Undergraduate Education in Cognitive Science:
Current Status and Future Prospects
Report of a Planning Workshop for the National Science Foundation
May 21-23, 1993
Send comments to Executive Summary
A two-day planning workshop was held in Washington DC on May 21-23, 1993 to consider the current state of undergraduate cognitive science education and to issue recommendations for future directions. The workshop was first suggested in the final report of the 1991 planning workshop for the cognitive science initiative at the National Science Foundation. It was attended by eighteen invited participants from colleges and universities and by eight officers of the NSF and other government agencies (Appendix 1). A sample of existing cognitive science programs was surveyed by questionnaire in conjunction with the workshop.
Current Status of Undergraduate Cognitive Science Education
The growth of cognitive science as a new area of scientific research and as a way of thinking about intelligent systems has been a major intellectual event in the second half of this century. Undergraduate education in cognitive science is a critical source of future researchers in the field and of a citizenry that appreciates the research. In addition cognitive science has considerable potential as a focus of integrative general education.
As a new field that draws from several established disciplines (including psychology, computer science, linguistics, philosophy, and biology) the establishment and development of a cognitive science program involves difficult interdepartmental negotiations. In spite of the hurdles over 40 undergraduate programs have been established at leading universities and colleges. Most programs are interdepartmental, although a few departments of cognitive science have been established.
Most programs impose a mixture of interdisciplinary content requirements (courses on language, cognitive psychology, artificial intelligence, philosophy of mind, and so on) and methodological requirements (courses in logic, statistics and experimental design, computer programming, and so on). Significant variation exists in the structure and content of the major at different institutions, but there is also a significant overlap among them. Many programs are highly dependent on preexisting departmental courses. Most programs contain an integrative introductory course. Several textbooks for such courses have been published. Capstone seminars or research projects for seniors are also common.
A number of problematic areas were identified. The use of preexisting departmental courses does not present most effectively either the distinctive computational perspective of cognitive science or its integration of disparate disciplinary sources. The breadth of the field and uncertainties about its core content can lead to a lack of depth in the major. Recent research developments have strongly reoriented the field, forcing undergraduate teachers, who were often trained prior to the new developments, to adjust. Although statistical data is lacking, the overall number of majors and the number of female and minority majors appear to be too small to provide a strong infrastructure for graduate programs and research in the field.
Recommendations for Cognitive Science Education
Undergraduate cognitive science stands at the threshold of more permanent national organization, a solid foundation of publicly available instructional models and materials, and levels of undergraduate interest and enrollment that are sustainable in the long term. Stronger agreement at the national level about the core content of the undergraduate major is needed to move the field over this threshold. In addition, a new round of funding from government and private foundations will be required. The funding will depend on a stronger recognition among cognitive scientists at a national level about what is needed and on the initiative of individuals and groups in making proposals for the development of curriculum and course materials. The highest priority should be on proposals to produce instructional materials that are widely usable.
The following areas of curricular experimentation and development are particularly worthy of support:
Integrative topic-oriented courses.
New approaches to teaching the formal, mathematical, and methodological content of cognitive science.
Courses and materials that incorporate new research developments in cognitive neuroscience, artificial neural networks, and situated action theory.
Courses and materials that increase the appeal of cognitive science to the beginning student.
New models of cognitive science teaching that are based on the results of research on cognition and instruction.
In April 1991, two years prior to the meeting reported here, a planning workshop for the cognitive science initiative at the National Science Foundation was held. The executive summary of the final report (Greeno, 1992) assigned "highest priority for new funding to programs that would extend the working corps of interdisciplinary cognitive scientists." The body of the report recommended that the NSF "support a study group to develop recommendations for programs of undergraduate study that would constitute preparation for graduate work in cognitive science, and encourage submission of proposals for curriculum development in cognitive science at the undergraduate level." The present workshop was a response to that recommendation.
The number of participants in the workshop was limited by funding and by the desire to promote a lively, wide-ranging discussion. The participants were chosen to represent the subdisciplines of cognitive science, a wide range of experience with undergraduate instruction, the Cognitive Science Society, and the earlier planning workshop. The conclusions reported here are largely the result of the discussion at the workshop, supplemented at various points by results from the accompanying survey.
In conjunction with the workshop a sample of 15 cognitive science programs was surveyed. A knowledgeable person in each program answered an open-ended questionnaire designed by the workshop organizer (Appendix 2). The survey answers expanded the base of information and opinion available for this report and served as a check on the conclusions developed at the workshop.
The primary purpose of the workshop was to assess the current status and future prospects of the undergraduate major in cognitive science. A particular concern was the implications for undergraduate programs of new developments in cognitive science. A session was devoted to strategies for attracting a diverse group of undergraduates into cognitive science. Finally, possible initiatives for enhancing undergraduate instruction in the field were discussed.
Historical Context of Cognitive Science Education
Cognitive science is conceived of by its practitioners as a basic science of intelligence in humans, other animals, and artificial systems. It includes the study of perception, learning, memory, knowledge, meaning, reasoning, language, attention, affect, consciousness, and the control of action. It is concerned less with the content of particular cognitive achievements (Shakespeares sonnets, the design of the paper clip, the strategies used in a particular business negotiation, and so on) and more with underlying principles of structure and process that make such achievements possible. The field is deeply rooted in the histories of philosophy, psychology, biology, linguistics, mathematics and other fields, but its contemporary shape emerged in the years following the second world war. Two characteristics are important to the present context: (1) The primary mode of explanation in cognitive science is the development of computational models of cognitive structures and processes; (2) A number of traditional disciplines have contributed important methods and insights to cognitive science, making research and the institutional organization of the field highly interdisciplinary.
