREE COMPUTER MODELS ADAPTABLE TO HEALTH INTERESTS
The three computer models which will be described represent a continuum from intra-individual decision-making to community to society at large. Results of various simulations and possible implications for public health education will be discussed.
Individual and Interpersonal Analysis The empirical theories of George Homans are embodied in a computer model of individual personality structure called HOMUNCULUS.19,20,22 Developed by John and Jeanne Gullahorn, HOMUNCULUS is a sophisticated attempt to explore the structure of personality and the dynamics of interpersonal interaction. 183
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
of the model is the memory structure (Figure 1) which is composed a series of lists interconnected on an elaborate grid. The grid resembles a decision tree, of which it is a more elegant variety called an EPAM (Elementary Perceiving and Monitoring) Network.21,25 An EPAM network is &dquo;built&dquo; by a special computer program which is capable of discriminating between two or more discrete elements. It then constructs the network by devising classes of elements and pathways within and between classes. An additional feature of the EPAM network is that classes are stacked, or arranged hierarchically. The memory structure of HOMUNCULUS is composed of three main classes or lists: identification, activities, and images. The identification list contains fairly generalized information about the personality being modeled. Included are a name, age, sex, verbal and mathematical skills, special talents, and any other traits which identify the individual. The basic structure may be altered considerably by varying the information on this list. Past behavior, actions taken by the individual, actions of others directed toward the individual, and those observed to happen to others are stored on the activities list. Each action and class of action has associated with it a value which expresses how rewarding those actions have been to HOMUNCULUS. The images list stores detailed information about the environment in which HOMUNCULUS lives. Concepts about the surroundings, of groups to which the individual belongs, and a composite &dquo;self-concept&dquo; form the major divisions of the list. Within each of these major lists are sublists and sub-sublists. At each of the increasingly circumscribed levels, there are numerous cross connections between the details at that level, at other levels, and on other lists. HOMUNCULUS operates by accepting a report of an activity by someone or something in its sphere of knowledge. This may be done by arbitrarily feeding it an action (e.g., &dquo;Your boss has just been taken to the hospital with a heart attack&dquo;), or by allowing a second model of HOMUNCULUS to generate the activity (e.g., HOMUNCULUS &dquo;A&dquo; is Sam and HOMUNCULUS &dquo;B&dquo; is Joe, members of a volunteer committee planning an anti-smoking program). The subsequent responses during the simulated interaction are based on a calculation of the advantages and disadvantages of each possible alternative action the individuals might take. The value associated with each action alternative also permits expressions of emotion in conjunction with the action chosen. In one simulation, acute animosity was generated between two different models, although they began with a neutral stance. Variations in behavior are possible as the simulation builds a record of- past successful choices in similar situations. This record of past
The
core
of
184
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
behavior also permits generalization of behaviors to handle new situations or new people in the environment. The major value of this computer model to public health education is the generalized personality structure. Unlike other models of individual behavior, this one is not based on a detailed analysis of one person or a small group of professionals, but represents a more general form of personality operating on the basis of enlightened rationality,I ,8 , 9,13,11,33 Thus, various types of personalities can be superimposed on the basic model and tested for correspondence with empirical observation. As has been mentioned, two models with no particular emotional set developed signs of &dquo;hatred.&dquo; In another simulation, the Gullahorns interacted a model they termed &dquo;cynical&dquo; with an &dquo;eternal optomist.&dquo; The result was frustration, anger, and a breakdown of communication between the two models.20 With modification to include beliefs, past experience, and health knowledge, similar simulations could be run. One such simulation might focus on fear arousal, and using various postulated personality types, study the differential effects of fear arousal. Results might lead to a more precise definition of &dquo;low,&dquo; &dquo;medium,&dquo; and &dquo;high&dquo; fear arousal. Alternatively, simulations might test the Health Belief Model and develop data on how &dquo;cues&dquo; to initiate health action are
generated.3 A further
of HOMUNCULUS is in the study of small appropriate modification, three or more models can be interacted and the dynamics of the process studied. If individual personality structures can be imposed on HOMUNCULUS,
potential
use
group behavior. With
then the members of real committees
or task groups could be studied. Potential process and task problems could be studied before any actual meetings and alternative means of intervening could be tested. This application has disturbing implications, however, since there would be no guarantee that a new, more manipulative era would not
be the result.
