Sunteți pe pagina 1din 16

Copyright

1994 by the American Psychological Association, Inc.


F
0096-1523/94/$3.00

Journal of Experimental Psychology:


Human Perception and Performance
1994, Vol. 20, No. 4, 783-798

Pictorial and Motion-Based Depth Information During Active Control of


Self-Motion: Size-Arrival Effects on Collision Avoidance

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Patricia R. DeLucia and Rik Warren


With computer simulations of self-motion, Ss controlled their altitude as they approached a floating
object and, after getting as close as possible to the object, tried to "jump" over it without collision.
Ss jumped significantly later for small objects, compared with larger objects that were approached
from equal distances at equal speeds and were positioned at equal clearance heights. This occurred
even when accretion-deletion information was present and when object width and length were
varied independently. Results were consistent with studies in which Ss judged a large far
approaching object to hit the viewpoint before a small near object that would have arrived sooner
(P. R. DeLucia, 199la, 1991b). Results suggest that pictorial information such as relative size
contributes to active collision-avoidance tasks and must be considered in models of perceived
distance and time-to-arrival.

whether two computer-generated cubes would collide with


each other: Compared to two same-sized cubes, differentsized cubes were predicted more often not to collide, even
when collision was imminent (DeLucia, 1991b).
Thus, ERSdepth and ERSarrjva, can occur under a variety of
display and judgment conditions. These effects, particularly
ERSdepth, can be considered related to size-distance relationships reported earlier (e.g., Epstein, Park, & Casey,
1961; Ittelson, 1951a, 1951b; Kilpatrick & Ittelson, 1951).
For ease of discourse we refer to ERS^^ as the sizearrival effect.
The size-arrival effect seems contrary to predictions that
are based on the general assertion that, in ordinary veridical
perception, motion-based depth information supersedes the
traditional static pictorial depth cues (e.g., Gibson, 1979).
That pictorial information may not always be superseded by
motion-based information has previously been demonstrated in the area of form perception (Dosher, Sperling, &
Wurst, 1986).
In the above studies, however, observers did not control
any aspects of the displays; in other words, the observers
were passive rather than active (Flach, 1990; Gibson, 1979;
Warren, 1990; Warren & McMillan, 1984; see also comments by Hochberg, 1982, p. 203). Some studies suggest
that active perceptual judgments are better than their analogous passive judgments (e.g., Gibson, 1962), but equivocal
evidence has been reported (Flach, Allen, Brickman, &
Hutton, 1992). In this article we demonstrate that the sizearrival effect that was reported with passive judgments
(DeLucia, 199la, 1991b) can occur when observers try
actively to avoid collisions with objects during simulations
of self-motion.

DeLucia (1989, 199la, 1991b) reported that when computer-generated floating objects approached a viewpoint, a
large far object appeared to be closer than a small near
object and appeared to hit the viewpoint before the small
object, which was specified to arrive sooner by time-tocontact information (tau; e.g., Lee, 1980a; see also Tresilian, 1991). Judgments were consistent with pictorial depth
information of relative size, rather than with time-to-contact
information, and occurred even when motion-based information was potentially above threshold. These findings
were referred to as the effects of relative size on judgments
of relative depth (ERSdepth) and arrival time (ERS.,^.,,),
respectively (DeLucia, 199la). Such effects were replicated
with textured objects, relatively long display durations, and
high-resolution photographic animations of approaching objects (DeLucia, 1991a, 1991b), suggesting that the effects
are not merely a result of using impoverished displays.
Furthermore, ERSdepth and ERS^^ occurred but were
reduced with simulations of self-motion as well as object
motion during self-motion (DeLucia, 199 Ib). Finally,
^ occurred when observers made judgments about

Patricia R. DeLucia, Armstrong Laboratory, Wright-Patterson


Air Force Base, Dayton, Ohio, and Department of Psychology,
Texas Tech University; Rik Warren, Armstrong Laboratory,
Wright-Patterson Air Force Base, Dayton, Ohio.
Portions of this research were completed while Patricia R.
DeLucia held a National Research Council-U.S. Air Force Armstrong Laboratory Research Associateship and were presented by
her at the 1992 Annual Meeting of the Association for Research in
Vision and Ophthalmology in Sarasota, Florida. We are grateful to
David N. Lee for generous discussions of his theory. We also
thank Jeff Maresh and Robert Todd of Engineering Solutions, Inc.,
for technical assistance and James R. Tresilian and three anonymous reviewers for comments on an earlier version of this article.
Correspondence concerning this article should be addressed to
Patricia R. DeLucia, Department of Psychology, Texas Tech University, Lubbock, Texas 79409-2051. Electronic mail may be sent
to djxpd@ttacsl.ttu.edu.

Experiment 1
The purpose of Experiment 1 was to determine whether
an observer moving at a constant speed toward an object
would "jump" sooner to avoid collision with a large stationary object, compared with smaller objects at the same
783

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

784

PATRICIA R. DELUCIA AND RIK WARREN

distance and clearance height. An analysis of visual information that is available to guide the act of jumping was
provided by Lee (1974, 1980b; see also Laurent, DinhPhung, & Ripoll, 1989; Lee, Lishman, & Thomson, 1982;
Warren, Young, & Lee, 1986).
For example, Lee (1974, 1980b) defined a maximally
efficient jump as one in which the zenith of the jump is
directly above the top of the obstacle. Such a jump requires
that a minimum energy, Emin, be expended at takeoff to
raise the center of gravity a height (h) just enough to clear
the wall. Emin is essentially proportional to h. The position
coordinates of any visible texture element are specified in
units of the observer's height above the ground; thus, bodyscaled spatial information about the size of the obstacle is
available in the optic flow field. It is assumed that such
body-scaled information could be used to judge Emin (1974,
p. 263).
The jump also must be initiated at the right time so that
the observer reaches the obstacle as the peak of the jump is
reached. In other words, the time to reach the obstacle
(measured from takeoff) must equal the time to reach the
zenith (D. N. Lee, personal communication, October 14,
1993). The time between takeoff and zenith is essentially
proportional to the square root of h and could be judged
using body-scaled information (Lee, 1974, p. 263). For any
given takeoff point, the time it would take to reach the
obstacle is the current time-to-contact with the obstacle,
which is specified optically by tau. To initiate the jump at
the right time, therefore, the jumper must monitor the timeto-contact with the obstacle (possibly via tau) and initiate
the jump just as the time-to-contact reaches the time from
takeoff to zenith (D. N. Lee, personal communication, October 14, 1993). Thus, Lee's analysis indicates that tau
provides potentially useful information to guide the timing
of a jump.
Lee's (1974) model does not predict that a jump would be
affected by the width (i.e., horizontal extent) or length (i.e.,
vertical extent) of the obstacle if the altitude of the top of the
obstacle is constant; perceived time-to-contact with the obstacle should be the same regardless of length, assuming
that the observer moves forward on a level path at a constant
horizontal velocity (D. N. Lee, personal communication,
October 14, 1993). In the displays used in our study, objects
of different sizes had the same time-to-contact and height.
One could argue that, if our observers timed their jumps on
the basis of optical information about time-to-contact as
described in Lee's analysis of jumping, they should start
their jump at the same time for objects of different sizes.
Our observers were not instructed explicitly to complete the
maximally efficient jump defined by Lee; rather, they were
instructed to get as close as possible to the object before
jumping in order to avoid collision (see "Procedure and
design" in the Method section). Nevertheless, such a jump
must be initiated at the right time to avoid collision, and this
time should not vary with object size when clearance height
and time-to-contact remain constant; optical time-to-contact
information (possibly tau) is a viable source of information
for such a task.

