Stellan Ohlsson, Ph.D.
How does the human mind change over time? The creative,
knowledge-constructing function of mind is the key to understanding human
cognition. This function manifests itself as cognitive development during the
first fifteen years of life; as skill acquisition and the attainment of
expertise via practice; and as belief revision and conceptual change. On a
societal scale, the knowledge-constructing function manifests itself in
scientific discovery, artistic achievement and technological innovation. The
overarching goal of Dr. Ohlsson's research program is to uncover the cognitive
change mechanisms that underlay these seemingly diverse forms of knowledge
construction. For descriptions of specific research projects, click on any of
the following: insight & discovery,
deep learning, skills & errors,
cognition, technology & real life.
Skills & Errors:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Ph
Background. People acquire skills by practicing.
Practice is paradoxical because it consists in attempting to solve a problem
that the learner or trainee has not yet mastered. The problem for cognitive
psychology is to understand why such attempts eventually lead to mastery. What
is the mechanism of improvement? How is the rudimentary skill of the novice
transformed into the competence of the expert?
It has been clear at least since the work of Edward
Thorndike in the 1890s that feedback from the environment plays a crucial role
in skill acquisition. Feedback includes information about errors, i.e., outcomes
that indicate that the action taken was inappropriate or less than useful. How
does a learner translate such information into appropriate revisions of his or
her current skill?
Objective and hypotheses. The objective of this
project is to describe the cognitive mechanisms of learning from error during
practice on sequential choice tasks. The latter category includes everyday
chores like making coffee and professional tasks such as equipment repair, as
well as so-called puzzles like the Tower of Hanoi task. The working hypothesis
is that to learn from error, the learner or trainee must have sufficient domain
knowledge, encoded as constraints on solutions, to be able to detect his or her
own errors. Once detected, the error is corrected by specializing the
relevant mental rule so that it no longer becomes active in situations in which
it causes errors.
Recent work. A series of computer simulations
has shown that this error correction mechanism is sufficient to construct
cognitive skills in numerical and scientific tasks. The next phase of this line
of research will attempt to apply this theory in realistic training scenarios.
Publications:
Ohlsson, S. (1996). Learning from performance errors. Psychological
Review, 103, pp. 241-262.
Ohlsson, S. (1996). Learning from error and the design of task environments.
International Journal of Educational Research, 25(5), 419-448.
Ohlsson, S., & Rees, E. (1991) The function of conceptual understanding in the
learning of arithmetic procedures. Cognition & Instruction, 8(2), pp. 103-179.
Deep Learning:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Ph
Background. Knowledge domains are organized
around fundamental ideas. Examples of such ideas include the concepts of force
and field in physics, natural selection in biology, germ in
medicine, representation in political theory, free market in
economy, cultural practice in social anthropology, jurisprudence
in law, and so on. Intuitive belief systems are also organized around
fundamental ideas, but they tend to be held implicitly and lack common labels
and hence are more difficult to describe briefly.
How do people acquire fundamental ideas? By what
process are such ideas constructed? Unlike other types of knowledge, fundamental
ideas cannot be acquired through discourse or concrete experience, because those
ideas are the very tools by which the mind interprets both discourse and
experience. Attempts to teach fundamental ideas by direct instruction or an
experiential example typically leads to distortion. And yet, people do sometimes
succeed in acquiring new fundamental ideas.
Objective and hypotheses. The objective of this
project is to describe the cognitive mechanism of deep learning. The working
hypothesis is that abstraction plays a crucial role. New fundamental ideas are
acquired by instantiating an abstract schema in a novel way; the new
instantiation gradually assimilates pieces of the relevant domain, until it has
effectively become the new center of that domain. Abstract schemas, in turn, are
generated by combining and transforming prior schemas.
Recent work. In a series of studies, students
were taught Darwin's theory of evolution through natural selection under a
variety of circumstances. Reading about the theory has little impact on
students' evolutionary explanations. However, reading an exposition of the
theory after first having acquired an abstract schema of variation-and-selection
in another domain does improve understanding of natural selection, as
evidenced by increased reliance on Darwinian ideas and decreased reliance on
non-Darwinian ideas (misconceptions) while constructing evolutionary
explanations. The next phase of this research will aim to replicate and
generalize this effect.
Publications:
Ohlsson, S., & Lehtinen, E. (1997). The role of
abstraction in the assembly of complex ideas. International Journal of
Educational Research.
Ohlsson, S. (1993) Abstract schemas. Educational
Psychologist, 28(1), 51-66.
Ohlsson, S. (1995). Learning to do and learning to
understand: A lesson and a challenge for cognitive modeling. In P. Reimann and
H. Spada, (Eds.), Learning in humans and machines: Towards an
interdisciplinary learning science (pp. 37-62). Oxford, UK: Elsevier.
Cognition, Technology & Real Life:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Ph
Cognitive psychology has implications for the design of
cognitive tools (e.g., computer software), the environments in which complex
tasks are performed (e.g., airplane cockpits) and instruction and training in a
variety of contexts. The following brief summaries describe three such
applications. Each application was developed in collaboration with another
laboratory.
Patient education
Non-compliance with physician's prescriptions has been
estimated to account for as much as 40% of the nation's health care costs.
Non-compliance is typically caused by lack of understanding. For example, a
patient might not follow instructions because he or she does not understand the
difference between acute and prophylactic medication. Physicians have little
time to explain such concepts to patients. A computer-based patient education
system was designed at the Intelligent Systems Laboratory at the University of
Pittsburgh. The system presented headache patients with basic information and
then answered follow-up questions on-line. The system used artificial
intelligence techniques to adapt answers to individual patients. Preliminary
evaluations were positive.
Publications:
Buchanan, B., Moore, J., Carenini, G., Forsythe, D.,
Ohlsson, S., & D., Banks, G. (1995). An intelligent interactive system for
delivering individualized information to patients. Artificial Intelligence in
Medicine, 7, 117-154
Cognitive diagnosis & training
Shrinking resources puts a premium on effective
training techniques in the armed services. In particular, the US Navy has to
train large numbers of radar operators for its destroyers and other war ships.
These operators are solving the complex task of interpreting a radar display
under severe time pressure. A project led by CHI Systems, a Philadelphia-based
software company, aims to use cognitive task analysis and modeling to improve
the feedback these operators receive during and after training sessions. The
immediate objective is to design a system that diagnoses the trainee's errors
on-line and generates recommendations to Navy trainers.
Virtual reality & education
New technologies for presenting interactive 3-dimensional
worlds have been developed at UIC's Electronic Visualization Laboratory (EVL).
This technology is a means for presenting students with alternative experiences
that contrast with everyday experience in educationally relevant ways. The objective
of this project is explore the potential of virtual reality to support deep learning.
During the fall of '97 and spring of '98, a pilot project will use virtual reality
to teach young children that the Earth is round, a concept that prior research has
shown is difficult to grasp. Future applications of virtual reality will focus on
more complex learning targets.
To The Cognitive Page.
