Perception & Psychophysics

2001, 63 (7), 1140-1152

The independence of size perception

and distance perception
RALPH NORMAN HABER
University of California, Santa Cruz, California
and University of California, Riverside, California
and
CHARLES A. LEVIN
Baldwin Wallace College, Berea, Ohio

Research on distance perception has focused on environmental sources of information, which have
been well documented; in contrast, size perception research has focused on familiarity or has relied on
distance information. An analysis of these two parallel bodies of work reveals their lack of equivalence.
Furthermore, definitions of familiarity need environmental grounding, specifically concerning the
amount of size variation among different tokens of an object. To demonstrate the independence of size
and distance perception, subjects in two experiments were asked to estimate the sizes of common ob-
jects from memory and then to estimate both the sizes and the distances of a subset of such objects
displayed in front of them. The experiments found that token variation was a critical variable in the ac-
curacy of size estimations, whether from memory or with vision, and that distance had no impact at all
on size perception. Furthermore, when distance information was good, size had no effect on distance
estimation; in contrast, at far distances, the distances to token variable or unknown objects were esti-
mated with less accuracy. The results suggest that size perception has been misconceptualized, so that
the relevant research to understand its properties has not been undertaken. The size–distance invari-
ance hypothesis was shown to be inadequate for both areas of research.

Theorists have assumed that similar processes are used (see Hershenson, 1999) has not received comparable at-
to perceive both the size of an object (how big a static tention in the research literature (including even size con-
object is in a metric sense in three-dimensional space) and stancy; see Epstein, 1961, 1977). As a consequence, ex-
the distance of an object (how far away a static object is planations of the sources of information and processing
from the observer in three-dimensional space). As a result, mechanisms that human observers depend on for the per-
theories of size and distance perception are so intertwined ception of size have been very unsatisfying and share
that questions of the independence of these two percep- nothing of the complexity and completeness of the com-
tual processes are typically unanswerable (see Gillam, parable descriptions for distance perception.
1995, 1998, for recent discussions). Furthermore, progress Why the disparity? We have argued (Haber & Levin,
in our understanding of these two processes is quite un- 1989) that theorists have failed to recognize that distance
balanced. perception and size perception not only serve very differ-
Nearly two centuries of research has focused on distance ent perceptual demands, but have failed to recognize that
perception, resulting in a nearly complete description of they depend on largely nonoverlapping sources of infor-
the sources of information and processing mechanisms mation. From these initial failures, theorists have been led
that human observers depend on for their perception of to relate the two processes inappropriately, to the great
distance (see Cutting & Vishton,1995; Haber, 1983; and detriment of understanding one of them.
especially, Sedgwick, 1986). In contrast, size perception With respect to perceiving the distance to an object,
we argued that its primary functions are to support visually
guided locomotion and to track moving objects in space
Part of this research was reported in a doctoral dissertation by C.A.L. (see also Cutting, 1998; Haber, 1986; Haber & Haber,
in 1990. Partial support for this research was provided by Research
1991). Moving observers need to perceive the locations
Grant NEI-07801 from the National Eye Institute to R.N.H. and by two
research contracts from the Veterans Administration’s Rehabilitation and distances to objects in the environment from an infi-
Research and Development Program (C448-R and C995-PA) to the Re- nite variety of ego-locations and distances so as to be able
habilitation Research and Development Center at the Hines VA Hospi- to move among them safely and surely and to be able to
tal, Hines, Illinois, and through contracts from the Hines VA Hospital avoid or catch those that are themselves moving. To do
to the University of Illinois at Chicago to R.N.H. We thank Lyn Haber
for her assistance in helping us match theory to data and words to
this, moving observers must depend on stimulus infor-
thoughts. Address correspondence to R. N. Haber, 730 Rimrock Drive, mation that is only available at the time of the perception,
Swall Meadows, CA 93514 (e-mail: haberhfc@[Link]). such as combinations of concurrent stimulation originat-

Copyright 2001 Psychonomic Society, Inc. 1140

 INDEPENDENCE OF SIZE AND DISTANCE 1141

ing from the observer’s physiological system, arising from racy and metrical scaling properties of distance and size
static and dynamic visual stimulation reaching the eyes responses. This is the purpose of the present experiments.
and from the consequences of observer-originated move- However, to better control the separation of these the-
ments made while looking at the scene. Furthermore, since oretical processes, the familiarity variable itself, as it ap-
human beings must be able to move about in this vast va- plies to size perception, must be described more com-
riety of different scenes, this combination of stimulation pletely than has been done in the past.
should make relatively little demand on prior experience Consider the following logical observations. For object
with the particular objects, paths, and obstacles in the familiarity to be useful in a size perception context, four
scene. In fact, skill at distance perception and the use of criteria must be met. (1) The current observation condi-
distance perception to control visually guided locomotion, tions must be sufficient to allow the observer to correctly
should develop and be available early in life and in tan- recognize (or identify) the token of the object presently
dem with maturation of locomotor abilities themselves: on display. (2) The observer must have retained in mem-
Learning, practice, experience, and memory should not be ory prior experience about the properties of the category
important components of explanations of distance percep- or class of object. (3) This prior experience must include
tion. Specifically, familiarity effects based on prior contact knowledge of the prototypic (average or typical) size of
or knowledge about the objects whose distances are the objects in this category. This means that upon dis-
being perceived should be irrelevant to determining the covering that the object being observed is an “X”, the ob-
accuracy of that perception. We must be (and are) able to server knows that the typical X is a given size. (4) This
move easily and accurately in unfamiliar as well as famil- prior experience must include knowledge of the amount
iar environments and among unfamiliar as well as famil- of size variation to be expected among possible tokens of
iar objects. There can be little substitute for concurrent the object. This means that the observer has to know
perceptual processing occurring in real time. whether all instances of X are the same size, and if not,
In contrast, we have argued that size perception should by how much they differ from each other.
function the other way around. Concurrent perceptual Most research on size perception that includes a famil-
processes should be irrelevant (other than to allow recog- iarity variable assumes (or assesses by questionnaire) that
nition or identification of the object and perhaps to per- the “familiar” objects chosen for judgment are known to
ceive its rigidity; see Hershenson, 1992). Once the object the observer. Thus, familiarity has been operationalized
is identified, the perceiver has access to most of its prop- almost exclusively in terms of prior contact (Criterion 2),
erties (its character, its trustworthiness, the consequences either throughout the lifetime of the observer, or through
of dropping or kicking it, its size, etc.). In this sense, training manipulated in the experiment. To our knowl-
“perceiving” the size of an object may be a misnomer: edge, it has never been studied in terms of knowledge of
Rather, the observer identifies the object and then “re- the prototypic size of the object (Criterion 3) or in terms
members” its size (along with its other relatively constant of knowledge of the amount of token variation (Crite-
characteristics) on the basis of prior encounters. Only for rion 4). The ability to recognize an object as being famil-
perception of completely novel objects that have never iar (Criterion 1) is also usually assumed in most research,
been encountered before would a concurrent perceptual though it can pose a problem for small objects viewed at
process be required. Since entirely novel objects become long distances.
increasingly rare as one acquires experience with the Although knowledge of prototypic size (Criterion 3)
world, we have argued that size “perception” is a cognitive might be part of the general knowledge picked up through
or memorial process, one that draws on quite different prior experience (Criterion 2), knowledge of token vari-
stimulus information and processing mechanisms for its ation among instances (Criterion 4) has never been de-
accuracy than those underlying distance perception. Fa- fined theoretically or addressed experimentally, though
miliarity with the object (acquired from the past) should its implications should be obvious from the work of
be the most important variable determining the accuracy Brunswik (1956) nearly a half century ago. Specifically,
of “perceiving” how big the object appears to be. an object whose tokens are known to occur in a wide
Familiarity as a variable in size perception has had a range of different sizes logically would not be very useful
long history (see Hershenson, 1999, for a recent discus- as a predictor of the size of the particular token on view at
sion; Hochberg, 1971, for an older one), and is typically the moment, regardless of how familiar observers are
accorded the most prominent role in how we “perceive” with the class of objects.
the size of an object. In contrast, familiarity as a variable To illustrate the potential interplay of Criteria 2, 3, and
is rarely mentioned with respect to distance perception 4 in size perception research, suppose that natural objects
(see Ittelson, 1960, for a few counter examples), except in the real world are classified into one of two broad cat-
in unusual and ecological atypical experimental settings egories, defined by characteristics of the tokens of the ob-
in which the normally predictive sources of distance in- jects. The first category contains token invariant objects,
formation are impoverished or absent. Given the sharp each of which has a highly constrained range of sizes
difference in the importance of the role of familiarity in among their exemplars. Regulation basketballs (except
these two kinds of perceptions, its manipulation should for women’s professional games) fulfill this definition:
provide a way to disentangle the determinants of the accu- The mean prototypic size (diameter) of regulation basket-
1142 HABER AND LEVIN