We gave observers practice with jumping in an attempt to


provide body-scaled information about height, that is, information about the relationship between control-stick input
and displayed output. We acknowledge that the observer's
"jump" was merely a simulation. Thus, we can only assume
that adequate body-scaled information about height was
available. Moreover, we assessed the effect of obstacle size
on jumping only when the altitude of the tops of the different-sized objects was the same.
Even though Lee's (1974) model does not predict an
effect of object size on jumping, it is possible that the size
of the object will influence perceived distance through the
same mechanism that operates in pictorial size-distance
relationships (e.g., Epstein et al., 1961; Ittelson, 195la,
1951b; Kilpatrick & Ittelson, 1951) and thus affect the
timing of the jump; that is, a size-arrival effect may occur.
If so, results would be consistent with the previously described passive studies and may indicate that pictorial size
can affect perception-action even when veridical time-tocontact information (tau) is available.
Method
Participants. Ten paid observers had normal or corrected visual acuity and were naive as to the experimental hypotheses.
Apparatus and displays. Computer simulations were created
with an MS-DOS 386/33MHz computer equipped with an XTAR
Falcon-4000 array processor and an XTAR PG-2000 display
board. Displays had 1024 X 768 pixel resolution and were shown
at a presentation speed of 30 frames/s on a 60 Hz, 48.26 cm
(19-in.) monitor. Scenes were based on the slow-expansion scene
described by DeLucia (199la) and are represented in Figure 1.
Simulations represented self-motion toward a black square that
was suspended above an islandlike ground plane. For ease of
description, we give most lengths in feet (meters), but rescaling to
other units is geometrically possible. The virtual ground plane was
10,000 ft wide X 20,000 ft long (i.e., about 3,030 m wide X 6,060
m long); was covered with randomly patterned, solid-colored
squares; and was located against a green background or "sea."
Above the sea was a blue sky; an explicit horizon line was present.
The virtual square was either 16 ft, 72 ft, or 128 ft (4.85 m, 21.82
m, or 38.78 m) in size, and was centered horizontally and vertically
along the principal line of sight (the forward or z axis). The virtual
eye (self) always began 128 ft (38.78 m) above the ground plane

\
\
Figure I. Experiment 1: Schematic representation of a selfmotion scene. (The actual scenes were in color.)

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE


(one eyeheight). At the beginning of each trial a single square was
located 1,280 ft (387.84 m) from the viewpoint along the forward
or z axis, and self-motion was forward at a constant speed of 150
ft/s (45.45 m/s). All squares were initially at the same distance, and
self-motion was always at the same speed toward all squares.
Procedure and design. Using a chin rest, participants viewed
the displays binocularly from the projectively correct viewing
position (36-cm distance). If participants did nothing, they would
"fly" straight and level and collide with a stationary square (except
when the top of the smallest square was aligned with the tops of
the medium-sized or largest square; see below). However, participants were instructed to "fly" over the square, first getting as close
as possible to the square before "launching" themselves to avoid
collision. Once past the square they were to lower their altitude. In
effect, they were to follow the side profile of the square. These
instructions were used to discourage observers from initially flying
high enough to clear the square and then remaining at that altitude
to avoid collision.
Participants controlled their simulated altitude but not their pitch
or forward speed: They pushed a control stick back in order to
ascend and forward in order to descend, as in an airplane, except
that the horizon always remained vertically centered on the screen;
the maximum vertical velocity was 100 ft/s (30.3 m/s). Altitude
control was possible at the start of the trial and lasted for 9.87 s,
during which altitude was sampled at 10 Hz. The time between the
beginning of the trial and potential contact with the square was
8.53 s.
It was expected that the height of the square would affect the
participant's initiation of the jump. Therefore, it was important to
compare jumps for different-sized squares only when the tops of
the squares were aligned (i.e., had the same height). Thus, additional scenes were created: For each scene in which a square was
centered on the principal line of sight, the other squares' heights
were adjusted, in separate scenes, so that all tops were aligned.
This is referred to as the square-top factor, which had three levels,
one for each square size. There were nine unique scenes, and each
scene was replicated 10 times in a random order for a total of 90
trials. Intertrial duration was self-paced, and feedback was not
provided. To familiarize the participants with the task and the
dynamics of the control stick, we provided 6 practice trials (more
on request); these were not included in the analyses.
The main objective of this study was to determine whether
participants would jump sooner to avoid collision with larger
squares than they would with smaller squares, presumably because
the larger squares appeared closer (DeLucia, 1989, 1991a, 1991b);
this was evaluated primarily with measures of launch time, defined
below. Other dependent variables were also obtained from the
sampled altitude time history:
1. Peak altitude is the maximum altitude that occurred after 6 s
from the start of the trial. This was done to exclude early false
jumps.
2. Time at peak altitude is the time at which peak altitude
occurred.
3. Launch time is the amount of time that elapsed after the start
of the trial when the jump was initiated. Launch time also reflects
the distance between the self and the square at the time of the
jump. That is, greater or later launch times indicate smaller distances between the self and square. To compute launch time, an
algorithm looked backward in time from the occurrence of peak
altitude and computed the percentage change in altitude at each
sampled point in time (i.e., the slope of the altitude). Launch time
was defined as the time at which the slope exceeded 1%. This
measure corresponded to estimates of launch time on the basis of
visual inspection of the mean time histories (e.g., Figure 2).