Community and Mass Communication Applications In expanding the scope of the model beyond the small group level,
a
referendum simulation derives from the election model discussed above., It is an attempt to develop more predictive ability using local information and is loosely based on the &dquo;two-step&dquo; flow of information theory with elements from the work of Hovland, the Sherifs, Heider, and Festinger. The base of the model is an actual random sample of the community under study. Approximately 500 persons are represented in the memory storage by the following information: demographic characteristics, attitudes and predisposing experiences concerning the referendum issue, frequency of exposure to various news or
community
185
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
information channels, attitudes toward persons or institutions in the community who might prove pivotal in the campaign, knowledge and acceptance of the standard arguments for and against the referendum issue, frequency of conversation about local politics, characteristics of conversational partners, initial interest in the issue, initial position on the issue, and voting history in local elections. Once this information has been gathered and stored, the simulation proceeds through two cycles which together represent a week during the referendum campaign. The cycles imitate the processes through which each individual may change his position. The first cycle in each week is exposure of the population to the communications media. The media are provided with inputs such as speeches by various individuals. A set of rules based on attitude toward the source, previous knowledge of the assertions made, congeniality of the assertion in relation to predisposition, and position on the issue determine which assertions will be accepted by each of the community members. As a result of media exposure, any individual represented may change the media assertions he accepts, his attitudes toward sources, his probability of future media exposure, and his interest or position in relation to messages received. The next cycle is to allow individuals to interact with each other. The potentiality of conversation is based on a match of demographic characteristics and of a common location for usual political discussions. Rate of conversation is based on the level of interest. As in the media cycle, assertions are made and accepted or rejected on the basis of the compatibility of conversation partners, previous knowledge or acceptance of the assertions, and position on the issue. The total simulation is composed of several weeks of campaigning culminating in a community vote. The model permits an active campaign or a dull one. The voting outcome is determined by the changes occasioned in the community and the numbers who participate in the voting. Although this model has not been fully tested in a real campaign, it holds much promise for public health education as the following example illustrates. A majority vote for a referendum issue is replaced by evaluation techniques to measure a decrease in automobile accidents. Then detailed plans for a media campaign to encourage seat belt use are devised as input during the simulation. Several different plans could be developed using various mixes of media messages. The simulation then could test which mix proved to be more effective in generating change, or it could show that mass media approaches would never produce measurable changes. The experiments might provide information on which messages prove more effective in reaching specific target audiences; and they might allow plans for an actual program on seat belt use to maximize expenditure of limited resources. 186
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
These are all issues public health education attempts to with limited precision.
answer now
Environmental and Population Applications The final model to be discussed considers the world as a whole. Drawing on his work with engineering systems, Jay Forrester of M.LT. developed the concept of continuous feedback loops. Applying this concept to ever larger systems, Forrester and his colleagues have described a model of the world based on the interaction of four
interlocking chains.’6 According to this model, the quality and quantity of life on this planet can be described by the interaction of loops representing use, and environmental pollution. Each of these four areas are connected to each other and to smaller loops within each major division. Population size, for example, is dependent on the interaction of the birth and death rates as well as the amount of food generated. Food, in turn, is dependent on the amount of capital expended in the agricultural field as well as the effects of pollution. Pollution depends on the way capital is expended and on the total size of the population. A community based group in Michigan has modified this model for use by its task forces in the study of regional problems of air and water pollution and solid waste disposal.30,31 Real data which reflect the actual conditions in the region have been gathered to form the base of the computer simulation. Using input from members of task forces, alternative measures to improve environmental quality are developed and goals set. The detailed programs thus generated are fed into the computer, which calculates the effects of these programs over
population, food supply, capital
a
twenty-year period.