Interdisciplinary research with a computational flavor began at a number of universities in the late 1940s and the 1950s (e.g., McCulloch and Pittss work on neural networks, Chomskys research on generative grammar and the associated early psycholinguistics, and Newell and Simons work on the Logic Theorist). Formally organized programs, departments, and curricula in cognitive science began to appear about 1970 (e.g., at the University of Chicago and at Hampshire College). The term cognitive science came into general use in the middle to late 1970s as a growing number of researchers from different disciplines embraced the computational approach to cognition (Bobrow & Collins, 1975, for example) . The Cognitive Science Society was founded in 1979 (Norman, 1981) , beginning its annual conferences and assuming control of the journal Cognitive Science, which had begun publication in 1977. The growth of cognitive science in the late 1970s and 1980s was facilitated by a special program of the Alfred P. Sloan Foundation, which made large grants to a number of universities to establish centers of research in the field. Sloan also funded three conferences on cognitive science education, held at the University of Rochester, Vassar College, and Hampshire College. A history of the field (Gardner, 1985) and the first textbooks (Johnson-Laird, 1988; Osherson et al., 1990; Posner, 1989; Stillings et al., 1995) appeared in the late 1980s. Second editions of the texts have begun to appear.
Research and teaching in cognitive science has been characterized by considerable intellectual ferment. The period from the middle 1950s through the 1970s is often referred to as the era of the cognitive revolution. During this period the use of computer simulation models to study thought processes developed rapidly, and the field of artificial intelligence, prefigured by Turing (1950) , was established. The computational approach to intelligence had a decisive influence on several traditional disciplines. In psychology the new approach replaced behaviorism, which had been the fields dominant paradigm in the United States for nearly fifty years. In linguistics generative grammar, which combined sophisticated computational models with an interest in the biological basis of language, supplanted traditional structural linguistics. Several areas of philosophy, including philosophy of mind, epistemology, and philosophy of language, were also deeply affected. The cognitive revolution profoundly changed the view of the mind in many quarters of academia, and its considerable reach into the popular imagination was confirmed when a book expounding the cognitive view in considerable depth became a best seller (Hofstadter, 1979).
During the period of the cognitive revolution the field was unified by a set of core ideas that are now often referred to as the classical view. The heart of the classical view is the physical symbol system hypothesis. The strong version of the hypothesis holds that computation over structured symbolic expressions is a fundamental property of intelligent systems. It is articulated clearly in Newell (1990) and Pylyshyn (1984). Much of the traditional research motivated by the hypothesis concerned the formal properties of symbolic systems, with little regard for either their environmental contexts or their physical implementations.
Undergraduate courses and curricula in cognitive science have traditionally emphasized the integrative, interdisciplinary nature of the new research and the changes it has worked in traditional pictures of the mind and behavior. The unifying, foundational character of the physical symbol system hypothesis was also often stressed.
In the 1980s and 1990s the leading ideas and research techniques within cognitive science have undergone rapid change. Three developments led to significant enlargements of and challenges to the classical view. First, research on artificial neural networks, which had declined during the 1960s and 1970s, was successfully revitalized (Rumelhart, McClelland, & The PDP Research Group, 1986) . When simulated on computers, these highly interconnected networks of simple neuron-like computing elements have shown powerful pattern recognition and learning capabilities, although they do not employ conventional algorithms or classical structured representations. The second development was the rapid advance of cognitive neuroscience. As cognitive science matured, researchers began to pay more attention to how cognitive capacities were instantiated in the brain. The combination of contemporary cognitive models, sophisticated experimental designs, and new technologies for neural localization (e.g., PET scanning and the analysis of event-related scalp potentials) has been very productive (Farah & Ratcliff, 1994 for example) . Finally, cognitive scientists have been paying increased attention to the role of ecological and social context in cognition, leading to a new emphasis on situated action (Special Issue, 1993).
Recently, those who teach in established cognitive science programs or who are beginning to develop programs, have faced the challenge of incorporating the new developments in the field into their curricula.
The Case for Undergraduate Cognitive Science
The emergence and continuing growth of cognitive science in the second half of this century has been an important development in scientific research and a major event in the general intellectual culture. The theories, findings, and methods of the field deserve a significant place in the undergraduate curriculum. That place, of course, has not opened up automatically. Cognitive science courses and programs have to be established by faculty groups who wish to bring their excitement with the field to undergraduates. Many such faculty groups have formed and have worked to overcome the institutional barriers that stand in the way of new programs that cross traditional curricular boundaries. Cognitive science now has a substantial presence in the American undergraduate curriculum.
Training Future Cognitive Scientists
A major area of scientific research requires a stable infrastructure in undergraduate education to supply successive generations of researchers. Undergraduate cognitive science programs are the ideal source of early training for future cognitive science researchers. Research in cognitive science can be conducted successfully only by people who have mastered multiple techniques and theoretical views that have their origins in more than one traditional discipline and who have developed a feel for the computational approach to questions about intelligence. Students who begin graduate study without a significant exposure to the questions and methods of the field face a burden of mastering a potentially overwhelming array of fundamentals while simultaneously trying to pursue a course of advanced study and research.