785

4. Altitude at launch is altitude at the time of launch.


5. Clearance is the difference in altitude between the top of the
square and the self, when the self was directly over the square.
Clearance indicates how far above the square the jump occurred.
6. Frequency of crashes is the number of trials in which the self
and square were ever at the same place at the same time; it was
measured as a percentage of the trials. This measure was not
affected significantly by any of the variables in Experiments 1-5.

Results
Figure 2 shows the mean time histories (the altitude as a
function of time) for each condition averaged across replications and participants. Figure 3 shows the means of each
dependent measure for each square size and top condition,
averaged across Experiments 1, 2, and 4; these experiments
shared the same pattern of results, as indicated by the results
of an analysis of variance (ANOVA). Each dependent variable in Experiment 1 was analyzed with a 3 X 3 (Square
Size X Altitude of Square Top) completely repeated measures ANOVA.1
Launch time. Only square size affected launch time,
F(2, 18) = 62.14, p < .0001, w2 = 40.16%, which increased as size decreased. In other words, people approached the smaller squares more closely before "launching" themselves to avoid collision. Conversely, people
launched themselves sooner when approaching larger
squares. Means were 7.80 s, 7.39 s, and 6.89 s for 16-, 72-,
and 128-ft (4.85-m, 21.82-m, and 38.78-m, respectively)
squares, respectively, with all comparisons significant, p <
.05. (Recall that collision with the square would occur 8.53
s after the beginning of the trial.)
Altitude at launch. Analyses indicated a main effect of
square top, F(2, 18) = 16.67, p < .002, to2 = 18.42%, and
square size, F(2, 18) = 13.32, p < .004, o>2 = 23.68%.
There was also a significant Square Top X Square Size
interaction, F(4,36) = 8.44, p < .001, 2 = 1.94%. Launch
altitude increased as size decreased and as the altitude of
square top increased.
Time at peak altitude. There was an effect only of
square size, F(2, 18) = 20.10, p < .0001, o>2 = 21.36%.
Time increased as square size decreased, with all pairwise
comparisons significant (p < .05), except between 16-ft
(4.85-m) and 72-ft (21.82-m) squares. Thus, participants
reached their peak altitude later for the smaller squares than
they did for the larger squares, which is not surprising
because launch time was also later for smaller squares.
Peak altitude. Analyses indicated a main effect of
square top, F(2, 18) = 290.51, p < .0001, to2 = 52.62%,
and square size, F(2, 18) = 16.36, p < .0002, <a2 = 4.40%,
but no interaction. As would be expected, peak altitude
increased as the altitude of square tops increased; Tukey's
honestly significant difference (HSD) tests indicated that all
pairwise comparisons were significant (p < .05). Peak
altitude increased as square size increased, with significant
1
All probability values from the results of the ANOVAs reflect
Greenhouse-Geiser corrections.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

786
PATRICIA R. DELUCIA AND RIK WARREN

8! *

,I

UJ

?!

1 ? !

(u)3oniuiv

91
*
8

(j) 3aruu.iv

(u) 3001111*

rt *-

(u)aoniuiv

(u) Mniuiv

8
M

8
~

8
8
8
8

W
M
M

(U) 30OLI1TV

CO

<o

8
-

(w) 3oniinv

821

(u) 3oniiriv

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE

differences among all sizes, p < .05, except between 16-ft


(4.85-m) and 72-ft (21.82-m) squares.
Clearance. Results indicated only a Square Top^ X
2
Square Size interaction, F(4, 36) = 3.32, p < .04, w =
1.46%.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Discussion
As shown by the means for launch time in the top left
panel of Figure 3, participants jumped sooner for larger
squares compared with smaller squares approached from
equal distances at equal speeds. Accordingly, the smaller
the square, the closer the approach before the launch was
initiated. This could have occurred because the small
squares appeared relatively farther away, as would be consistent with the previously described passive judgments
(DeLucia, 1989, 1991a, 1991b).
In addition, participants began their jumps at lower altitudes for larger squares than they did for smaller squares
(see the top middle panel of Figure 3), which could have led
to relatively earlier launch times and peak altitudes for
larger squares (but see Experiment 5). Lower launch altitudes for larger squares occurred even when altitudes of all
square tops were aligned and thus may indicate a misperception of height due to square size; this is consistent with
informal observations made during the development of the
displays that smaller squares appeared higher. Once participants began their ascent and attained more optical information, however, they may have realized that they had
overestimated the small square's height and made an adjustment; this would account for the lower peak altitudes for
smaller squares than for larger squares (see the bottom left
panel of Figure 3).

Experiment 2
The earlier launch times for the large squares may have
occurred because the larger squares occluded the ground
plane at lower positions compared to the smaller squares,
thus providing information that the large squares were
nearer. In Experiment 2, the ground plane was eliminated.

Method
Ten new participants having the same characteristics as the
participants in Experiment 1 completed Experiments 2 and 3. Half
of the participants completed Experiment 2 first. Computer simulations were the same as in Experiment 1 except that the squares
were located against a solid blue background; neither a ground
plane nor a horizon line was present. The participant's task was
identical to that in Experiment 1.

787

Results and Discussion


The F, p, and o>2 values from the ANOVA are shown in
Table 1. The pattern of results was the same as in Experiment 1, except for clearance: Only square size affected
clearance, which increased as size increased; the difference
between 72-ft (21.82-m) and 128-ft (38.78-m) squares was
not significant. Mean launch times were 7.94 s, 7.49 s, and
7.15 s for the 16-ft (4.85-m), 72-ft (21.82-m), and 128-ft
(38.78-m) squares, respectively, with all pairwise comparisons significant (p < .05). These results suggest that the
differences among the positions of the squares' occlusions
of the ground texture do not account for the data from
Experiment 1. The effect of size on clearance in Experiment
2 may indicate that the potential misperception of square
height, discussed in Experiment 1, is exacerbated when
background texture is not present.