Table 1 ory of the tightly constrained prototypic size. In contrast,

Measured Heights and Cognitive Estimates if the observers determine that an object is of the token
of Heights of 30 Common Objects (all Units in Feet) variable class, recall of the prototypic size provides much
Measured Cognitive Estimates less size information of the exemplar by virtue of the ob-
Object Height SD Height SD servers’ past experience, even though the amount of fa-
Token Invariant Objects miliarity with specific objects in the second class may be
CARD TABLE 2.33 0.00 3.13 0.62 equivalent to that of the first (by Criteria 2 and 3). If “size
BASEBALL BAT 2.79 0.04 2.86 0.58 perception” depends on a memorial process, estimates
BASKETBALL 0.75 0.00 0.95 0.19 of object size should be less accurate for token variable
MILK BOTTLE 0.83 0.00 0.98 0.23
DOOR 6.51 0.05 6.55 0.87
objects.
BEER BOTTLE 0.63 0.04 0.63 0.16 The discussion above expands the definition of famil-
BIKE 3.04 0.06 3.18 0.78 iarity in two directions: It distinguishes among objects
POP BOTTLE 1.00 0.00 1.11 0.24 themselves in terms of their token size variation as a char-
SKI POLE 4.00 0.08 3.78 0.68 acteristic of the natural world, and it includes as a char-
FILE CABINET 2.26 0.04 2.94 0.73
BOWLING BALL 0.50 0.00 0.80 0.22 acteristic of observers’ knowledge about objects both
AXE 2.93 0.07 2.59 0.78 their knowledge of prototypic size and their knowledge
TENNIS 2.26 0.02 2.12 0.51 of token size variation.
GUITAR 3.26 0.04 3.13 0.65 In this paper, we report the results of two experiments
JEANS 2.74 0.07 3.05 0.46
that examined implications of this definition of familiar-
Token Variable Objects ity. In Experiment 1, we asked subjects to make size judg-
CHRISTMAS TREE 5.85 1.18 6.82 1.25 ments of familiar objects that differed in their token vari-
KIT. TRASH 2.37 0.57 2.48 0.70
SPEAKER 1.96 0.77 2.54 0.84
ation. The objects were not on view. The results validated
KIT. CHAIR 3.05 0.53 3.30 0.75 the distinction among familiar objects on the basis of
GLOBE 1.31 0.82 1.56 0.70 their token variation. In Experiment 2, we asked subjects
ROAD CONE 1.80 0.76 2.01 0.68 to make both size and distance estimations of both unfa-
TABLE LAMP 2.50 0.46 1.97 0.78 miliar and familiar objects that differed in their token
TEDDY BEAR 1.44 0.36 1.24 0.67
ROCK. CHAIR 3.81 1.23 4.01 1.24 variation, all of which were on view. In Experiment 2,
TRICYCLE 2.53 0.66 2.23 0.70 we also varied viewing conditions to determine whether
TV 2.02 0.69 2.18 0.73 those conditions impacted accuracy of distance and size
SAW HORSE 3.06 0.64 3.41 1.71 estimates in similar ways. The results showed that the
RED WAGON 1.48 0.49 1.72 1.21
HOUSE PLANT 1.78 0.91 2.06 1.48
underlying processing mechanisms of size and distance
SCREEN 4.60 1.08 5.17 2.20 perception are quite different.

EXPERIMENT 1
balls is 0.75 ft, and the standard deviation among exem-
plars is virtually zero (see Table 1 for a number of other Can object classes be subdivided into those that are
examples). The very small size variance among tokens token invariant and token variable? Do observers have
means that the prototypic size is likely to be representative and use this knowledge in making objective size estima-
of any exemplar encountered. The second category con- tions about such objects? To show the former, physical
tains token variable objects, each of which can also be size measurements were made of at least 10 instances of
described by a prototypic (average) size, but each object each of a large number of classes of familiar objects. To
has a much less constrained size range among its tokens show the latter, subjects were asked to make objective
(e.g., stereo speakers, Christmas trees, teddy bears—see size judgments of these kinds of objects from memory in
Table 1 for other examples). Thus, knowing the prototypic the absence of any concurrent perceptual information.
size of token-variable objects does not predict the size of (We followed a procedure first used by Bolles & Bailey,
any particular exemplar. 1956, though they did not consider or manipulate token
The above classification of token variability is depen- variation).
dent on only the distribution of tokens in the environ-
Method
ment—an ecological variable determined by natural Stimuli. The subjects were presented with a list of names of 50
processes, manufacturers’ choices or regulations and stan- common portable objects. A much longer list of potential stimuli
dards. When observers become familiar with an object, was drawn up that the authors believed would be familiar to college
their experience (Criterion 2) should allow them to learn students and would contain some objects that had very small token
or determine the prototypic size of the object (Criterion 3) variation and others that would have large token variation. The list
and should allow them to learn or determine the amount was pared to 50 by asking a group of experimental psychologist
judges to rate the objects on familiarity (without reference to token
of the object’s token size variation in the environment variability). To include an object on the final list, these judges had
(Criterion 4). If the observers determine that an object is to agree that an object would be highly familiar to undergraduate
of the token invariant class, size “perception” may sim- students. The resulting 50-item list contained all of those objects
ply involve object recognition and then recall from mem- that produced high agreement among the judges.
 INDEPENDENCE OF SIZE AND DISTANCE 1143

