harnad@phoenix.Princeton.EDU (Stevan Harnad) (12/05/90)
PSYCOLOQUY Tue, 4 Dec 90 Volume 1 : Issue 16 Assistant Professor, Swarthmore College Cognitive Psychologist, Bureau of Labor Statistics Cognitive Psychology, Georgia Tech Assistant Professor, Mercer University Distinguished Professorship, Ball State University Lecturer, University of British Columbia ---------------------------------------------------------------------- From: "NAME \"Jeanne Marecek\"" <MARECEK%campus.swarthmore.edu@pucc> Subject: Assistant Professor, Swarthmore College Swarthmore College anticipates a one-year appointment at the beginning Assistant Professor level in 1991-92 in clinical/abnormal psychology. Applications should include a vita, 3 recommendations, a statement of teaching and research interests, and selected re- or preprints. Address correspondence to Search Committee, Department of Psychology, Swarthmore College, 500 College Ave., Swarthmore, PA 19081-1397. Women and minority candidates are especially encouraged to apply. ------------------------------ From: Jonathan Vaughan <jv10+@andrew.cmu.edu> Subject: Cognitive Psychologist, Bureau of Labor Statistics POSITION AVAILABLE FOR PSYCHOLOGIST WITH EXPERIENCE IN COGNITIVE MODELLING, AI, AND/OR EXPERT SYSTEMS. The Collection Procedures Research Lab (CPRL) of the Bureau of Labor Statistics, Dept. of Labor is seeking a research psychologist with experience in cognitive modelling, AI, and/or expert systems. The CPRL is a center for multidisciplinary research on both theoretical and applied issues concerning written and oral communications that relate to survey data collection. The position is career-conditional, with a starting salary in the range of $48,592-$63,172 (GS-14) depending on experience. This person would have considerable latitude in determining his/her own program of research within the broad interests of the Bureau. Activities might involve the modelling of various aspects of the question-answering process including, for example, autobiographical retrieval and differences between self and proxy respondent knowledge representations. Opportunities also exist to design and program interactive interviewer training systems. If you are interested, please contact: Marie van Mellis-Wright, Ph.d. Bureau of Labor Statistics GAO Building, Rm.2126 441 G St. N.W. Washington, DC 20212 (202) 523-3605 Please do not respond by Email. ------------------------------ From: billman%pravda@gatech.edu (Dorrit Billman) Subject: Cognitive Psychology, Georgia Tech The School of Psychology at the Georgia Institute of Technology is searching for faculty in cognitive psychology as part of our interdisciplinary thrust in cognitive science. We seek candidates with vigorous research programs in any area of cognitive psychology; Special consideration will be given to candidates with interest in interdisciplinary collaboration and with research in learning, reasoning, problem solving, knowledge representation, language, or vision. Level of appointment is open and the position(s) would begin in the fall of 1991. Responsibilities include maintenance of strong programmatic research, supervision of graduate student research, and classroom instruction. Georgia Tech is an Equal Opportunity/Affirmative Action Employer and a member institution of the University System of Georgia. To apply, send vita and references to: Dorrit Billman, School of Psychology, Georgia Institute of Technology, Atlanta, GA 30332. ------------------------------ From: Frank Dane <FDANE@uga.cc.uga.edu> Subject: Assistant Professor, Mercer University MERCER UNIVERSITY: Tenure-track Assistant Professor in Cognitive/Developmental/Experimental Psychology beginning August 1991. Ph.D. required, prior teaching experience preferred. Courses taught may include Introductory, Developmental, Cognitive, Sensation & Perception, Research Methods, or other courses in the individual's specialty area to make a total load of 8 course equivalents over 3 quarters. Faculty in the department are committed to excellence in teaching and maintaining an active research program that engages undergraduates. Applicants should submit a cover letter describing their research and teaching interests, a copy of their vita, preprints or reprints, and three letters of recommendation to: Dr. Francis C. Dane, Chair, Psychology Department, Mercer University, Macon, Georgia, 31207. Review of applications will begin February 20 and continue until the position is filled. Mercer University is an Equal Opportunity/Affirmative Action Employer that encourages applications from women and minority members. ------------------------------ From: 00MEKITE%BSUVAX1.BITNET@pucc Subject: Distinguished Professorship, Ball State University DISTINGUISHED PROFESSORSHIP SOCIAL AND BEHAVIORAL SCIENCES The Department of Psychological Science and the Honors College of Ball State University invite nominations or applications for the Reed Voran Honors Distinguished Professor in Social and Behavioral Sciences. We are seeking a person with a sincere interest in undergraduate teaching, a willingness to participate in the programs of the Honors College, and a distinguished program of research in any area of psychology. The particular specialty area is open and applications form persons with interdisciplinary interests are welcomed. The successful applicant will be expected to participate in the undergraduate teaching program of the Psychological Science department and the Honors College, and may be involved in the graduate program if desired. We expect that the Distinguished Professor will be a role model for other faculty in both teaching and research. The appointment carries salary, teaching load, and research support commensurate with such a prestigious position. Applications or nominations should be send to David Hines, Chairperson, Department of Psychological Science, Ball State University, Muncie, IN 47306. Inquires are also welcome (317) 285-1690 or 00MEKITE@BSUVAX1. Formal review of the applications will begin on December 15,1990 and continue until the position is filled. ------------------------------ From: Daniel_Perlman@mtsg.ubc.ca Subject: Lecturer, University of British Columbia The University of British Columbia School of Family and Nutritional Sciences 1991-92 Sessional Lecturer FAMILY SCIENCE The School of Family and Nutritional Sciences at the University of British Columbia requires a full-time, sessional lecturer to teach in the area of Family Science. The candidate will be expected to teach six 3.0 credit courses (e.g., courses having three 50 minute lectures per week for 13 weeks), especially likely to include the Contemporary North American Family in Societal Context, Family Resource Management, Marital and Family Interaction in North America, and Family Research plus some of the following: Families in the Canadian Economy, Introduction to Family Science, Economic Roles of Women, the Development of Relationships, and Housing for the Family. The Appointment will be from September 1, 1991 until May 31, 1992. Salary: $30,000. In accordance with Canadian Immigration requirements, priority will be given to Canadian citizens and permanent residents. The University of British Columbia is committed to the federal government's equity program and encourages applications from all qualified individuals. Preference will be given to candidates with the following qualifications: a Ph.D. plus experience teaching Family Science and Human Development courses. Please submit a resume, including the names of three references who may be contacted. Applications will be reviewed on or about MARCH 15TH, 1991, but applications will be accepted until the position is filled. Inquiries and resumes should be addressed to: Dr. Daniel Perlman, Director School of Family and Nutritional Sciences University of British Columbia Vancouver, Canada V6T 1W5 Phone: (604) 228-6518 Fax: (604) 228-5143 BitNet: USERHDP@UBCMTSG ------------------------------ End of PSYCOLOQUY Digest ******************************
harnad@phoenix.Princeton.EDU (Stevan Harnad) (12/05/90)
PSYCOLOQUY Tue, 4 Dec 90 Volume 1 : Issue 16 Paper: Optimal Utilities, Gordon Becker ---------------------------------------------------------------------- [Editor's Note: This paper has been refereed by a member of Psycoloquy's Editorial Board and has been accepted for "skywriting" discussion. All subsequent discussion elicited on this topic will appear under the heading: "Optimal Utilities/Becker." Comments as well as discussion papers on other topics are invited. All contributions will be refereed.] From: "Gordon Becker" <becker@zeus.unomaha.edu> Subject: Paper: Optimal Utilities The following paper argues that human suffering can be eliminated, expected utility maximized, and the basic problems of statistical decision theory avoided, by changing "tastes" (values/preferences) rather than improving "beliefs" (understanding). The paper questions the basic axioms of economics and decision theory. The de-emphasis on intellectual processes challenges the cognitive psychology paradigm, and the emphasis on feelings challenges behaviorism. This change in emphasis blinds many scientists to the logical and empirical content of the paper and their reactions are then based more on emotional and "religious" rather than scientific grounds. The author would like to get a critical discussion of the paper by social scientists who are not restricted by a narrow religious paradigm. OPTIMAL UTILITIES Gordon Becker1 University of Nebraska at Omaha Omaha Nebraska becker@unoma1.BITNET ABSTRACT This paper considers a modification of expected utility theory: Instead of considering only the decision maker's long-range utilities over entire life histories, the decision maker's current short range utilities over limited time-spans are considered, and the action space of the decision maker is expanded to include actions that allow the decision maker to modify hir current short-range utility function. It is shown that there exists an optimal utility function that maximizes expected utility, minimizes decision costs and considerably simplifies the decision making problem. Changing a current, non-optimal, short-range utility function to the optimal may, however, incur such high costs that other actions are preferable. 