Experiment 3
In Experiments 1 and 2, the height of the bottoms of the
squares varied when square size and the altitude of square
top varied. In Experiment 3 we determined whether effects
of size on collision avoidance would occur when the altitude
of either square top or square bottom remained constant.
Method
The 10 participants from Experiment 2 performed the same task
as in Experiment 1. The scenes were similar to Experiment 1 in
that a textured ground plane and horizon line were present. However, the altitude or height of the top of the square, or of the bottom
of the square, was systematically kept constant while the size of
the square varied. In three separate scenes, the top of the square
was always 148 ft (44.84 m) above the ground and the size of the
square was either 36 ft (10.91 m), 90 ft (27.27 m), or 144 ft (43.63
m); on such trials the altitude of the square's bottom also varied.
In three additional scenes the bottom of the square was held
constant at 100 ft (30.3 m) above the ground, and the size of the
square was either 36 ft (10.91 m), 90 ft (27.27 m), or 144 ft (43.63
m); on such trials, the altitude of the square's top varied. Participants completed a total of 6 unique scenes and 60 trials in a
random order.

Results
Mean time histories are shown in Figure 4. Data were
analyzed with a 2 X 3 completely repeated measures
ANOVA. In this design there were three square sizes and
two levels of a factor referred to as "fixed." The two levels
of the fixed factor were altitudes of square top and square
bottom.
Launch time. Only square size affected launch time,
F(2, 18) = 56.34, p < .0001, <o2 = 51.03%. As in Experiments 1 and 2, launch time increased or occurred later as

Figure 2 (opposite). Experiment 1: Altitude as a function of time, averaged over replications and participants, for each square size and
square top condition. (Vertical lines represent a side view of the squares.)

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

788
PATRICIA R. DELUCIA AND RIK WARREN

Hme-to-cMUct (s)
o 52

1 i]

Ilil

O>

o s?

i
s

i 9ft

<

<

H-tt

(s)etuu.ipunei

00

SJ

IS

T1me-to-cmtct (s)

2
8

() 8CXIBIB8IO

.w
*

i-S
> 3

il

8.

N
wf

^
||

1|
8-1

.8-

23
"3
1- C 00

S'C
S g
S 2

.
s

ill

789

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE


Table 1
Results of Experiments 2 and 4
Independent
variables
Square size
F(2, 18)
P<
2

(a (%)

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Square top

Launch
time

Time at
peak
Launch
altitude
altitude
Experiment 2

Peak
altitude

Clearance

35.05
.0001

11.44
.007
13.00

11.58
.002
17.35

18.07
.002
6.40

21.75
.0002
21.19

35.77
.0001
24.45

NA
NA

ns

529.7
.0001
62.77

NA
NA

7.13
.002
1.1%

NA
NA

NA
NA

NA
NA

33.53

F(2, 18)

ns

P<

NA
NA

OT (%)

Top x Size
F(4, 36)
P<
2

<a (%)

Square size
F(2, 18)
P<

or (%)
Square top
F(2, 18)
P <2 <u (%)
Top X Size

ns

NA
NA

ns

ns

ns

ns

Experiment 4
241.77
.0001
79.84

741.46
.0001
45.97

14.17
.002
24.74

11.02
.006
3.50

24.83
.0002
24.40

NA
NA

ns

1672.11
.0001
50.63

ns

NA
NA

ns

ns

ns

1829.46
.0001
74.87
ns

NA
NA
ns

ns

Note. NA = not applicable.

size decreased; means were 7.71 s, 7.21 s, and 6.79 s for


36-ft (10.91-m), 90-ft (27.27-m), and 144-ft (43.63-m)
squares, respectively, with all comparisons significant,
p < .05.
Altitude at launch. Analyses indicated an effect of the
fixed factor F(l, 9) = 17.71, p < .002, w2 = 11.43%, and
a Square Size X Fixed Factor interaction, F(2, 18) = 14.30,
p < .002, co2 = 10.9%. Launch altitude increased as size
decreased when the altitude of square top was fixed but
increased as size increased when square bottom was fixed;
the latter is not surprising because, in this condition, square
tops began relatively higher for larger squares than they did
for smaller squares. Finally, launch altitude was lower when
the top was fixed, compared to the bottom, except for the
scene with the smallest square.
Time at peak altitude. Results indicated no significant
effects.
Peak altitude. Analyses indicated a main effect of the
fixed factor, F(l, 9) = 420.96, p < .0001, to2 = 16.04%,
and square size, F(2, 18) = 160.89, p < .0001, o2 =
38.60%; and a significant Fixed Factor X Square Size
interaction, F(2, 18) = 199.11, p < .0001, w2 = 18.71%.
Peak altitude increased as size increased, but differences
were greater when square bottom remained constant, compared to square top. Peak altitude was smaller when the top
was fixed, compared to the bottom, except for the scene
with the smallest square.
Clearance. No significant effects were found.

Discussion
Results indicate that when the altitude of the top of the
squares, or of the bottom of the squares, remained constant,
launch time increased or occurred later as size decreased.
Thus, it would seem that the results of Experiments 1 and 2
were not due to the covariation between square size and the
altitude of squares' tops or bottoms. Such an interpretation
is not without a caveat, however, because in Experiment 3
the altitude of both the square's top and its bottom could not
be kept constant as size varied; when size varied either top
or bottom was constant (see Warren & Owen, 1982).

Experiment 4
Although participants were free to control their altitude as
they wished in Experiments 1-3, they did not necessarily
have to vary altitude much in order to complete the task
successfully. Furthermore, instructions tacitly encouraged
observers to get as close as possible to the square at the
initial altitude before jumping. Thus, it is not known
whether participants would have responded differently if
additional visual information were made available by varying altitude as they approached the square, which would
result in accretion-deletion of the ground texture by the
square. In Experiment 4, participants varied their altitude
during the approach in order to provide such information
and were informed that the information would be present.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

790
PATRICIA R. DELUCIA AND RIK WARREN

3?

UJ *

sx

0)
N
H
CO

t)

8
Pi
8
(M
8
W
8

I ?!
8

(U) 30fUU1V

I 2 I ?
I
g

8
M

8
CM

8
8

(w) aoruuiv

S.

c
o

<y>
O

u
I.

