Journal of Experimental Psychology: General © 2010 American Psychological Association

2010, Vol. 139, No. 4, 625– 637 0096-3445/10/$12.00 DOI: 10.1037/a0020977

Situated Naı̈ve Physics:

Task Constraints Decide What Children Know About Density

Heidi Kloos Anna Fisher

University of Cincinnati Carnegie Mellon University

Guy C. Van Orden

University of Cincinnati

Children’s understanding of density is riddled with misconceptions— or so it seems. Yet even preschool-
ers at times appear to understand density. This article seeks to reconcile these conflicting outcomes by
investigating the nature of constraints available in different experimental protocols. Protocols that report
misconceptions about density used stimulus arrangements that make differences in mass and volume
more salient than differences in density. In contrast, protocols that report successful performance used
stimulus arrangements that might have increased the salience of density. To test this hypothesis, the
present experiments manipulate the salience of object density. Children between 2 and 9 years of age and
adults responded whether an object would sink or float when placed in water. Results indicated that
children’s performance on exactly the same objects differed as a function of the saliency of the dimension
of density, relative to the dimensions of mass and volume. These results support the idea that con-
straints—rather than stable knowledge— drive performance, with implications for teaching children
about nonobvious concepts such as density.

Keywords: physics misconceptions, perceptual learning, preschoolers

A naı̈ve mind appears riddled with misconceptions— overly The argument put forward in this article is that misconceptions
simple ideas about the world that were not explicitly taught. originate in the competition between immediately available con-
Misconceptions exist about biology, astronomy, chemistry, and straints (cf. Kloos & Van Orden, 2009, 2010). The term constraint
physics, to name just a few domains, and they exist among a wide refers to a relation between actor and task that changes the avail-
range of individuals, including infants, preschoolers, high-school able degrees of freedom for task responses (Bernstein, 1967;
students, college students, and adults.1 Yet the nature of miscon- Flach, Dekker, & Stappers, 2008; L. B. Smith, Thelen, Titzer, &
ceptions is still debated. Why, for example, do children and adults McLin, 1999; Turvey, 1990; Turvey & Carello, 1981). Constraints
mistakenly equate the density of a block with its heaviness? Are are neither a reflection of the task protocol alone nor a reflection
there cognitive limitations when it comes to differentiating density of the capacities of the actor alone. In other words, performance is
from heaviness? Or is ordinary exposure to the consequences of neither a mere reflection of the environment nor a half-silvered
density (e.g., sinking vs. floating) too limited? Or are there other mirror on the actor’s mental structures. Instead, it is the coming
reasons altogether? together of the immediate task context and an actor’s prior history,
which reduces the available options leading to a particular perfor-
mance.
Illustrative of this principle is the task in which a child searches
This article was published Online First September 20, 2010.
Heidi Kloos and Guy C. Van Orden, Center for Cognition, Action, and for a toy hidden in one of two locations: A or B (Piaget, 1963).
Perception, Department of Psychology, University of Cincinnati; Anna Factors that determine the location at which the child would search
Fisher, Department of Psychology, Carnegie Mellon University. include the task instructions, the attractiveness of the toy, and the
This work was supported by National Science Foundation Grant NSF child’s capacity to remember where the toy was seen last. Other,
DRL 723638 and National Institute of Child Health & Human Develop-
ment Grant NICHD HD055324 awarded to Heidi Kloos and the National
1
Science Foundation’s Behavioral and Cognitive Sciences Grant BCS Sample references include Chinn & Brewer, 1993; Dunlop, 2000;
0642716 and Dynamics of Human Behavior Grant DHB 0728743 awarded Eryilmaz, 2002; Ferrari & Chi, 1998; Flavell, 1999; Gelman & Raman,
to Guy C. Van Orden. We thank Daniel Baum, Chris Erb, Adrienne 2002; Gopnik & Meltzoff, 1997; Krist, Fieberg, & Wilkening, 1993;
Frazier, and Ahn Thu Inman for their help with materials and stimuli, and Mazens & Lautrey, 2003; McKinnon & Geissinger, 2002; McKinnon,
we thank Michael Adom, Becky Fenstermaker, Cathy Odar, and Anna Geissinger, & Danaia, 2002; Mikkilä-Erdmann, 2001; Nakhleh, 1992;
Silverman for their help with data collection. Anna Haussmann provided Nguyen & Rosengren, 2004; Ozay & Oztas, 2003; Pine, Messer, &
valuable input on drafts of this article. Godfrey, 1999; Pine, Messer, & St. John, 2001; Riemeier & Gropengiesser,
Correspondence concerning this article should be addressed to Heidi 2008; Springer, 1995; Van Dooren, De Bock, Hessels, Janssens, & Ver-
Kloos, Department of Psychology, University of Cincinnati, 230 Dyer schaffel, 2004; Walz & Kerr, 2007; Wellman & Gelman, 1992; Wilkening
Hall, Cincinnati, OH 45221-0376. E-mail: heidi.kloos@uc.edu & Martin, 2004; Yip, 1998.

625
626 KLOOS, FISHER, AND VAN ORDEN

less obvious, factors include the spatial distance between the two mass and volume (e.g., Kloos & Van Orden, 2005; C. Smith,
hiding places A and B, the temporal delay between hiding and Carey, & Wiser, 1985). For example, children are presented with
searching, the details of the hiding events, changes in the child’s pairs of objects for which the heaviest objects have the lowest
coordination of reaching, and probably many others (e.g., L. B. density. In these task contexts, children perform poorly (cf. Hew-
Smith et al., 1999). The immediate relations between all these son & Hewson, 1983; Penner & Klahr, 1996; Piaget, 1963; C.
factors constitute the constraints on the child’s performance at a Smith et al., 1985). For instance in C. Smith et al. (1985), 80% of
given time. 3-year-olds, 70% of 4-year-olds, 81% of 5-year-olds, and 62% of
The constraints view has implications for the study of miscon- 6- to 7-year-olds incorrectly concluded that a less dense object was
ceptions far beyond a mere acknowledgment that a task context made out of “heavier kind of stuff” (p. 197). And when an
can affect a child’s performance. First, this view presupposes aluminum block was heavier than a steel block, only 33% of 8- to
interdependence between a child’s environment and mental con- 9-year-olds could categorize the blocks correctly on the basis of
structs. Like woven threads that become cloth, “knowledge is density.
fundamentally a coproduction of the mind and world, which like a Such findings have been taken to suggest that children start out
woof and warp need each other to produce and to complete an with some mistaken idea about density, maybe that density is
otherwise incoherent pattern” (Hung, 2002; p. 394). In other equivalent to mass (cf. Carey, 1985). Nonetheless, there are also
words, knowledge and task are not separate causal factors but reports of competent density performance in young children. Most
mutually codetermining factors of a child’s performance. As a notable is Kohn’s (1993) study, in which children responded
consequence, it is not possible to empirically isolate specific whether objects would sink or float. No feedback was provided,
mental structures or deficiencies that might elicit misconceptions and no object was ever placed in water. Yet even preschoolers
(e.g., a competence that is yet blind to differences between density performed above chance when they had to distinguish between
and heaviness). Neither knowledge nor the lack of it can be sinkers and floaters, without being misled by mass or volume.
isolated empirically. Instead, misconceptions—like every cogni- Similar success was reported in a steel-and-aluminum sorting task
tive performance— entangle the factors that underlie the expressed used in the C. Smith et al. (1985) study mentioned previously. The
misconception. Successful and unsuccessful performances both proportion of incorrect responses, for the same objects, was cut in
entail the relevant congruent and opposing constraints present in a half when objects were judged one by one rather than in pairs
“knowledge” response. pitting mass against density.
Second, an emphasis on constraints makes unique predictions One explanation for this apparent contradiction is that success-
about how to support veridical understanding and conceptual ful performance may stem from an implicit rather than explicit
change. It underscores the importance of aligning task constraints understanding of density (Kohn, 1993). Another explanation is that
so as to favor veridical performance from the beginning. In con- children are more successful on less complex tasks than on more
trast, a conventional way to correct a misconception is to create a complex tasks (C. Smith et al., 1985). Both of these explanations
task context in which mistaken performance is elicited first (e.g., assume the primacy of knowledge about mass and volume over
Kloos & Somerville, 2001). The idea of such an approach is to density. But this assumption is difficult to reconcile with instances
change a misconception by highlighting its shortcomings through in which people fail to ignore density, even with explicit instruc-
juxtaposition. However, giving priority to the mistaken perfor- tions to do so. For instance, when presented with two equally
mance might, in fact, be counterproductive if it increases the heavy objects that differ in size, children and adults judge the
salience and possible coherence of constraints that support making bigger object as being heavier than the smaller object, a finding
an error. Instead, a child may be made less susceptible to irrelevant known as the size–weight illusion (Pick & Pick, 1967; Robinson,
features of the task if the task context reliably tips the balance of 1964; Stevens & Rubin, 1970). In particular, because a heaviness
constraints in the direction of veridical performance. judgment requires the participant to lift the object, and thus to
Before these ideas can be explored, the importance of immedi- rotate it around a joint (the wrist, the elbow, or the shoulder),
ately available constraints needs to be demonstrated empirically. perceived heaviness is determined by the object’s rotational iner-
With this goal in mind, we focus on children’s misconceptions tia, which is a function of object density (Amazeen & Turvey,
about object density. 1996; Kloos & Amazeen, 2002). Apparently, some task conditions
make it difficult to perceive heaviness without density intrusions;
Misconceptions in Density Tasks yet other task conditions make it difficult to perceive density
without heaviness intrusions.
Material density is defined by the ratio of an object’s mass over
volume. Therefore, any manipulation of density is also a manip- Constraints on Density Performance
ulation of the relation between mass and volume, neither of which
predicts density by itself. It is commonly assumed that density has As we suggested earlier, rather than reflecting stable miscon-
to be inferred from the relation of mass to volume, making density ceptions (or stable veridical knowledge, for that matter), a stable,
something of a hidden feature, compared with the more obvious repeatable performance reflects a stable configuration of con-
features of mass (e.g., heaviness) or volume (e.g., size). This is one straints in the relations between participant and task. That is to say,
reason that density is a concept covered in school (e.g., Ohio misconceptions are neither necessarily a reflection of different
Department of Education, 2009), making it relevant for investiga- types of knowledge (implicit vs. explicit knowledge) nor a reflec-
tion. tion of different cognitive demands (easy vs. difficult tasks).
Tasks that test for understanding of density often involve sce- Instead, both successful and unsuccessful performances might
narios in which variations in density are pitted against variations in reflect the tipping point of a ratio of opposing constraints (for a
 SITUATED NAIVE PHYSICS 627