During the
course of these experiments, non-obvious alternatives have been discovered and then explored in subsequent simulations. By these means a detailed proposal for solid waste disposal has been developed for implementation by local governmental agencies. The proposal also includes plans for an annual evaluation to monitor the progress of the implementation, to discover any problems or side effects, and to propose corrective measures to keep the program on
target. SOME IMPLICATIONS FOR PUBLIC HEALTH EDUCATION
Granting that computer models and simulations will continue to be developed, and that they will be used increasingly as the models become more sophisticated, there are many non-trivial implications for public health education which might be discussed, only two of which will be - examined here. 187
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
Training of future health educators will undoubtedly need to change. What training a professional has necessarily impacts on the kind of role that professional will assume in his practice. A world which requires elaborate preparation in mathematical theory-building, computer programming, and more sophisticated methods of analysis of communities will require much restructuring of the public health education curriculum as we now know it. The question is whether the expertise now exists in our academic programs to provide these new areas of study. Even if public health education programs begin to move toward these areas, there is the question of whether a new role of public health educator as community simulation expert will be accepted by other health professionals. Recent attempts to rid the profession of old expectations of public health educators have met with resistance. Too many health professionals still cast public health educators into technician molds of pamphlet-producer, resident audiovisual specialist, annual report writer, or health fair coordinator. An alternative, of course, is to encourage others to become the community health simulation experts and to apply the materials and research in our practice. If we choose this alternative, must be prepared to deal with other consequences, the most basic of which is an ethical concern. The question now is not if the technology will exist to manipulate human behavior, but how soon. Before we reach this new age of social engineering, we must consider how we will interface with the constructors of human system models.7> 2?,28 Otherwise, who will insure that human behavior and particularly health behavior computer models include assurances of privacy for the individual, protection against even subtle coercion, or the option of saying &dquo;no&dquo; to even the most personally hazardous health decision? The issue for public health educators is not whether we participate in modeling and simulations but will we be involved soon enough to influence the basis on which models are developed and the assumptions about the dignity of the persons to be affected. Public health educators have long been advocates for voluntary decision making and behavioral change. It is becoming imperative that we act on our ethical premise. We must open ourselves to computer uses such as simulation and involve ourselves in examining the benefits and costs. We must begin to train ourselves and our future colleagues. Finally, we must begin a dialogue with simulation scientists to insure the ultimate freedom of each person to decide his own health behavior. we
REFERENCES 1. Abelson
RP, Bernstein A: A computer simulation model of community referendum controversies. Public Opin Q 27:93-122, January 1963.
188
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
2. Baker W: The internal organization of computer models of cognitive behavior. Behav Sci 12 :156.161, March 1967. 3. Becker MH (ed): The health belief model and personal health behavior. Health Educ Monogr 2:Winter 1974. 4. Benfari RC: PROBE: A computer instrument for field surveys of psychiatric disorders. Arch Gen Psych 23:352-358, October 1970. 5. Bernstein AL: Innovation in educational research: Implications for school health. J Sch Health 39:385-391 June, 1969. 6. Binder S: Computers in public health. Clin Pharm and Therapeutics 8:162-169, January-February 1967. 7. Brooks FP: The Mythical Man-Month. Reading, Mass., Addison-Wesley Publishing Co., 1975. 8. Colby KM: Computer simulation of changes in personal belief systems. Behav Sci 12:248-253, May 1967. 9. Colby KM, Watt J, Gilbert JG: A computer method of psychotherapy: Preliminary communication. J Nerv Ment Dis 142:148-152, 1966. 10. Computers in public health. W.H.O. Chron 23:556-563, December, 1969. 11. Costello RM, Manders KR: Locus of control and alcoholism. Br J Addict 69:11-17, March 1974. 12. Dowell LJ: The relationship between knowledge and practice. J Educ Res 62:201-205, May 1969. 13. Dutton J, Starbuck W (eds): Computer Simulation of Human Behavior. New York, John Wiley and Sons, Inc., 1971. 14. Etzioni A, Etzioni-Halevy E (eds): Social Change: Sources, Patterns, and Consequences, second edition. New York, Basic Books, 1973. 15. Felton B, Kahana E: Adjustment and situationally-bound locus of control among institutionalized aged. J Geront 29:295-301, May 1974. 16. Forrester J: World Dynamics. Cambridge, Mass., Wright-Allen, Inc., 1971. 17. Greenberg S: GPSS Primer. New York, John Wiley and Sons, Inc., 1972. 18. Greenes RA, Sidel VM: The use of computer mapping in health research. Health Sci Res 2:243-258, Fall 1967. 19. Gullahorn JT, Gullahorn JE: Computer model of elementary social behavior, in