Size measurement procedure. Next, we obtained physical size to college students, some truly had no variance among a
(height) measurements of as many of the 50 objects as we could large sample of their tokens, whereas others had sub-
reasonably sample. The heights of actual instances of these objects stantial variation from instance to instance.
(to the nearest quarter inch) in their normal orientation were mea-
How accurate are size estimations from memory?
sured. We found the objects in a minimum of three different stores
and measured at least 10 different instances of each object and The 30 size estimates made by each subject were com-
(wherever possible) found at least 10 different brands, all in local pared with the respective 30 means of the physical size
retail stores in the greater Chicago area. The 30 objects are listed in measurements of the actual objects using multiple regres-
Table 1, along with the mean and SD of each of their measured sizes. sion analyses, one analysis for each of the 109 subjects.1
The 20 objects not measured were excluded from subsequent The proportion of variance accounted for in cognitive es-
study because we could not find a suff icient number of representa- timated size by true size is
tive samples of them to measure. A few were also excluded when
we realized (during the subsequent design phase of Experiment 2) cognitive estimated size = B (true size),
that we would not be able to transport instances of them (e.g., a re-
frigerator) for the second experiment. a measure of the accuracy of estimating the sizes of these
Subjects and size estimation procedures. The subjects were 30 objects.2 This proportion is given by the multiple cor-
109 male undergraduate students enrolled in an introductory psy- relation R2. The mean R2 averaged over 109 subjects was
chology course at the University of Illinois at Chicago. The sub- .94 (SE = 0.02). The very high value indicates that the
jects participated as partial fulfillment of a course requirement. The
stimulus questionnaire containing the names of the items to be es-
subjects’ size estimates for each object were very similar
timated was included in a mass testing packet along with question- to the prototypic (average) measured size of each object.
naires constructed by five other experimenters that pertained to dif- The B value represents the slope of the regression equa-
ferent experiments. Questionnaire packets were distributed at the tion. Mean B = 1.04 (N = 109, SE = 0.03). We tested the
end of a class session and were returned within 48 h. slope against a null estimate of 1.00, and found we could
On the questionnaire, the 50 objects were listed in random order, not reject the null hypothesis ( p . .10).
each followed by a line on which the subjects were to write the es-
Therefore, the answer to the second question is also
timated height of that object. The instructions for answering the
questionnaire were as follows: “Listed below are a number of com- yes: College students know the prototypic size of com-
mon objects. Please estimate the height of each of these objects to mon objects with great accuracy. This great accuracy is
the nearest inch. Please make your estimates in feet and inches. For found even though the objects were not observed at the
objects such as a baseball bat or a broom, please estimate the height time of size estimation.
of them standing on end.” Are subjects sensitive to an object’s token varia-
tion? The between-subjects SDs of the size estimates of
Results each object (each based on N = 109) are found in column
Do tokens vary among common objects? Token 4 of Table 1. The range of these SDs among the 15 token
variability was estimated from the SDs among the 10 to invariant objects was 0.16 to 0.87 ft, with a mean = 0.51 ft,
25 physical measurements of each of 30 of the objects on which was not significantly different from zero (p .
the questionnaire that were measured. The SDs ranged .10). In contrast, the 15 token variable objects ranged in
from a minimum of 0.00 ft (for five objects) to a maxi- SDs from 0.67 to 2.20 ft, with a mean = 1.04 ft, which was
mum of 1.23 ft. (Because subjects were asked to make significantly greater than zero (p , .05). The means of the
their estimates in feet and inches, all measurement data two sets of SDs were significantly different from each
are also reported in those units.) Post hoc examination of other [t (28) = 3.904, p , .001].
the SDs of these multiple height measurements showed Therefore, the answer to the third question is also yes:
that 15 of the 30 objects had SDs among the measured in- The subjects’ responses differed depending on the token
stances of between 0.00 and 0.08 ft. The standard devia- variability of the objects whose size they were estimating.
tions are listed in the second column of Table 1 in the first They vary less among themselves in estimating the sizes
15 rows (reordered from the actual questionnaire to re- of objects that have little token variation.
flect this post hoc distinction). The remaining 15 objects Do subjects estimate the size of token invariant ob-
had SDs among measured instances of between 0.36 and jects more accurately? The 109 regression equations
1.26 ft. These are listed in the remaining 15 rows. (based on all 30 objects) were recomputed separately first
Using this post hoc determination, we labeled the first for the 15 objects classified as token invariant and then
15 objects, those objects with the near-zero variances, for the 15 objects classified as token variable. These 218
token invariant objects and the remaining 15, those with equations were first modeled nonlinearly and then redone
larger variances, token variable objects. following a linear model when none of the 218 intercep-
As it turned out, there was no difference between the tions or exponents were found to differ from zero or unity,
mean prototypic sizes of the two classes of objects. The respectively. Repeated measures analysis of variance
mean measured heights were 2.38 ft versus 2.62 ft (as is (ANOVA) revealed that the accuracy (R2) of the cognitive
shown in the second column of Table 1), which was not size estimates for the token invariant objects (mean R2 =
significant [t (28) = 0.474, p . .10]. .97) was significantly greater than the accuracy of the cog-
Therefore, the answer to the first question is yes: We nitive size estimates for the token variable objects (mean
found that, among common objects likely to be familiar R2 = .92) [F(1,108) = 38.79, p , .001]. The two mean B
1144 HABER AND LEVIN

(slope) coefficients (1.02 and 1.09, respectively) were that the variation among the size estimates made by the
not significantly different from each other ( p . .10), nor subjects was significantly lower for objects with low token
were they different from unity ( p . .10). variance.
Therefore, the answer to the fourth question is also yes: The latter finding suggests that when a group of sub-
When token variability is low, the subjects were signifi- jects are asked to estimate the size of a familiar object
cantly more accurate (higher R2) in estimating the proto- (elicited by its name) whose tokens have been demon-
typic size of such objects from memory. strated to be invariant in size, all of the subjects seem to
Does the scaling of remembered objective size esti- have the same sized objects in mind when they make their
mations follow a ratio scale? The relationship between estimates. This produced the lower between-subjects SD
measured size and cognitively estimated size is linear (ex- in the size estimations. In contrast, when estimating the
ponent of unity) with an intercept of zero and a regression size of an object whose tokens have been demonstrated to
coefficient of unity. Furthermore, there were no differ- be more variable in size, one of two different processes
ences between the means of the measured physical sizes might be occurring. One is that each subject has a some-
and the corresponding cognitive sizes of the objects. Be- what different sized instance of the prototype in mind
cause it can be assumed that measured (true) size follows and produces that value when asked to make an estimate.
a ratio scale, the findings conjointly indicate that the sub- The other is that all subjects have a less precise sense of
jects were using a ratio scale of cognitive size that is equal the prototypic (average) size of token variable objects, so
to true feet and inches in making their cognitive size esti- that there is more variation between subjects. The latter
mations. This is found even though the cognitive size es- alternative would suggest also more within-subjects vari-
timations were made from memory, without the objects’ ation for the token variable objects if they were asked to
being visually accessible. repeat this task several times.
Whatever the explanation, the finding is robust: Size
Discussion estimations are more accurate for token invariant objects
We briefly consider five topics. than for token variable objects.
Token variation is logically important. The common Do subjects know that objects differ in token vari-
objects selected for the present study could be distin- ation? We did not ask the subjects directly to classify
guished by the magnitude of their token variation among these objects into categories on the basis of their estimated
the large number of exemplars that were measured. We token variances. Even so, the data from the regression
easily found objects that could be classified into two analyses suggest that the subjects could have performed
groups: a group containing objects whose tokens were all this task with great accuracy. However, it is possible that
about the same size, and a group containing objects whose the subjects had only a single token of an object in mind
tokens varied substantially among themselves. This when they made their size estimations and that they were
finding concerns a property of the distribution of tokens not aware that token-to-token variation differed among
of object types—an ecological variable. It shows that the objects. If that was the case, since the token variable
token variability (Criterion 4) is logically relevant, since objects exist in a range of sizes, the single tokens in mind
objects can be distinguished by the amount of their token among the subjects would differ among themselves,
variation. thereby producing a greater between-subjects variation in
We make no claim as to the proportion of common ob- estimations. We failed to eliminate this possibility.
jects that are in the two classes or to their discontinuity, Size “perception” or size cognition? Because these
only that it was easy to find a number of exemplars of results were all obtained in the absence of any concurrent
both classes. visual experience with the objects, they support the analy-
College students know the prototypic sizes of fa- sis of size “perception” as a cognitive or memorial pro-
miliar objects. This is shown by the overall very high R2 cess. When subjects are familiar with an object, they use
values found from the regression analyses and from the that knowledge to estimate its size and can do so with very
evidence of a ratio scaling of the estimated size judg- high accuracy even without seeing any exemplars of the
ments. Even without having had an opportunity to exam- objects. It is difficult to think of Experiment 1 as a percep-
ine the objects visually, the subjects made size estimates tual task.
whose variance was almost entirely accounted for by the In Experiment 2, we considered how the sizes of un-
true sizes of the objects, producing correlations between familiar objects are estimated.
estimated and true size approaching unity. In summary, the results support extending the concept
This result also provides evidence that when Criterion 2 of familiar size to knowledge of prototypic size and knowl-
(prior exposure) underlying the concept of familiarity is edge of token variance. Not only are subjects knowledge-
satisfied, Criterion 3 (knowledge of prototypic size) is also able about both aspects of an object, but furthermore, token
satisfied. variance is shown to affect the accuracy with which pro-
Token variation affects the accuracy of size estima- totypic size is known. This implies that even when ob-
tions. This conclusion is supported by two results. We jects are in view, cognitive size information for token vari-
found that the accuracy of estimating the prototypic size able objects would be somewhat less useful in estimating
of an object (as assessed by R2) is significantly greater object sizes than for token invariant objects (this impli-
for objects with very low token variance. We also found cation is explicitly tested in Experiment 2 below). Finally,
 INDEPENDENCE OF SIZE AND DISTANCE 1145