review, see Van Orden, Kloos, & Wallot, 2009). If so, then what
constraints are prominent in performing density judgments?
Density tasks involve category judgments (e.g., “Does it sink or
float?”; “Which one is made of the heavier kind of stuff?”).
Nonobvious constraints in this case might be supplied, in part, by
the relative salience of features that vary in the task context.2
Indeed, in tasks that elicited successful density performance, mass,
volume, and density were distributed across objects in a way that
made it difficult, if not impossible, to compare objects on any
dimension other than density. For example, in Kohn’s (1993)
study, children were never given an opportunity to compare ob-
jects side by side, and differences in mass or volume were never
brought to children’s attention. Conversely, in tasks that elicit
unsuccessful density performance, children were either explicitly
encouraged to focus on mass and volume in isolation (e.g., by
asking children to point to the heavier of two objects), or objects Figure 1. Combination of object pairs and predicted density judgments.
differed in many salient properties other than density, including Pairs of cubes differ in volume (shown as different sizes), mass (the heavier
shape, function, and color (Piaget, 1963). Thus, the latter task of the two objects is underlined), and/or density (the denser of the two
contexts were such as to maximize constraints that favored a focus objects has a darker outline). Combination 4 was used in Experiment 1 (as
on properties other than density. part of the pair-presentation trials) and in Experiment 3 (as part of the
This difference in how tasks were implemented—rather than an density-unconfounded pairs).
implicit versus explicit density understanding (or high vs. low
cognitive demand)— could account for the differences in density dimension that functions as a category marker. That is to say, the
performance just discussed. The experiments we report here tested salience of density as a category marker in a pair trial is reduced
this hypothesis explicitly. Does a nonobvious manipulation of con- with respect to the salience of mass or volume as a category
straints change the apparent knowledge children exhibit in density marker, at least so long as mass is pitted against density within a
tasks? We hypothesized that salience of object properties determines pair. Therefore, unless mass and density are correlated positively,
whether a child performs correctly— even for the same objects. density will be less salient in paired presentations than when cubes
One intuitive way to manipulate the salience of objects’ prop- are presented one by one. The goal of Experiment 1 was to test
erties would be to manipulate the distribution of mass and volume whether such a difference in the salience of density affects chil-
factorially with the distribution of density. However, given that dren’s performance on density tasks.
density is a ratio of mass and volume, any variations in density are
also variations in mass or volume. It is therefore not possible to
construct a within-object factorial manipulation. Instead we ma-
Experiment 1
nipulated presentation conditions and how objects’ properties were As in Kohn’s (1993) study, participants were asked to predict
distributed across the set of objects that were presented to partic- whether cubes would sink or float in water. Unlike in previous
ipants. A set of cubes was constructed such that half of the cubes research, however, we changed the experimental protocol to ma-
had one density (they floated) and the other half of the cubes had nipulate the salience of object density. For pair-presentation trials,
another density (they sank). Mass and volume, on the other hand, objects appeared in pairs for which density was not a salient
varied incrementally, such that there were seven different levels of category marker. In particular, mass was either held constant while
mass and six different levels of volume. Adjacent values of mass volume and density varied (see Combination 2 in Figure 1), or
and volume increased in relatively small increments (i.e., approx- mass and volume were pitted against density (see Combination 4
imately 60%) compared with the single increment of density (i.e., in Figure 1). For single-presentation trials, the same objects were
400%). As a result of this manipulation, variations in density presented one by one as part of a larger set of objects for which
demarcate a category (e.g., floaters vs. sinkers), whereas neither density was the most salient category marker across presentations.
variations in mass nor variations in volume could function as clear Children between 2 and 9 years of age and adults participated. We
category markers. Therefore, across the entire set of cubes, with predicted that even young children could judge whether a cube will
each cube presented one by one, density is likely to be the most sink or float, as long as variation in object density is made salient.
salient dimension in a context in which category membership
needs to be determined.
2
Such one-by-one presentation of cubes is contrasted by presen- Note that salience—like any constraint more generally—is never a
tation of the same cubes in pairs. Figure 1 shows possible combi- function of task context alone, nor is it a function of a person’s history
nations of objects in schematic form. As in the symbolism of C. alone. Even large changes in a task context will be ignored if the actor’s
propensities to perform are not aligned with the changes in the task context.
Smith et al. (1985), the heavier object in a pair is underlined, and
For example, adults instructed to focus on the ball handling in a basketball
the denser object has the darker outline. Pairs of objects could game will fail to notice a man in a gorilla suit pounding his chest on the
differ in mass and density (Combination 1), in volume and density basketball court (Simons & Chabris, 1999). On the other hand, even a tiny
(Combination 2), in mass and volume (Combination 3), or in all change in a task context can affect performance if the learner has become
three dimensions (Combinations 4 and 5). In pairs that differ in attuned to the relevant dimensions, as documented in numerous instances
density as well as in mass or volume, density is no longer the only of perceptual learning (Gibson, 1963; Hall, 1991).
628 KLOOS, FISHER, AND VAN ORDEN

In contrast, we expected that even older children would make

mistakes when the salience of density is relatively low.

