Building a Science of Teaching
Edward F. Redish
Department of Physics
University of Maryland

Why should physicists (and other non-education-college academics) be interested in finding out about (and doing) educational scholarship?
The times they are a-changin’:
new professions
shifting boundaries
new technology
Who do we want to be in the 21st century?
What is our research role?
What is our educational role?

Figuring out education
Scientists have a number of valuable insights about how to understand the workings of the real world reliably.
Our knowledge comes from
careful observation
analysis and synthesis
Community activity
confirming experiments
challenge and confrontation of ideas
peer review, publication, conferences and seminars

Building a community consensus

Why do disciplinary scholars
need to study education?
We need to improve
the success of our instruction.
The education schools have other fish to fry.
Discipline independent results are interesting (when they exist) but too limited
at the university level.
No one else has the disciplinary expertise needed.
The benefits accrue to the disciplines.
It’s intellectually extremely interesting!

Why begin with
cognitive psychology?
If we want to understand a physical system
(like a student), we better understand something
about how that system functions!
“The whole of science is nothing more than a refinement
of everyday thinking.  It is for this reason that the critical thinking of the physicist cannot possibly be restricted
to the examination of concepts from his own specific field. 
He cannot proceed without considering critically
a much more difficult problem, the problem of analyzing
the nature of everyday thinking.”
Albert Einstein, “Physics and Reality,” J. of the Franklin Institute 221 (1936).

A Model of Student Learning

A Better Model
from Cognitive Science

A model of thinking
relevant to instruction: Principles
Long-term memory can exist
in (at least) 3 stages of activation
primed (ready for use),
active (immediately accessible)
Memory is associative and productive
Activating one element leads (with some probability)
to the activation of associated elements.
Activation and association are context dependent
What is activated and subsequent activations
depend on the context, both external and internal
(other activated elements).

A Hierarchy of structures
Patterns of association
(the basic structure – viz. neural nets)
Primitives / facets
Mental models

Physical reasoning maps primitive elements onto specific situations

Visual system
Primitive: a book is an object
Facet: interpreting a given visual pattern
         on the retina as a book
Physics phenomenology
Primitive: bigger is stronger
Facet: larger objects sink (incorrect generalization)
Facet: when a truck hits a car, the truck exerts
         a bigger force on the car than the car exerts
         on the truck (wrong)

This is a …?

Slide 14

This is a…

Which picture has more information? 
Which is easier to interpret?

We interpret what we see by matching
to templates and patterns
that exist in our long-term memory.
The pattern is not a recorded instance. 
We can interpret objects we have
never exactly seen before.
The closer the input is to an existing pattern, the easier it is to interpret.

"A set of four 3x5..."
A set of four 3x5 cards is dealt on a table
as shown above. Each card has a letter
on one side and a number on the other.
The dealer proposes that these 4 cards satisfy the rule:
“If there is a vowel on one side of the card,
then there is an odd number on the other.”
What is the smallest number of cards you have to turn over to be sure the rule is satisfied?  Which ones?

"You are acting as bouncer..."
You are acting as bouncer at the local pub.  It is your job to check ID’s for the servers.
One server has placed four 3x5 cards on the bar, describing the customers at a table in the back.
On one side of the card is his best guess of the patron’s age, on the other, what they are drinking.
Should you go to the back to check some ID’s?  Whose?

A small problem: What is 3 ½ divided by ¼?

Different contexts may trigger students
to reach for different resources.
Resources situated in students’ everyday experiences
are often much easier for them to use than formal ones.
Transfer is non-trivial.  Linking situated and formal methods may be particularly difficult for students.
What may look simple to someone accustomed to a context may be hard for someone new to that context.

(A problem that looks like Gin and Coke to you
may look like K2A7 to your students!)

Student responses
depend on context*
Exam problem:
A steel ball resting on a platform
is being lowered at a constant speed.

"FCI 18*: An elevator..."
FCI 18*: An elevator is being lifted up at a constant speed. (Ignore friction and air resistance)
The upward force on the elevator
by the cable is greater than
the downward force of gravity.
The amount of upward force
on the elevator by the cable equals
the downward force of gravity.
The upward force on the elevator
by the cable is less than
the downward force of gravity.
It goes up because the cable is being
shortened, not because of the force being exerted by the cable on the elevator.
The upward force on the elevator
by the cable is greater than the downward force due to the combined effects of air pressure
and the force of gravity.

Exam Problem
90% give the correct answer
the normal force on the ball is equal to the downward force due to gravity
FCI 18
54% choose the correct answer:
the upward force on the elevator by the cables
equals the downward force due to gravity
36% choose a common misconception:
the upward force on the elevator by the cables
is greater than the downward force due to gravity

Students’ responses
may depend on context.
It not only matters that they “know”
the physics, it matters when
they naturally bring it up.
“Physics problems” may cue
different resources from
“ordinary life situations”.

Key Ideas
1. Knowledge is associative
2. Learning is productive / constructive.
The brain tries to make sense of new input
in terms of existing mental structures.
We learn by analogy / metaphor
-- New constructions tend to be built from old.
3. Cognitive response is context dependent.
The productive response depends on the context in which new input is presented, including the student’s mental state.
Students can use multiple models
-- Confusion about appropriate context can make it appear
as if students hold contradictory ideas at the same time

Characteristics of Schemas
Schemas are the basic associational patterns that activate or prime a chain of connections. (spreading activation)
Schemas can be
context dependent
School-based schemas may be less robust and effective than life-experience-based schemas (situated cognition)

Organizing Long-Term Memory
The fact that some bit of knowledge
or know-how is “in there” doesn’t help much
if it doesn’t come up when you need it.
What’s important is not just
what knowledge you have
but its functionality --
how appropriately you access it
how well you can use it.