>
o
o

CD

8
-

(u) aorutrw

8 8

(u)3oruin
ca 73

3 H

^s

S aj

!
ll

o ,

00 g

791

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE

8 *

"-'

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

,8

I 8 I ? *

8 I ? *

<u) soniinv

(u) Mfuuiv

(U) 30rUllTV

so,

1
CX
T3

.1

u
o

|8

8 i ? *

(u) xjniuiv

8 8 8 ?

(U) 30fUUTV

1"
3
C w
O J

<o

CO

^2
u

.jj

'
I
I

3 -O
&C C

1 2 I ?

(U) 3Q01U1V

9L

(U) SOtUllTY

ZL

821

792

PATRICIA R. DELUCIA AND RIK WARREN

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Method
Ten new participants had the same characteristics and viewed
the same computer simulations as in Experiment 1. Participants
completed the task described in Experiment 1, but they were
instructed first to perform three additional maneuvers as rapidly as
possible: (a) bring the top of the square below the horizon, (b)
bring the bottom of the square above the horizon, and (c) center the
square on the horizon. Participants were told that by completing
these maneuvers, the square would cover and uncover the texture
on the ground plane, which would provide information about the
distance of the square that could be used to complete the jump.
After completing the three maneuvers, participants completed the
collision-avoidance task described in Experiment 1.

Results and Discussion


Mean time histories are shown in Figure 5, and results of
the ANOVA are shown in Table 1. The same pattern of
results occurred as in Experiment 1, except that the Square
Size X Square Top interaction (on launch altitude) was not
significant. Also, clearance measures indicated the same
patterns as in Experiment 2. Mean launch times were 8.22 s,
7.71 s, and 7.27 s for 16-ft (4.85-m), 72-ft (21.82-m), and
128-ft (38.78-m) squares, respectively, with all comparisons
significant (p < .05). Even when participants were informed
that, by varying their altitude, accretion-deletion information would become available and provide information about
the distance of the square, they still jumped later for the
smaller squares than they did for the larger squares. Such
results suggest that accretion-deletion information was not
as effective as pictorial size in this task.

Optical Analyses: Experiments 1-3


Even though tau and its first derivative were the same for
different-sized squares, participants jumped significantly
sooner for large squares than for smaller squares, suggesting
that such optical information was not the dominant optical
basis for the timing of the jump. Additional optical variables
of magnitude of visual angle, absolute change in visual
angle, rate of change in visual angle (i.e., rate of expansion),
and elevation angle were estimated from mean launch times
and launch altitudes for the smallest and largest squares in
each square top condition of Experiments 1-3.2 Elevation
angle refers to the angle between the horizon line and top of
the square, and it seemed to be a viable source of information about square height. For each of these variables we
computed the ratio of large-to-small squares at the beginning of the trial and again at launch time.3
If participants waited for a particular optical variable to
reach the same value for the small and large squares before
jumping, then the ratio of the value for small to large
squares would approach 1 at launch time. As represented in
Figure 6 (left panel), the ratios for visual angle, change in
angle, and rate of expansion were relatively closer to unity
at launch time compared to the start of the trial, whereas

the ratio for time-to-contact information began at, and


then diverged from, unity. This may suggest that visual
angle of the square, change in visual angle, and/or rate of
expansion affected the timing of the jump more than did
time-to-contact information (but see the General Discussion section).
Furthermore, the mean ratio for elevation angle remained
most stable across time, possibly because the squares' tops
began at the same altitudes. We note that launch altitudes
were higher and launch times were later for the smaller
squares than for the larger squares, possibly indicating a
misperception of height and distance, respectively. More
important, higher launch altitudes coupled with later launch
times would correspond to a decrease and an increase,
respectively, in elevation angle, thereby potentially accounting for a stable elevation angle. Thus, participants may have
adjusted their altitude to a perceived optimal value prior to
the jump, possibly based on elevation angle and the square's
apparent height. We hypothesize further that once participants began their jump, they realized that the small object
was lower than they initially perceived and then compensated, thereby accounting for the lower peak altitudes obtained with smaller objects.
The same pattern of results occurred for all experiments
except in Experiment 3, when square bottoms began at the
same altitude (refer to Figure 6, right panel): the ratio for
elevation angle showed a decrease at launch time. In other
words, participants increased their altitude more before
jumping over the larger squares, which is not surprising
because the tops of the larger squares started higher than
those for the smaller squares.

Experiment 5
In Experiments 1-4, square size was varied by increasing
the width and length of the square by the same magnitude.
In Experiment 5 we tested whether size-arrival effects
would occur when either object width or object length alone
was varied. The results of the experiment demonstrated that
such parameters do not both have to change for size-arrival
effects to occur.

Method
Ten new participants viewed the same displays as in Experiment
1, except that the object's width and length were varied independently while the altitude of the object's top remained fixed at 164
ft (49.69 m). The virtual object was either 16 ft (4.85 m), 72 ft
(21.82 m), or 128 ft (38.78 m) wide and fixed in length, or was 16
ft (4.85 m), 72 ft (21.82 m), or 128 ft (38.78 m) long and fixed in
width; width and length factors were completely crossed. Details
2
Visual angles were estimated from the full size of the virtual
object even though, at launch time, the bottom of the mediumsized and largest squares were typically beyond the bottom of the
computer screen.
3
The start of the trial is defined here as 1 s after the trial began
had there been no change in altitude; time-history plots indicated
minimal change.

793

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE

Opted Vartabte
----- \*uelAn>

Opted VWM*
--"-- \feuelAn0e

--O- Rfecf&qparebn
v

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

---- TtnetoCortact

--D---" BwefcnArga
-- TTmMContacJ

\
i s
-"
IsAter TriaTs Onset

-te

AtTmecfLaixh

TITO

1 tAterTrtafsOnset

AtTmeofLaurrh

Trre

Figure 6. Magnitude of several optical variables at 1 s after the beginning of the trial and at mean
launch time, expressed as the ratio of the largest and smallest squares. Left: Representative example
for Experiment 1 (square top = 164); right: Experiment 3 (fixed square bottoms).
of the displays, procedure, and data analyses were the same as
those described in Experiment 1.