Method
Participants. Twenty-one adults (14 women, 7 men) and 58
children between 2 and 9 years of age (25 girls, 33 boys) partic-
ipated in this experiment. Adults were recruited through the sub-
ject pool of introductory psychology classes and received partial
course credit for participation. Children in this and all subsequent
experiments were recruited from urban and suburban middle-class
day care centers and elementary schools. Five additional children
were tested but omitted from the analyses because they lost interest
and failed to complete the task.
Materials. Ten wooden cubes were constructed according to
the specifications of mass, volume, and density shown in Figure 2
(adapted from Kohn, 1993): All objects were one of two densities
(2.0 g/cm3 or 0.5 g/cm3), one of seven masses (between 30 g and
503 g), and one of six volumes (between 39 cm3 and 405 cm3).
Neither mass nor volume alone was absolutely predictive of the
cube’s sinking behavior: Cubes with a mass of 77 g, 123 g, or
197 g and cubes with a volume of 64 cm3, 104 cm3, 166 cm3, or
262 cm3 could either sink or float. This full set of cubes, presented
one at a time, was used for the single-presentation trials.
A subset of six cubes (from the total of 10) were combined into
pairs for the pair-presentation trials, portrayed with connecting
lines in Figure 2. Within a pair, one cube would sink if placed in
water (density ⫽ 2 g/cm3) and the other would float (density ⫽ 0.5
g/cm3). In addition, cubes within a pair could differ either in
volume (solid lines) or in mass and volume (dashed lines). Given
that mass either did not correlate with density or correlated nega-
tively, we predicted lower accuracy on these trials than on single-
presentation trials (see Combinations 2 and 4, respectively, in
Figure 1).
Cubes were hollowed out and filled with lead and wood putty
until the desired mass was obtained. Care was taken to distribute Figure 2. Masses, volumes, and densities of objects used in Experiments
mass equally throughout each cube. Once the cube was filled and 1–3. Values displayed on cubes reflect their volume (in cm3). The lines
closed, individual sides were painted in bright colors (each side a connecting cubes show how cubes were combined into pairs for the
different color), such that cubes could not be distinguished from pair-presentation trials of Experiment 1, with density correlating either
each other on the basis of color alone. A string of number digits, with volume alone (solid line; see Combination 3 in Figure 1) or with both
mass and volume (dashed line; see Combination 4 in Figure 1). Cubes
added surreptitiously to a side of the cube, made it possible for the
inside the dashed box were used in both conditions.
experimenter to distinguish between the cubes.
Procedure. Participants were tested in a quiet room (either at
their school or in the lab) by hypothesis-blind experimenters. The collection for this and all subsequence experiments was supported
cover story involved a character named Wump, who found special by SuperLab Pro 2.0 software.
rocks on a faraway planet and wanted to know whether the rocks Throughout the experiment, there were no practice trials, no
would sink or float in earth water. For single-presentation trials, cube was ever placed in water, and children did not receive
participants were presented with one cube at a time, each of the 10 feedback on their performance. Therefore, learning across an ex-
cubes occurring twice in random order. Participants were asked to perimental session was unlikely. Participants were first presented
pick up the cube and report whether it would sink or float. For with pair-presentation trials and then with single-presentation tri-
pair-presentation trials, participants were presented with a pair of als, because we predicted better performance in the latter kind of
cubes at a time, each of the six pairs occurring twice in random trials. Had the single-presentation trials been presented first, chil-
order, with counterbalanced left–right arrangement. For each pair, dren might have been sensitized to density variations in the entire
participants were asked to pick up the two cubes and report set, inflating performance on pair-presentation trials. Therefore,
whether one cube would sink and the other float, both cubes would the ordering of the tasks used in Experiment 1 runs counter to our
sink, or both cubes would float. Participants could hold the cubes predictions, because children could potentially become attuned to
for as long as they wanted, and they were prompted to simply mass and volume during pair-presentation trials and subsequently
guess if they hesitated to respond. Randomization of trials and data underperform during single-presentation trials.
 SITUATED NAIVE PHYSICS 629

Results and Discussion children’s proportion of correct responses (across trials) as a function
of age. There was an overall improvement across age, r(56) ⫽ .50,
The first preliminary analysis pertained to children’s perfor-
p ⬍ .001. As can be seen in the figure, the youngest participants (2-
mance on pair-presentation trials, when children had to decide
to 3-year-olds) performed at chance (choosing the correct option in
whether one of the cubes in a pair would sink, both would sink, or
fewer than 15 out of 20 trials; binomial probability p ⬎ .12, assuming
both would float. On a majority of these trials (73%), participants
a chance probability of .5). However, performance improved with
determined that one cube would sink and the other would float.
age: Five 4- to 5-year-olds (out of 15; 33%) performed consistently
The answer option both sink was chosen on 18% of trials, whereas
above chance, and by 6 years of age, the majority of children (at least
the answer option both float was chosen on 9% of trials.
80% per age group) performed above chance. No systematic pattern
Were children affected by the particular combination of cubes
(Combination 2 vs. Combination 4; see Figure 2)? To answer this of performance was apparent when contrasting cubes within the
question, we collapsed the proportion of correct responses across subset (e.g., the heaviest vs. lightest cubes).
the three pairs of Combination 2 and across the two pairs of To compare performance on pair-presentation trials with per-
Combination 4. A repeated-measures analysis of covariance, with formance on single-presentation trials, we rescored performance
age as a covariate, revealed no main effect or interaction of on pair-presentation pairs to gauge performance on each individual
combinations (all Fs ⬍ 0.5, ps ⬎ .51). We therefore collapsed each cube. This allowed us to equate chance probability across the two
child’s performance across the pair-presentation trials. Figure 3A types of trials at 50% (children could be either correct or incorrect
displays individual children’s proportion of correct responses on for each particular cube). Furthermore, we restricted our analysis
pair-presentation trials as a function of age. There was no signif- to those cubes that appeared in both the pair- and the single-
icant improvement across age, r(56) ⫽ .16, p ⬎ .23, with only 48% presentation trials (shown in the dashed box of Figure 2). Figure 4
of children (28 out of 58) performing significantly above chance shows the mean proportion of correct performance as a function of
(choosing the correct option on six or more of the 10 trials; trial type and the following age groups: 2- to 3-year-olds (n ⫽ 12),
two-tailed binomial probability p ⬍ .05, assuming a chance prob- 4- to 5-year-olds (n ⫽ 15), 6- to 7-year-olds (n ⫽ 18), 8- to
ability of .25). 9-year-olds (n ⫽ 13), and adults (n ⫽ 21).
The second preliminary analysis pertained to children’s perfor- A 2 ⫻ 5 mixed-design analysis of variance (ANOVA) was
mance on single-presentation trials, when children had to decide conducted, with trial type as the within-subject factor and age
whether a cube would sink or float. Figure 3B shows individual group as the between-subjects factor. The analysis revealed a

Figure 3. Individual children’s proportion of correct performance on pair-presentation trials (A) and single-
presentation trials (B). Solid lines represent chance performance ( p ⫽ .25 for the four-choice pair-presentation
trials; p ⫽ .50 for the two-choice single-presentation trials). Dashed lines demarcate performance that is different
from chance using the binomial probability test (above chance for A: performing correctly on at least six out of
10 trials; above chance for B: performing correctly on at least 15 out of 20 trials; below chance for B: performing
correctly on less than five out of 20 trials).
630 KLOOS, FISHER, AND VAN ORDEN

1.0
Density-Less-Salient
Pair Presentation
*
.9 Single Presentation
Density-More-Salient *

.8
*

Mean % Correct
.7

.6

.5

.4

.3
2- to 3-year-olds 4- to 5-year-olds 6- to 7-year-olds 8- to 9-year-olds Adults

Figure 4. Mean proportion of correct judgments in Experiment 1 for cubes that were presented as part of a pair
(pair-presentation trials) or alone (single-presentation trials), separated by age group. Chance performance is .5.
Error bars represent standard errors, and asterisks represent reliable differences between trial types (pair
presentation vs. single presentation).