this suggests that token variance should be considered as ferent quadrant of the field. Each subject was randomly assigned to
an integral part of any theory that uses familiar size as a begin at one of the three quadrants.
variable. The subjects were given a pen and a clip board, on which was at-
tached a scoring sheet with the names of 15 objects listed in a random
order. The names given for the token invariant and token variable ob-
EXPERIMENT 2 jects were chosen so as to be descriptive without conveying addi-
tional size information (e.g., bicycle not ten speed bicycle). The un-
Subjects were asked to view actual instances of the known objects were named by color and shape (e.g., blue triangle).
same objects displayed at various distances in a visually Two blank lines followed each of the 15 object names. One col-
rich natural scene. The two independent variables in- umn of blank lines was labeled “distance” and the other was labeled
cluded type of object (15 objects that were familiar token “height”. The subjects were instructed to estimate the objective dis-
tance from themselves to each object and to enter that estimate on
invariant, 15 that were familiar token variable, and 15 the first blank line after the object’s name. The estimate of the ob-
that were unfamiliar) and viewing distance (half the ob- jective height above the ground of each object at the tallest part of
jects were placed in a near viewing range from 10 to the object was to be entered on the second blank line after the ob-
150 ft and half in a more distant range from 150 to over ject’s name. The subjects were told that they could make the two es-
300 ft). The two dependent measures were the subjects’ timates in either order.
estimated objective sizes and the estimated objective dis- When all 9 subjects had made all 30 estimates (15 heights and 15
distances) of the objects in one quadrant (object type), the experi-
tances of all of the objects. menter told them which quadrant they were to observe next, and
they turned at the same time, repeating the procedure with a new
Method scoring sheet.
Subjects. Nine male college students, all volunteers, screened After all subjects had estimated all objective distances and ob-
for normal or corrected-to-normal acuity, binocular, and color vi- jective sizes, they returned to the van area, and, with their backs to
sion, served as subjects in Experiment 2. None had participated in the field, they filled out two questionnaires. The first asked the sub-
Experiment 1. jects to rate the familiarity of each of the token invariant and token
Stimuli. The 45 stimuli to be judged included the 15 token in- variable objects that they had just viewed on a scale of 1 (not at all
variant objects and the 15 token variable objects tested in Experi- familiar) to 10 (highly familiar). The second questionnaire asked
ment 1. Elongated objects (e.g., a bike) were displayed with the for cognitive size estimates of the 30 objects, using the same format
long axis perpendicular to the subject’s line of regard. In addition, as that employed in Experiment 1.
15 objects of unknown size were used. These consisted of flat card-
board cutouts of three geometric shapes (ovals, rectangles, and tri- Size Perception Results
angles), each painted in one of seven different colors (orange, blue, Are the “familiar” objects familiar? First, we veri-
black, yellow, green, white, and red), which were matched in size to
the heights of the two other object types. Each of the 45 objects was fied that the objects on the questionnaire in Experiment 1
supported vertically by a hidden stand when necessary. were also familiar to the new set of subjects. In response
Placement of stimuli. The experiment was conducted on a sin- to the familiarity questionnaire, the lowest rating given
gle large grassy field, approximately 1,000 3 650 ft, which was by any subject for any object was 7 (on a 10-point scale).
bounded by trees on three sides. Each set of 15 objects of a similar Averaging across subjects, no object had a mean famil-
type was arrayed in a separate quadrant of the field, radiating out iarity rating significantly different from 10 (range, 8.1–10,
from the center of the field, in such a way that the subjects (who
stood in the center) could view only the 15 objects of a single type
M = 8.86, p . .10). Furthermore, the mean familiar rat-
at a time. The fourth quadrant of the field was empty, so that sub- ing for the two groups of objects did not differ from each
jects could walk to the center without seeing any of the objects. The other (8.92 and 8.80, respectively, p . .10). Therefore,
scene beyond the perimeter of each quadrant of the field that con- these 30 objects were highly familiar to the new subjects.
tained stimuli was similar: irregular grass for about 300 ft beyond How accurately do subjects estimate object sizes
the farthest object in the quadrant, terminating in a visual horizon while looking at the objects? Fifty-four regression
of dense forest.
equations of the form
The objects were randomly assigned to the true viewing distance
from the center of the field with the following constraints: About estimated size = (B) true size
half of them were located between 10 and 150 ft and the remaining
half between 150 and 300 ft from the subject; there was no occlu- were calculated to determine the source and amount of
sion among objects from the viewing position of the subject, and variance accounted for in size estimations, one equation
the visual angle among the set of 15 objects in any one quadrant for each of the 9 subjects for each of the three object types
was less than 30º from the subject’s viewing position. The correla- for each of the two distance ranges. The grand mean (over
tion between object height and viewing distance was trivial.
Procedure. The 45 objects were placed on the field prior to the the 9 subjects, three object types, and two viewing dis-
arrival of the subjects. The subjects arrived by van, stood with their tances) is R2 = .95. Hence, the true size of an object ac-
backs to the field, and were given complete instructions before they counted for virtually all of the variance in estimated ob-
walked onto the field. Then they were told to walk single file with jective size by the subjects. The results, broken down by
their heads and eyes down. They followed the experimenter, who object type and viewing distance (discussed below) are
led them to the center of the field through the fourth quadrant and displayed in Figure 1.
turned them so that they faced the direction that they had just come
from. In this way, the subjects could not see the objects arrayed
Does the object’s token variability affect size esti-
around them until the testing began. mation accuracy? The mean R 2 for the three object
All 9 subjects were tested at the same time. During the testing, classes (.97, .94, and .94, respectively, as is shown in
the subjects stood in groups of 3, each group facing toward a dif- Figure 1) were significantly different from one another
1146 HABER AND LEVIN