Results and Discussion


Figure 7 shows the means of launch time, time at peak
altitude, and peak altitude for each width and length condition. Dependent measures were analyzed with a 3 X 3
(Object Width X Object Length) completely repeated measures ANOVA.
Launch time. Analyses indicated effects of object width,
F(2, 18) = 27.55, p < .0001, w2 = 3.4%, and object length,
F(2, 18) = 9.9l,p < .0049, w2 = 7.80%, but no interaction.
Launch time increased as object width decreased, with all
pairwise comparisons significant (p < .05). Launch time
increased as object length decreased from 128 ft (38.78 m)
to either 16 ft (4.85 m) or 72 ft (21.82 m; p < .05); other
comparisons were not significant. Mean launch times were
7.14 s, 7.01 s, and 6.87 s for the 16-ft (4.85-m), 72-ft
(21.82-m), and 128-ft (38.78-m) widths, respectively.
Means were 7.06 s, 7.19 s, and 6.77 s for corresponding
object lengths.
Altitude at launch. No significant effects occurred.
Time at peak altitude. There were significant effects of
object width, F(2,18) = 10.84, p < .0029, w2 = 4.20%, and
object length, F(2, 18) = 7.25, p < .0082, co2 = 11.29%.
Time at peak altitude increased as object width decreased
from 128 ft (38.78 m) to 16 ft (4.85 m), p < .05, and as
object length decreased from 128 ft (38.78 m) to 72 ft
(21.82 m;/7 < .05); other comparisons were not significant.
Peak altitude. Results indicated effects of object width,
F(2, 18) = 16.70, p < .0002, w2 = 1.24%, and object
length, F(2, 18) = 11.51, p < .0045, w2 = 3.25%. Peak
altitude increased as object width increased from 16 ft (4.85
m) to 72 ft (21.82 m) or 128 ft (38.78 m; p < .05). Peak

altitude increased as object length increased from 16 ft (4.85


m) or 72 ft (21.82 m) to 128 ft (38.78 m) (p < .05); other
comparisons were not significant.
Clearance. Results indicated a significant effect of object width only, F(2, 18) = 5.44, p < .0246, w2 = .57%;
clearance increased as object width increased, with significant differences between 16-ft (4.85-m) and 128-ft
(38.78-m) widths only.
The results of Experiment 5 are consistent with those of
Experiments 1-4; size-arrival effects can occur when either
an object's width or length varies. Furthermore, the lack of
interactive effects of width and length on performance suggests that such effects combine additively in our collisionavoidance task. Indeed, the amount of variance in mean
launch time attributed to object size in Experiments 1-4
was much greater than that attributed to either length or
width in Experiment 5. The relatively smaller effects in
Experiment 5 can be observed by comparing the top left
panels of Figures 7 and 3. These results may reflect an effect
of the object's area on launch time, but further study is
needed before such a conclusion can be drawn. Finally,
object width or length did not affect launch altitude, as did
square size in Experiments 1-4. Such results suggest that
the lower launch altitudes for larger squares that occurred in
the previous experiments do not account for the corresponding earlier launch times and peak altitudes, which also
occurred in Experiment 5.

General Discussion
The present results demonstrate robust size-arrival effects
on active collision-avoidance tasks. Participants jumped
earlier and higher to avoid collision with large objects,
compared with smaller objects. Such results occurred when

794

PATRICIA R. DfiLUCIA AND RIK WARREN

10
Object Width
D

Smal

A fcbcim

Lage

Time of Contact
0
S3

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

1S3
2S3
16

72

128

CbpctLengfri

260

Object Width

O Srel
A- (Aim
- Lap

160

16

72

138

Object Length (ft)


Figure 7. Experiment 5: Results for each object length and object width condition. Error bars
indicate 1 standard error of the mean. Top, left: Launch time; top, right: Time at peak altitude;
bottom: Peak altitude.

time-to-contact and the clearance heights of the objects


remained constant as object size varied. The same results
were obtained when objects varied in length or width, or
both, and occurred even when accretion-deletion information was available and noted to participants. The present
findings are consistent with previous studies in which observers made passive judgments about the relative arrival
time of two computer-generated approaching objects (DeLucia, 1991a, 1991b, 1992). We consider several potential
explanations of such effects.
Motion-Based Information Is Below Threshold
It could be argued that motion-based information such as
optical expansion or tau was below threshold for smaller
objects compared with larger objects, perhaps due to computer aliasing or because of smaller optical sizes.4 Such an
explanation does not seem tenable for several reasons. First,

as early as 1 s after a trial began, the change in the visual


angle that the objects projected to the eye was above traditional thresholds for displacement of a target (20 s arc;
Easier, 1906; cf. Graham, 1965) and for detection of size
change of a simulated object moving in depth (1 min arc;
Hills, 1975). DeLucia (1991a) also reported that size-arrival
effects occurred when such thresholds were exceeded. Second, in Experiments 1-4 mean launch times increased as
object size decreased, and all pairwise comparisons of size
were significant. Thus, a threshold-based explanation would
imply that time-to-contact information for the mediumsized object was above threshold when it resulted in a lower
4
DeLucia (199la) concluded from empirical data that potential
effects of computer aliasing on tau and motion thresholds do not
account for size-arrival effects in passive tasks. (Here, aliasing
refers to effects of the computer's digitization process rather than
to technical meanings assigned by signal-analysis theory.)

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE

mean launch time than the smallest object, but that the same
time-to-contact information was below threshold when the
medium-size object resulted in a greater mean launch time
than the largest object. Such an argument is not internally
consistent. Finally, results of previous studies suggest the
importance of critical perceptual-motor adjustments in the
last 4 s prior to contact, and die nonuse of tau when the latter
is beyond 10 s (Schiff & Detwiler, 1979). In Experiments
1-4, mean launch times occurred when time-to-contact was
between .32 s and .85 s for the smallest object and between
1.21 s and 1.82 s for the largest object, when tau was well
within its "usable" range. It would seem, therefore, that
size-arrival effects occurred when motion-based information was above threshold.