Enhancing General Education
The need to train future cognitive scientists is only part of the case for undergraduate cognitive science, for the field has a great deal to bring to general education. The potential contributions fall into four areas. First, cognitive science addresses some of the large questions that animate intellectual discourse: What is the nature of meaning; are perception and knowledge objective; what is the relationship between mind and brain; what is consciousness; and so on. Such questions are central to undergraduate education, and they are of great interest to contemporary undergraduates. The relatively arcane details of current scientific research can be historically grounded in various intellectual traditions. The presence in most programs of philosophers and scholars trained in other fields who have an interest in historical and foundational issues makes this grounding relatively easy to accomplish. Cognitive sciences empirical approach to these enduring questions provides a bridge between science and areas of academic pursuit that have become increasingly distant from science. Second, cognitive science can be a premier area of general science education. In addition to being exposed to a range of highly accessible empirical research, students can encounter questions about the status of scientific knowledge in an open context in which all forms of human knowledge are investigated empirically. This naturalistic epistemology has pedagogical advantages in a cultural milieu where a priori assertions of the special status of scientific knowledge are often met with equally dogmatic denials. Cognitive science opens up a rich vein of discourse among people with very different views of science. The naturalistic approach is also one of the most successful current approaches to the philosophy of science (Giere, 1988; Thagard, 1988) . Third, cognitive science is a superb field for attracting students into an encounter with formal intellectual disciplines. Within the context of the big questions just mentioned the student encounters such disciplines as experimental design and statistics, formal logic, the theory of algorithms, and the theory of dynamical systems. Students who become fascinated with artificial neural networks as freshmen, for example, discover that they must take calculus and linear algebra to pursue their interest further. Finally, cognitive science is being applied in important areas of public policy and technology. Issues in human-computer interaction, workplace organization, reading remediation, mathematics education, cognition and aging, and the reliability of traumatic memories, among many others, are being pursued by cognitive scientists. Increasingly, some familiarity with cognitive science will be a condition for informed citizenship. Some of these general educational advantages, of course, are available to students who major in one of the disciplines that contributes to cognitive science, but no other field integrates them all in such a compelling manner.
The Institutional Context of
Undergraduate Cognitive Science
Faculty groups who wish to establish or sustain a cognitive science program almost invariably face serious institutional hurdles. Interested faculty members typically come from more than one department. The established departments must be persuaded to let faculty members take on new responsibilities, assume joint appointments, or even leave in order to join a newly-formed cognitive science department. The departments may fear competition from the new program for students, operating funds, space, and new appointments. Interested faculty members may begin to fear that their interest in starting an innovative program is actually jeopardizing their chances for tenure, promotion, or adequate resources.
Once a program is approved its faculty faces the challenge of developing a curriculum in a rapidly evolving interdisciplinary field with little history of undergraduate instruction. They often find that they must seek outside funds for course and laboratory development. If the program remains an interdepartmental arrangement, rather than achieving departmental status, then teaching commitments and budgets must often be renegotiated frequently. New appointments tend to be hard to come by.
Meeting the Challenges Locally
Given the obstacles, a significant number of stable cognitive science programs have been established over the last twenty years or so. A program is started by a dedicated group of faculty members who have a vision of cognitive science at their institution and enough political savvy to propose feasible plans that are attractive to potential allies in the larger faculty and the administration. At universities the initial establishment of cognitive science usually comes at the graduate level, while at colleges undergraduate programs must be proposed directly with no track record of success at the graduate level to call on.
The consequences of the extreme sensitivity to local conditions are that cognitive science has been institutionalized in a variety of ways at the undergraduate level, the range of faculty expertise has varied considerably, and the contents of curricula have been far from standard.
The availability of external funds typically plays a role in the establishment of a cognitive science program. At universities funds for research and graduate training often play the leading role with the establishment of an undergraduate major coming later. At colleges the availability of outside funds for undergraduate course and laboratory development is often critical to the acceptance of a proposal for a cognitive science program.
Characteristics of Current Programs
Number of Programs and Institutional Status
This project did not include an attempt to locate all undergraduate programs in the United States, but existing lists contain over forty programs, many of which were confirmed during the organization of the current workshop and the conduct of the accompanying survey. Each institution has at least a published, formalized mechanism through which an undergraduate can achieve certification in cognitive science (a major, minor, or the equivalent). There are some academic departments of cognitive science (e.g., at Brown, MIT, and the University of California at San Diego). Most programs, however, are interdepartmental. The level of faculty activity and institutional commitment vary widely among them, ranging from near departmental status to catalog listings overseen by a few designated faculty members. A list of the programs that came to the attention of the workshop organizers appears in Appendix 3.
In addition to institutions that have established formal undergraduate majors or minors, there are many other institutions that offer cognitive science courses and that allow undergraduates to concentrate in cognitive science through some general mechanism for designing an independent or interdepartmental major.
Structuring the Undergraduate Major
Putting together an undergraduate program in a rapidly changing interdisciplinary field that lacks an established history of undergraduate instruction is a formidable challenge. Faculty groups designing a major confront a number of issues.
Boundaries of the curriculum.
Largely because of its interdisciplinary origins, the question of what is and what is not part of cognitive science is less easy to answer than it is in other fields. Different institutions have made different decisions about what to include and how much to include. Undergraduate curricula at universities tend to be more narrowly defined, in part because they are influenced by the strengths of parent graduate programs, in part because many specialized undergraduate courses can be offered in a university, and in part because specialized programs are capable of drawing a respectable number of students from a large undergraduate population. College programs tend to be somewhat more inclusive because they draw from smaller faculties and student bodies. They often reach further into the contributing fields of psychology, philosophy, biology, and so on. Additional breadth can also result from a concern with the general education issues sketched earlier. The variation in curricular boundaries is benign in the sense that there is a strong family resemblance among programs. All of the programs discussed at the workshop and sampled in the survey showed substantial overlap. The variation at the margins can be viewed as healthy curricular experimentation.