1. INTRODUCTION The behaviorist position predominate among current day economists and psychologists represents tastes by fixed numbers or fixed distributions that do not vary with time or as a function of the decision situation and are not under even the partial control of the decision maker. Even the procedural rationality advocated by Simon (1982) leaves the decision maker predominately (if not completely) under the control of the environment. The actions considered by the decision maker are essentially actions on the environment which only indirectly impact on the decision maker hirself. Similarly, the multiple "selfs" of the decision models of Arrow and Raynaud (1986) and of Keeney & Raiffa (1976) depend on the context or state of the world as determined by the environment rather than by the decision maker. The tastes in these models are "given" at the start of the problem and the decision maker has no control of what goes on "inside" hir skin. There have been serious objections to these formulations both as descriptions and prescriptions (Becker & McClintock, 1967; Einhorn & Hogarth, 1981; Ellsberg, 1961; Kahneman & Tversky, 1979; Sen, 1982; Simon, 1982; Slovic, Fischoff & Lichenstein, 1977), and a "call" has been made to "open the preferences" (Etzioni, 1985. see also Albanese 1982, 1985, 1987). This "call" has been largely to repair the descriptive flaws in the theory. Although the emphasis in this paper is on the prescriptive side, it is, like Etzioni and Albanese, based on the belief that preferences are modifiable and under at least the partial control of the decision maker. 2. CHANGING PREFERENCES Savage (1954) and von Neumann and Morgenstern (1944) used second-order meta-preferences as the basis for their models. However, the utilities derived in most experimental tests are based on observable, first-order preferences. Thus the term "utility" is used in different senses by decision theorists and experimenters. Stigler and Becker (1977) showed that there was an underlying set of meta-preferences that could account for the changes they found in first-order preferences. They emphasized the stability of the underlying long range utility function, but their results clearly demonstrate that first-order preferences (short range utilities) change. It is well known that in real life, our tastes (utilities) not only change but can be deliberately modified by us and by others. Cigarettes, cigars, beer, wine, liquor and other foods often taste horrible the first time(s) they are consumed ... but with social pressure and the desire to appear "sophisticated" many people continue to use these distasteful products until, with repeated use they became enjoyable (even addictive). Moreover, after acquiring a desire for such products, many people retrain themselves to dislike them when they learn that the product endangers their health. Such sequences of learning to like and then dislike a particular object or experience clearly show that utility can be changed. Decision models that deny the decision maker the opportunity to modify hir own short range utilities (first-order preferences) are not only unrealistic but, as we shall see, eliminate the "most rational" solutions to real-life decision problems. 3. OPTIMIZING TASTES Given a specific decision maker with specific utility function(s) and specific probabilities over outcomes, what change(s) in utility (preference) would increase hir expected utility? If the utility of any possible outcome were increased, that increase would raise the expected utility for the decision maker. The more likely the outcome, the greater the increase in expected utility. Thus if the rank order of the utilities of the outcomes matched the probabilities of occurrence of the outcomes, that ranking would maximize expected utility for that set of probabilities. This result suggests a new corollary to the old maxim that a rational person does not modify hir probabilities (beliefs) to match the desirability of the outcomes. We can now add a new maxim: It is rational to modify one's tastes (preferences) according to one's beliefs (expectancies); and in fact to rank preferences for outcomes according to their likelihood of occurrence. Note that the larger the increase in the desirability of the outcome, the greater the raise in the expected utility. The maximum increase, for a given utility function anchored at the high and low ends, occurs when the utility of an outcome is raised to the same desirability level as the most preferred outcome. This is true for every outcome regardless of its likelihood of occurrence. Thus the maximum expected utility is achieved when the utility of every outcome is raised to the utility of the most desired outcome. No further improvement can be made, given the original anchors. Therefore, for a given set of utility anchors, the optimal utility function is the one in which the utilities of all outcomes are equal to the utility of the outcome that was originally most desired2. When all outcomes have the same utility, another interesting result occurs that is advantageous to the decision maker - There is no need to choose among the available actions since all have the same expected utility. Accordingly, there is no need to compute expected utilities. Nor is there any need to revise expectancies about the likelihood of events or outcomes. In fact there is no need to set up a decision table, collect any information about possible actions and outcomes nor perform any of the information gathering or processing activities usually associated with rational decision making. Thus with optimal utilities, the decision maker avoids all the costs of decision making that accompany rational decisions with non-optimal utility functions. Moreover, a person with optimal utilities creates no problems for others since (s)he is always willing to take whatever action the other(s) decide to be best. No changes in the social welfare function need to be made when (s)he enters or leaves a group, since hir flat function does not change the social welfare function. People with non-optimal utility functions create problems for themselves and for others by making it necessary to engage in costly procedures to revise the social welfare function for any group they join or leave. Thus the optimal utility function not only maximizes utility for the holder, but reduces decision costs for everyone else with whom (s)he interacts. It is desirable both from an individual and societal perspective. Changes in preferences (utilities) generally require time and effort, and not all changes are feasible. The specific costs incurred in changing from one set of preferences to another depend on the individual's ability to modify hir tastes, as well as the specific changes that are to be made. Hence, although the optimal utility function is the best function a person or group can have, changing from a non-optimal to an optimal function may incur such high costs that it may not be the best course of action. Consider an ordinary mortal with such strong likes and dislikes that it would take thirty years of disciplined living, with many frustrations to change hir utilities to the optimal no- preference function, and who might die before completing the change, or so soon after completing it that the thirty year costs outweigh the gains. There is no guarantee to anyone that (s)he will succeed in achieving the optimal function even after long and strenuous effort. Given a limited lifespan and the costs of changing tastes, some people will attain more satisfaction during their life by not trying to change from non-optimal to optimal preferences. Whether one should even attempt to modify the utility function in the direction of aligning utilities with outcome probabilities or increasing appreciation for any outcome thus depends not only on the increase in expected utility but also on the cost of such effort. 