Tau Is Less Detectable for Smaller Objects


The following explanation of size-arrival effects was
suggested by an anonymous reviewer5: Time-to-contact is
given optically as the ratio of image size to the temporal
derivative of image size. For two different-sized objects
with the same time-to-contact, the image size and temporal
derivative of image size is smaller and therefore necessarily
less detectable for a smaller object. Therefore, tau is relatively less detectable for a small object that has the same
objective time-to-contact as a larger object. According to
this argument, detectability is considered a continuous function, that is, optical size and rate of expansion are above
threshold but are relatively less detectable for smaller
objects.
This explanation must be worked out further before it can
be evaluated completely; Lee's (1980a) model of time-tocontact does not address how time-to-contact information
may be extracted or detected from the optic array (for a
recent discussion of this issue, see Tresilian, 1993). Rather,
the model shows that information about time-to-contact is
available directly in the optic array and need not be computed from distance and velocity. As such, parsimony is a
primary strength of Lee's model. In contrast, a detectability
model, as we understand it, suggests that apparent time-tocontact is based on two separate functions: detectability of
an object's optical size and detectability of an object's rate
of optical expansion, both of which determine the detectability of tau. It is difficult to see how parsimony can be
maintained with such a hypothesis. Moreover, previous
studies suggest that observers are sensitive to tau independently of optical size or rate of expansion (e.g., Regan &
Hamstra, 1991; Todd, 1981; see also Wang & Frost, 1992).
Finally, we note that small fast-moving objects may have
relatively less detectable optical sizes but more detectable
rates of expansion compared with large slow-moving objects that have more detectable optical sizes but less detectable rates of expansion, and which have the same time-tocontact. It is not clear what a detectability model would
predict in such cases. In conclusion, although we cannot
rule out a detectability interpretation of our results, further
specification of such a model is needed before it can be
evaluated completely.

795

Floating Objects Present Misleading


Optical Information
It has been argued that under ordinary circumstances,
optical contact of an object with the ground specifies the
object's physical contact with the ground unless there is
information to indicate otherwise (Gibson, 1950; Sedgwick,
1983). Furthermore, the relative distance of any two locations along the ground (e.g., the bottom of two objects) is
specified optically by the ratio of the respective angles
between the horizon line and the line of regard to a location
on the ground (for small angles; see Sedgwick, 1983, p. 452,
Figure 22). It could be argued, therefore, that because the
objects in the present study were floating, such optical
information about the relative depth of the different-sized
objects was deceptive in specifying that the larger objects
were nearer (see Sedgwick, 1983, Figure 6). When the
bottoms of the objects were at a fixed altitude as size varied
in Experiment 3, however, such angles specified that all
three distances were the same; nevertheless, size-arrival
effects occurred. Moreover, size-arrival effects occurred in
Experiment 5 when the altitude of the objects's top and
bottom were fixed as width varied. This argument, therefore, does not seem to explain our data.
Larger Objects Appear to Require Higher Jumps
J. R. Tresilian (personal communication, November 2,
1992) offered the following explanation of our results: At
the same time-to-contact, larger objects appear to require
higher jumps than smaller objects; thus, the latter result in
later jumps. This is consistent with our finding that participants jumped higher for larger objects (see results regarding peak altitude). However, it is unclear why larger objects
would appear to require higher jumps than do smaller objects; the tops of the objects were at the same altitude when
the object's width and/or length varied.
Nevertheless, we measured judgments of time-to-contact
as a function of object size in a control experiment in which
the jumping task was eliminated; such judgments could not
be based on the perception that larger objects require higher
jumps per se. Specifically, 10 students from Texas Tech
University viewed computer simulations of self-motion toward a square object. They were instructed to press a key
when they thought they would collide with the object
(Schiff & Detwiler, 1979; see also DeLucia, 1991a, Experiment 2). Displays were created with an MS-DOS 4867
50MHz computer and were presented in 640 X 350 pixel
resolution at a speed of 35 frames/s on a 35.56 cm (14-in.)
monitor. The parameters of the scenes were as described in
5
The same reviewer suggested that size-arrival effects may
occur because people are just more nonchalant about running into
something small (i.e., less worried about it and therefore less
careful). This would imply that our participants made inferences
about the objects' virtual sizes and did not use the same criteria in
following the instructions for the three objects. Although our
results do not provide evidence for such hypotheses, our experiments were not aimed to address these issues.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

796

PATRICIA R. DELUCIA AND RIK WARREN

Experiments 1-2 (e.g., the same three levels of square size


and altitude of square top), except that the objects consisted
of white square outlines inscribed with a cross and were
shown against a black background. The scene was terminated after 7 s so that observers did not see when the object
moved completely beyond the edges of the computer screen.
The time between the beginning of the trial and keypress
was recorded.
A 3 X 3 repeated measures ANOVA indicated an effect
of square size only, F(2, 18) = 6.17, p < .03. Estimates of
time-to-contact increased as square size decreased; means
were 8.77 s, 8.47 s, and 8.29 s, respectively. Tukey's HSD
tests indicated a significant difference between only the
smallest and largest squares, p < .05. Thus, consistent with
the results of Experiments 1-5, observers reported relatively
earlier collision times for larger objectseven when the
jumping task was eliminated. Such results make it difficult
to apply the argument that earlier launch times occurred for
larger objects in Experiments 1-5 because they appeared to
require higher jumps. The results of the control experiment
indicate that larger squares were perceived as closer in time
than smaller squares, which may account for the earlier
launch times in Experiments 1-5.

Apparent Size-Distance Couplings


It could be argued that participants jumped earlier for
large objects because they projected a relatively larger image, and because observers assumed the objects were the
same virtual size; thus, larger objects appeared closer than
did the smaller objects. Such an explanation is consistent
with our observations of the displays and with traditional
studies of apparent size-distance relationships (e.g., Epstein
et al., 1961; Hastorf, 1950; Ittelson, 1951a, 1951b; Kilpatrick & Ittelson, 1951). In addition, DeLucia (1991a,
1991b) reported that when computer-generated floating objects approached a viewpoint, observers judged a large far
object as closer than a small near object that would have
arrived sooner.
In the present study, familiar size information about the
objects was not available. Pictorial relative size information
was present: At any given value of time-to-contact in the
different scenes, the smaller object projected a smaller image than the larger object.6 Accordingly, if both objects
were assumed to be equal in virtual size, the object that
projected the smaller visual angle would have appeared
farther, thereby potentially accounting for later jumps. Further analyses suggest, however, that such an explanation
cannot account for our collision-avoidance data: A large
approaching object that projects a relatively large visual
angle, but that is assumed to be the same virtual size as a
smaller object, should appear at a relatively smaller distance
from the viewpoint at every point in time. However, the
apparent change in the large object's distance (velocity) also
should appear smaller by the same proportion. Accordingly,
the larger object should appear relatively closer but slower,
and apparent time-to-contact should remain the same for
different-sized objects (assuming perceived time-to-contact