Choosing the requirements for the major (or minor) has been a serious issue for most programs. It is difficult to select from the many concepts, methods, and prerequisite intellectual skills that seem central to the field. Nevertheless, although there is variation in both content and the degree of choice accorded the student, nearly all the programs discussed at the workshop or sampled in the survey impose considerable structure on the students program. Typically, all students are required to take one or more specific courses, to select courses from each of two or more menus, or both. Some programs contain tracks, which require advanced students to develop strength in some topic or disciplinary area (e.g., language or artificial intelligence). Some programs also require an undergraduate thesis or similar intensive independent study project. Finally, most programs allow students to elect some courses freely from the entire curriculum.
Content area requirements.
Most programs require students to take one or more courses in each of several content areas, which are often defined in terms of traditional disciplines. Thus, a program might require the student to take one course in each of psychology, linguistics, computer science, philosophy, and neuroscience. These breadth requirements are an accurate reflection of the interdisciplinary origins of cognitive science, and they reflect the strongly held belief that one of the things that distinguishes the cognitive scientist is a familiarity with the perspectives of several disciplines. The requirements also reflect limited resources to design new courses and the desire of the traditional departments to preserve their enrollment base.
Formal and Methodological requirements.
The belief that the best cognitive science research combines methods and formal analytical skills that first arose in various disciplines is reflected in a common requirement that undergraduate students study more than one methodology or formal skill. Such requirements are implemented either by requiring certain courses or by constraining choices from menus so that the student must choose more than one formally or methodologically oriented course. The most common requirements appear to be statistics and laboratory-based experimental design. This requirement reflects the importance in cognitive science of empirical research employing behavioral measures. A second common requirement is a course that involves computer programming. The rationale for this requirement is twofold. First, many people believe that the experience of programming instills an appreciation of the notion of a formal system. Second, programming skills are required for advanced work in several areas of cognitive science, including artificial intelligence, the development of simulation programs, the use of computers to control empirical experiments, and in some cases for data analysis. Requirements for one or more courses in logic, syntax, formal semantics, or theory of computation are also relatively common. Such requirements are justified by the centrality of language to cognitive science and by the ubiquity of theories that employ predicate calculus-like or language-like representations. Finally, many programs require some continuous mathematics, such as courses in calculus and linear algebra. These courses provide a foundation for the study of artificial neural networks and are probably being included in cognitive science curricula with increasing frequency. The requirement of a laboratory course in neuroscience is also likely to become more common than it now is.
This list of methodological desiderata could easily command seven or eight courses, or well over half of a students major requirements. Most programs reduce the load by giving students some choices and by allowing some courses to count toward both topic and method requirements. For example, a course in cognitive psychology that includes a laboratory might meet both the psychology and experimental methods requirement, or a course in syntax might meet both the language and the formal methods requirement. Or, students might be given a choice between logic and computer programming on the theory that either course gives them a feeling for formal systems.
Required introductory and capstone courses.
Many programs have felt the need to give their majors a common experience in an introductory course and sometimes in a senior seminar or project course as well. The availability of the texts mentioned above eases the task of mounting an introductory course, although new programs usually require some planning support for the development of the course. Introductory courses typically try to introduce the range of questions, findings, and methods in cognitive science in an engaging way. They also try to introduce the computational approach to cognition, including the notions of levels of analysis, formal system, representation, and some of the basic concepts of artificial neural networks. Some courses include demonstrations and laboratory experiences as well. The common experience for seniors can take many forms, ranging from a seminar that goes over advanced readings, to a group research project, to a colloquium in which seniors present their thesis projects.
Some Problematic Areas
Any group of cognitive scientists who consider the issues just discussed and make decisions that meld the content of the field with their local circumstances will come up with a program that is intellectually sound and recognizably similar to most established programs. There are, however, several common problematic areas.
Traditional disciplinary vs. newly designed courses.
Many curricula appear to be heavily dependent on preexisting courses from the contributing departments. It would be less than ideal to force each generation of students to recreate cognitive science from its disciplinary sources. Without studying syllabi and talking to instructors it is difficult to assess how much the traditional disciplinary content of the standard courses has been altered, but it seems fair to say that many programs have not had the support needed to develop intermediate and advanced courses that were designed from scratch to embody the cognitive science approach. Appropriate textbooks, software packages, and other curricular materials for such courses are still scarce.
Examples are not hard to generate. One can imagine a student burdened with integrating material on language from courses in syntax (linguistics department), psycholinguistics (psychology department), and natural language processing (computer science). Much of the material could be tied together in a one or two term course on language, with additional material from philosophy and research on neural networks thrown in. A course on formal methods in cognitive science could make an integrated presentation of appropriate material from discrete mathematics, theory of computation, syntax, logic and formal semantics, and programming. Such courses are difficult to design and sometimes fail, but good examples exist, and further experimentation is very much in order.
Breadth vs. depth.
Faculty members who staff undergraduate programs often comment that it is difficult to structure a major in such a way that students achieve both adequate breadth and depth in cognitive science. Graduate faculty sometimes complain that students coming out of cognitive science programs are broadly knowledgeable but lack depth (whereas students coming out of traditional departments have depth in their discipline but do not know many of the things that a cognitive scientist has to know). Requiring advanced students to select a topical or subdisciplinary track is one strategy for achieving adequate depth. A labor intensive but highly effective strategy is to require advanced students to do research under the guidance of a faculty mentor. Many colleges and some universities offer research experiences to at least some of their undergraduate cognitive science majors.
The conundrum of breadth versus depth is, of course, a permanent feature of undergraduate education, but in cognitive science it may also result from a failure to think more deeply about what is truly essential to the field and from losses in efficiency that arise from cobbling together a curriculum from preexisting courses.