4. OTHER CONSEQUENCES OF OPTIMAL UTILITIES The flat utility function that characterizes optimal utilities violates a basic axiom3 of decision theory that there be at least some consequence(s) or act(s) that is (are) preferred. This fifth axiom of Savage (1954) is appropriately referred to as the "nontriviality assumption" (Fishburn, 1975, p292) since without it, as we have seen, any decision is as good as any other in terms of expected utility or risk. However, maximum satisfaction is achieved when one "trivializes" the problem, and hence the fifth axiom of Savage and the fixed preference assumption should be rejected as "irrational" since satisfying them forces the individual to lower hir own satisfaction, increase the costs to others, unduly complicates the decision problem, unnecessarily restricts the decision maker, and generally results in the decision maker experiencing some type of "regret" that could be avoided by a less rigid, more accepting attitude than that required by these two "non-rational" assumptions. The average person's limited ability to maximize von Neumann-Morgenstern expected utility has been interpreted as evidence of man's bounded rationality (Simon 1982), but this failure may be more an indication of our theorists' unbounded ability to invent, employ, or apply needless complexities rather than indications of limited rationality on the part of real-life decision makers. If optimal solutions to decision problems do not require the high powered computing skills dictated by those models, the limitations found in our experiments are not necessarily reflections of "bounded rationality", but rather only bounded "information processing capacity". This type of limitation is much less serious than limited rationality. The fact that the optimal utility function is one in which all outcomes have equal desirability raises ethical problems regarding some of our current practices. For example, the industrial consultant today does not question the value system of hir client nor does (s)he try to get the client to "improve" hir values even though the change in utility function could increase the client's expected utility over that which would result from maximizing hir current utilities. Changing other people's values is considered poor taste in today's world and certainly requires social skills not always associated with expertise in decision theory. What should the industrial consultant do now when (s)he knows that the client does not have optimal utilities? Advertising presents an interesting paradox. On the one hand it increases the desirability of an alternative and thus raises expected utility. However, the increased "desire" for that alternative, lowers the satisfaction associated with present conditions and increases preferences for one brand over another, both of which make the utility function less optimal. The changes brought about by current advertizing practices may well be "dangerous to the mental health" of the consumer, the nation, and the world...today and for future generations. Should such practices be made illegal? Should schools teach everyone to have flat utility functions? Do people have a right to be non-optimal, and to raise their children to have non-optimal tastes? 5. THE FEASIBILITY OF ATTAINING OPTIMAL UTILITIES There is also a question about whether or not it is possible for anyone to attain an optimal (no-preference) function even after long and arduous effort. It so happens that one of the first social scientists in India, Sakyamuni4, while studying "suffering", claimed to have achieved this optimum (Walshe, 1987). He found that suffering is a psychological problem caused by a rejection of realty, by desiring that which does not exist at the moment, by wanting what you do not have, or wanting to get rid of something you do have. He claimed that had ended his own suffering by accepting and appreciating everything as it actually is. Moreover he claimed that anyone could attain that ideal state through a mental training program, "the noble eightfold path", in which the individual learns to be aware and appreciative of reality every changing moment. In the more than 2500 years since Sakymuni's discovery, millions of people in every country of the world have been trained to follow his Eightfold Path, and many claim to have attained optimal utilities. Not everyone who has begun the training claims to have achieved the optimal. And many who claim to have achieved the optimal, have not been recognized as having done so by others. Many find the training difficult and most people do not achieve complete success, but there are many that have completely freed themselves from suffering and hence there is some evidence that the optimal utility function is realizable. It should be noted that Sakymuni Buddha did not establish a religion nor claim that he was any kind of diety. In fact he insisted that his findings be tested by each person and validated by their own experience rather than accepted as revealed truths. Modern texts on personality are devoting more and more space to buddhism (Frager and Fadiman, 1984; Engler, 1990), but most western psychologists continue to ignore eastern psychologies and to equate them with religion, despite the strong praises for their scientific merit by respected western psychologists such as William James (1961), Gordon Allport (1961), and Garner Murphy (1968). In addition to those who have attained optimal utilities by following the Buddhist program, others have been recognized as having achieved such a state through religious and "transpersonal psychology" practices (Mikulas, 1987). Since many people have attained this ideal state, in every culture, in every age, including the present, optimal utilities appear to be a feasible as well as optimal human condition. 6. REFERENCES Albanese, Paul J. (1982). Toward a methodology for investigating the formation of preferences. Unpublished Ph.D. dissertation, Harvard University, Cambridge, MA. __________________ (1985). Comments on Amitai Etzioni's "Opening the preferences A socio-economic research agenda". The Journal of Behavioral Economics. 14, 207-208. ____________________ (1987). The nature of preferences: An exploration of the relationship between economics and psychology. Journal of Economic Psychology 8, 3-18. Allport, Gordon (1961). Pattern and Growth in Personality. New York: Holt, Rinehart, and Winston. Arrow, Kenneth J. & Raynaud, Herve (1986). Social Choice and Multicriterion Decision-Making, Cambridge Mass.: The MIT Press. Becker, Gordon M., Degroot, Morris H. & Marschak, Jacob (1963). Stochastic models of choice behavior. Behavioral Science, 8:41-55. Becker, Gordon M., & McClintock, Charles G. (1967). Value: Behavioral decision theory. Annual Review of Psychology, 18:239-86. Einhorn, Hillel J. & Hogarth, Robin M. (1981). Behavioral Decision Theory: Processes of Judgment and Choice. Annual Review of Psychology, 32:53-88 Ellsberg, Daniel (1964). Risk, ambiguity, and the Savage axioms. Quarterly Journal of Economics. 75:643-49. Engler, B. (1990). Personality Theories. Boston: Houghton Mifflin Co. Etzioni, Amitai. (1985). Opening the preferences: A socio-economic research agenda. The Journal of Behavioral Economics. 14, 183-205. Frager, R. & Fadiman, J. (1984). Personality and Personal Growth. New York: Harper and Row. Fishburn, Peter C. (1975). A theory of subjective expected utility with vague preferences. Theory and Decision, 6:287-310. James, W. (1961). The varieties of religious experience. New York: Crowell-Collier. Hahn, Frank, (1982). On some difficulties of the utilitarian economist, pp 187-198. In Sen, Amartya & Williams, Bernard, (1982). Utilitarianism and Beyond, Cambridge: Cambridge University Press. Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47:263-91. Keeney, Ralph L. and Raiffa, Howard. (1976). Decisions with multiple objectives: Preferences and value tradeoffs. New York: John Wiley and Sons. Mikulas, William L. (1987). The Way Beyond: An overview of spiritual practices. Wheaton IL: The Theosophical Publishing House. Murphy, G. and Murphy, L. (1968). Asian Psychology. New York: Basic Books. Ramsey, Frank, P. (1931). The Foundations of Mathematics and Other Logical Essays, New York: Harcourt, Brace and Co. Savage, Leonard J. (1954). The Foundation of Statistics, New York: John Wiley. Sen, Amartya (1974). Some debates in capital theory. Economica. 41:328-35. ___________ (1982). Choice, Welfare and Measurement, Oxford: Basil Blackwell Publisher. Simon, Herbert A, (1982). Models of Bounded Rationality. Volume 2. Behavioral Economics and Business Organization, Cambridge Mass: The MIT Press. Slovic, Paul, Fischoff, Baruch & Lichtenstein, Sarah, (1977). Behavioral Decision Theory. Annual Review of Psychology, 28:1-39. Stigler, George J. and Gary S. Becker. (1977). De gustibus non est disputandum. American Economic Review 67:76-90. von Neumann, John, & Morgenstern, Oskar. (1944). Theory of Games and Economic Behavior, Princeton: Princeton University Press. Walshe, Maurice (1987). Thus Have I Heard. Lom Publication Society. 7. FOOTNOTES 1. I am grateful to Steve Evans for his constant encouragement and critical, good natured help. This research was supported in part by the University of Nebraska at Omaha. An earlier version of the paper was presented July 1988 at the Behavioral Economics Convention, San Diego CA. 2. The "satisfaction" for the individual would also increase if the anchors were higher on an absolute scale but consideration of such a change requires psychometric considerations beyound the scope of this paper. 3. Frank Ramsey (1931), John von Neumann and Oskar Morgenstern (1944), and Leonard J. Savage, 1954) assumed not only that the decision maker had a fixed set of preferences, or "utilities", that reflect the "tastes" of the individual, but also that these fixed constants satisfied the strong transitivity requirements dictated by the Weak Axiom of Revealed Preferences (Hahn, 1982; Sen, 1982). 4. The conversation between Sakyamuni and Sabhuti (Sen, 1974) is of questionable origin and should not be used to question Buddha's credentials as a scientist (or economist). "Buddha's" closing statement, as reported by Sen, "And do not think of the fruit of action. Fare forward." has the flavour of the present paper, and the hint of an Oxford-Bostonian accent. ------------------------------ PSYCOLOQUY is sponsored by the Science Directorate of the American Psychological Association (202) 955-7653 Co-Editors: (scientific discussion) (professional/clinical discussion) Stevan Harnad Perry London, Dean, Cary Cherniss (Assoc Ed.) Psychology Department Graduate School of Applied Graduate School of Applied Princeton University and Professional Psychology and Professional Psychology Rutgers University Rutgers University Assistant Editors: Malcolm Bauer John Pizutelli Psychology Department Psychology Department Princeton University Rutgers University End of PSYCOLOQUY Digest ******************************
harnad@phoenix.Princeton.EDU (Stevan Harnad) (12/05/90)