Feigenbaum ES, Feldman J (eds): Computers and Thought. New York, McGraw-Hill Publishing Co., 1963. 20. Gullahorn JT, Gullahorn JE: Computer Simulation of Human Interaction in Small Groups. Proceedings, Joint Computer Conference, Spring 1964. M, Stoll CS: Simulation and Gaming in Social Science. New York, The Free Press, 1972. 22. Loehlin J: Computer Models of Personality. New York, Random House, 1968. 23. McLeod J: International aspects, in Sackman H, Borko H (eds): Computers and the Problems of Society. Montvale, NJ, AFIPS Press, 1972. 24. Newell A: Learning, Generality and Problem-Solving. Santa Monica, The RAND Corp., Memorandum Rm-3285-1-PR, February 1963. 25. Newell A, Simon H: Simulation of human thought, in Dutton J, Starbuck W (eds): Computer Simulation of Human Behavior. New York, John Wiley and Sons, Inc., 21. Inbar
1971. 26. Pool
I, Abelson RP, Popkin SL: Candidates, Issues and Strategies. Cambridge, Mass., MIT Press, 1965. 27. Sachman H, Borko H (eds): Computers and the Problems of Society. Montvale, NJ, AFIPS Press, 1972. 28. Schneider BR: Travels in 29. 30.
Computerland
or,
Incompatibilities
and Interfaces.
Reading, Mass., Addison-Wesley Publishing Co., 1974. Skinner BF: The Technology of Teaching. New York, Appleton-Century-Crofts, 1968. Stockton E: Planning Implementation Strategies for Environmental Programs Using 189
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
System Dynamics, unpublished paper, National Sanitation Foundation, 1972. E: Application of Systems Simulation to Environmental Quality Management, unpublished paper, National Sanitation Foundation, 1972. Tompkins SS, Messick S (eds): Computer Simulations of Personality: Frontiers of Psychological Theory. New York, John Wiley and Sons, Inc., 1962. Waldorf W, Coleman J: Analysis and simulation of reference group process, in Dutton J, Starbuck W (eds): Computer Simulation of Human Behavior. New York, John Wiley and Sons, Inc., 1971. Wessel MR: Freedom’s edge: The computer threat to society. Reading, Mass., Addison-Wesley Publishing Co., 1974.
31. Stockton 32.
33.
34.
190
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
Pontiac, Michigan Three levels of computer simulation of individual and aggregate human behavior are examined and applications to health education are reviewed General implications of this use of computers for the training and future role of public health educators are suggested .
The intent of this discussion is to introduce an area of computer usage which may have immediate application to the practice of public health education, but which certainly will have professional, ethical, and practical implications for the future of public health education and public health educators. It is not the purpose of this discussion to review what we already know about computers: how we use them; what benefits they have occasioned; or what threats they
pose.7,28,34 BASIS FOR CONCERN WITH COMPUTER SIMULATIONS
A basic question which concerns behavioral scientists engaged in research or attempting to apply research findings is how people make decisions. Health education programs are often designed to encourage decisions that will lead to behavior that will enhance health, but too often they appear to produce no effect that is measurable or generalizable from one population to another. Several models which are creating interest in health education programming include the Health Belief Model, Diffusion models, and Locus of Control models.3,11>14>1’ Each of these begins with different, but sometimes overlapping, assumptions; and each provides a partial explanation of how people structure decisions. Yet none is sufficient and necessary for the prediction of behavior. Similar problems abound with other models which have not been cited. Thus, the scientist or practitioner must structure his own decision about which one to use from among the competing theories. Once a model or combination of models has been chosen, the problem shifts to one of testing; and with testing, the concern with separating the trivial influences from the non-trivial in evaluating results. It is simplistic to state that there is almost an infinite number 181
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
of variables which could or should be considered; but no matter how obvious, the real problem will not disappear. This leads to the two fundamental reasons for concern with computer models and simulation of human behavior: (1) simulation permits experimentation with partial human systems without the ethical quandary of unexpected, undesirable, or morally questionable effects on people; (2) models can be structured to embody one or more theoretical concepts, and in more easily controlled situations, test the assumptions and dynamics of the concepts more exhaustively than is possible in real human systems. DEFINITION AND EXAMPLES OF SIMULATION
To simulate
an
object,
person,
or
a
process,
according
to