New developments in cognitive science.
For the first time in the short history of the field new research programs that are not straightforward extensions of past work have arisen. Undergraduate programs face the challenge of incorporating the new work in cognitive neuroscience, artificial neural networks, and situated action into their curricula. Room must be found in already overstuffed curricula for the new material. Further, since the new work calls the classical foundations of the field into question in various ways, it cannot simply be tacked onto old curricula. Courses must be redesigned, discarding some old material and reconceptualizing other material to integrate it with new developments. This kind of course and program redesign requires support. The necessary support must in many cases include a component of faculty development, since most faculty members in cognitive science programs received their graduate training before the new lines of research began.
The recruitment of students is a key problem in undergraduate science education today. The number of baccalaureates in science and engineering, already too low, declined in recent years for both men and women Foundation, 1991 #18 . The decline was not explained by the overall decline in the population of 22-year-olds. Minority recruitment is also a continuing problem, which will assume new policy dimensions with the demographic shifts in the ethnic composition of the U.S. population anticipated in the coming years. The behavioral sciences have resisted the overall trend, showing substantial increases in undergraduate enrollment. Cognitive science is a behavioral science in certain respects and a biological and computational science in other respects and therefore might be expected to hover somewhere between the general decline and the behavioral upswing.
Undergraduate cognitive science does not have a separate category in standard national statistical surveys, and no attempt has been made within the field to track enrollments. Comment at the workshop and answers on the survey suggest that enrollments in recent years have been modest and stable. Typical programs appear to attract 1% or 2% of each class into the major. Most programs would like more students, to preserve their institutional viability, to raise enrollments in the more innovative and intellectually exciting cognitive science courses, and to increase the sense of community among majors. Recruitment into the major is a crucial issue at colleges with enrollments under 2,000 students, where the yearly number of majors may be distributed around a mean of 4 or 5, and years with 0-3 majors can occur with some frequency. Faculty groups proposing new undergraduate programs should pay careful attention to the question of potential enrollments and should be willing to put considerable energy into cultivating enrollments after the program is established.
It is probably fair to say that it would be good for most programs and good for the field if the number of majors went up to 2%-4% of graduates, which would be a 50%-100% increase at most institutions. This number is not unrealistic since it is comparable to or above the numbers for physical sciences at many institutions and well below the numbers for blockbuster majors, such as psychology or government. Although it is reasonable to hope for increased enrollments, it is also clear that there are strong constraints on growth. At most institutions the cognitive science major essentially competes for students with overlapping majors in departments of psychology, biology, philosophy, and computer science, which have physical facilities and name recognition on their side. Cognitive science is also a difficult field with a relatively narrow reach into postgraduate job markets.
There are several strategies for increasing course enrollments and the number of majors. The first is to offer attractive, integrative entry-level courses to beginning students. The second is to teach the formal parts of cognitive science in an accessible way, so that required formal courses do not intimidate and filter out too many students. The third is to put more emphasis on applied cognitive science to combat a common student perception that the field is hopelessly abstract and impractical. A fourth strategy is to have introductory cognitive science courses listed for the fulfillment of campus-wide distribution or general education requirements. A fifth strategy is to try to recruit potential cognitive science students at the time of admission. Implementations of this strategy range from cooperating with other science programs on new undergraduate admissions literature to outreach into secondary schools.
There is no data, even of the anecdotal kind above, on the percentages of women and minority students in undergraduate programs. It seems reasonable to assume, however, that the statistical picture is similar to that in the other sciences and therefore unsatisfactory. Strategies for increasing enrollment generally should have a positive effect on diversity. In addition, strategies aimed specifically at increasing diversity were discussed at the workshop.
It is generally agreed that in the professorial ranks women and minorities are statistically underrepresented relative the United States population. Thus, there is a need for faculty role models for women and minority undergraduates. Every effort should be made to ensure the success of young female and minority scholars in graduate school, post-doctoral positions, and junior faculty positions. Departments should see to it that young scholars take advantage of existing programs at federal agencies and private foundations and that those programs are supplemented by local efforts in affirmative action recruiting, mentoring, and research support.
The diversity of undergraduate enrollments can potentially be increased by two further strategies. First, curricular connections can be increased between the cognitive science program and other programs that have higher female and minority enrollments. Two examples on some campuses might be premedical studies and education, both of which have intellectual connections to cognitive science. If students in such areas are provided with natural entry points into cognitive science, some of them will take more courses or become cognitive science majors. A somewhat different strategy is to teach more about topics that are likely to be of interest to underrepresented students. This approach has its dangers, since it can be perceived as demeaning by the very students it is designed to attract, and it can in fact degenerate into pandering. The dangers can be avoided, however, and teachers who adopt it often report strong initial effects on enrollment.
Application of cognitive science to cognitive science instruction.
The rapidly growing literature on cognition and instruction has produced new models for effective leaming and teaching, and it does not strongly support the lecture and examination format that is still the norm in undergraduate education, including cognitive science programs. Cognitive scientists should experiment with applying findings in cognition and education to their own teaching.
The four undergraduate curricula summarized here are representative of mature programs that have enjoyed strong support from their institutions and from government and private foundations. Three of them come from a range of institutional scales: the large public research university (UCSD), private university (Brown), and liberal arts college (Vassar). UCSD and Brown are unusual in having departments of cognitive science, but this distinction makes their programs more instructive in a sense because they represent choices that were made under unusually favorable conditions. The fourth program, at Hampshire College, an experimenting liberal arts college, demonstrates that cognitive science can thrive in unconventional contexts.