PSYCOLOQUY Tue, 4 Dec 90 Volume 1 : Issue 16 Paper: Monkey Friendship, Dennis R. Rasmussen ---------------------------------------------------------------------- [Editor's Note: This paper has been refereed by a member of Psycoloquy's Editorial Board and has been accepted for "skywriting" discussion. All subsequent discussion elicited on this topic will appear under the header: "Monkey Friendship/Rasmussen." Comments as well as discussion papers on other topics are invited. All contributions will be refereed.] From: "Dennis R. Rasmussen" <PIPIAPAN@vms.macc.wisc.edu> Subject: Paper: Monkey Friendship This article bears on the following areas of research: Cognition; Comparative Psychology; Deception; Demography; Ethology; Primatology; Sociobiology. I would like to solicit comments from readers who are interested in all these fields. In particular, I would value the comments of my colleagues who study cognition in nonhuman animals; I may not agree with them but wish to try to understand their views more thoroughly. Like Skinner, I think we only see the behavior of other individuals. Hence the study of thoughts must always be studies of behavior; in humans the behavior may be very complex verbal and written behavior but it is still behavior. This is the most "cognitive" research article I've ever produced; it deals with "meanings" of behavior. I hope psychologists will be struck by the immense amount of potential research on the relationships between various types of affiliation: specific combinations of "affiliation" may each have a different meaning. Primatologists, myself included, have only scraped the surface of this level of understanding of how one nonhuman and nonverbal primate communicates with another. Not only do we need to study amounts and contexts of behavior, such as grooming, but we need to study how these amounts are associated with other patterns of behavior. It is not difficult to imagine a coding scheme where combinations and intensities of dozens of forms of affiliative behavior convey messages on the type and degree of an overall affiliative relationship. I would also value the comments of colleagues who work at the interface between Psychology and Demography. Sadly this is a somewhat undeveloped area. Some of the differences I have observed in human behavior in contracepting and noncontracepting, low birth rate and high birth rate populations seem parallel to those I've observed in the rhesus groups. ------------------------------------------------------------------------------ CONTEXTUAL VARIATION IN RECIPROCITY AND COMPLEMENTARITY OF RHESUS FRIENDSHIP: COVARIATION IN MEASURES OF AFFILIATION IN A REPRODUCTIVE AND A NONREPRODUCTIVE GROUP Dennis R. Rasmussen Wisconsin Regional Primate Research Center, University of Wisconsin, and Animal Behavior Research Institute, Madison, Wisconsin 53715 U.S.A. ABSTRACT The subjects of this research were in two groups of 11 rhesus (_Macaca_ _mulatta_) matched by sex, age, and previous housing experience. Mature males in one group were vasectomized so females did not conceive. In this group there were higher rates of sexual behavior because of the females' repeated nonpregnant estrous cycles and the conception and pregnancy of the mature females in the reproductive group. Seven measures of affiliation were found to assess similar aspects of affiliation across social contexts represented by subcategories of individuals within groups and the reproductive and nonreproductive groups. There was, however, considerable variation across interaction categories in both the concordance of variables within groups and interrelationships between variables across groups. Intrasexual interactions were, for example, more discordant in the nonreproductive group. Intrasexual competition in the sexually active nonreproductive group was the apparent immediate cause of the more discordant patterns of intrasexual affiliation. In our interactions, overlap in means of conveying friendship is important: If a person says they like you, nods their head when saying this, and writes expressing friendship, the consistency of these sources of information increase confidence in the friendship. Inconsistency in sources of information may make us less confident of friendship and, perhaps, suspect deception. Among rhesus there is also variation in consistency of affiliative interactions, variation that may convey information on degree of affiliation, relative status of individuals, and that may deceive. INTRODUCTION Affiliation is often defined by elaboration of the variables that are used for its assessment in primatology (O'Keeffe et al., 1983; Baker & Estep, 1985; Ehardt & Bernstein, 1987). The most universally used measures of affiliation are proximity and grooming (Rasmussen, 1984; Byrne et al., 1989). Measures of behavioral patterns associated with proximity (such as approaches, leaves and follows) and grooming presents, presents, and mounts are sometimes used to assess affiliation (Rasmussen, 1984; Chadwick-Jones, 1989). Vocalizations may also been used to assess affiliation (Biben et al., 1986; Masataka & Biben, 1987). Measures used to assess primate affiliation generally have three characteristics: They are associated with either distance reduction or proximity maintenance between individuals, they do not evoke escape responses, and they are not agonistic behavioral patterns. Affiliation is an intervening variable (MacCorquodale & Meehl, 1948; Miller, 1959; Hinde, 1985) since it is a tendency that is measured by many variables fitting the above characteristics. Few studies have focused on the degree of association between measures of affiliation; such studies help determine the utility of an intervening variable (Hinde & Datta, 1981). The ambiguous distinction between affiliative and sexual behavior provides an additional reason for analyses of the associations between measures of affiliation: sexual interactions may fit all characteristics of affiliation. Sexual behavior is functionally defined as behavior that has been associated with conception during the phylogenetic past of an organism (Scott, 1956; Tinbergen, 1965; Rasmussen, 1984). Presents and mounts have received the most theoretical and empirical attention as measures of affiliation perhaps because they do not neatly fit into any single behavioral category. This is particularly true for intrasexual presents and mounts since they are not sexual within a functional definition of sexual behavior. Intrasexual presents and mounts have been interpreted as affiliative (Reinhardt et al., 1986). Presents and mounts have also been interpreted as agonistic since they may be used during submission, appeasement, tension reduction, or the enlistment of the aid of others in aggressive interactions (Chadwick-Jones, 1989). A few studies have reported on the influence of social context on relationships between measures of affiliation. Heterosexual affiliation has, for example, been found to increase among rhesus (_Macaca_ _mulatta_) during breeding seasons (Hill, 1986). In comparisons between matched groups of rhesus and bonnet (_M_. _radiata_) macaques, heterosexual affiliation has been found more frequent in the groups with higher rates of sexual activity (Rasmussen, 1984, in prep.; Rasmussen & Goy, 1988). The age and sex of interacting individuals provide a social context that may influence their affiliative behavior. Male => male* or female => female mounts could have a different degree of association with other affiliative variables than male <=> female mounts. It is also possible that male => female mounts differ in use from female => male mounts. The validity of measures for assessment of affiliation as an intervening variable is therefore analyzed as a function of the age and sex of the individuals involved in the interactions. ------------------------------------------------------------------------------ * <=> symbolizes bidirectional behavior or a measure of the distance between a dyad. When, for example, analyses are focused on grooming of males by females and grooming of females by males this is symbolized as male <=> female grooming. => symbolizes directional behavioral interactions. Male grooms of females are therefore symbolized as male => female grooming. Interactions directed by one individual to another are referred to as directional dyads. Affiliative interactions directed by the oldest male to the oldest female therefore constitute a directional dyad. ------------------------------------------------------------------------------ The influence of social context is further evaluated by comparison of the relationships between the affiliative measures in a reproductive and a nonreproductive group. There were significantly higher rates of sexual behavior in the nonreproductive group (Rasmussen, in prep.; Rasmussen & Goy, 1988). Mature females in that group had repeated nonpregnant estrous cycles and engaged in sexual behavior during each of the cycles (Michael & Zumpe, 1988). Sexually mature females in the reproductive group conceived and were pregnant throughout the duration of observations. They therefore engaged in less sexual behavior (Wilson et al. 1982; Hill, 1986). METHODS Subjects and Housing The nonreproductive group was composed of a 5 year old vasectomized male, a 4 year old vasectomized male, an intact 2 year old male, four 4 year old females, one 3 year old female, and three 2 year old females. The reproductive group was sex and age matched. Reproduction was prevented with vasectomy since it has the minimum direct effects on the hormones and behavior of the sterilized animal (Phoenix, 1973) and it could be used on the fewest subjects. Subjects were the rhesus available who were most closely matched by age, weight and housing history (the amount of time they were housed at the Vilas Park Zoo in Madison, in group cages, as pairs, by