reliable linear trend of age, F(4, 74) ⫽ 7.15, p ⬍ .01, with The younger age group included ten 2- and 3-year-olds (Mage ⫽
participants performing better with increasing age. More impor- 3.4 years), and the older age group included ten 4- and 5-year-olds
tantly, the effect of trial type was statistically reliable, F(1, 74) ⫽ (Mage ⫽ 5.01 years).
8.1, p ⬍ .01. Children performed better on single-presentation Materials and procedure. Materials and procedures were
trials (M ⫽ .75, SD ⫽ .23) than on pair-presentation trials (M ⫽ identical to those used during the single-presentation trials of
.66, SD ⫽ .27). Planned paired-sample t tests revealed an effect of Experiment 1. The only difference was the number of trials. This
trial type for the three older age groups (6- to 7-year-olds, 8- to time there were three blocks of trials (10 trials per block), and each
9-year-olds, and adults), paired-sample ts ⬎ 2.1, ps ⬍ .05, but not cube was presented once per block (resulting in a total of 30 trials).
for the younger children between 2 and 5 years of age ( ps ⬎ .6).
In particular, children 6 years and older performed better on the
single-presentation trials than on the pair-presentation trials, Results and Discussion
whereas children 5 years and younger failed to produce a reliable
difference. How well could children sort cubes into sinkers versus floaters
Consistent with our prediction, adults and children 6 years and when it was the first task they were asked to complete? Perfor-
older were affected by the trial-type manipulation. Participants mance of the 2- to 3-year-olds did not exceed chance level,
who performed above chance were more likely to do so for whether considering their performance during a single block of 10
single-presentation trials than for pair-presentation trials. This was trials (MBlock 1 ⫽ .53, MBlock 2 ⫽ .53, MBlock 3 ⫽ .50; SDs ⬍ .16)
true even though the same cubes were considered in both types of or across the total of 30 trials (M ⫽ .52, SD ⫽ .07), single-sample
trials. Experiment 2 takes a second look at younger children’s ts ⬍ 0.9. Only one of the ten 2- to 3-year-olds performed above
ability to pick up on difference in density, whereas Experiment 3 chance in one block of trials (choosing the correct answer in eight
investigates whether extraneous factors of the pair-presentation, out of 10 trials, binomial probability p ⬍ .05). All other 2- to
other than the targeted difference in salience, could explain chil- 3-year-olds performed at chance, whether considering only the
dren’s low performance. first block of trials or across all 30 trials.
Many 4- and 5-year-olds, on the other hand, were successful in
Experiment 2 this task. Six of the ten 4- to 5-year-olds performed above chance
during the first block, and five children performed above chance
An unexpected finding of Experiment 1 pertained to the rela- across all three blocks (choosing the correct answer in at least 21
tively weak performance of preschool children on single- of the 30 trials; binomial probability p ⬍ .05). Figure 5 shows 4-
presentation trials. Previous research (e.g., Kohn, 1993) found to 5-year-olds’ performance in each block of trials, separated by
above-chance performance for this age group, yet in our experi- whether we considered all 10 cubes presented or just the subset of
mental setup, children younger than 6 years of age performed at
cubes that were the focus of analysis in Experiment 1. Performance
chance. It is possible that children’s performance on single-
was above chance in both cases, whether we considered Block 1 only,
presentation trials suffered from fatigue or boredom, or some other
Mall cubes ⫽ .72, Msubset of cubes ⫽ .77, SDs ⫽ .24; single-sample
confounded factor, given that these trials always occurred last. In
ts(9) ⬎ 2.85, ps ⬍ .02, or considered all 30 trials, Mall cubes ⫽ .70,
Experiment 2, we tested this hypothesis, presenting children be-
Msubset of cubes ⫽ .69, SDs ⬍ .22; single-sample ts(9) ⬎ 2.92, ps ⬍ .02.
tween 2 and 5 years of age with single-presentation trials only (as
Importantly, performance in Block 1 was better than performance of
was done in Kohn, 1993).
4- to 5-year-old children on single-presentation trials of Experiment 1,
independent-sample t(23) ⫽ 1.79, p ⬍ .05.
Method
Taken together, the results show that 4- to 5-year-old children,
Participants. Children between 2 and 5 years of age (9 girls, as a group, perform successfully when presented with only single-
11 boys) participated, none of whom participated in Experiment 1. presentation trials, replicating Kohn’s (1993) finding. Apparently,
 SITUATED NAIVE PHYSICS 631

1.0 All Cubes cessfully. It is this ever-present context sensitivity of performance

.9 Subset of Cubes that is the heart of our hypothesis.
Mean % Correct

.8

.7 Experiment 3
.6 In Experiment 1, the salience manipulation was confounded— by
.5 necessity— by differences in whether objects were presented singly
.4 (single-presentation trials) or in pairs (pair-presentation trials). Mak-
ing a judgment about two cubes (pair-presentation trials) versus
.3
making a judgment about one cube might pose different cognitive
Block 1 Block 2 Block 3
demands on children. This difference in demand could account for
Figure 5. Mean proportion of correct judgments of 4- to 5-year-olds in the difference in performance reported in Experiment 1. Experi-
Experiment 2 for single cubes (single-presentation condition), separated by ment 3 was designed to test for this possibility: Are sink or float
session. Means displayed were computed both across all cubes and across judgments for a pair of cubes inherently difficult?
the subset of cubes that were entered into the analysis in Experiment 1. To answer this question, we presented children with pairs of
Chance performance is .5. Error bars represent standard errors. cubes only (see Figure 6). For one type of pair, mass correlated
positively with density, yielding density-confounded pairs. The
heavier object in a pair was the denser one, either with volume held
4- to 5-year-olds’ performance in Experiment 1 suffered from constant (double-dashed lines; see Combination 1 in Figure 1) or
presenting the single-presentation trials last, after the pair- with an added manipulation of volume (double solid lines; see
presentation trials. Of course, whatever factor was confounded Combination 5 in Figure 1). We expected competent performance
with order in Experiment 1 is yet another nonobvious and seem- concerning these pairs because children could perform correctly
ingly trivial factor that determines whether children perform suc- merely by paying attention to the difference in mass (while ignor-

Figure 6. Pairs of cubes used in Experiment 3. In density-confounded pairs (gray lines), mass and density
correlate positively, either with changes in volume (solid double lines; see Combination 5 in Figure 1) or with
volume fixed (dashed double lines; see Combination 1 in Figure 1). In density-unconfounded pairs (black lines),
mass and density correlated either negatively (dashed lines; see Combination 4 in Figure 1) or not at all (solid
curves; see Combination 3 in Figure 1). Children were exposed to one of the two density-confounded
combinations and one of the two density-unconfounded combinations. Comb. ⫽ combination.
632 KLOOS, FISHER, AND VAN ORDEN