Other programs around the country contain many variations on the themes present here. Faculty groups who are revising their curricula or planning new programs should attend to the general points above and should seek further information from the Cognitive Science Society, the organizers of this workshop, or some of the other programs listed in appendix 3. In addition to their general curricular organization, most programs also include one or more innovative courses, whose syllabi, readings, assignments, and so on are of interest and may be obtainable directly from the programs offices or faculty members.
University of California at San Diego
The major at UCSD is distinguished by an extensive set of integrative core courses and by rigorous formal and methodological requirements. All majors must complete several lower-division courses, including a three-quarter calculus sequence, logic and statistics, Lisp and symbolic programming, and neurobiology of cognition. The upper-division requirements comprise ten quarter-length core courses and three electives (six for students seeking the B.S. degree).
The core courses are organized into three year-long sequences, one in fundamental cognitive phenomena, one in cognitive neuroscience, and one on modeling cognitive phenomena. The cognitive phenomena sequence covers material that would be encountered traditionally in courses on cognitive psychology, cognitive development, syntax, semantics, and cognition and culture. The modeling sequence includes considerable experience with programming both symbolic and neural network models.
Students in the B.S. program may declare a specialization in neuroscience, computation, human cognition, or clinical aspects of cognition. In this case four of the six electives must be chosen from a list of courses approved for the specialty. Many of the electives are offered by other departments.
At the beginning of their senior year, majors may apply for admission to an honors program, which requires a senior thesis project, a minimum 3.5 grade point average, and the completion of a cognitive science graduate course.
Cognitive science minors or concentrations are offered to students who are majoring in other fields. A lower-level year-long introductory course (Minds, Brains, and Computers) is offered to allow students to partially fulfill minor, concentration, or general education requirements.
The Department of Cognitive and Linguistic Sciences at Brown offers AB. and Sc.B. degrees in cognitive science. Each degree has a distinct character. The A.B. program is intended for students with only a secondary interest in cognitive modeling or cognitive neuroscience, whereas the Sc .B. program in intended for students with a strong interest in either or both of these areas. Generally, the curriculum balances required courses with a system of electives governed by menus and by student faculty consultation.
The department offers an introductory course in cognitive science that is a prerequisite to the concentration. All concentrators are required to take three methodological courses: a laboratory course (chosen from six laboratory courses offered by the department), a basic computation course (chosen from three courses offered by the department or by computer science), and a statistics course (chosen from two courses offered by the department or by mathematics). Concentrators are also required to take four content-area courses: cognitive psychology, language, perception, and cognitive neuroscience. The final required course is a capstone senior seminar.
Four elective courses can be chosen from an extensive list offered by the department as well as by computer science, engineering, neuroscience, philosophy, psychology, and anthropology. A program of electives is expected to show coherence and depth in one or more of the subject-matter focuses of the program. Students in the A.B. program are encouraged to take at least one term of independent study.
Students in the Sc.B. program are required to take at least one term of independent study and must also complete a coherent selection of four additional courses in the life sciences, physical sciences, or mathematics.
Students may apply to an honors program and are accepted at faculty discretion. The program requires a year-long thesis project.
At Vassar the cognitive science major is offered by a multidisciplinary program. The faculty are drawn from cooperating departments, currently comprising psychology, computer science, anthropology (linguistics), philosophy, and biology.
The cognitive science faculty offers five integrative courses that are required of all majors: An introductory course; perception and action, which includes some hands-on experience with robotics; knowledge and cognition; methods in cognitive science, which focuses on a specific research problem each term; and an upper-level seminar in cognitive science, which also focuses on varying topics. In addition, all majors are required to take a course in statistics and experimental design offered by the psychology department.
Beyond the requirements, four elective tracks are offered: formal models; cross- species studies; language; leaming and development. Each track specifies seven further courses offered by the cooperating departments. Students must select a single track, although they can modify the track requirements with the approval of the faculty.
A senior thesis project is required of all majors.
The cognitive science program at Hampshire reflects the colleges ongoing experiments with interdisciplinary undergraduate education. All cognitive science faculty are housed within a single interdisciplinary school. Course offerings change frequently. The program is strongly shaped by the Colleges emphasis on student initiative and on the narrative assessment of portfolios and examinations (Hampshire has neither grades nor credit hours). All cognitive science students are required to complete a coherent concentration in the field followed by a year-long thesis project. The course work for the concentration and the thesis project are negotiated with, supervised by, and evaluated by two-person faculty committees. The requirement that all students complete a substantial research project mandates a high level of rigor in each individual program. The variation in students programs is comparable to that at Vassar.
Recommendations for Undergraduate Cognitive Science
If the training of future graduate students is to be improved and if cognitive science is to realize its potential contribution to undergraduate education, the issues outlined above must be addressed.
Increased Standardization of the Core Curriculum
In spite of the strong family resemblance among undergraduate programs there is arguably too much variability in their core content. Ideally, a degree in the field should be a more reliable indicator of a students knowledge and skills. The trade- offs between breadth and depth that now face many programs could be eased if a small number of courses could be devoted to developing core material. A stronger national identity for the field might also make it easier to establish, expand, and raise funds for individual programs.
Movement toward a common core should take place relatively gradually, since there is a need for vigorous curricular experimentation to address the other problematic areas discussed earlier. Informal mechanisms such as the present workshop and sessions at the Cognitive Science Society meetings could be followed by a set of formal recommendations by the Society.