themselves, and with their mothers). All females were nulliparous and thus did not vary in parturitional, lactational, or infant rearing experience. The groups were housed in identical indoor pens measuring 6.7 m in length, 2.5 m in width and 2.6 m in height. The pens were separated by a minimum distance of 1.2 m and in a room with two additional identical pens containing breeding rhesus groups. Lights were automatically turned on at 06.00 hours and turned off at 18.00 hours. Two frosted windows next to the pens let in ambient light. Behavioral Sampling The data were collected from January 2, 1987 until the day before the birth of the first infant in the reproductive group, June 1, 1987. I collected the observations for 5 days each week from 15.00 to 18.00 hours. Individual group members were the focus of 14 min sampling sessions. Focal subjects were selected sequentially from a list of all individuals in both groups. Focal sampling (J. Altmann, 1974) was used for the variables requiring constant monitoring of the subject. Concurrent samples (Hausfater, 1974; Chapais, 1986) were collected on all occurrences of variables (Martin & Bateson, 1986) that could be simultaneously observed for all group members. Data were recorded on a Tandy 102 computer. An auditory cue preceded 2 min intervals by 15 sec. to start focus of attention on the variables sampled instantaneously. Data reduction and analysis were conducted with SPSS/PC+ (Norusis, 1988a, 1988b) and UNIX. The analyses are based on 2445 2 min interval samples collected during 14 min sampling sessions. Behavioral Variables: Sampling Method The unit of observation and the method used to sample each variable are summarized in Table 1. ------------------------------------------------------------------------------- Table 1. Variables, Observation Method and Sampling System ------------------------------------------------------------------------------- Variable: Unit of Observation Sampling Method ------------------------------------------------------------------------------- 1. Nearest Neighbor Distance individual-focal instantaneous 2. Close Distance to Nearest individual-focal instantaneous Neighbor 3. Approaches, Leaves & individual-focal frequency Follows (ALF) 4. Grooming all individuals 1/0 5. Grooming Presents individual-focal frequency 6. Presents individual-focal frequency 7. Mounts all individuals frequency ------------------------------------------------------------------------------- Analytic Strategy Several methods are used to describe and analyze the relationships between measures of affiliation and differences in these relationships between the two groups. The methods are first applied to all group members to determine broad patterns. Patterns in affiliative interactions between subpopulations of individuals in each group are then conducted. For example, mature female => mature male interactions are analyzed as a subpopulation of interactions between all group members. As in Fisher's protected t-test, tests of significance were not conducted on subcategories unless the tests conducted on all dyads, and on the immediately higher category of interaction, were significant (Cohen & Cohen, 1983; Rasmussen, 1984). Statistical Control of Days Together before Group Formation Some pairs of individuals spent more time together than others before group formation. Differential amounts of time spent together before observation could alter the relationships between the variables. For example, individuals who spent more time together before observations might engage in less sexual behavior but groom more often (Takahata, 1982a, 1982b). Groups of nonhuman primates cannot yet be as closely matched as, say, groups of inbred rodents. Previous studies in which matched groups are compared have seldom used subjects as closely matched as those in this project, matching made possible by the large population of rhesus maintained by the Wisconsin Regional Primate Center. Previous studies (Boccia et al., 1982; Rasmussen, 1984; Boccia, 1989) have also not attempted the statistical control of differences possible with the methods developed for this purpose. Here influences of days dyads spent together before group formation were statistically controlled by the regression of the affiliative directional dyadic interactions, as assessed by each variable, on this nuisance variable. Transformations were used, when appropriate, to normalize residuals from regression. Linear and quadratic fits to Days Together were tried for every dependent variable. For all dependent variables, except Grooming, the quadratic aspect of Days Together did not appreciably increase R; only linear fits were therefore used. 1/0 Grooming Rate was regressed on both the linear and quadratic aspects of Days Together. The regession of the variables on Days Together were not tested for statistical significance since control of a nuisance variable is not predicated on either the magnitude or significance of the influence (Cohen & Cohen, 1983). All descriptive statistics and analyses are based on the residualized variables. SPEARMAN CORRELATIONS BETWEEN VARIABLES: Spearman rank order correlations (Siegel & Castellan, 1988) were calculated between the variables in each group. These correlations are used solely for description of the way directional dyads are ranked by pairs of affiliative variables; they are not tested for statistical significance. If, for example, monkey "A" Grooming Presents at a high rate to monkey "B" does it also tend to Approach, Leave or Follow monkey "B" at a high rate? Spearman correlations were used so that curvilinear relationships between the variables would not influence the magnitude of their association. There were 21 correlations between variables (the half matrix minus the diagonal). Description of linear and curvilinear fits between variables would be tedious and their complex interpretation would obscure the focus of this paper. The distance of monkey "A" to monkey "B" is necessarily the same as the distance of "B" to "A". There were therefore 55 ([11 x 10]/2) unique nondirectional dyads on which Nearest and Close Neighbor Distance could be calculated in each group. Behavioral variables are directional: Monkey "A" Mounting "B" may have a different social significance than monkey "B" Mounting "A". The behavioral variables were therefore calculated for the 110 (11 x 10) directional dyads in each group. Duplicate values of neighbor distances were matched with directional behavioral interactions so their association could be described. For example, the rate of Grooming of the 4 year old male by the 5 year old male in the nonreproductive group was paired with the same distance value used with the rate of Grooming of the 5 year old male by the 4 year old male. Although Nearest and Close Neighbor Distance are not unidirectional, correlations with these variables are included in analyses of these interactions, such as those directed by females towards males, so that the results are comparable to the other analyses. MATRIX CORRELATIONS: A Pearson correlation was calculated between the Spearman correlation matrices from the groups. This correlation between paired values of correlation coefficients is referred to here as a "matrix correlation". The matrix correlation describes the similarity of the variables' interrelationships in the two groups. Like all correlations (Cohen & Cohen, 1983), matrix correlations are not influenced by linear transformations of the coefficients in either matrix. They therefore assess similarities in relative values of correlation coefficients. The significance of the matrix correlations was determined with a Monte Carlo test based on 10000 permutations of the matrices (Dow et al., 1987). The Association between Measures of Affiliation within each Group: Distances to neighbors decrease with greater affiliation, and all the behavioral measures increase with greater affiliation. If the measures assess affiliation as a unitary intervening variable, then measures of distance to neighbor should be positively correlated; the behavioral measures of affiliation should be positively correlated, and the measures of distance to neighbor and the behavioral measures should be negatively correlated. The signs of the 21 correlations between the measures of affiliation within each group are therefore used as an initial index of whether they tend to assess the same underlying affiliative tendency. The degree of similarity in the way directional dyads were ranked by the affiliative variables is then assessed with the Kendall Coefficient of Concordance, W (Siegel & Castellan, 1988). The coefficient of concordance provides a measure of the consistency with which the variables rank directional dyadic interactions. The coefficient of concordance was calculated by ranking Nearest Neighbor Distance and Close Nearest Neighbor in descending order and the other variables in ascending order. The initial assumptions about the way variables evaluate affiliation are based on their usual interpretation in many studies of nonhuman primates (Rasmussen, 1984; Bernstein & Ehardt, 1986). Each pattern of affiliation is therefore initially interpreted as reciprocal (Hinde, 1987): if, for example, monkey "A" Grooming Presents more to monkey "B", "A" is assumed to more often Groom "B". This might be thought of as a form of a monkey "golden rule". Patterns of behavior in which monkey "A" is the actor and monkey "B" the recipient are positively associated with patterns in which monkey "A" is the recipient and monkey "B" the actor. If Monkey "A" Grooms monkey "B" more, Monkey "A" also Grooming Presents to monkey "B" more. If Monkey "A" Mounts monkey "B" more, Monkey "A" also Presents to monkey "B" more and so on. More complex, and