Organization of Long-Term Memory: Schemas

Slide 30

Interview Response of 2 “Grad Students”

Some guidelines for teaching
Students’ responses depend on context –
including the state of their mind at the time.
Hands on activities are not enough.
They have to be brains on, as well.
(Learning environments need to be designed
to prime the students’ states of mind.)
Connections count – not just the content.
Evaluations must focus on functional learning,
not on the “presence” of the knowledge in a (presumed) unstructured box..
Learning is a growth – not a transfer.
Students have to make connections many times
before they “stick” (synapses grow).

Some cognitive goals
In addition to having students
master the physics content,
our cognitive considerations suggest
that we also want to consider
the extent to which students
have a conceptual understanding of the physics
(see the physics as “making sense”)
the extent to which students can access the correct knowledge appropriately)
the way the students organize their knowledge (develop a coherent and consistent view of the physics they are learning).

Some research-based instructional environments in physics
Interactive Lecture Demonstrations
(Sokoloff & Thornton)
Peer Instruction
(Van Heuvelen)
(McDermott et al.)
Group Problem Solving
(Heller & Heller)
RealTime Physics
(Laws, Thornton, & Sokoloff)
Problem Solving Labs
(Heller & Heller)
Full Studio
Physics by Inquiry
(McDermott et al.)
Workshop Physics
Studio Physics
(Wilson & Cummings)
(Beichner & Risley)

The UW Tutorial Model*
Tutorials have a number of critical elements:
facilitator training session
tutorial with research-
based worksheets
and Socratic facilitators
tutorial homework
exams have
a tutorial question
some tutorials (those
developed at UMd) use
data acquisition.
Lectures (and labs) unchanged.

Workshop Physics*
In a WP room
Students use powerful computer tools
for observation and modeling.
guided inquiry model of instruction.
can flexibly restructure groups.
instructor in the room’s center
can see all computer screens
at once.
class can easily switch
from small to large group

Evaluating Concept Learning:
The Force Concept Inventory (FCI)*
30 item multiple-choice probe of student's understanding of basic concepts in mechanics.
The choice of topics is based on careful thought
about the fundamental issues and concepts
in Newtonian dynamics.
The questions are framed in (semi-)real life contexts
in common speech rather than physics jargon.
The distractors (wrong answers) are malicious. 
They are based on research that probes
the students' most common responses.

Some preliminary results
A study of 60 classes around the country by Dick Hake* shows that across a wide range of initial states the fraction of the possible gain is similar for classes of a similar structure.
For traditional classes he finds
h » 0.20 ± 0.05

Can research-based instructional models produce better conceptual gains?
We tested a change in our instruction*
in calculus-based physics for engineers.
recitation is replaced by a group-learning
concept-building activity (tutorial).
trained TA’s help students learn qualitative reasoning with research-based worksheets.
Half the lecture classes had recitations,
half tutorials.  Students were tested
with pre- and post-FCI.

Research Context
Introductory calculus-based physics
~90% of population are engineers
Course occupies 3 semesters
3 hrs of lecture/wk (100-200 students)
1 hr small group (25-30 students)
2 hrs lab/wk in semesters 2 & 3 (24 students)
Small group sessions have 2 options
recitation (TA led problem solving)
tutorial (UW model)

Tutorials produced significantly higher gains than recitations

Extension to many schools*
This study was extended to 14 colleges and universities teaching calculus-based physics using 4 instructional models:
traditional with recitation
traditional with tutorial
traditional with group problem solving
workshop physics
(a small class active-engagement model).
Both primary and secondary implementations of the research-based curricula were observed.

Slide 43

Interactive environments are not enough.
RPI attempted to extend the
Workshop Physics idea to a large class.
The class is broken into groups
of 50 in WP-like sessions.
Materials are written
by physics faculty.
Cummings et al. gave pre-/post
FCI to calculus-based students.
The environment is technology rich
and highly interactive.
They compared traditional materials
to research-based (ILD's)
in random sections.

Building a community consensus
of education

Building a community consensus
of education

The research effort in university level PER has grown substantially
over the past decade.
· Air Force Academy
· American U.
· Arizona State*
· Boise State
· Carnegie-Mellon
· Dickinson.
· Harvard
· Iowa State U.*
· IUPU Fort Wayne
· Kansas State U.*
· Montana St.*
· N. C. State*
· NE Louisiana U.
· Northwestern

E. F. (Joe) Redish
David Hammer
John Layman (emeritus)
     (  joint with Education)
Grad Students
Rebecca Lippmann
David May (visiting from OSU)
Laura Lising
Andy Elby*
Apriel Hodari**
(*PFSMETE Fellow
 ** Congressional Fellow)
Seth Rosenberg (CCNY)
Recent Graduates, Associates
and Visitors

Jeff Saul, U. Cent. FL (Ph.D. 1998)
Michael Wittmann, U. ME (Ph.D. 1998)
Mel Sabella, UW (Ph.D. 1999)
Bao Lei, OSU (Ph.D. 1999)

Richard Steinberg§ (CCNY)
Beth Hufnagel* (Anne Arundel CC)
  (§ NAE Spencer Fellow
         *PFSMETE Fellow)

Pratibha Jolly (India)
Gilli Shama (Israel)
Jonte Bernhard (Sweden)
ZuYuan Wang (China)