ing differences in density). Or they could perform correctly be- across pairs presented twice within each combination—no differ-
cause the correlation between mass and density might heighten ence was found between different pairs within Combinations 1, 3,
attention to both features (cf. Fisher & Tokura, 1996; Morgan, and 5 (see Figure 1; repeated-measure one-way ANOVA, Fs ⬍
Meier, & Newport, 1987). 1.3, ps ⬎ .34). For the density-unconfounded pairs from Combi-
Conversely, for the other type of pair, mass and volume were nation 4, it is noteworthy that children were much more likely to
pitted against density, yielding density-unconfounded pairs. Mass perform incorrectly when the cube pair had maximal mass differ-
and volume were either negatively correlated with density (i.e., the ence (M120-g mass difference ⫽ .28) than when the mass difference
lighter and smaller object in a pair was always the denser one; was smaller (M46-g mass difference ⫽ .58; M74-g mass difference ⫽ .58),
dashed lines; see Combination 4 in Figure 1) or were uncorrelated planned-contrast repeated-measure one-way ANOVA, F(8) ⫽ 2.1,
with density (density was the same despite changes in mass and p ⬍ .05. This suggests that children are highly affected by the
volume; solid curves; see Combination 3 in Figure 1). For both
perceived difference in mass when cubes are combined in pairs.
combinations, we expected children to have difficulty correctly
We nevertheless combined performance on pairs within Combi-
identifying sinkers and floaters. Although the present experiment
nation 4 for subsequent analyses.
does not test children’s ability to attend to density, it does allow us
Children’s most common response was that one cube of the pair
to determine whether trials that present paired objects are inher-
would sink and the other would float, when that was the correct
ently difficult and thus could explain the low performance in
Experiment 1. response (see Combinations 1, 4, and 5 in Figure 1). However,
children had no difficulty per se with the response options both
float and both sink. For Combination 3, when both cubes floated or
Method both cubes sank, children chose these two response options in
almost half of the trials. Figure 7 shows the mean proportion of
Participants. Participants were twenty-five 4- and 5-year-old
response choices (both sink or both float vs. one sinks, one floats),
children who did not participate in the previous experiments (17
girls, 8 boys; Mage ⫽ 5.08 years). This age group might be best separated by combination. There was a significant difference in the
suited to test a possible effect of cognitive demand: Children at this proportion of one sinks, one floats response options between
age are sensitive to differences in density (see Experiment 2), yet Combination 3 (pairs for which both cubes sink or both cubes
their performance is more likely to suffer under the possibly higher float) and Combinations 1, 4, and 5 (pairs for which one cube sinks
cognitive demand of a pair-presentation context, compared with and the other floats), planned-contrast between-groups one-way
the performance of older children and adults. ANOVA (assuming each combination as a between-subjects fac-
Materials. Figure 6 shows how the cubes were combined in tor), F(1, 46) ⫽ 3.44, p ⬍ .03.
Experiment 3 to create density-unconfounded pairs and density- To address the main question of the experiment, Figure 8 shows
confounded pairs. Pairs of cubes corresponded to four of the children’s proportion of correct responses for density-confounded
combinations shown in Figure 1 (Combinations 1, 3, 4, and 5). For pairs and density-unconfounded pairs, either overall across trials or
two of these combinations (Combinations 1 and 5; depicted in gray for cubes that were shared across types of trials for a child. For
double lines), mass was positively correlated with density, yielding comparison purposes, we also displayed 4- to 5-year-olds’ perfor-
eight density-confounded pairs. These two combinations differ in mances on the pair-presentation trials of Experiment 1 and on the
whether volume was varied (solid double lines; see Combination 5 single-presentation trials of Experiment 2.3
in Figure 1) or not (dashed double lines; see Combination 1 in Two mixed-design ANOVAs were conducted— one for the trial
Figure 1). For the other two combinations (Combinations 3 and 4; data and one for the shared-cubes data—with pair type as the within-
depicted in black single lines), mass was pitted against density, subject factor (density-confounded, density-unconfounded) and group
yielding nine density-unconfounded pairs. These two combina- as the between-subjects factor (Group 1: children presented with
tions differed in whether mass and density correlated negatively Combinations 3 and 5; Group 2: children presented with Combina-
(dashed lines; see Combination 4 in Figure 1) or not at all (solid tions 1 and 4). There was a main effect of pair type, Fs(1, 24) ⬎ 10.5,
curves; see Combination 3 in Figure 1).
ps ⬍ .01, with children performing worse on density-unconfounded
Procedure. The procedure was identical to the pair-
trials than on density-confounded trials, whether we considered their
presentation procedure of Experiment 1 except for the different
performance on trials (Munconfounded ⫽ .46, Mconfounded ⫽ .72) or on
combinations of pairs that were presented. Some children were
shared cubes (Munconfounded ⫽ .60, Mconfounded ⫽ .80). Neither the
presented with Combinations 1 and 4 (see Figure 1), and other
children were presented with Combinations 3 and 5. This division main effect of group nor the interaction reached significance, Fs(1,
ensured that every child performed on density-confounded trials 24) ⬍ 1.7, ps ⬎ .20.
(Combination 3 or 4) as well as on density-unconfounded trials The high performance of preschoolers on the density-
(Combination 1 or 5), without creating an excessively long exper- confounded trials rules out the possibility that the difference be-
imental protocol. tween trial types reported in Experiment 1 was due to differences
in cognitive demand (e.g., due to pair judgments being inherently

Results and Discussion

3
This comparison was appropriate, given the order effect we found
Preliminary item analyses revealed no difference between the between Experiment 1 and Experiment 2 for 4- to 5-year-olds’ perfor-
first and second presentation of each unique pair (each unique pair mances in the density-salient condition (see the Results and Discussion
was presented twice), ts ⬍ 1.6, ps ⬎ .10, and—after collapsing section of Experiment 2).
 SITUATED NAIVE PHYSICS 633

both sink or both float Our findings make superfluous a discussion about what task
one sinks, one floats might reflect a child’s “real” competence. If available constraints
1.00
matter to successful performance, then successful performance in
Mean % of Response Types

a single-presentation condition is not an overestimation of a child’s

.75
competence. And unsuccessful performance in a pair-presentation
condition is not an underestimation of a child’s competence. Fur-
.50
thermore, and consistent with the well-known effect ⫽ structure
fallacy (e.g., Lakoff, 1990), the current findings question the need
.25
to postulate the existence of a mental structure to explain an effect.
Successful performance does not imply the existence of one kind
.00
of structure (e.g., implicit knowledge). And inaccurate perfor-
Comb. 5 Comb. 1 Comb. 4 Comb. 3 mance does not imply the existence of (or the lack of) a different
Figure 7. Mean relative frequency of choosing answer options one sinks,
kind of structure (e.g., explicit knowledge). Instead, differences in
one floats versus both float or both sink, separated by combination. Error performance in the current task can be explained by differences in
bars represent standard errors. Comb. ⫽ combination. the salience of density versus mass or volume. When density was
made salient, even preschool children performed well. And when
density was made less salient, performance decreased even among
more difficult than judgments about single objects). Children eas- adults. It remains to be seen whether differences in the salience of
ily distinguished between sinkers and floaters in a pair condition relevant dimensions could explain other context-dependent differ-
when density and mass correlated positively. This finding supports ences in performance (cf. Diamond, 1998; Ferretti & Butterfield,
the idea that children’s low performance in the pair-presentation 1986; Gigerenzer & Richter, 1990; Kaiser, Proffitt, & Anderson,
trials of Experiment 1 is due to our salience manipulation. 1985; Keen, 2003; Krist, Fieberg, & Wilkening, 1993; Levin,
Siegler, & Druyan, 1990; McCloskey, 1983).
General Discussion Note that similarity in performance does not always imply
We examined whether children’s understanding of density is similarity in constraints. Take for example the equivalent perfor-
constrained by the distribution of mass, volume, and density in the mance on single-presentation trials (Experiment 1) versus density-
immediate context of the task. Our hypothesis was that children confounded trials (Experiment 3; both Ms ⫽ .72). Despite similar
would correctly predict whether an object sinks or floats when performance, the constraints available in single-item presentations
variations in density are sufficiently salient. Results of Experi- are likely to differ from the constraints available in the paired-item
ments 1 and 2 support our predictions. By 5 years of age, partic- presentations. At the minimum, density was uncorrelated with
ipants performed above chance when objects were part of the set mass in the single-item presentation, and neither mass nor volume
for which density was distributed saliently. Performance decreased serves as a dichotomous category marker across these trials. Con-
when salience of density was reduced relative to the salience of versely, in successful pair trials, density correlated positively with
mass and volume. Results of Experiment 3 suggest that this dif- mass, making available an additional distinction to affect perfor-
ference was not merely due to differences in cognitive demand mance. The mere difference in whether there is more than one
when judging paired items versus single items. viable stimulus dimension may change the relation between the

Across Shared Cubes

Across Trials
1.0
Density-Unconfounded
Density-Less-Salient
.9 (Exp. 2: Single Presentation)
Density-Confounded
Mass/Density-Salient
.8
Mean % Correct

(Exp. 1: Pair Presentation)

.7
.6
.5
.4
.3
Comb. 4 vs. 1 Comb. 3 vs. 5 Comb. 4 vs. 1 Comb. 3 vs. 5 Exp.1 vs. Exp.2

Type of Pair

Figure 8. Mean proportion of correct judgments of 4- to 5-year-olds in Experiment 3, separated by whether

cubes were part of a density-unconfounded pair or a density-confounded pair (repeated-measure design). Data
represent performance across pairs of a certain type or for cubes shared across the pairs in a particular
combination. Corresponding means from Experiment 1 (pair-presentation trials) and Experiment 2 (single-
presentation trials) are displayed for comparison purposes. Error bars represent standard errors. Comb. ⫽
combination; Exp. ⫽ experiment.
634 KLOOS, FISHER, AND VAN ORDEN

