FEd August 1998 Newsletter - Cognitive Psychology and Learning Theory: A Web Tou

August 1998



Previous Newsletters

Current Issue

Contact the Editor

Cognitive Psychology and Learning Theory: A Web Tour

Samuel Bowen

As many of us in the physics community examine various indications that our traditional courses may not be effectively reaching many students, it is worthwhile asking if any other sources of study would be helpful. Since there have recently been a number of new developments in the areas of cognitive psychology, neurophysiology, and learning theory, I thought it timely to provide a web-based tour of some of these developments. The following tour makes not claims to completeness, but is an opening to some possibly useful theories and disciplines.

Modern current cognitive models

Most current cognitive models have essentially a three part model, for example, as found in the information processing model (http://tiger.coe.missouri.edu/~t377/IPTheorists.cfm). These three parts are: a sensory register, a sort term memory, and a long term memory (LTM). The sensory register is the collection of sensory inputs with a some small amount of processing. Signals from the sensory register or the LTM pass through the short term memory which has a capacity of 4 to 7 "chunks". Changes in the long term memory and increases in speed of retrieval from it constitute learning. The question is how do these three elements operate and how are persistent and retrievable LTM items most efficiently created.

An old model of the learning process viewed the long term memory as analogous to a warehouse into which items are moved and from which those items can be retrieved, independently of the context in which they were stored. Current models of the learning process make several distinctions:

  1. Memory storage is associative, items are stored by connections to previously stored memories and thus are strongly history dependent.
  2. Items are stored by associations with past memories; contexts and actions. Many cognitive psychologists also argue that knowledge is stored in the memory in a fashion which depends integrally on the sensory mechanisms and contexts by which it was learned and retrieved. In other words, some are now arguing that there is not separate storage of information independent of the context and actions accompanying learning.. One implication of this distinction is that knowledge gained in an isolated context, and not well connected to well established earlier LTM items, will quickly decay and not be generalized to other contexts.
  3. The strength of an item is greater, in terms of recall and long term persistence, the larger the number of associations to previously stored LTM items.

A more detailed model by Roger Shank of Northwestern University of the learning process (htttp://www.ils.nwu.edu/~e_for_e/nodes/NODE-25-pg.cfm) suggests that placing new items in long term memory requires a sequence of three steps:

  1. Adopting a goal (wanting to understand or generalize something),
  2. Generating a question, and
  3. Developing an answer.

This view of learning asserts that effective long term storage of knowledge requires that the learning stage first be set by a question and that the answer be generated by the student actively asking and answering questions. Shank argues that our memories are most effectively constructed out of cases (memories and experiences) which we collect together and organize in generalizations arise out of the process of asking questions and developing answers.

Many of the current physics reform efforts are responsive to these new ideas of learning and long term memory. The Interactive-engagement methods are constructed around many of these ideas of how mental models are made and how items are stored in long term memory. The initial contact of many students with the tools and modes of thinking in physics appears to reflect a lack of previous experiences on which to build LTM. It seems possible to me that almost any method of instruction that is based on these emerging models of learning should be capable of effecting stable and useful long term memory processes and facts if we can make contact with the pre-existing memories. Our common goal should be to develop in our students the integrated memories from which they can learn and apply physics in a wide variety of contexts.

The Short Web Tour of Learning Theory

If you search the web under category of learning theory you will encounter three quite different types of sites: Education Colleges, Cognitive Psychology, and Artificial Intelligence Departments.

The colleges of education sites are primarily for practitioners of pre-college education and are not very strong on references and supporting data. A site at the University of Hawaii (http://www. hcc.hawaii.edu/education/facdev/AdultLearning.cfm) is a good example of a listing of practical principles. Also on this site is a learning inventory which attempts to allow students to classify the best way in which they learn. (http://www. hcc.hawaii.edu/education/facdev/lsi.cfm). There are a number of such inventories and they deserve examination by physicists as we examine seriously how we can get our subject across to a wider variety of students, especially those who might be future pre-college teachers. A site which contains a large collection of such learning style inventories and tests is (http://www.oise.utoronto.edu/~ggay/lstests.cfm). A more general site that provides a brief description of more popular theories of learning is (http://www.funderstanding.com). A similar site with more scholarly summaries is Theory into Practice Database (http://www.gwu.edu/~tip.index.cfm). One of the best sites for my gaining a broad overview of learning theory was the site (http://www.coh.arizona.edu/inst/edp512s97/learningtheory.html). Of similar value was the site at (http://tiger.coe.missouri.edu/~t377)

The most satisfying site for me and, I suspect, for other physicists, is the Cognitive Architecture site at the University of Michigan. Here, a group of students in a computational learning theory class has constructed a site that discusses all of the current computer based models of learning. This website is a carefully constructed collection of documents from which one can gain not only a good understanding of computational learning theory models, but also data on human learning and how these computer based systems are designed to model human learning. The discourses in this site offer much explicit and detailed connections between theories and observations.

In this tradition is the very valuable, but long, book on the web which has been written by Larry Shank at Northwestern University(htttp://www.ils.nwu.edu/~e_for_e/nodes). This document provides a very detailed and closely argued model of human learning that is close to my introductory summary. The book is a focused argument for computer based simulations as a powerful tool for learning (from the particular to the general), but his general presentation of modern learning theory is excellent.

A very different, but useful site is CogSci (http://humanitas.ucssb.edu/users/steen/CogSci ) at UCSB. This site is operated by Francis F. Steen in the Department of English. He attempts to being together the best of cognitive psychology as it applies to memory, learning and imagery. Even though this site is primarily for non-scientists, much of its information could be extremely valuable. One section entitled "Talking about Memory" at (http://humanitas.ucssb.edu/users/steen/CogSci/Memory.html) reviews two books in cognitive neuroscience by M.S. Gazzaniga, and by R. E. Cytowic on data showing that memories are stored inseparably with their sensory modality, action, and context. The site has an excellent bibliography and articles which contain sections on "naïve physics", "numerancy", and "folk biology". These sections especially contain descriptions of the intuitive notions (and decidedly non-Newtonian) that students bring to class from their experience of growing up.

Finally, there is another interesting area, the study of the formation of mental models, what many of us would identify as the "real physics". There was an excellent review paper by R. Koffijberg that appeared on the web at (http://www.to.utwendt.nl/prj/Min/Papers/Mental.Model). It reviewed the studies to distinguish the nature of mental models and their acquisition by computer simulations. The bibliography is an excellent introduction to this area.

While reviewing all of these and many other sites and sources, I was struck by the paucity of physics related references under the areas of mental models, learning, and problem solving. Much of this literature on problem solving was reviewed by David Maloney in the Fall Newsletter. In this web search there were many applications of these ideas in mathematics and computer science, but few physics applications. At the very least, there should be more involvement of scholars from the fields of cognitive psychology, learning theory, and allied fields in our physics education research and vice versa. If readers have further sources in this area, the editors would be pleased to learn about them.