found ( p . .10): Both viewed size and remembered size

predictors accounted for the same (high) amount of vari-
ance when compared directly or when compared in step-
wise additions.
Taken together, these results show that the subjects
were equally accurate in making size estimations from
memory and from concurrent perception.
Does distance information, true distance, or esti-
mated distance affect size estimations? We used the
variable of distance range to manipulate the quality of
distance information, since there is overwhelming evi-
dence that objective distances are estimated more accu-
rately for objects in near distance when compared with
far distance (see Gillam, 1995; Sedgwick, 1986, for re-
views). As can be seen in Figure 1, there was no main ef-
fect of range distance on size estimation accuracy (p .
.10) and no interaction of distance range and object class
( p . .10). The estimations of object sizes were just as ac-
curate when the objects were far away as when they were
near. This result suggests that distance information is not
used to improve size estimations.
Figure 1. Mean R2 from regression analyses using true size to To determine whether either true distance or estimated
predict estimated size as a function of object type and distance distance might predict estimated size, the true distance to
range. each object and the estimated distance to each object were
added as additional terms to the regression equations just
examined. However, for none of the 54 equations did
[F(2,16) = 6.59, p , .005]. Scheffé comparisons indicated adding either predictor significantly improve the predic-
that this effect was due to the higher R2 for the token in- tion of estimated size ( p . .10), nor did they improve pre-
variant objects compared with either the token variable dictions for any of the three object classes considered
objects ( p , .05) or with unknown objects ( p , .05), separately.
with the latter two not differing between themselves. Thus, considering all 54 size estimation equations or
This result suggests that if the prototypic size is known subgroups of them by object class or distance, adding pre-
and token variance is minimal, size estimations are more dictors based on adequacy of distance information, true
accurate, as compared with when token variance is larger distance, or estimated distance (or signed error in size
or prototype size and token variance are unknown. estimation, or absolute error in size estimation, which we
Does viewing the objects improve the accuracy of also verified) had no effect on predicting size estimation.
estimating their sizes? We compared the 9 subjects in In this experiment, no variable involving distance impacted
Experiment 2 who looked at the objects with the 109 sub- size estimations. This was also the case for the unfamil-
jects in Experiment 1 who did the size estimation task en- iar objects.
tirely from memory. We also compared visually based es- How accurately are the sizes of unfamiliar objects
timates with the memory-based estimates of the same 9 estimated? Even though there is no remembered size in-
subjects in Experiment 2, since they made their esti- formation available for the 15 unfamiliar objects, the
mates both ways. subjects were also quite accurate in estimating those
The R2 values based on the 9 subjects in Experiment 2 sizes, though significantly less so than for token invariant
under visual conditions (R2 = .97 and .94, respectively) familiar objects. The high mean R2 for the unfamiliar ob-
were not significantly different ( p . .10) from the com- jects (mean R2 = .94) surprised us; we had expected it to
parable ones based on the 109 subjects in Experiment 1 be much lower than the mean R2 = .97 for familiar token
(R2 = .97 and .92, respectively), in which there were no invariant objects, following the assumption that the lack
visual access to the objects. of remembered size information would impair size esti-
To determine whether viewed size (estimates made mation accuracy. The significant difference suggests that
while standing in the field) or remembered size (estimates this lack of information does impair performance, but
taken from the questionnaire responses) of the 9 subjects not by much. This result in isolation shows that subjects
was a better predictor of estimated size, 36 of the above can estimate the sizes of even totally unfamiliar objects
regression equations were recomputed using stepwise quite accurately. It also suggests that token variable ob-
combinations of the two predictor terms (the cognitive jects are estimated no better than the totally unfamiliar ob-
size term could not be included in the model statement jects (mean R2 = .94 in each case). We will return to the
for the unfamiliar objects because those objects have no implications of these results after considering those for
known remembered sizes). No difference in mean R2 was distance estimations.
 INDEPENDENCE OF SIZE AND DISTANCE 1147

Is accuracy of distance estimation affected by dis-

tance information and by size information? These two
questions are considered together because there is an
interaction between the independent variables of object
distance and object type. Figure 3 shows the results of the
54 equations above broken down by distance range (the
manipulation of the quality of distance information) and
object class (the manipulation of familiar size informa-
tion). The apparent interaction is significant [F(2,48) =
4.44, p , .05]. Although the two main effects are also sig-
nificant ( p , .05), those effects are obviously due to the
strength of the interaction.
Planned comparisons revealed that for the near distance
range, in which distance information is assumed to be
quite good (white bars), change in object type had no ef-
fect on the accuracy of the distance estimates ( p . .05).
For the far distance range, in which distance information
is assumed to be poorer (dark bars), all three object types
differed among themselves: The accuracy of estimating
the distances to token invariant objects was significantly
higher than to token variable objects ( p , .05), and those
Figure 2. Mean regression coeff icients (B) from regression were significantly higher than for unknown objects ( p ,
analyses using true size to predict estimated size as a function of
object type and distance range.
.05). Further, for token invariant objects, distance range
had no effect on the accuracy of the distance estimates (p .
.05). For both token variable objects and unknown objects,
the far distance range did show a significant reduction in
Scaling properties of visually estimating size. All re- the accuracy of the distance estimates ( p . .01).
gression equations were linear (exponents were not dif- Summarizing these results, true distance accounts for
ferent from unity), and all had a true zero point (intercepts virtually all of the variance in estimated distance; the
were not different from zero). The remaining variable is amount of variance accounted for was unaffected by the
the regression coefficient or scaling factor (B) from each object type when distance information was good; and the
equation, which, given the zero intercept and unity ex- amount of variance accounted for by true distance was re-
ponent, provides an estimate of the slope of the linear re-
gression function, or the amount of over- or underesti-
mation of size. The mean coefficient on true size did not
differ from unity for any of the three object types (mean
B = 1.03, 0.98, and 1.14, respectively, p . .10), a finding
comparable to that for sizes estimated from memory in
Experiment 1. Nor did they interact with distance range.
However, the three slopes do differ among themselves
[F(2,16) = 5.77, p , .05], with Scheffé tests showing that
the unknown objects (which were not measured in Ex-
periment 1) were significantly more overestimated in size
than were the token variable objects ( p , .05). These re-
sults are displayed in Figure 2.

Distance Perception Results

How accurately can subjects estimate distance
from themselves to objects? Fifty-four linear regression
equations of the form
estimated distance = (B) true distance
were computed, comparable with those already reported
for size perception (9 subjects 3 3 object types 3 2 dis-
tance ranges). The mean R2 averaged over all 54 equa-
tions is .97. This shows that true distance accounted for Figure 3. Mean R2 from regression analyses using true distance
nearly all of the variance in the subjects’ estimates of ob- to predict estimated distance as a function of object type and dis-
jective distance. The subjects were therefore very accurate. tance range.
1148 HABER AND LEVIN

types differed in slope (mean B = 0.734, 0.676, and 0.456,

respectively) [F(2,51) = 9.80, p , .001], with each of
these significantly less than unity ( p , .05). Scheffé tests
also showed that the slope for token invariant objects was
significantly higher (less underestimation of distance)
than was the slope for token variable objects ( p , .05),
with the latter slope significantly higher than that for the
unknown objects ( p , .01).
Therefore, size information effects distance underesti-
mation, with more underestimation occurring as the qual-
ity of information specifying object size is reduced.