child and the task protocol, yielding a different basis for perfor- initial focus on mass and volume actually impedes acquiring a
mance. formal understanding of density.
Although the current manipulation pertained to changes in the A better alternative might be to avoid false beliefs altogether,
task context, it is important to stress that constraints pertain more namely by structuring the learning environment in such a way as
broadly to relations between participant, instructions, procedures, to make the to-be-taught concept more salient (cf. McNeil, Uttal,
and stimuli that reduce or increase the degrees of freedom for Jarvin, & Sternberg, 2009; Son, Smith, & Goldstone, 2008; Uttal,
particular response actions. Thus, constraints are not found in the Scudder, & DeLoache, 1997). For example, one could adopt a
body, brain, or task, each working in isolation. They are discov- pedagogy in which density, rather than mass and volume, was
erable only in relations between the participant and task (e.g., most salient. The “dots-per-box” teaching model is a point in case
Havas, Glenberg, & Rinck, 2007). In the current experiments, (Maclin, Grosslight, & Davis, 1997; C. Smith, Maclin, Houghton,
changes in the composition of stimuli changed the landscape of & Hennessey, 2000; C. Smith & Unger, 1997; Snir, Smith, &
constraints for a child’s judgments of sinking and floating. If the Grosslight, 1993). Students were presented with schematic repre-
task were changed such that the instructions were to “throw the sentations of what an object would look like on the inside, assum-
cubes at a target,” the new task ecology would exclude previous ing mass is expressed as the number of dots and volume is
response options while expanding the degrees of freedom for arm
expressed as the number of same-sized boxes (hence, dots-per-box
trajectories in throwing. And if the cubes were too heavy for the
model). Density, therefore, is captured schematically through the
child to lift, the task ecology would change yet again.
visible distance between mass units within the available volume.
Along the same line, if a person’s knowledge (or more gener-
Middle-school children were found to benefit from this visualiza-
ally, a person’s prior history) were to be changed, then a new task
tion to understand density.
ecology would emerge. Recall, for example, that infants’ perfor-
Of course, given that knowledge is always grounded in task
mance in the A-not-B task, mentioned earlier, is a function (in
constraints, successful performance need not necessarily lead to an
part) of their motor history (i.e., the number of reaches to one of
the hiding locations prior to the crucial change in hiding location) abstract concept of density that will generalize to new contexts.
and their strength of memory traces for the hidden objects (i.e., Making density salient, such as through a dots-per-box model,
their knowledge about where the object was hidden). Similarly, allows children to perceive density directly, on the basis of atten-
children’s performance in the current experiments is a function (in tional processes that tune in to the most salient dimension. How-
part) of a child’s prior history with objects. For example, a child’s ever, such direct perception might not instill an understanding of
experience of heaviness as a downward force might heighten the the mathematical relation between mass, volume, and density.
salience of this dimension when it comes to judging falling or Such a lack of an abstract understanding of density might leave the
sinking behaviors (for a discussion, see Kloos & Van Orden, learner prone to mistakes when a misleading task context makes
2005).4 In other words, just as a change in task context changes inappropriate dimensions salient.
task ecology, a difference in the child’s prior history changes the In order to bootstrap a more formal understanding of the concept
task ecology. What kind of history would promote constraints to of density, one may need to combine a supportive task context with
embody a veridical concept of density? the protracted weaning of students from the scaffold available in
the supportive task context. This might be accomplished by pro-
viding children with diverse instances (O’Reilly & Munakata,
Learning About Density 2000; Reeves & Weisberg, 1994) or by explicitly contrasting a
supportive task context with one that leads to mistakes in perfor-
How can constraints be employed to instill a sophisticated
mance (cf. Gentner, Loewenstein, & Thompson, 2003; Loewen-
understanding of a physics concept such as density? Common
teaching practices designed to replace misconceptions focus on stein & Gentner, 2001; see Dixon & Bangert, 2004, for a compar-
eliciting a child’s false beliefs before the child is exposed to ison of the two approaches). Both approaches might allow a
science instruction (e.g., McDermott & Redish, 1999; Pfundt & student to come to embody stable and reliable biases, not unlike a
Duit, 1991). The idea behind such practices is that false beliefs will young child gaining stronger memory for hidden objects in an
trump formal instruction if their shortcomings are not pointed out A-not-B task. These embodied changes, though not empirically
explicitly (e.g., Kloos & Somerville, 2001). In the context of apparent, are likely to decrease the probability of making density
density instruction, such a procedure of initially engaging false errors in a misleading task context—the same way a stronger
beliefs might involve a protocol in which children are asked to memory for a hidden object decreases the likelihood of searching
relate differences in mass and volume to outcomes that are affected for it at the wrong location. The effects of task contexts can never
by density. be eliminated, however. For example, there are circumstances in
Results from our study suggest that such a procedure, rather than which even adults commit the A-not-B error. Likewise, a suffi-
calling to mind existing beliefs, might in fact create and perpetuate
them. For example, testing environments that make mass salient 4
In the current set of experiments, children did not show a bias toward
can lead children to rely incorrectly on mass to predict outcomes
focusing on mass alone. When mass was pitted against density (density-
such as buoyancy (cf. Hall, 1991). Of course, once beliefs are unconfounded pairs), mean performance was at chance, with a majority of
created, they are hard to correct, not unlike a persistent priming children failing to base their judgment on mass alone (and thus failing to
effect (cf. Gershkoff-Stowe, Connell, & Smith, 2006) or illusory perform below what would be predicted by chance alone). However, in
correlation effect (cf. Hamilton & Rose, 1980; Johnston & Jacobs, different contexts, children were indeed found to focus exclusively on mass
2003; Lord, Ross, & Lepper, 1979). It is therefore plausible that an (e.g., C. Smith et al., 1985).
 SITUATED NAIVE PHYSICS 635