Merrian-Webster, is &dquo;to effect the appearance of’ that object, person, or process. A simulation, then, is a representation of something else. A photograph could be considered a simulation; Einstein’s equation, E = me 2, is a simulation of the process through which matter and energy change form. The definition of simulation to be used here is based on McLeod, who differentiates between a computer model and the assumptions, theory, program, and data needed to implement the model on the computer.23 The simulation is the dynamic process of carrying out experiments with the model on the computer. Sophisticated computer models of some behavior already are widely used. The most familiar is the election model .26 First developed in 1960 to estimate the effect of religious beliefs on the outcome of the presidential election, the basic model has been refined to the point that it can make accurate predictions of election results from small samples of votes within minutes of their tabulation. A less familiar use of simulation devolves from queuing theory and applications to industrial processes. A high level computer language, General Purpose Simulation System (GPSS), could be used to study and plan medical care delivery systems.l7 For example, an outpatient clinic may be defined by the elements of its facilities, staff, tasks performed, time needed for each task, and the pattern of utilization by patients. Varying these elements in simulation will yield data for choosing among alternative staffing patterns, patient flow, physical layout, and other configurations in the organization of the clinic. Such computer use does not, however, address the decisions of people to attend the clinic; whether they will enter the system early in the course of a disease episode; or if they will appear for regular health maintenance. The remainder of this discussion will focus on several computer models and simulations which do attempt to address these issues. The choice of simulations is entirely personal. They are 182
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
Figure l. Memory structure of permission of the publisher. not the
they
HOMUNCULUS Source: Loehlin, 1968. Reprinted by
only models which could be examined, nor is it claimed the most representative or the best models.
that
are
THREE COMPUTER MODELS ADAPTABLE TO HEALTH INTERESTS
The three computer models which will be described represent a continuum from intra-individual decision-making to community to society at large. Results of various simulations and possible implications for public health education will be discussed.
Individual and Interpersonal Analysis The empirical theories of George Homans are embodied in a computer model of individual personality structure called HOMUNCULUS.19,20,22 Developed by John and Jeanne Gullahorn, HOMUNCULUS is a sophisticated attempt to explore the structure of personality and the dynamics of interpersonal interaction. 183
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
of the model is the memory structure (Figure 1) which is composed a series of lists interconnected on an elaborate grid. The grid resembles a decision tree, of which it is a more elegant variety called an EPAM (Elementary Perceiving and Monitoring) Network.21,25 An EPAM network is &dquo;built&dquo; by a special computer program which is capable of discriminating between two or more discrete elements. It then constructs the network by devising classes of elements and pathways within and between classes. An additional feature of the EPAM network is that classes are stacked, or arranged hierarchically. The memory structure of HOMUNCULUS is composed of three main classes or lists: identification, activities, and images. The identification list contains fairly generalized information about the personality being modeled. Included are a name, age, sex, verbal and mathematical skills, special talents, and any other traits which identify the individual. The basic structure may be altered considerably by varying the information on this list. Past behavior, actions taken by the individual, actions of others directed toward the individual, and those observed to happen to others are stored on the activities list. Each action and class of action has associated with it a value which expresses how rewarding those actions have been to HOMUNCULUS. The images list stores detailed information about the environment in which HOMUNCULUS lives. Concepts about the surroundings, of groups to which the individual belongs, and a composite &dquo;self-concept&dquo; form the major divisions of the list. Within each of these major lists are sublists and sub-sublists. At each of the increasingly circumscribed levels, there are numerous cross connections between the details at that level, at other levels, and on other lists. HOMUNCULUS operates by accepting a report of an activity by someone or something in its sphere of knowledge. This may be done by arbitrarily feeding it an action (e.g., &dquo;Your boss has just been taken to the hospital with a heart attack&dquo;), or by allowing a second model of HOMUNCULUS to generate the activity (e.g., HOMUNCULUS &dquo;A&dquo; is Sam and HOMUNCULUS &dquo;B&dquo; is Joe, members of a volunteer committee planning an anti-smoking program). The subsequent responses during the simulated interaction are based on a calculation of the advantages and disadvantages of each possible alternative action the individuals might take. The value associated with each action alternative also permits expressions of emotion in conjunction with the action chosen. In one simulation, acute animosity was generated between two different models, although they began with a neutral stance. Variations in behavior are possible as the simulation builds a record of- past successful choices in similar situations. This record of past
The
core
of
184
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