GENERAL DISCUSSIO N

Empirical Demonstrations
of an Ecological Definition of Familiarity
Several criteria to meet a definition of familiarity were
proposed: (1) recognizability, (2) prior experience,
(3) knowledge of prototypic size, and (4) knowledge of
token variation. The present results provide a start in op-
erationalization of these definitions.
Figure 4. M ean regression coefficient (B) from regression Criterion 1 was not manipulated in the present re-
analyses using true distance to predict estimated distance as a search, though we were convinced in pretesting that all of
function of object type and distance range. these objects could be seen and identified by name or by
color and shape even at 300 ft. However, it is still impor-
tant to verify that observers have adequate perceptual op-
duced when distance information was reduced. However, portunity to identify an object before any effect of its fa-
when distance information was poorer, estimations of miliarity can become apparent.
distances improved when the objects provided better fa- To have had prior experience with an object (Crite-
miliar size information. Thus, size information was used rion 2) so as to be able to recognize or identify it, has
to improve distance estimations when distance informa- usually been the only criterion for familiarity considered
tion itself was inadequate. in past research. The present results strongly suggest that
It should be noted that while the R2 values are very when this criterion is satisfied, the much more important
high, they are not at ceiling: As the results indicate, it is Criterion 3 is also satisfied: Prior experience imparts
still possible to demonstrate significant variation in accu- knowledge of the prototypic size of the object. From mem-
racy as either distance information or size information is ory alone, without viewing the object, subjects can pro-
degraded. vide size estimates that are highly accurate using a ratio
Scaling properties of visually estimating distance. scale of size.
As with size estimation, the mean intercept (A) is not sig- Criterion 4, knowledge of token variation, introduces
nificantly different from zero, and the mean exponent (C) a new dimension, which we have manipulated for the first
is not significantly different from unity, so that the func- time. The present results strongly indicate the importance
tion relating true distance to estimated distance is linear of this criterion. Some familiar objects have little token
with a true zero point. Figure 4 shows the means of the variation and others have a great deal, and these differ-
slopes, given by the coefficient (B) of true distance, av- ences in token variation affect the estimates of size and
eraged over all 54 equations. The overall mean was .62 distance made by subjects. The present results indicate
(which was significantly less than unity—p , .001), in- that object token size variation and subjects’ knowledge
dicating that even though linear, there was a substantial of that variation should be measured and manipulated in
underestimation of distance. experiments in which familiarity with the stimuli is im-
Underestimation of distance was not affected by the portant.
quality of distance information. Although previous results
(e.g., Sedgwick, 1986) suggest that more underestimation Overall Levels of Accuracy
of distance should be found at far than at near distances, of Size and Distance Perception
an ANOVA failed to replicate this finding: There was no Because the subjects were always asked to provide ob-
significant difference in the amount of underestimation jective physical size and distance estimations in the pres-
between the near and the far distance (mean B = 0.61 and ent experiments, their match to true sizes and distances
0.63, respectively, p . .10). can be assessed. We assumed that physical measurements
However, underestimation of distance was affected by of true size and true distance follow ratio scales. The sub-
object type. An ANOVA revealed that the three object jects in both experiments, both for sizes and distances,
 INDEPENDENCE OF SIZE AND DISTANCE 1149

produced estimations that were very close to these ob- Even so, for the perception of both size and distance,
jective physical dimensions. For size, this was especially the amount of token variation of the familiar objects af-
true: The scaling of estimated size to true size follows a fected accuracy.
ratio scale (zero intercept, unity slope and unity power).
Physical size ranged from less than a half foot to nearly How do Subjects Estimate the Sizes
6 feet; yet the average absolute error in the subjects’ es- of Token Variable Familiar Objects So Well?
timates was less than 2 in. Distance estimations were less Again, although token variable objects were estimated
accurate only in the sense that the subjects underestimated with less accuracy than were token invariant objects,
distances systematically; but even so, their estimates were their accuracy levels were still impressively high. The
almost perfectly correlated with true distance. Hence, logic underlying the ecological-based importance of
across all conditions, the subjects were able to give highly token variation would suggest that size memory for ob-
accurate size and distance estimates. jects with substantial token variation should provide
Our asking the subjects to provide estimates using nat- much less help in estimating their sizes accurately. So
ural units (e.g., feet and inches) is not a common exper- why did the subjects do so well?
imental procedure. The majority of experiments require One possible explanation is that some familiar objects
unit-free estimates (e.g., magnitude estimation), compar- with high token variability might contain some familiar
ison judgments (e.g., which is larger), or nonverbal esti- subcomponents that are token invariant. For example,
mates (e.g., adjustment of a standard), all of which require Christmas trees have high token variation, even when
more complex assumptions in order to allow conclusions considering just those for home consumption. However,
about accuracy. We expected that adult subjects could their needle length and needle diameter is close to token
make direct estimations in natural units (the present re- invariant for any given type of pine tree. Hence, by ob-
sults strongly support us), and, by requesting responses serving the needles, observers have access to some size
in natural units, we learned much more about what the invariant information, even though the overall height of
subjects knew of the stimuli and the task (for further dis- the tree varies unpredictably. With this token invariant in-
cussion, see Levin & Haber, 1993; Toye, 1986). formation imbedded in a token variable object, observers
may be able to determine the overall size more accurately
The Role of Prototypic Size Knowledge than they could do otherwise.
and Token Variation in the Perception What is needed to explore this possibility is a thor-
of Size and of Distance ough examination of the visual information contained in
With respect to size estimation, the results of Experi- objects and in scenes that specify object sizes.
ment 2 show that the sizes of token invariant objects
were more accurately estimated than were those of token Why do Subjects Estimate
variable familiar objects and totally unfamiliar objects. the Sizes of Unfamiliar Objects so Well?
With respect to distance estimations, the results of Ex- Although Experiment 2 showed consistently that the
periment 2 show that the distances to token invariant ob- subjects estimated the sizes of the unfamiliar objects
jects were estimated more accurately than those to token with less accuracy than those of familiar ones, the analy-
variable or unfamiliar objects, but only for larger distances ses also showed that their estimations were still quite ac-
where distance estimations were less accurate under all curate. The scaling between true and estimated size still
conditions. Hence, when information to distance was less followed a ratio scale (zero intercept, unity power, and
than complete, familiarity with object size helped, espe- with a slope only slightly above unity), and the mean
cially if the object had little token size variation. multiple correlation was R2 = .94. The average absolute
For the distance estimation results, we found a three- error between estimated size and true size for the unfamil-
way difference: Token invariant familiar objects provided iar objects, regardless of viewing distance, was only 4 in.
more information to distance than did token variable fa- The 15 unfamiliar objects were cardboard cutouts in
miliar objects, both of which provided more information the shape of squares, rectangles, and ovals. These objects
than unfamiliar objects. This allowed for separation of two conveyed no overall familiar size information: The rela-
effects. First, previous experience, which, by the results of tively accurate size estimates could not have been pro-
Experiment 1 provided highly accurate information about duced from memory based on prior contact with the ob-
prototypic size, aided in the estimation of distance, re- jects. So why did the subjects do so well? We consider and
gardless of whether token sizes were highly variable or reject four possibilities, the first two of which are based
not. Second, low token variation provided some additional on familiarity, the last two on perception.
information that could be used to help estimate distances One could argue that our unfamiliar objects still con-
accurately. veyed some familiarity, based on fortuitous shape–size
The comparable three-way difference was not found correlations. But this seems unlikely. They were cut from
for size estimation. Familiarity did not increase the accu- very large sheets of heavy cardboard into regular geo-
racy of size estimations unless the object had size invari- metric shapes of different sizes and were painted with
ant tokens. different flat colors. There was no correlation between ei-
1150 HABER AND LEVIN

