Intuitive Physics: Across Tasks and Age

1
Intuitive Physics Across Tasks and Age
Cedar Riener, Dennis Proffitt, Timothy
Salthouse Department of Psychology
1. Intuitive physics performance does reflect a
unitary disposition.
Do you remember the material covered in your
high school physics course? Many dont, but
would nonetheless assume that most adults
(including themselves) could accurately predict
the behavior of simple physical systems such as a
ball rolling off a table or water moving within a
tilted glass. A number of tasks have been
adopted to test just this peoples intuitive
understanding of physics. These tasks, called
intuitive physics tasks, have met with surprising
results. Instead of demonstrating comprehension
of basic physics, many studies in the field of
intuitive physics have shown that a large
percentage of adults display a systematic
misunderstanding about the simple principles
which govern physical events 1,2. Much of the
previous work in the field of intuitive physics
has focused on two major approaches to explain
these unexpected results. The first approach
asserts that the systematic errors observed on
intuitive physics tasks are due to systematically
incorrect implicit theories of physics3. Jean
Piaget showed that young children were not simply
failing at certain tasks, but instead showing
systematic biases4. This led him to propose
that children went through stages of theoretical
development which reflected a non-incremental
reorganization of knowledge. McCloskey observed
that when adult participants were given the
c-shaped tube task (see Intuitive Physics Items
on right) they behaved as if they possessed
something akin to medieval impetus theory5.
Those participants that drew a curved path for a
ball exiting a c-shaped tube seemed to hold a
belief that the curvilinear impetus imparted to
the ball by the tube gradually dissipates after
the ball exits the tube. There are two sources
of evidence against a theoretical explanation of
intuitive physics performance. First of all,
participants are not internally consistent6.
People may give two different responses to the
same question when it is presented twice.
Secondly, the surface characteristics of the
problem can affect performance. For example,
while many participants erroneously report that a
ball exiting a c-shaped tube will continue to
curve, very few also believe that water exiting a
curved hose will also continue to curve7. The
second approach has sought to analyze the
specific problem or task and explain errors based
on characteristics of each task. McAfee and
Proffitt explained errors on the water level task
by applying a universal perceptual bias 8. These
authors pointed out that the water-level task is
not necessarily a problem which tests ones
knowledge of whether water remains horizontal
regardless of the orientation of the container.
Performance on the water level task could reflect
how one represents the problem either based on
an environment-centered coordinate frame or a
container-centered coordinate frame. While this
explanation explains the errors in the water
level task, it does not apply well to other
tasks, such as the c-shaped tube problem.
Proffit and Gilden explained errors in natural
dynamics tasks as related to the complexity of
the problem, not the principles tested9. This
approach is broad in its scope (it applies to
many physical events) but it is limited in its
precision, in that it makes a binary distinction
between simple and complex problems. Our
research is aimed at answering questions at a
more fundamental level. First of all, if there
are compelling explanations for individual
physics tasks, do we still have reasons to
believe that intuitive physics tasks belong to a
single category? In search of evidence for or
against the unity of intuitive physics task, our
research addresses the question Does intuitive
physics performance reflect a unitary and
persistent general mechanical reasoning ability?
The answer to this question will not help to
judge explanations for performance on individual
tasks, or address whether or not participants
hold incorrect theories about physics. However,
our experiment will gather evidence which will
inform the worth of considering intuitive physics
a single literature.
Assessments
Sample
b. Factor analysis
a. Tests of Reliability

• Intuitive Physics Items
• 5 pairs of standard intuitive physics tasks (see
  below)
• Standardized battery of Mechanical Reasoning
  Items
• 6 items from the Woodcock Johnson Mechanical
  Reasoning subtest
• Standardized Battery of Fluid Intelligence
  Items
• Average of the z-scores from the Woodcock
  Johnson Analysis-Synthesis and Wechsler Block
  Design
• Standardized Battery of Crystallized
  Intelligence Items
• Average of the z-scores from the Wechsler
  Vocabulary subtest and the Woodcock Johnson
  Picture Vocabulary subtest

• The intuitive physics items are strongly
  correlated with the mechanical reasoning items,
  suggesting that they reflect the same underlying
  ability.
• Cronbachs coefficient a
• Analogous to testing for measurement error
• Adjusted for binary values
• On a scale from 0 to 1
• Intuitive Physics coefficient a 0.74
• Nunnally (1978) sets criteria at 0.70

• Even if factors do exist, they are weak.
• 5 factors with eigenvalues greater than 1.
• Only the first factor accounts for more than 6
  of variance (19.1)
• all five together only account for 36.2 of
  variance
• Most of these factors are actually bloated
  specifics
• the same problem phrased two different ways

• 204 participants
• Ages 20-91
• Broad cross section of healthy adults
• only 7 in school
• Mean years of education 16