ciently misleading task context can lead otherwise competent cognitive development. Human Development, 45, 61– 68. doi:10.1159/
adults to misjudge the density of a material. 000048152
Gentner, D., Loewenstein, J., & Thompson, L. (2003). Learning and
transfer: A general role for analogical encoding. Journal of Educational
Conclusion Psychology, 95, 393– 405. doi:10.1037/0022-0663.95.2.393
Gershkoff-Stowe, L., Connell, B., & Smith, L. (2006). Priming overgen-
Our results have demonstrated that minimal changes in available eralizations in two- and four-year-old children. Journal of Child Lan-
constraints can influence children’s and adults’ knowledge of guage, 33, 461– 486. doi:10.1017/S0305000906007562
density in predictable ways. These findings resolve a previous Gibson, E. J. (1963). Perceptual learning. Annual Review of Psychology,
conflict between reports about children’s naı̈ve understanding of 14, 29 –56. doi:10.1146/annurev.ps.14.020163.000333
density. The resolution stems from the idea that task performance, Gigerenzer, G., & Richter, H. R. (1990). Context effects and their inter-
successful or not, directly reflects the situated relations between action with development: Area judgments. Cognitive Development, 5,
the participant’s history and the details of the experimental proto- 235–264. doi:10.1016/0885-2014(90)90017-N
col. Consistent performance across age reflects consistently avail- Gopnik, A., & Meltzoff, A. N. (1997). Words, thoughts and theories,
able constraints rather than intrinsically stable beliefs or compe- Cambridge, MA: MIT Press.
tence. The demonstrated consequences of constraints underscore Hall, G. (1991). Perceptual and associative learning. New York, NY: Oxford
the importance of a scaffolded structure in the learning environ- University Press. doi:10.1093/acprof:oso/9780198521822.001.0001
Hamilton, D. L., & Rose, T. L. (1980). Illusory correlation and the
ment. Our results support learning and testing approaches that
maintenance of stereotypic beliefs. Journal of Personality and Social
include environmental support for competent performance and
Psychology, 39, 832– 845. doi:10.1037/0022-3514.39.5.832
only later weaning performance of its supportive structure. Havas, D. A., Glenberg, A. M., & Rinck, M. (2007). Emotion simulation
during language comprehension. Psychonomic Bulletin & Review, 14,
References 436 – 441.
Hewson, M. G., & Hewson, P. W. (1983). The effect of instruction using
Amazeen, E. L., & Turvey, M. T. (1996). Weight perception and the haptic students’ prior knowledge and conceptual change strategies on science
size-weight illusion are functions of the inertia tensor. Journal of Ex- learning. Journal of Research in Science Teaching, 20, 731–743. doi:
perimental Psychology: Human Perception and Performance, 22, 213– 10.1002/tea.3660200804
232. doi:10.1037/0096-1523.22.1.213 Hung, D. (2002). Situated cognition and problem-based learning: Implica-
Bernstein, N. A. (Ed.). (1967). The coordination and regulation of move- tions for learning and instruction with technology. Journal of Interactive
ments. London, England: Pergamon Press. Learning Research, 13, 393– 414.
Carey, S. (1985). Conceptual change in childhood. Cambridge, MA: Brad- Johnston, K. E., & Jacobs, J. E. (2003). Children’s illusory correlations: The
ford Books. role of attentional bias in group impression formation. Journal of Cognition
Chinn, C. A., & Brewer, W. F. (1993). The role of anomalous data in and Development, 4, 129 –160. doi:10.1207/S15327647JCD0402_01
knowledge acquisition: A theoretical framework and implications for Kaiser, M. K., Proffitt, D. R., & Anderson, K. (1985). Judgments of natural
science instruction. Review of Educational Research, 63, 1– 49. and anomalous trajectories in the presence and absence of motion.
Diamond, A. (1998). Understanding the A-not-B error: Working memory Journal of Experimental Psychology: Learning, Memory, and Cogni-
vs. reinforced response, or active trace vs. latent trace. Developmental tion, 11, 795– 803. doi:10.1037/0278-7393.11.1-4.795
Science, 1, 185–189. doi:10.1111/1467-7687.00022 Keen, R. (2003). Representation of objects and events: Why do infants look
Dixon, J. A., & Bangert, A. S. (2004). On the spontaneous discovery of a so smart and toddlers look so dumb? Current Directions in Psycholog-
mathematical relation during problem solving. Cognitive Science, 28, ical Science, 12, 79 – 83. doi:10.1111/1467-8721.01234
433– 449. doi:10.1207/s15516709cog2803_6 Kloos, H., & Amazeen, E. L. (2002). Perceiving heaviness by dynamic
Dunlop, J. (2000). How children observe the universe. Publications of the touch: An investigation of the size-weight illusion in preschoolers.
Astronomical Society of Australia, 17, 194 –206.
British Journal of Developmental Psychology, 20, 171–183. doi:
Eryilmaz, A. (2002). Effects of conceptual assignments and conceptual
10.1348/026151002166398
change discussions on students’ misconceptions and achievement re-
Kloos, H., & Somerville, S. C. (2001). Providing impetus for conceptual
garding force and motion. Journal of Research in Science Teaching, 39,
change: The effect of organizing the input. Cognitive Development, 16,
1001–1015. doi:10.1002/tea.10054
737–759. doi:10.1016/S0885-2014(01)00053-3
Ferrari, M., & Chi, M. T. H. (1998). The nature of naive explanations of
Kloos, H., & Van Orden, G. C. (2005). Can a preschooler’s mistaken
natural selection. International Journal of Science Education, 20, 1231–
beliefs benefit learning? Swiss Journal of Psychology, 64, 195–205.
1256. doi:10.1080/0950069980201005
Ferretti, R. P., & Butterfield, D. C. (1986). Are children’s rule assessment doi:10.1024/1421-0185.64.3.195
classifications invariant across instances of problem types? Child De- Kloos, H., & Van Orden, G. C. (2009). Soft-assembled mechanisms for the
velopment, 57, 1419 –1428. doi:10.2307/1130420 unified theory. In J. P. Spencer, M. Thomas, & J. McClelland (Eds.),
Fisher, C., & Tokura, H. (1996). Acoustic cues to grammatical structure in Toward a unified theory of development: Connectionism and dynamic
infant-directed speech: Cross-linguistic evidence. Child Development, systems theory re-considered (pp. 253–267). Oxford, England: Oxford
67, 3192–3218. doi:10.2307/1131774 University Press. doi:10.1093/acprof:oso/9780195300598.003.0012
Flach, J. M., Dekker, S., & Stappers, P. J. (2008). Playing twenty questions Kloos, H., & Van Orden, G. C. (2010). Voluntary behavior in cognitive and
with nature (the surprise version): Reflections on the dynamics of motor tasks. Mind and Matter, 8, 19 – 43.
experience. Theoretical Issues in Ergonomic Science, 9, 125–154. doi: Kohn, A. S. (1993). Preschoolers’ reasoning about density: Will it float?
10.1080/14639220601095353 Child Development, 64, 1637–1650. doi:10.2307/1131460
Flavell, J. H. (1999). Cognitive development: Children’s knowledge about Krist, H., Fieberg, E. L., & Wilkening, F. (1993). Intuitive physics in action
the mind. Annual Review of Psychology, 50, 21– 45. doi:10.1146/ and judgment: The development of knowledge about projectile motion.
annurev.psych.50.1.21 Journal of Experimental Psychology: Learning, Memory, and Cogni-
Gelman, S. A., & Raman, L. (2002). Folk biology as a window onto tion, 19, 952–966. doi:10.1037/0278-7393.19.4.952
636 KLOOS, FISHER, AND VAN ORDEN