behavior also permits generalization of behaviors to handle new situations or new people in the environment. The major value of this computer model to public health education is the generalized personality structure. Unlike other models of individual behavior, this one is not based on a detailed analysis of one person or a small group of professionals, but represents a more general form of personality operating on the basis of enlightened rationality,I ,8 , 9,13,11,33 Thus, various types of personalities can be superimposed on the basic model and tested for correspondence with empirical observation. As has been mentioned, two models with no particular emotional set developed signs of &dquo;hatred.&dquo; In another simulation, the Gullahorns interacted a model they termed &dquo;cynical&dquo; with an &dquo;eternal optomist.&dquo; The result was frustration, anger, and a breakdown of communication between the two models.20 With modification to include beliefs, past experience, and health knowledge, similar simulations could be run. One such simulation might focus on fear arousal, and using various postulated personality types, study the differential effects of fear arousal. Results might lead to a more precise definition of &dquo;low,&dquo; &dquo;medium,&dquo; and &dquo;high&dquo; fear arousal. Alternatively, simulations might test the Health Belief Model and develop data on how &dquo;cues&dquo; to initiate health action are
generated.3 A further
of HOMUNCULUS is in the study of small appropriate modification, three or more models can be interacted and the dynamics of the process studied. If individual personality structures can be imposed on HOMUNCULUS,
potential
use
group behavior. With
then the members of real committees
or task groups could be studied. Potential process and task problems could be studied before any actual meetings and alternative means of intervening could be tested. This application has disturbing implications, however, since there would be no guarantee that a new, more manipulative era would not
be the result.
Community and Mass Communication Applications In expanding the scope of the model beyond the small group level,
a
referendum simulation derives from the election model discussed above., It is an attempt to develop more predictive ability using local information and is loosely based on the &dquo;two-step&dquo; flow of information theory with elements from the work of Hovland, the Sherifs, Heider, and Festinger. The base of the model is an actual random sample of the community under study. Approximately 500 persons are represented in the memory storage by the following information: demographic characteristics, attitudes and predisposing experiences concerning the referendum issue, frequency of exposure to various news or
community
185
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
information channels, attitudes toward persons or institutions in the community who might prove pivotal in the campaign, knowledge and acceptance of the standard arguments for and against the referendum issue, frequency of conversation about local politics, characteristics of conversational partners, initial interest in the issue, initial position on the issue, and voting history in local elections. Once this information has been gathered and stored, the simulation proceeds through two cycles which together represent a week during the referendum campaign. The cycles imitate the processes through which each individual may change his position. The first cycle in each week is exposure of the population to the communications media. The media are provided with inputs such as speeches by various individuals. A set of rules based on attitude toward the source, previous knowledge of the assertions made, congeniality of the assertion in relation to predisposition, and position on the issue determine which assertions will be accepted by each of the community members. As a result of media exposure, any individual represented may change the media assertions he accepts, his attitudes toward sources, his probability of future media exposure, and his interest or position in relation to messages received. The next cycle is to allow individuals to interact with each other. The potentiality of conversation is based on a match of demographic characteristics and of a common location for usual political discussions. Rate of conversation is based on the level of interest. As in the media cycle, assertions are made and accepted or rejected on the basis of the compatibility of conversation partners, previous knowledge or acceptance of the assertions, and position on the issue. The total simulation is composed of several weeks of campaigning culminating in a community vote. The model permits an active campaign or a dull one. The voting outcome is determined by the changes occasioned in the community and the numbers who participate in the voting. Although this model has not been fully tested in a real campaign, it holds much promise for public health education as the following example illustrates. A majority vote for a referendum issue is replaced by evaluation techniques to measure a decrease in automobile accidents. Then detailed plans for a media campaign to encourage seat belt use are devised as input during the simulation. Several different plans could be developed using various mixes of media messages. The simulation then could test which mix proved to be more effective in generating change, or it could show that mass media approaches would never produce measurable changes. The experiments might provide information on which messages prove more effective in reaching specific target audiences; and they might allow plans for an actual program on seat belt use to maximize expenditure of limited resources. 186
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
These are all issues public health education attempts to with limited precision.