sometimes less nice, relationships are possible and probable (Seyfarth, 1977). These initial interpretations are therefore null hypotheses about possible relationships between measures of affiliation and are referred to as the expected reciprocal direction. MEAN SQUARED CORRELATIONS: The mean strength of agreement between each measure of affiliation and all others is described with the mean of its squared correlations. Correlations between variables opposite to that expected are given a value of 0 when calculating the mean. The mean squared correlation may therefore be interpreted as the mean proportion of affiliative variation shared between dyads ranked by one variable and all others. Differences in mean squared correlations between groups are used to pinpoint the measures of affiliation that are most dissimilar in relationships with others and, therefore most sensitive to the contextual differences between groups. RESULTS The positive and significant matrix correlations for every category of interaction and significant coefficients of concordance for every category of interaction in both groups were striking outcomes of the analyses (Table 2). Since the coefficients of concordance were significant, the variables do tend to measure similar aspects of behavior across the social contexts of the two groups and across interaction categories. This suggests that affiliation is a useful intervening variable and the variables used for its measure in previous studies were correctly chosen. There were positive, strong, and significant matrix correlations for every category of interaction. The ways in which the variables assessed affiliation therefore tended to remain similar in the two groups across categories of interactions. There was, however, variation in the matrix correlations, the number of correlations in the expected reciprocal direction, the coefficients of concordance, and differences between groups. The utility of measures for the assessment of affiliation is thus partially contingent on who is interacting with whom and the group in which the interactions occur. ------------------------------------------------------------------------------ Table 2. Summary of the variables used to describe the interrelationships between the affiliative variables for each analyzed dyadic composition. ------------------------------------------------------------------------------ Category of N Matrix Concordance r/N >1 "Golden Rule" Interactions r RG NG RG NG ------------------------------------------------------------------------------ all 110 +.95* +.55* +.50* .86 21 21 all sexually mature 42 +.92* +.55* +.45* .86 19 21 all male => male 6 +.65* +.84* +.51* .86 17 21 all female => female 56 +.90* +.49* +.48* .43 19 21 mature female => female 20 +.75* +.55* +.36* .86 16 21 male <=> female 48 +.97* +.55* +.53* .67 21 21 male => female 24 +.93* +.55* +.56* .29 21 21 mature male => female 10 +.78* +.49* +.71* .00 21 18 female => male 24 +.97* +.58* +.54* .57 20 21 mature female => male 10 +.93* +.60* +.67* .33 21 21 ------------------------------------------------------------------------------ 1 r/N > Proportion mean squared correlations greater in the Reproductive Group * Statistically significant, P<.05 ------------------------------------------------------------------------------ Matrix Correlations: Interrelationships between Variables The matrix correlations were less than +.90 (Table 2) for three interaction categories; contextual differences in the two groups therefore had the greatest influence on the relationships between the affiliative variables for these categories. The smallest matrix correlation was for male => male interactions, the next smallest correlation for mature female => female interactions, and the third smallest correlation for mature male => female interactions. The contextual differences between the two groups therefore had the greatest influence on the variables' interrelationships for isosexual interactions and on interactions directed by mature males to females. Greater intrasexual competition in the more sexually active nonreproductive group seem the likely cause of the decreased similarities between the correlation matrices composed of male => male and mature female => mature female interactions. The variables in the nonreproductive group appeared to be used more frequently in agonistic contexts and to convey information on agonisitc rank. Intrasexual Presents were, for example, more frequently used for appeasement and less often for affiliation. The use of Presents for appeasement resulted in a negative correlation between male => male Grooming Presents and Presents in the nonreproductive group. Males of higher rank "requested" Grooming with Grooming Presents and were Presented to by the males from whom they "requested" Grooming. Information on relative status was therefore conveyed by the complementary direction of affiliative interactions. In contrast, Grooming Presents and Presents were strongly and positively correlated in the reproductive group: Intrasexual affiliation was therefore more reciprocal and less related to status. Reproductive males were, for example, more likely to Groom a male to whom they Grooming Presented. A similar difference, but of lower magnitude, existed between groups for mature female => female interactions. Mature male => female interactions had the third lowest matrix correlation. This was the result of the stronger concordance of affiliative variables in the nonreproductive group. The stronger concordance was paralleled with a stronger mean squared correlation for every variable in the nonreproductive group. The stronger concordance and mean squared correlations reflect the much greater reciprocity of mature male => female interactions in the nonreproductive group. Nonreproductive adult males had, for example, a strong tendency to Groom most frequently the adult females to whom they most frequently Grooming Presented whereas there was a faint tendency in the opposite direction in the reproductive group. The greater reciprocity of mature male => female interactions in the nonreproductive group may have been the result of the greater synchronization of affiliative interactions occurring between sexually interacting pairs and the more solicitous and tolerant behavior (Carpenter, 1942) of the males in those pairs. Correlations with Grooming Presents differed most between groups in the analyses conducted on the three interaction categories with the lowest matrix correlations. The social use of Grooming Presents is therefore particularly strongly influenced by contextual differences between groups. In the nonreproductive group, intrasexual Grooming Presents were directed to those with whom appeasement was not necessary, to those of lower agonistic rank. Intrasexual grooming Presents were more reciprocal in the reproductive group. Heterosexual Grooming between sexually active pairs was more reciprocal in the nonreproductive group. Grooming Presents were therefore strongly correlated with Grooming; these variables were not associated in the reproductive and less sexually active group. Depending on social context, Grooming Presents may therefore be used in either complementary or reciprocal affiliative interactions. Coefficients of Concordance There was stronger concordance in the way the variables ranked directional dyadic interactions in the reproductive group for 7 of the 10 interaction categories. Male => male interactions had the greatest difference in coefficients of concordance, mature male => female interactions had the second greatest, and the third greatest difference was in mature female => female interactions. The three interaction categories with the greatest difference in concordance were also those with smallest matrix correlations. Two of the three greatest differences in the coefficient of concordance arose for intrasexual interactions. Intrasexual competition associated with higher rates of sexual behavior is, again, the probable cause of the lower concordance between affiliative variables in the nonreproductive group. Decreased reciprocity of intrasexual interactions in the nonreproductive group is sometimes a form of intrasexual competition: one individual in the dyad may receive decreased social resources compared to the other. If, for example, a nonreproductive male more frequently Presented for Grooming to another male he was less likely to Groom that male. Greater discordance between measures of affiliation suggests less consistent and more conflicting social signals, an inconsistency that may grade into deception. In the nonreproductive group, for example, a strong tendency for female "A" to Groom "B" had close to no covariation with rate at which female "A" Grooming Presented to "B". Analyses of other patterns of behavior, such as aggression, would permit further detection of intrasexual deception in the nonreproductive group. For example, ALF rate, Grooming, and agonistic behavior appeared to be more positively associated in the nonreproductive group, a deceptive blend of affiliative and agonistic interactions: Instead of "An I'll scratch your back if you will scratch mine" pattern of interaction; those in the nonreproductive group might more closely approximate a pattern summed up as "I'll let you scratch my back and then bite yours". Sequential data analyses would help differentiate this possibility from conciliation and reconciliation (de Waal & Yoshihara, 1983). It is, of course, not impossible that conciliation and reconciliation could deceptively mask agonistic behavioral patterns. The severity of wounds could far exceed the positive benefits of conciliation and reconciliation; the total valence of interactions might be overwhelmingly negative. The second greatest