Lakoff, G. (1990). Women, fire, and dangerous things: What categories the size-weight illusion. Journal of Experimental Child Psychology, 5,
reveal about the mind. Chicago, IL: University of Chicago Press. 362–371. doi:10.1016/0022-0965(67)90064-1
Levin, I., Siegler, R. S., & Druyan, S. (1990). Misconceptions about Pine, K. J., Messer, D. J., & Godfrey, K. (1999). The teachability of
motion: Development and training effects. Child Development, 61, children with naı̈ve theories: An exploration of the effects of two
1544 –1557. doi:10.2307/1130763 teaching methods. British Journal of Educational Psychology, 69, 201–
Loewenstein, J., & Gentner, D. (2001). Spatial mapping in preschoolers: 211. doi:10.1348/000709999157671
Close comparisons facilitate far mappings. Journal of Cognition and Pine, K., Messer, D., & St. John, K. (2001). Children’s misconceptions in
Development, 2, 189 –219. doi:10.1207/S15327647JCD0202_4 primary science: A survey of teachers’ views. Research in Science &
Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and Technological Education, 19, 79 –96. doi:10.1080/02635140120046240
attitude polarization: The effects of prior theories on subsequently con- Reeves, L. M., & Weisberg, R. W. (1994). The role of content and abstract
sidered evidence. Journal of Personality and Social Psychology, 37, information in analogical transfer. Psychological Bulletin, 115, 381–
2098 –2109. doi:10.1037/0022-3514.37.11.2098 400. doi:10.1037/0033-2909.115.3.381
Maclin, D., Grosslight, L., & Davis, H. (1997). Teaching for understand- Riemeier, T., & Gropengiesser, H. (2008). On the roots of difficulties in
ing: A study of students’ preinstruction theories of matter and a com- learning about cell division: Process-based analysis of students’ concep-
parison of the effectiveness of two approaches to teaching about matter tual development in teaching experiments. International Journal of
and density. Cognition and Instruction, 15, 317–393. doi:10.1207/ Science Education, 30, 923–939. doi:10.1080/09500690701294716
s1532690xci1503_2 Robinson, H. B. (1964). An experimental examination of the size-weight
Mazens, K., & Lautrey, J. (2003). Conceptual change in physics: Children’s naive illusion in young children. Child Development, 35, 91–107.
representations of sound. Cognitive Development, 18, 159–176. Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: Sustained
McCloskey, M. (1983). Naive theories of motion. In D. Gentner & A. L. inattentional blindness for dynamic events. Perception, 28, 1059 –1074.
Stevens (Eds.), Mental models (pp. 299 –324). Hillsdale, NJ: Erlbaum. doi:10.1068/p2952
McDermott, L. C., & Redish, E. F. (1999). Resource letter: PER-1: Physics Smith, C., Carey, S., & Wiser, M. (1985). On differentiation: A case study
education research. American Journal of Physics, 67, 755–767. doi: of the development of the concept of size, weight, and density. Cogni-
10.1119/1.19122 tion, 21, 177–237. doi:10.1016/0010-0277(85)90025-3
McKinnon, D., & Geissinger, H. (2002). Interactive astronomy in elementary Smith, C., Maclin, D., Houghton, C., & Hennessey, M. G. (2000). Sixth-
schools. Journal of Educational Technology and Society, 5(1), 124–128. grade students’ epistemologies of science: The impact of school science
McKinnon, D., Geissinger, H., & Danaia, L. (2002). Helping them under- experience on epistemological development. Cognition and Instruction,
stand: Astronomy for Grades 5 and 6. Information Technology in Child- 18, 349 – 422. doi:10.1207/S1532690XCI1803_3
hood Education Annual, 14, 263–275. Smith, C., & Unger, C. (1997). What’s in dots-per-box? Conceptual
McNeil, N. M., Uttal, D. H., Jarvin, L., & Sternberg, R. J. (2009). Should bootstrapping with stripped-down visual analogs. Journal of the Learn-
you show me the money? Concrete objects both hurt and help perfor- ing Sciences, 6, 143–181. doi:10.1207/s15327809jls0602_1
mance on mathematics problems. Learning and Instruction, 19, 171– Smith, L. B., Thelen, E., Titzer, R., & McLin, D. (1999). Knowing in the
184. doi:10.1016/j.learninstruc.2008.03.005 context of acting: The task dynamics of the A-not-B error. Psychological
Mikkilä-Erdmann, M. (2001). Improving conceptual change concerning Review, 106, 235–260. doi:10.1037/0033-295X.106.2.235
photosynthesis through text design. Learning and Instruction, 11, 241– Snir, J., Smith, C., & Grosslight, L. (1993). Conceptually enhanced sim-
257. doi:10.1016/S0959-4752(00)00041-4 ulations: A computer tool for science teaching. Journal of Science
Morgan, J. L., Meier, R. P., & Newport, E. L. (1987). Structural packaging Education and Technology, 2, 373–388. doi:10.1007/BF00694526
in the input to language learning: Contributions of prosodic and mor- Son, J. Y., Smith, L. B., & Goldstone, R. L. (2008). Simplicity and
phological marking phrases to the acquisition of language. Cognitive generalization: Short-cutting abstraction in children’s object categoriza-
Psychology, 19, 498 –550. doi:10.1016/0010-0285(87)90017-X tions. Cognition, 108, 626 – 638. doi:10.1016/j.cognition.2008.05.002
Nakhleh, M. B. (1992). Why some students don’t learn chemistry: Chem- Springer, K. (1995). Acquiring a naive theory of kinship through inference.
ical misconceptions. Journal of Chemical Education, 69, 191–196. doi: Child Development, 66, 547–558. doi:10.2307/1131596
10.1021/ed069p191 Stevens, J. C., & Rubin, L. L. (1970). Psychophysical scales of apparent heaviness
Nguyen, S. P., & Rosengren, K. S. (2004). Parental reports of children’s and the size-weight illusion. Perception & Psychophysics, 8, 225–230.
biological knowledge and misconceptions. International Journal of Be- SuperLab Pro (Version 2.0) [Computer software]. San Pedro, CA: Cedrus.
havioral Development, 28, 411– 420. doi:10.1080/01650250444000108 Turvey, M. T. (1990). Coordination. American Psychologist, 45, 938 –953.
Ohio Department of Education. (2009). Academic content standards. Col- doi:10.1037/0003-066X.45.8.938
umbus, OH: Author. Turvey, M. T., & Carello, C. (1981). Cognition: The view from ecological
O’Reilly, R. C., & Munakata, Y. (2000). Computational explorations in realism. Cognition, 10, 313–321. doi:10.1016/0010-0277(81)90063-9
cognitive neuroscience: Understanding the mind by simulating the Uttal, D. H., Scudder, K. V., & DeLoache, J. S. (1997). Manipulatives as
brain. Cambridge, MA: MIT Press. symbols: A new perspective on the use of concrete objects to teach
Ozay, E., & Oztas, H. (2003). Secondary students’ interpretations of mathematics. Journal of Applied Developmental Psychology, 18, 37–54.
photosynthesis and plant nutrition. Journal of Biological Education, 37, doi:10.1016/S0193-3973(97)90013-7
68 –70. Van Dooren, W., De Bock, D., Hessels, A., Janssens, D., & Verschaffel, L.
Penner, D. E., & Klahr, D. (1996). The interaction of domain-specific (2004). Remedying secondary school students’ illusion of linearity: A
knowledge and domain-general discovery strategies: A study with sink- teaching experiment aiming at conceptual change. Learning and Instruc-
ing objects. Child Development, 67, 2709 –2727. doi:10.2307/1131748 tion, 14, 485–501. doi:10.1016/j.learninstruc.2004.06.019
Pfundt, H., & Duit, R. (1991). Bibliography: Students’ alternative frame- Van Orden, G. C., Kloos, H., & Wallot, S. (2009). Living in the pink:
works and science education (3rd ed.). Kiel, Germany: Institute for Intentionality, wellbeing, and complexity. In C. A. Hooker (Ed.), Phi-
Science Education. losophy of complex systems: Handbook of the philosophy of science (pp.
Piaget, J. (1963). The psychology of intelligence. New York, NY: Rout- 639 – 683). Amsterdam, The Netherlands: Elsevier.
ledge. Walz, K. A., & Kerr, S. C. (2007). “Holes” in student understanding:
Pick, H. L., & Pick, A. D. (1967). A developmental and analytic study of Addressing prevalent misconceptions regarding atmospheric environ-
 SITUATED NAIVE PHYSICS 637

mental chemistry. Journal of Chemical Education, 84, 1693–1696. doi: Yip, D. Y. (1998). Teachers’ misconceptions of the circulatory system.
10.1021/ed084p1693 Journal of Biological Education, 32, 207–215.
Wellman, H. M., & Gelman, S. A. (1992). Cognitive development: Foun-
dational theories of core domains. Annual Review of Psychology, 43,
337–375. doi:10.1146/annurev.ps.43.020192.002005
Wilkening, F., & Martin, C. (2004). How to speed up to be in time: Received March 8, 2009
Action-judgment dissociations in children and adults. Swiss Journal of Revision received June 30, 2010
Psychology, 63, 17–29. doi:10.1024/1421-0185.63.1.17 Accepted July 1, 2010 䡲

New APA Editors Appointed, 2012–2017

The Publications and Communications Board of the American Psychological Association an-
nounces the appointment of 9 new editors for 6-year terms beginning in 2012. As of January 1,
2011, manuscripts should be directed as follows:
● Emotion (http://www.apa.org/pubs/journals/emo), David DeSteno, PhD, Department of Psy-
chology, Northeastern University, Boston, MA 02115
● Experimental and Clinical Psychopharmacology (http://www.apa.org/pubs/journals/pha), Su-
zette M. Evans, PhD, Columbia University and the New York State Psychiatric Institute, New
York, NY 10032
● Journal of Abnormal Psychology (http://www.apa.org/pubs/journals/abn), Sherryl H. Good-
man, PhD, Department of Psychology, Emory University, Atlanta, GA 30322
● Journal of Comparative Psychology (http://www.apa.org/pubs/journals/com), Josep Call,
PhD, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany
● Journal of Counseling Psychology (http://www.apa.org/pubs/journals/cou), Terence J. G.
Tracey, PhD, Counseling and Counseling Psychology Programs, Arizona State University,
Tempe, AZ 85823
● Journal of Personality and Social Psychology: Attitudes and Social Cognition (http://
www.apa.org/pubs/journals/psp), Eliot R. Smith, PhD, Department of Psychological and
Brain Sciences, Indiana University, Bloomington, IN 47405
● Journal of Experimental Psychology: General (http://www.apa.org/pubs/journals/xge), Isabel
Gauthier, PhD, Department of Psychology, Vanderbilt University, Nashville, TN 37240
● Journal of Experimental Psychology: Human Perception and Performance (http://
www.apa.org/pubs/journals/xhp), James T. Enns, PhD, Department of Psychology, Univer-
sity of British Columbia, Vancouver, BC V6T 1Z4
● Rehabilitation Psychology (http://www.apa.org/pubs/journals/rep), Stephen T. Wegener,
PhD, ABPP, School of Medicine Department of Physical Medicine and Rehabilitation, Johns
Hopkins University, Baltimore, MD 21287
Electronic manuscript submission: As of January 1, 2011, manuscripts should be submitted
electronically to the new editors via the journal’s Manuscript Submission Portal (see the website
listed above with each journal title).
Manuscript submission patterns make the precise date of completion of the 2011 volumes uncertain.
Current editors, Elizabeth A. Phelps, PhD, Nancy K. Mello, PhD, David Watson, PhD, Gordon M.
Burghardt, PhD, Brent S. Mallinckrodt, PhD, Charles M. Judd, PhD, Fernanda Ferreira, PhD, Glyn
W. Humphreys, PhD, and Timothy R. Elliott, PhD will receive and consider new manuscripts
through December 31, 2010. Should 2011 volumes be completed before that date, manuscripts will
be redirected to the new editors for consideration in 2012 volumes.