answer now
Environmental and Population Applications The final model to be discussed considers the world as a whole. Drawing on his work with engineering systems, Jay Forrester of M.LT. developed the concept of continuous feedback loops. Applying this concept to ever larger systems, Forrester and his colleagues have described a model of the world based on the interaction of four
interlocking chains.’6 According to this model, the quality and quantity of life on this planet can be described by the interaction of loops representing use, and environmental pollution. Each of these four areas are connected to each other and to smaller loops within each major division. Population size, for example, is dependent on the interaction of the birth and death rates as well as the amount of food generated. Food, in turn, is dependent on the amount of capital expended in the agricultural field as well as the effects of pollution. Pollution depends on the way capital is expended and on the total size of the population. A community based group in Michigan has modified this model for use by its task forces in the study of regional problems of air and water pollution and solid waste disposal.30,31 Real data which reflect the actual conditions in the region have been gathered to form the base of the computer simulation. Using input from members of task forces, alternative measures to improve environmental quality are developed and goals set. The detailed programs thus generated are fed into the computer, which calculates the effects of these programs over
population, food supply, capital
a
twenty-year period.
During the
course of these experiments, non-obvious alternatives have been discovered and then explored in subsequent simulations. By these means a detailed proposal for solid waste disposal has been developed for implementation by local governmental agencies. The proposal also includes plans for an annual evaluation to monitor the progress of the implementation, to discover any problems or side effects, and to propose corrective measures to keep the program on
target. SOME IMPLICATIONS FOR PUBLIC HEALTH EDUCATION
Granting that computer models and simulations will continue to be developed, and that they will be used increasingly as the models become more sophisticated, there are many non-trivial implications for public health education which might be discussed, only two of which will be - examined here. 187
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
Training of future health educators will undoubtedly need to change. What training a professional has necessarily impacts on the kind of role that professional will assume in his practice. A world which requires elaborate preparation in mathematical theory-building, computer programming, and more sophisticated methods of analysis of communities will require much restructuring of the public health education curriculum as we now know it. The question is whether the expertise now exists in our academic programs to provide these new areas of study. Even if public health education programs begin to move toward these areas, there is the question of whether a new role of public health educator as community simulation expert will be accepted by other health professionals. Recent attempts to rid the profession of old expectations of public health educators have met with resistance. Too many health professionals still cast public health educators into technician molds of pamphlet-producer, resident audiovisual specialist, annual report writer, or health fair coordinator. An alternative, of course, is to encourage others to become the community health simulation experts and to apply the materials and research in our practice. If we choose this alternative, must be prepared to deal with other consequences, the most basic of which is an ethical concern. The question now is not if the technology will exist to manipulate human behavior, but how soon. Before we reach this new age of social engineering, we must consider how we will interface with the constructors of human system models.7> 2?,28 Otherwise, who will insure that human behavior and particularly health behavior computer models include assurances of privacy for the individual, protection against even subtle coercion, or the option of saying &dquo;no&dquo; to even the most personally hazardous health decision? The issue for public health educators is not whether we participate in modeling and simulations but will we be involved soon enough to influence the basis on which models are developed and the assumptions about the dignity of the persons to be affected. Public health educators have long been advocates for voluntary decision making and behavioral change. It is becoming imperative that we act on our ethical premise. We must open ourselves to computer uses such as simulation and involve ourselves in examining the benefits and costs. We must begin to train ourselves and our future colleagues. Finally, we must begin a dialogue with simulation scientists to insure the ultimate freedom of each person to decide his own health behavior. we
REFERENCES 1. Abelson
RP, Bernstein A: A computer simulation model of community referendum controversies. Public Opin Q 27:93-122, January 1963.