ther the objects’ sizes and their construction (shape, color, were just the kinds of objects that the size–distance in-
thickness, or surface texture) or with their placement in variance hypothesis (SDIH) was designed to describe:
the field. Nor could the subjects have seen or interacted When the objects are unfamiliar and their properties are
with the objects before making their judgments. Even unknown, distance information should help subjects es-
though the subjects could identify the different shapes and timate their sizes. Yet, it did not do so.
colors, they had no way to use that to inform themselves There is a further problem with a dependence on dis-
about the objects’ sizes. tance variables to inform size perception. The stimulus
Following the argument about token variable familiar information examples just suggested (illumination, ob-
objects, perhaps these objects contained some token in- ject surface texture), plus most others that could be men-
variant familiar details that could be used to estimate their tioned (e.g., stereopsis, motion perspective), undergo
overall sizes. However, we could neither think of any pos- substantial degradation with distance, especially over the
sibilities nor find any after the fact by examination of the distance ranges used here. There is clear evidence of this
painted cardboards. in the distance estimations: lower accuracy and greater un-
While all of the 45 objects are on display in the same derestimation at the greater distances. But size estima-
grassy field, the objects of different familiarity types are tions were not degraded at the farther distances!
located in different 90º quadrants, with a 30º empty gap We do not know how the subjects arrived at the size
between the nearest objects of different familiarity. The estimations for the unfamiliar objects in this experiment.
subjects had been instructed to face only one quadrant at All we can say is that they did not do it in the same way as
a time and to not look across to any of the others while re- they did for the distance estimations. This ignorance re-
sponding to one of them. We did not observe any subject flects a general ignorance about the perceptual variables
doing this. Therefore, it is unlikely that the subjects could underlying size perception. Most of the theoretical dis-
have compared the distances to familiar and unfamiliar cussions about size perception appeal to familiarity (as
objects in the same or immediately successive glances, do we) and ignore any other variables. But there must be
and used the distance information to help them estimate some others, and size perception theorists have to identify
sizes of the unfamiliar objects. Furthermore, because dis- and demonstrate them (see Gillam, 1995, for some hints).
tance information did not predict size estimation accu-
racy, it does not appear that “peeking” would have helped. The Perception of Object Size
Finally, there was no interaction in the accuracy of es- is Independent of Object Distance
timating the sizes of unfamiliar objects as a function of In Experiment 2, we failed to find any hint that the ac-
whether the subjects judged the familiar or unfamiliar curacy of size perception of an object was affected by the
objects first. true distance to the object, by the amount of distance in-
Could there have been any perceptual information formation in the scene, by the subject’s estimated distance
available while the subjects looking at the objects that to the object, or by the accuracy or error made in distance
was predictive of their sizes? Several examples can be estimation. The ability to estimate the size of objects lo-
imagined. If there is a single source of illumination (es- cated from 10 to 100 ft was unaffected by any variable
pecially a distant one) lighting a group of objects in a involving distance.
scene (as was the case), the length of an object’s shadow This conclusion is just as valid for unfamiliar objects
cast on the ground will be correlated with the object’s (see above) about which the subjects had no prior knowl-
height. If observers take the singularity of illumination edge of their token or prototypic size, as for highly famil-
into account, they will have access to a scalar for each iar token invariant objects. Since the unfamiliar objects
object’s size, without having any familiarity with the ob- were estimated less accurately in size than were familiar
jects. As a different example, if a number of the objects ones in Experiment 2, the failure of distance to aid size
share similar surface texture (which was the case) and are perception cannot be attributed to a ceiling effect.
located close enough to the observer so that the texture Although the results regarding the role of distance in
density is perceivable (unlikely for the very distant ob- the perception of object size may be surprising to those
jects), then the extent (or number) of the textural ele- who believe that distance information is used to perceive
ments will be correlated with the extent of the object it- size, we feel they make good sense. Most of the objects
self. Again, if perceivers take the singularity of surface we encounter in the natural world are known objects with
qualities into account, they will have access to a partial invariant or moderately variable tokens. For these object
scalar for object size. types, the accuracy of the perception of their sizes should
The problem with the examples of potential sources of be primarily dependent on the accuracy of identification,
perceptual information and all similar ones, is that these the degree of familiarity, and knowledge of their token
are just the kinds of stimulation that are also used for dis- variability. The role of distance cues in aiding the percep-
tance perception. Yet, there is no correlation between the tion of the sizes of known objects may be far less impor-
accuracy of size estimations and distance estimations. If tant (as long as the object is close enough to identify or
the subjects were using a source of information for size recognize unambiguously).
perception that also underlies distance perception, there Size is a property of objects themselves, not of their
should have been a correlation between these distance location, nor of the location of perceivers. Given that we
variables and size estimations. But there is none. These live in a world primarily made up of rigid objects, size is
 INDEPENDENCE OF SIZE AND DISTANCE 1151