difference in the coefficients of concordance between groups was for mature male => female interactions. Exactly opposite to the intrasexual interactions, there was a much stronger coefficient of concordance in the nonreproductive group. The greater concordance was paralleled by larger mean squared correlations for every variable in the nonreproductive group. As with the low matrix correlation, the greater reciprocity of mature male => female interactions appeared to cause the difference in the coefficients of concordance between the two groups. Number of Correlations in the Expected Reciprocal Direction The number of correlations in the expected reciprocal direction differed across groups and across interaction categories (Table 2). In the reproductive group all but one interaction category had all 21 correlations in the expected reciprocal direction. Only half of the interaction categories had all correlations in this direction in the nonreproductive group. As a whole, the monkey "Golden Rule" was followed in more interactions in the reproductive group than in the nonreproductive group: if, for example, monkey "A" Groomed "B" at a high rate, monkey "A" also tended to Grooming Present to "B" at a high rate. This difference between groups was absolute for intrasexual interactions. The only deviation from the trend was for mature male => female interactions; for these, all correlations were in the expected reciprocal direction in the nonreproductive group whereas 18 of 21 were in this direction in the reproductive group. Proportion of Mean Squared Correlations Greater in the Reproductive Group The three interaction categories with the lowest proportion of mean squared correlations greater in the reproductive group were all heterosexual: mature male => female, male => female, and mature female => male. The greater synchronization of affiliative interactions between sexually interacting pairs appears to be responsible for these results. One of these three interaction categories, mature male => female interactions, was found most different from the others in matrix correlations, number of correlations in the expected reciprocal direction and differences between coefficients of concordance. By all measures of relationships between the variables, mature male => female interactions were markedly influenced by contextual differences between groups: these interactions were more reciprocal in the nonreproductive and more sexually active group. CONCLUSIONS The differential concordance of the measures of affiliation suggest that partially redundant sources of information are used by monkeys to signal their affiliative tendencies towards each other and that discordance of affiliative interactions might be used to deceive. In human exchange we tend to feel more confident say, that another person means yes when they nod yes, say yes and write yes. When a person nods yes, says yes and writes no or, vice versa, then meaning becomes more difficult to interpret and we may feel deception of ourselves, or others, is possible if not probable. Analogous concordance and discordance of affiliative interactions occur in rhesus. These analyses have only concentrated on the agreement between 7 measures of affiliation. Further analyses are necessary to determine how these measures are associated with unambiguously sexual and agonistic interactions. Analyses of the variability in degree reciprocity between dyads for each variable would provide further insight into the ways patterns of affiliation are used. Such analyses will be necessary for a more thorough understanding of the subtleties in use of affiliative interactions between rhesus macaques. Further analyses are also necessary on the use of affiliative interactions, such as Grooming Presents, in status interactions. When affiliative interactions are used in status interactions they may simultaneously convey both agonistic and affiliative information. Analogously, bowing between people, and the depth of the bows, may simultaneously signal affiliation and differences in status. ACKNOWLEDGEMENT Drs. R. Goy, F. De Waal and V. Reinhardt made useful suggestions on aspects of the theory and analyses. Dr. V. Reinhardt performed the vasectomizations and helped select the animals for the study. Dr. S. Sholl wrote the communication program, COMX, used to transfer data between computers. Paul DuBois programmed my data collection method for use on Tandy 102 lap top computers. This research was supported by NIMH National Research Service Award 1 F32 MH09419-01 RERA, NIH Grant RR00167 and NSF Grant 880414. REFERENCES [available from author; deleted from email version -- ed.] ------------------------------ PSYCOLOQUY is sponsored by the Science Directorate of the American Psychological Association (202) 955-7653 Co-Editors: (scientific discussion) (professional/clinical discussion) Stevan Harnad Perry London, Dean, Cary Cherniss (Assoc Ed.) Psychology Department Graduate School of Applied Graduate School of Applied Princeton University and Professional Psychology and Professional Psychology Rutgers University Rutgers University Assistant Editors: Malcolm Bauer John Pizutelli Psychology Department Psychology Department Princeton University Rutgers University End of PSYCOLOQUY Digest ******************************
harnad@phoenix.Princeton.EDU (Stevan Harnad) (12/05/90)
PSYCOLOQUY Tue, 4 Dec 90 Volume 1 : Issue 16 ML91 Call for papers: Eighth International Machine Learning Workshop New Flier: Support for Database Activities Psycoloquy: Call for Abstracts ---------------------------------------------------------------------- From: Stevan Harnad (harnad@clarity.princeton.edu) Subject: Call for Abstracts for Psycoloquy (and sci.psychology.digest) Readers of Psycoloquy are invited to post abstracts of preprints, technical reports, forthcoming and recently published papers. Please indicate how the paper can be retrieved, either electronically or by Usmail (and where it will be or has been published, if known). If you wish to elicit "skywriting" discussion of the paper in Psycoloquy, accompany the abstract with a self-contained synopsis of the paper, the points you wish to discuss, and what areas of expertise would be relevant for potential commentators. ------------------------------------------ From: Lawrence Birnbaum <birnbaum@fido.ils.nwu.edu> Subject: ML91 Call for papers: Eighth International Machine Learning Workshop ML91 The Eighth International Workshop on Machine Learning Call for Papers The organizing committee is please to announce that ML91 will include the following workshop topics: Automated Knowledge Acquisition Computational Models of Human Learning Learning Relations Machine Learning in Engineering Automation Learning to React/in Complex Environments Constructive Induction Learning in Intelligent Information Retrieval Learning from Theory and Data Papers must be submitted to one of these workshops for consideration. The provisional deadline for submission is February 1, 1991. Papers to appear in the Proceedings must fit in 4 pages, double column format. More details about the constituent workshops, including submission procedures, contact points, and reviewing committees, will be forthcoming shortly. ML91 will be held at Northwestern University, Evanston, Illinois, USA (just north of Chicago), June 27-29, 1991. On behalf of the organizing committee, Larry Birnbaum and Gregg Collins ------------------------------ From: Fred Stollnitz <fstollni%NSF.GOV@pucc> Subject: New Flier: Support for Database Activities ====================== A N N O U N C E M E N T ===================== The Division of Instrumentation and Resources at the National Science Foundation has recently published a flier describing the possibilities of support for "Database Activities in the Biological, Behavioral, and Social Sciences." Copies of the flier (NSF 90-70) may be obtained by sending a request (including your surface mail address) to the address below. Email requests and telephone inquiries are welcome. ====================================================================== Robert J. Robbins Program Director, Database Activities Phone: (202) 357-9880 Biological, Behavioral, and Social Sciences FAX: (202) 357-7745 National Science Foundation 1880 G Street, Room 312 InterNet: rrobbins@note.nsf.gov Washington, DC 20550 BitNet: rrobbins@nsf ====================================================================== ------------------------------ PSYCOLOQUY is sponsored by the Science Directorate of the American Psychological Association (202) 955-7653 Co-Editors: (scientific discussion) (professional/clinical discussion) Stevan Harnad Perry London, Dean, Cary Cherniss (Assoc Ed.) Psychology Department Graduate School of Applied Graduate School of Applied Princeton University and Professional Psychology and Professional Psychology Rutgers University Rutgers University Assistant Editors: Malcolm Bauer John Pizutelli Psychology Department Psychology Department Princeton University Rutgers University End of PSYCOLOQUY Digest ******************************
harnad@phoenix.Princeton.EDU (Stevan Harnad) (12/05/90)