188
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
2. Baker W: The internal organization of computer models of cognitive behavior. Behav Sci 12 :156.161, March 1967. 3. Becker MH (ed): The health belief model and personal health behavior. Health Educ Monogr 2:Winter 1974. 4. Benfari RC: PROBE: A computer instrument for field surveys of psychiatric disorders. Arch Gen Psych 23:352-358, October 1970. 5. Bernstein AL: Innovation in educational research: Implications for school health. J Sch Health 39:385-391 June, 1969. 6. Binder S: Computers in public health. Clin Pharm and Therapeutics 8:162-169, January-February 1967. 7. Brooks FP: The Mythical Man-Month. Reading, Mass., Addison-Wesley Publishing Co., 1975. 8. Colby KM: Computer simulation of changes in personal belief systems. Behav Sci 12:248-253, May 1967. 9. Colby KM, Watt J, Gilbert JG: A computer method of psychotherapy: Preliminary communication. J Nerv Ment Dis 142:148-152, 1966. 10. Computers in public health. W.H.O. Chron 23:556-563, December, 1969. 11. Costello RM, Manders KR: Locus of control and alcoholism. Br J Addict 69:11-17, March 1974. 12. Dowell LJ: The relationship between knowledge and practice. J Educ Res 62:201-205, May 1969. 13. Dutton J, Starbuck W (eds): Computer Simulation of Human Behavior. New York, John Wiley and Sons, Inc., 1971. 14. Etzioni A, Etzioni-Halevy E (eds): Social Change: Sources, Patterns, and Consequences, second edition. New York, Basic Books, 1973. 15. Felton B, Kahana E: Adjustment and situationally-bound locus of control among institutionalized aged. J Geront 29:295-301, May 1974. 16. Forrester J: World Dynamics. Cambridge, Mass., Wright-Allen, Inc., 1971. 17. Greenberg S: GPSS Primer. New York, John Wiley and Sons, Inc., 1972. 18. Greenes RA, Sidel VM: The use of computer mapping in health research. Health Sci Res 2:243-258, Fall 1967. 19. Gullahorn JT, Gullahorn JE: Computer model of elementary social behavior, in
Feigenbaum ES, Feldman J (eds): Computers and Thought. New York, McGraw-Hill Publishing Co., 1963. 20. Gullahorn JT, Gullahorn JE: Computer Simulation of Human Interaction in Small Groups. Proceedings, Joint Computer Conference, Spring 1964. M, Stoll CS: Simulation and Gaming in Social Science. New York, The Free Press, 1972. 22. Loehlin J: Computer Models of Personality. New York, Random House, 1968. 23. McLeod J: International aspects, in Sackman H, Borko H (eds): Computers and the Problems of Society. Montvale, NJ, AFIPS Press, 1972. 24. Newell A: Learning, Generality and Problem-Solving. Santa Monica, The RAND Corp., Memorandum Rm-3285-1-PR, February 1963. 25. Newell A, Simon H: Simulation of human thought, in Dutton J, Starbuck W (eds): Computer Simulation of Human Behavior. New York, John Wiley and Sons, Inc., 21. Inbar
1971. 26. Pool
I, Abelson RP, Popkin SL: Candidates, Issues and Strategies. Cambridge, Mass., MIT Press, 1965. 27. Sachman H, Borko H (eds): Computers and the Problems of Society. Montvale, NJ, AFIPS Press, 1972. 28. Schneider BR: Travels in 29. 30.
Computerland
or,
Incompatibilities
and Interfaces.
Reading, Mass., Addison-Wesley Publishing Co., 1974. Skinner BF: The Technology of Teaching. New York, Appleton-Century-Crofts, 1968. Stockton E: Planning Implementation Strategies for Environmental Programs Using 189
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
System Dynamics, unpublished paper, National Sanitation Foundation, 1972. E: Application of Systems Simulation to Environmental Quality Management, unpublished paper, National Sanitation Foundation, 1972. Tompkins SS, Messick S (eds): Computer Simulations of Personality: Frontiers of Psychological Theory. New York, John Wiley and Sons, Inc., 1962. Waldorf W, Coleman J: Analysis and simulation of reference group process, in Dutton J, Starbuck W (eds): Computer Simulation of Human Behavior. New York, John Wiley and Sons, Inc., 1971. Wessel MR: Freedom’s edge: The computer threat to society. Reading, Mass., Addison-Wesley Publishing Co., 1974.
31. Stockton 32.
33.
34.
190
Downloaded from heb.sagepub.com at Purdue University on June 18, 2015