Three critical issues must be addressed before a strong set of national recommendations for undergraduate curricula could be issued. First, more detailed agreement is required about the core knowledge and techniques of the field. It is possible that two or more distinct undergraduate specializations will have to be recognized. Second, a record of shared experience and some agreement should be established about how to teach more of the core material integratively rather than through the juxtaposition of traditional disciplinary courses. Third, standards for curricula should not be developed without attention to the need to make cognitive science attractive and accessible to undergraduates at different levels of study.
Needed Areas of Curricular Experimentation
Following the discussion of problematic areas above there is a need for several types of curricular experimentation. Such experimentation is already occurring, but there needs to be more of it, and the results must be shared with the wider cognitive science community in the form of conference presentations or publication.
Integrative Topic-oriented Courses
There is a need for integrative courses in areas such as language, reasoning and problem solving, vision, and theories of cognitive architecture.
New Methods Courses
Experimentation with methods courses that are specifically designed to meet the needs of the cognitive science student should be encouraged. One example is an course that introduces computational styles of thought and introduces formal methods by presenting material in theory of computation, programming, and logic.
Responses to New Developments in Research
New developments in cognitive neuroscience, artificial neural networks, and situated action must be integrated into the cognitive science curriculum more fully.
Courses that Increase the Appeal of Cognitive Science
Building undergraduate enrollments requires courses that will draw more students to investigate the field. Introductory courses that are pitched at a lower level than those that are currently typical and courses that have a strong general education flavor should be investigated. Courses on applied topics could be added to many programs. There are also many topics that have a special appeal to undergraduates and that can be used to present substantial chunks of cognitive science. They include consciousness, animal cognition, creativity, and truth and meaning. The development of curricular materials that could be used in pre-college education should also be encouraged.
Using Cognitive Science Research to Increase the Effectiveness of Instruction
Cognitive scientists should study the effectiveness of their own teaching and should use the results of research on cognition and instruction to strengthen their curricula.
Increased Information Exchange
Communication among undergraduate cognitive science programs must be increased. Useful immediate steps would be the institution of an annual session on education at the Cognitive Science Society meetings and the establishment of a news group or information repository on the internet.
Increased Publication of Books and Software
Books and educational software that are tailored for cognitive science instruction are still in relatively short supply. The most successful local curricular innovations should lead to instructional materials that can be widely used.
The Role of External Funding
Having achieved a noticeable place in American higher education, undergraduate cognitive science now stands at the threshold of more permanent national organization, a solid foundation of publicly available instructional models and materials, and levels of undergraduate interest and enrollment that are sustainable in the long term. Funding will be required to cross the threshold. The development of new courses and curricular materials that are exportable and that capture the special perspective of cognitive science requires faculty time and in some cases equipment, software, and technical or clerical assistance that is not normally available in program budgets.
Securing the needed funding depends jointly on the initiative of cognitive science teachers and on the recognition by government and private foundations that undergraduate cognitive science is worthy of support. Both the initiation of proposals and their success could be enhanced by further specification of what is needed and further information sharing at the national level.
Bobrow, D. G., & Collins, A. (Eds.). (1975). Representation and Understanding: Studies in Cognitive Science. New York: Academic Press.
Farah, M. J., & Ratcliff, G. (Eds.). (1994). The Neuropsychology of High-Level Vision. Hilldale, NJ: Erlbaum.
Gardner, H. (1985). The Minds New Science. New York: Basic Books.
Giere, R. N. (1988). Explaining Science: A Cognitive Approach. Chicago: University of Chicago Press.
Greeno, J. G. (Ed.). (1992). To Strengthen American Cognitive Science for the Twenty-First Century. Washington, D.C.: National Science Foundation.
Hofstadter, D. R. (1979). Gbdel, Escher, Bach: An Eternal Golden Braid. New York: Basic Books.
Johnson-Laird, P. (1988). The Computer and the Mind: An Introduction to Cognitive Science. Cambridge, MA: Harvard.
National Science Foundation, (1991). Science and engineering baccalaureates declined for both men and women from 1986 to 1989. Mosaic, 22(3), back cover.
Newell, A. (1990). Unified Theories of Cognition. Cambridge, MA: Harvard.
Norman, D. A. (Ed.). (1981). Perspectives on Cognitive Science. Hillsdale, NJ: Erlbaum.
Osherson, D., et al. (Eds.). (1990). An Invitation to Cognitive Science (3 volumes). Cambridge, MA: MIT Press.
Posner, M. I. (Ed.). (1989). Foundations of Cognitive Science. Cambridge, MA: MIT Press.
Pylyshyn, Z. W. (1984). Computation and Cognition: Toward a Foundation for Cognitive Science. Cambridge, MA: MIT Press.
Rumelhart, D. E., McClelland, J. L., & The PDP Research Group (Eds.). (1986). Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Cambridge, MA: MIT Press.
Special Issue: Situated Action, (1993). Cognitive Science, 17(1), 1-133.
Stillings, N. A., Weisler, S. E., Chase, C. H., Feinstein, M. H., Garfield, J. L., and Rissland, E. L. (1995). Cognitive Science: An Introduction (2nd ed.). Cambridge, MA: MIT Press. (1st edition, 1987)
Thagard, P. (1988). Computational Philosophy of Science. Cambridge, MA: MIT Press. Turing, A. M. (1950). Computing machinery and intelligence. Mind, 59, 433-460.
Appendix 1: Workshop Participants
Invited faculty members from academic institutions
Neil Stillings, Organizer and Chair, Cognitive Science, Hampshire College
Steve Weisler, Co-organizer, Cognitive Science, Hampshire College
Beverly Adams, Psychology, Univ. of Virginia
Mark Bickhard, Cognitive Science Program, Psychology, Lehigh Univ.