a constant for any specific instance of an object. The in- This is just the condition in which the SDIH predicts that
stances of objects of a given name (e.g., basketballs) have familiar size should be effective in aiding distance per-
a prototypic size and a token size variation that are de- ception.
pendent solely on the distribution of such objects. We sug- However, the above finding is a between-conditions
gest that our knowledge of the world and not our percep- effect, one that depends on the selection of objects. It fails
tion of the world determines our estimates of the size of as a within-conditions effect. For example, change true
objects in the world. Specifically, for most objects, the size to perceived size in the SDIH equation so it now reads
perception of their sizes is independent of distance per-
perceived distance
ception. Furthermore, the token variability of objects
should exert a stronger effect on their size perception = perceived size / tangent of retinal size.
than should cues to their distance.
Hence, the formulation of the size–distance invariance In Experiment 2, we found no support for this form of
hypothesis (SDIH) as the SDIH: Neither perceived size nor accuracy in estimat-
ing size accounted for any variance in any of the regres-
perceived size = distance 3 tangent of retinal size sion equations predicting distance. Hence, even though
the token variation status of the objects themselves af-
not only fails to describe the data for any interpretation of
fected the accuracy of estimating their distance, the per-
distance inserted into the equation but it seems irrelevant
ception of their size did not.
for explaining how we interact with objects.
This failure would seem to violate the computational
One could argue that had we used a greater range of
distances, or impaired distance information in some implications of the SDIH. It is not clear how subjects use
the knowledge gained from prototypic size or from token
other way, perhaps we would have found an effect of dis-
variance to help them perceive an object’s distance, but it
tance on size perception. But this argument is backwards.
is not by substituting into an equation their knowledge of
Distance information would be expected to be helpful,
its prototypic size, their knowledge of its token variance,
under the SDIH, when it is very accurate, not when it is
nor their estimate of the object’s size while looking at it.
inaccurate or degraded. Hence, the impact of distance in-
formation should have been found for the objects in the Again, it looks as if familiarity is not a perceptual variable.
As with the absence of distance information or dis-
near distance, those seen under ideal conditions.
tance perception in the prediction of size perception, the
If the SDIH does not account for size perception, what
absence of size perception in the prediction of distance
does? For familiar objects, these experiments provide a
perception seems reasonable to us. Because human be-
clear answer—the estimation of the size of a familiar ob-
ings are highly mobile organisms, distances always vary.
ject is determined primarily from the observer’s knowl-
edge of its size acquired from prior experience with the Therefore, the distance to an object must be perceived at
the time of observation and cannot be determined by ref-
object class and its tokens
erence to prior experience with distances or to prior ex-
perceived size = true (known) size. perience with the objects whose distances are being per-
ceived. Such logic suggests that the perception of distance
This conclusion is strengthened by the findings that
is determined heavily by environmental sources of in-
the size estimation results obtained in Experiment 1 (with-
formation that describe distance and not by the familiar-
out perception) are the same as those in Experiment 2 ity of objects or other kinds of knowledge carried by the
(with perception). Hence, the component of size “per-
perceiver.
ception” based on familiarity with prototypic size for ob-
Therefore, we expect that access to familiar size infor-
jects with low token size variation should be more parsi-
mation about objects in a scene does not provide much as-
moniously described as a memory, prior knowledge or
sistance in the perception of the distance to such objects
experience effect, and not perceptual at all.
and certainly not when there is already good distance in-
formation.
The Perception of Object Distance The moral is clear. We need to study size perception in
is Independent of Object Size the same way that we have studied distance perception
At first blush, the SDIH, which assumes an explicit for the past 200 years: We need to consider the physio-
dependence of distance on size perception (see Epstein, logical, visual, environmental, and ecological sources of
1961; Kilpatrick & Ittelson, 1953), fared well in the form information that specify object sizes. We have a good un-
perceived distance = true size/tangent of retinal size. derstanding of distance perception from this approach.
With an equivalent focus, size perception should catch up.
The critical support comes from the finding that when
the quality of distance information is reduced (objects REFERENCES
located from 150–300 ft, as compared with 10–150 ft)
and overall accuracy of distance perception was poorer Bolles, R. C., & Bailey, C. (1956). Importance of object recognition
in size constancy. Journal of Experimental Psychology, 51, 222-
(validating that distance information was reduced in qual- 225.
ity), the loss in accuracy was not found for the familiar Brunswik, E. (1956). Systematic and representative design in experi-
objects, especially those with low token size variance. mental psychology. Berkeley: University of California Press.
1152 HABER AND LEVIN

Cut ting, J. E. (1998). Information from the world around us. In Levin, C. A., & Haber, R. N. (1993). Visual angle as a determinant of
J. Hochberg (Ed.), Perception and cognition at century’s end (pp. 71- perceived interobject distance. Perception & Psychophysics, 54, 250-
97). San Diego: Academic Press. 259.
Cutting, J. E., & Vishton, P. M. (1995). Perceiving layout and know- Sedgwick, H. (1986). Space perception. In K. R. Boff, L. Kaufman, &
ing distances: The integration, relative potency, and contextural use J. P. Thomas (Eds.), Handbook of perception and human performance
of different information about depth. In W. Epstein & S. Rogers (Vol. 1, pp. 1-57). New York: Wiley.
(Eds.), Perception of space and motion (pp. 71-118). San Diego: Aca- Toye, R. C. (1986). The effect of viewing position on the perceived lay-
demic Press. out of space. Perception & Psychophysics, 40, 85-92.
Epstein, W. (1961). The known-size-apparent-distance hypothesis.
American Journal of Psychology, 74, 333-346. NOTES
Epstein, W. (1977). Stability and constancy in visual perception. New
York: Wiley. 1. In all regression analyses reported in this research, we started with
Gillam, B. J. (1995). The perception of spatial layout from static opti- a nonlinear model, one for each subject; for example,
cal information. In W. Epstein & S. Rogers (Eds.), Perception of
space and motion (pp. 23-70). San Diego: Academic Press. cognitive estimated size = A + B (true size)C
Gillam, B. J. (1998). Illusions at century’s end. In J. Hochberg (Ed.), and then tested the intercept A against 0.00 and the exponent C against
Perception and cognition at century’s end (pp. 97-137). San Diego: 1.00. In each case, we accepted the null hypothesis of zero and unity, re-
Academic Press. spectively, based on the appropriate statistical tests and dropped these
Haber, R. N. (1983). Stimulus information and processing mechanisms terms. Thus, the final equation was the linear form
in visual space perception. In J. Beck, B. Hope, & A. Rosenfeld (Eds.),
Human and machine vision (pp. 157-235). New York: Academic Press. cognitive size estimate = B (true size),
Haber, R. N. (1986). Toward a theory of the perceived layout of scenes. one equation for each subject. Furthermore, residual plots for each sub-
In A. Rosenfeld (Ed.), Human and machine vision II (pp. 109-148). ject revealed no heteroscedasticity, and for no subject was there a sig-
New York: Academic Press. nificant correlation between the absolute residual and the predictor.
Haber, R. N., & Haber, L. (1991). Why mobile robots need a spatial 2. It is possible to model equations that predict, for each object, the
memory. In P. S. Schenker (Ed.), Sensor fusion III: 3D perception error in its estimated size, defined as the difference between the esti-
and recognition (Proceedings of SPIE, Vol. 1383, pp. 411-424). mated size judgment made by the subject and the mean measured pro-
Bellingham, WA: SPIE Press. totypic size of the object, as reported in column 1 of Table 1. Modeling
Haber, R. N., & Levin, C. A. (1989). The lunacy of moon watching: the error in estimation is mathematically equivalent to modeling the es-
Some preconditions on explanations of the moon illusion. In M. Her- timation directly. Therefore, we interpret the magnitude of the R2 as a
shenson (Ed.), The moon illusion (pp. 299-318). Hillsdale, NJ: Erl- measure of accuracy: The greater the amount of variance accounted for
baum. in predicting estimated size by true size, the greater the subject’s accu-
Hershenson, M. (1992). Size-distance invariance: Kinetic invariance racy in estimating size. Furthermore, we used three measures of size es-
is different from static invariance. Perception & Psychophysics, 51, timation as dependent variables in the regression analyses: estimated
541-548. size, absolute error in size (true size minus estimated size), and signed
Hershenson, M. (1999). Visual space perception. Cambridge, MA: error in size. Since these are so highly correlated, analyses are reported
MIT Press. only for estimated size. The other two produce the same pattern of find-
Hochberg, J. E. (1971). Perception: II. Space and movement. In J. W. ings. Therefore, when results are described in terms of accuracy, the
Kling & L. A. Riggs (Eds.), Woodworth and Schlosberg’s Experi- same conclusions apply to descriptions of error.
mental psychology. New York: Holt, Reinhart & Winston.
Ittleson, W. H. (1960). Visual space perception. New York: Springer-
Verlag.
Kilpatrick, F. P., & Ittelson, W. H. (1953). The size-distance invari- (Manuscript received December 30, 1999;
ance hypothesis. Psychological Review, 64, 188-192. revision accepted for publication September 21, 2000.)