Intuitive Physics Items
Water Level Problem
Pendulum Problem
C- Shaped Tube Problem
3. Intuitive physics performance is strongly
correlated with fluid intelligence.
2. This disposition is persistent across the
adult lifespan.
On the left is a glass of water sitting on the
table. On the right, the glass has been tilted.
Assume that the glass and water are now
stationery. Draw the line showing the top
surface of the water, given that it ends at the
indicated point.
The c-shaped tube you see below is lying flat on
a table (this is an aerial view). A ball is
placed in one end and launched so that it
proceeds through the tube and out the other side.
Draw the path that the ball will take once it
exits the tube, starting at the given point.
The pendulum below is swinging from side to side.
The string is cut when the pendulum reaches its
highest point (indicated by the scissors). Draw
the path of the pendulum bob as it falls to the
ground.
Archimedes Principle Problem
Falling Object Problem
This airplane is carrying a canister of supplies
as it flies over a field. The plane drops the
canister. Draw the path that the canister will
follow before it hits the ground.
The picture on the left shows a styrofoam disc
floating in a tub of water. A clay ball is
attached to the top of the disc with glue. The
water level on the tub is marked with an X.
The disc and the ball are then removed from the
tub and placed back into the tub, upside down (as
shown). Where will the water level be relative
to the case on the left? (above, the same, or
below the X) (circle one)
The performance of participants in our study
offers compelling evidence that intuitive physics
performance is indeed reflective of a general
mechanical reasoning ability. Furthermore, this
ability is stable across the adult lifespan, as
well as highly correlated with measures of fluid
cognitive ability. Before we situate our
evidence in the context of the current intuitive
physics literature, however, it is important to
state that the existence of a mechanical
reasoning ability does not rule out in principle
either a theory approach or individual problem
explanations. Rather, it can be seen as a
possible source of confirmation and clarification
for certain explanations within these accounts.
Future researchers can design intuitive physics
studies knowing that a certain amount of the
noise that they see in their performance data
can be attributed to individual differences in a
mechanical reasoning ability. By assessing this
ability prior to administering the intuitive
physics problems, this noise could be filtered
out, allowing other patterns in the data to be
more easily recognizable. Applying our findings
to the two major approaches discussed in the
introduction, the evidence for an underlying
mechanical reasoning ability seems to lend
support to the account discussed by Proffitt and
Gilden (89). Our evidence suggests an expansion
on this account. Rather than splitting intuitive
physics tasks between easy and hard categories
along the lines of problem complexity, perhaps
items can be placed on a continuum of difficulty.
This scale would complement the scale of
intuitive physics performance described in our
results. Finally, we must now question our
original motivation for studying intuitive
physics tasks. If everyone has some level of
this ability, and some problems are harder than
other, perhaps the surprising errors observed
on intuitive physics tasks in the past is only
surprising to those who get them right.

• Does intuitive physics performance reflect a
  unitary mechanical reasoning disposition?
• Is this disposition persistent across the adult
  lifespan?
• How is intuitive physics related to other
  cognitive abilities?

Mechanical Reasoning Items
The Mechanical Reasoning Items are made up of 6
multiple choice items which require mechanical
reasoning using physical concepts such as center
of mass, gravity, and collision dynamics.
Fluid and Crystallized Intelligence Tests
Fluid intelligence was measured by taking an
average of the z-scores from the Woodcock Johnson
Analysis-Synthesis and Wechsler Block Design
subtests. These subtests provide an approximate
measure of ones facility with processing given
information, regardless of experience or stored
knowledge. Crystallized intelligence was
measured by taking an average of the z-scores
from the Woodcock Average of the z-scores from
the Wechsler Vocabulary subtest and the Woodcock
Johnson Picture Vocabulary subtest. These
subtests provide an approximate measure of the
level of stored knowledge that an individual has
acquired by testing vocabulary and picture
recognition.

• Champagne, A. B., Klopfer, L. E. Anderson, J.
  H. (1980). Factors influencing the learning of
  classical mechanics. American Journal of Physics.
  48, 1074-1079
• Caramazza, A., McCloskey, M., Green, B. (1981).
  Naive beliefs in "sophisticated" subjects
  Misconceptions about trajectories of objects.
  Cognition, 9(2), 117-123.
• McCloskey, M. (1983) Intuitive physics.
  Scientific American. 248(4), 122-130.
• Piaget, J., Inhelder, B. (1956). The childs
  conception of space. London Routledge and Kegan
  Paul.
• McCloskey, M., Kohl, D. (1983). Naive physics
  The curvilinear impetus principle and its role in
  interactions with moving objects. Journal of
  Experimental Psychology Learning, Memory,
  Cognition, 9(1) 146-156.
• Cooke, N. J., Breedin, S. D. (1994).
  Constructing naive theories of motion on the fly.
  Memory Cognition, 22(4), 474-493.
• Kaiser, M. K., Jonides, J., Alexander, J.
  (1986). Intuitive reasoning about abstract and
  familiar physics problems. Memory Cognition,
  14(4), 308-312.
• McAfee, E. A., Proffitt, D. R. (1991).
  Understanding the surface orientation of liquids.
  Cognitive Psychology, 23(3)
• Proffitt, D. R., Gilden, D. L. (1989).
  Understanding natural dynamics. Journal of
  Experimental Psychology Human Perception
  Performance, 15(2), 384-393.

This research was supported by funds from the
Virginia Aging Initiative directed by Timothy
Salthouse