Roy 0. Elveton, Cognitive Science program, Philosophy, Carleton College
James Greeno, Education, Stanford Univ., Institute for Research on Leaming
Alan Lesgold, Learning Res. & Development Center, Intelligent Systems Prog,, Psychology, Univ. of Pittsburgh
Kenneth Livingston, Cognitive Science Program, Psychology, Vassar College
Dan Lloyd, Cognitive Science Program, Philosophy, Trinity College
Margery Lucas, Cognitive Science Program, Psychology, Wellesley College
Michael McCloskey, Cognitive Science, Johns Hopkins Univ. (Unable to attend)
Gary Olson, Cognitive Science & Machine Intelligence Laboratory, Psychology, Univ. of Michigan
Paula Schwanenflugel, Educational Psychology, Univ. of Georgia
Kathryn Spoehr, Cognitive & Linguistic Sciences, Brown Univ.
Mark Steedman, Cognitive Science Program, Linguistics, Univ. of Pennsylvania
Thomas Wasow, Symbolic Systems Program, Linguistics & Philosophy, Stanford Univ.
Bonnie Webber, Cognitive Science Program, Computer Science, Univ. of Pennsylvania
David Zipser, Cognitive Science, UCSD
Representatives from government agencies
Lida Barrett, Education & Human Resources, NSF
Susan Chipman, Office of Naval Research
Oscar Garcia, Director, Interactive Systems Program, NSF
Howard Kurtzman, Director, Basic Behavioral & Cognitive Sciences Research, NIMH
Richard Louttit, Acting Director, Social, Behavioral & Economic Research, NSF
Paul Rodriquez, Education & Human Resources, NSF
Joseph Young, Director, Human Cognition & Perception Program, NSF
Terry Woodin, Education & Human Resources, NSF
Appendix 2: Survey Questionnaire
1. Name and address of your program.
2. Describe the institutional status of your program. Is it an academic department or the equivalent? Is it an interdepartmental program whose faculty members are appointed in cooperating departments, such as psychology, linguistics, or philosophy? Is it organized in some other way?
3. How many faculty members are in the program? If faculty members time is divided between the cognitive science program and departmental programs, can you characterize the typical division of responsibilities between the two programs?
4. Characterize the options that are open to students who wish to pursue cognitive science as a significant component of their undergraduate program. Is cognitive science offered as an undergraduate major or the equivalent? Is a cognitive science minor or certificate available? Can students design independent majors that include a significant amount of cognitive science?
5. Characterize the availability of cognitive science to the beginning or initially curious student. For example, do you offer introductory courses in cognitive science? Are any courses in cognitive science included in a general education or liberal arts distribution requirement? Do students mainly find their way to cognitive science via offerings in other programs, such as psychology or linguistics?
6. Can students do an undergraduate thesis project in your program? If so, is the thesis a requirement, an option chosen by the student, or an honor?
7. How many students in your undergraduate program in the last three years (89- 90, 90-91, 91- 92) went on to graduate programs in cognitive science or its contributing disciplines (psychology, linguistics, etc.).
8. Has your program produced outstanding students who are making contributions to cognitive science as faculty members, post-docs, or mature graduate students. If possible, give names, graduate programs, current institutional affiliations, and research areas.
9. If you have a formal cognitive science major, concentration, or minor, what is the structure of the educational program, i.e. course requirements, electives, and so on? Feel free to submit printed descriptions to supplement your answer to this question.
10. Briefly characterize your institutions approach to cognitive science as a field of undergraduate study. Feel free to refer to existing descriptions that you are providing to the survey. Some of the issues that typically arise in designing cognitive science programs are the following: (1) Resolving a tension between an inclusive "cognitive studies" approach and a tightly focused core cognitive science approach; (2) Deciding how much formal material to require, e.g. mathematics, computer science, logic, syntax, or statistics; (3) Deciding whether to require all students to get a broad exposure to cognitive science or whether to allow them to specialize early in a subfield, such as natural language processing or connectionist modeling.
11. Can you say anything about institutional political hurdles that your program has faced and how they were resolved.
12. Can you briefly describe significant funding that your undergraduate program has received from your institutions administration, foundations, government agencies, or private donors. How important a role has this funding played in establishing and sustaining your program?
14. Do you have any ideas about how undergraduate cognitive science education could be strengthened nationally?
Questions 15- 18 are for institutions that also have a graduate program in cognitive science:
15. Can you comment on the general quality and variability of the undergraduate training received by your incoming graduate students?
16. Have you accepted students with undergraduate degrees, minors, or certificates in cognitive science? If so, can you comment on the preparation of these students for graduate work?
17. Do you have suggestions for changes in undergraduate education in the cognitive sciences that would improve the overall level of preparation for graduate work?
18. Has an advanced degree in cognitive science provided a good career path for graduates of your program? That is, has their training benefited them in the search for post-doctoral fellowships, academic or industrial jobs, support for their research programs, and so on?
Appendix 3: A Sample List of Institutions with Undergraduate Programs in Cognitive Science
Notes: This list is not exhaustive. Programs range from full-scale departments to interdepartmental minors. Some programs are not called "cognitive science," e.g., Symbolic Systems at Stanford.
Univ. of California at Berkeley
Univ. of California at Irvine
Univ. of California at Los Angeles
Univ. of California at San Diego
University of Colorado
University of Florida
Georgia Institute of Technology
Johns Hopkins University
Massachusetts Institute of Technology
Memphis State University
University of Michigan
University of Minnesota
University of Pittsburgh
University of Pennsylvania
University of Rochester
Trinity College, Connecticut
Trinity University, Texas
University of Virginia
Washington & Lee University