PSYCOLOQUY Tue, 4 Dec 90 Volume 1 : Issue 16 Access to the phonological lexicon (S. Cassidy) Query: Research Organizations (A. Kendall) Response to D.S. Stodolsky's "Consensus Journals" (G. Becker) ---------------------------------------------------------------------- From: steve@comp.vuw.ac.nz (Steve Cassidy) Subject: Access to the phonological lexicon A question about access to the phonological lexicon. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ As a preliminary to my question let me describe the context in which I ask it. I am involved in building a computational model of the development of word recognition. In outline, my model is similar to that described by Seymour et al. (1989): the child starts reading using a purely visual strategy (Seymour calls this Logographic) and later develops a strategy involving recoding words with letter to sounds rules and recognising them via the phonological (speaking/listening) lexicon. My approach is to try and show how each process used within the various strategies can develop from the child's existing capabilities, within the environment in which the child learns. Thus my model of visual recognition is such that I can argue that it can develop from object recognition processes and a growing knowledge of the shapes of letters. As the model acquires more vocabulary items, it can learn more about how best to structure its lexicon for reading. The model as it stands is described in Cassidy (1990a, 1990b). I am now thinking about how the reader starts to use phonological information. The standard account says that the written form is re-coded using some grapheme to phoneme procedure (analogy or non-lexical rules are examples), the resulting phonological form is then recognised using auditory word recognition procedures. The use of non-lexical rules begs the question as to where they came from. One possible solution is that the child can using letter sounds or letter names as proto-rules to do this re-coding. If this is the case then the resulting phonological form will more than likely be incorrect, but it may have about the right consonants in the right places. Analogy might offer a better candidate phonological form but is a very ill defined procedure; I do have ideas as to how it could work but the suffer similar problems of vowel ambiguity. My question concerns access to the phonological lexicon; that is, finding a candidate phonological lexical item given a partial phonological form. How much information is needed and how immune to variation is the access procedure? Which parts of the phonological representation are important -- vowels, consonants, onset, rime? Is there some particular structure, such as onset+rime, which will facilitate access? Is there any evidence that children (or adults) are able to access the lexicon given incomplete descriptions of the target word? If so, does this give any insight into the above problem? Tasks such as rhyme production and alliteration might be suggested but I can imagine another process which could achieve these tasks in a 'generate and test' manner that doesn't require access from partial descriptions. Models of spoken word recognition such as the Cohort model and the Logogen model have very little to say about the information content and structure of the phonological lexicon. In fact they see the lexicon as a collection of processing units, one per word, which sum features of the stimulus until the word is recognised. This could be seen to be equivalent to a lexicon full of descriptions of words which are matched by some process or set of processes against the stimulus. These models don't tell us much about what is in the lexical descriptions, rather they concentrate on how information is collected and matched and how `recognition units' compete for attention. The cohort model might be viewed in this way: the lexicon is a collection of word descriptions indexed under a number of features. When a stimulus arrives the first features available (be they phonetic features, phonemes syllables or whatever) are used to select a subset of the lexicon (those words indexed on that feature), as more features become available they are used to further reduce the size of the subset until there is only one candidate which can then be responded. Under this view it would be possible to use this structure to access lexical items from incomplete phonological descriptions, such as those obtained from a crude letter to sound translation. Seymour, P.H.K. and Evans, H.M. and Kinnison, S.E.C. (1989) Logographic, alphabetic and orthographic processes in early reading development. In Press, Department of Psychology, The University, Dundee, Scotland. Cassidy, S. (1990a). Early Reading Development: A computational view. Technical Report CS-TR-90/8, Dept. Computer Science, Victoria University, Wellington, New Zealand Cassidy, S. (1990b). Substitution Errors in a computer model of early reading. Paper presented to the First Australasian Cognitive Science Conference, Sydney, Australia. Steve Cassidy ========================================================== Computer Science, steve@comp.vuw.ac.nz Victoria University of Wellington, ==================== Box 600, Voice: +64 4 715328 Wellington, New Zealand. Fax: +64 4 712070 ---------------------------------------------------------- ------------------------------ From: "Arthur Kendall" <AJK%NIHCU@pucc> Subject: Query: Research Organizations As part of an effort to consider total-quality management (TQM) approaches in our agency, I am trying to identify organizations whose work is (research, policy studies, evaluation, "think tank", etc.) where there is an explicit management approach that is (humane, nonauthoritarian, professional, total-quality, democratic, participative). If you know of any such organizations, please contact me. If you know people who might have information, please forward this note to them. _________________________________________________________ Arthur J. Kendall National Security & International Affairs Division United States General Accounting Office (GAO) Washington, DC 20548 USA Commercial phone: (202) 275-8455 FTS: 275-8455 BITNET: AJK@NIHCU GAO is an oversight agency in the legislative branch. It is not the same as GSA which is an executive branch agency. *** DISCLAIMER **** This is not an official communication. Only official communications represent official findings, results, or opinions. Any opinions expressed are solely those of the sender as an individual. ------------------------------ From: "Gordon Becker" <becker@zeus.unomaha.edu> Subject: Response to D.S. Stodolsky's "Consensus Journals" The statistical procedure that Stodolsky proposes for evaluating articles and for selecting new authors would make it even more difficult than it is now to publish articles that deviate from the dominant paradigm. Current peer-review practices have been highly criticized for their rejection of new ideas. Consensus assures conservatism and Stodolsky's procedure for selecting authors and papers would give technologists conducting minor mop- up experiments more power than scientists breaking new ground. Some of us fear that we are already stifling science with the present old-boy review. We need some way to break out of conservatism, not perpetuate it, and certainly not strengthen it. Stodolsky's method also perpetuates the use of anonymity with all of its disadvantages. Not only does anonymity permit the reviewer to make false and unsupported statements, to be unnecessarily disparaging, personal and disrespectful, all with impunity; it also prevents opening or continuing a fruitful exchange and cooperative effort with helpful reviewers. The claim that reviews would be less critical and candid without anonymity, have been shown to be untrue in the many real life situations where open review has been used. The consensus journal also perpetuates the current practice of considering publication itself as an index of one's "scientific contribution" rather than considering the actual impact of the publication on science. A much better index of one's contribution is the citation index. Perhaps we should let articles remain in the electronic media until they are cited in other electronic articles sufficiently to justify publishing them in hard copy for others not actively engaged in the particular specialty of the article. Thus electronic journals would be for scientists actively engaged in the specialized area of the article and published journals would be for those in related or other areas. A basic problem with any statistical procedure such as that advocated by Stodolsky to evaluate new articles is the need for his assumption of a steady state condition. Research is designed to change the state of knowledge, to generate new ways of organizing data. Significant break-throughs deviate from the past and are generally rejected by the majority when first presented. Any evaluation method that does not consider changes in state, and that gives equal votes to all reviewers is likely to reject the really significant paper. Let's make it easier for anyone to publish a paper in the electronic media, encourage open review, and give more weight to the actual use of the article. The problem for reviewers then becomes one of verifying that the citations are actually justified and relevant rather than inflated by friends and cohorts. It might be necessary to limit the number of citations permitted in an article. The task of reviewers might then be to verify the accuracy and relevancy of citations rather than the worth of the article itself. The reviewers task would be easier and more objective task than the present one, and well suited as a task for graduate students and new professionals. Gordon Becker UNO Omaha NE 68182 becker@unoma1.BITNET ------------------------------ PSYCOLOQUY is sponsored by the Science Directorate of the American Psychological Association (202) 955-7653 Co-Editors: (scientific discussion) (professional/clinical discussion) Stevan Harnad Perry London, Dean, Cary Cherniss (Assoc Ed.) Psychology Department Graduate School of Applied Graduate School of Applied Princeton University and Professional Psychology and Professional Psychology Rutgers University Rutgers University Assistant Editors: Malcolm Bauer John Pizutelli Psychology Department Psychology Department Princeton University Rutgers University End of PSYCOLOQUY Digest ******************************