is based on perceived distance and perceived velocity). We


would thus expect the same launch times for all objects,
which is inconsistent with our data. We acknowledge that
this explanation assumes that apparent size and apparent
distance are coupled. Such an assumption may not always
be valid, as demonstrated by the moon illusion: Although
the zenith and horizon moons project the same visual angle,
the moon appears larger and closer, instead of farther (for a
summary, see Hershenson, 1989). In any case, our analyses
suggest that the apparent size-distance couplings considered here cannot adequately account for size-arrival effects.
Finally, we acknowledge that projected size, change in
projected size, and rate of expansion were all relatively less
for the smaller objects, and thus we cannot tease apart their
effects. We prefer to consider explanations of our data in
terms of projected relative size because it is more closely
tied to apparent distance, whereas changes in projected size
or rate of expansion are more closely tied to apparent
velocity (see also Bootsma & Oudejans, 1993).
Effective Information Varies Over
the Viewing Period
Thus far, it seems that tau and pictorial relative size
information cannot adequately explain size-arrival effects.
Our previous explanations were based on the assumption
that such information was used by observers over the entire
viewing period. It is possible, however, that the visual
information that determined performance varied over time.
For example, during the initial part of the trial, the smaller
object may have appeared farther due to pictorial size information. Then, as time progressed and time-to-contact
became smaller, tau may have become more influential;
participants may have realized that the small object was not
as far as originally perceived. This would be consistent with
the results of our optical analyses of Experiments 1-3: If
either tau or optical size alone was used by participants, then
launch times would have occurred when such information
reached the same value for large and small objects. Analyses indicated, however, that although the ratio of optical size
for large to small objects decreased dramatically from 8.0 at
the trial's start to 3.7 at launch time, it did not reach 1. In
addition, although the ratio of time-to-contact for large to
small objects began at 1, it increased to 2.1 at launch time;
even so, it was closer to a ratio of 1 than was the ratio for
optical size. Thus, it is plausible that both pictorial size and
tau contributed to performance but that the influence of each
varied over the course of the trial. Such an explanation
maintains both pictorial relative size and tau as viable
sources of information that contribute to size-arrival effects
(see also DeLucia, 199la, 1991b) and that vary in effectiveness throughout the task.
6
Apparent size-distance relationships also can occur when different-sized objects are presented successively (Epstein, 1961),
and when objects are dissimilar in shape (Epstein & Franklin,
1965; for a relevant study on time-to-arrival, see also Caird &
Hancock, 1992).

SIZE-ARRIVAL EFFECTS ON COLLISION AVOIDANCE

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Conclusion
The results of the present experiments extend DeLucia's
(1989, 1991a, 1991b) findings that size-arrival effects can
occur under a wide range of display and passive task conditions. We show that such effects can also occur with
active collision-avoidance tasks and when different-sized
objects are shown successively rather than simultaneously.
The previous results obtained with passive tasks were extended in the present study with an active task that could be
considered closer to ordinary perceptual conditions (e.g.,
see Gibson, 1979). Further study must be done before
size-arrival effects can be explained adequately. Neither
pictorial size (on the basis of an assumption of equal-sized
virtual objects) nor tau alone can explain our results if it is
assumed that such information was used continuously over
the course of the trial. Alternatively, observers may have
used pictorial size in the early part of the trial, until tau
reached a certain value and then increasingly influenced
performance.
Previous studies in which action, or judged arrival time,
was correlated with geometric time-to-contact information
such as tau, have been taken to indicate that observers gear
performance to such information. This argument is based
partly on the observation that such a strategy is relatively
simple and economical, that is, it is not necessary to perceive either distance or velocity in order to perceive timeto-contact (Lee, Young, Reddish, Lough, & Clayton, 1983;
see also Savelsbergh, Whiting, & Bootsma, 1991; Tresilian,
1991). However, it also has been suggested that, under
certain conditions, visual information other than tau may be
used to guide timing judgments or actions (Cavallo &
Laurent, 1988; DeLucia, 1991a; Schiff & Oldak, 1990;
Tresilian, 1991, 1994). The results of the present study are
consistent with this assertion: Pictorial relative size as well
as tau may contribute to the jumping tasks we studied. In
addition, our optical analyses suggest that observers may
have used elevation angle to adjust altitude prior to the
jump, because elevation angle may provide useable information about height. If relative size is indeed used in
collision-avoidance tasks, it becomes critical to measure the
limits and relative strengths of both pictorial and motionbased depth information when considering models of depth
and arrival-time perception (DeLucia, 1991a, 1991b, 1992).

References
Basler, A. (1906). Uber des Sehen von Bewegungen. I. Die
Wehrnehmung kleinster Bewegungen. Arch. ges. Physiol., 115,
582-601.
Bootsma, R. 1, & Oudejans, R. R. D. (1993). Visual information
about time-to-collision between two objects. Journal of Experimental Psychology: Human Perception and Performance, 19,
1041-1052.
Caird, J. K., & Hancock, P. A. (1992). Perception of oncoming
vehicle time-to-arrival. Proceedings of the Human Factors
Society 36th Annual Meeting, 2, 1378-1382.
Cavallo, V., & Laurent, M. (1988). Visual information and skill
level in time-to-collision estimation. Perception, 17, 623-632.
DeLucia, P. R. (1989). Even while approaching in depth, small

797

near objects may appear farther than large far objects. Supplement to Investigative Ophthalmology and Visual Science, 30,
250.
DeLucia, P. R. (1991a). Pictorial depth cues and motion-based
information for depth perception. Journal of Experimental Psychology: Human Perception and Performance, 17, 738-748.
DeLucia, P. R. (1991b). Small near objects can appear farther than
large far objects during object motion and self motion: Judgments of object-self and object-object collisions. In P. J. Beek,
R. J. Bootsma, & P. C. W. van Wieringen (Eds.), Studies in
perception and action: Posters presented at the Vlth international conference on event perception and action (pp. 94-100).
Amsterdam: Rodopi.
DeLucia, P. R. (1992). Even during active control of self-motion,
collision avoidance for small objects can occur later than for
large objects of equal distance and speed. Investigative Ophthalmology and Visual Science, 33, 1144.
Dosher, B. A., Sperling, G., & Wurst, S. (1986). Tradeoffs between stereopsis and proximity luminance covariance as determinants of perceived 3D structure. Vision Research, 26, 973990.
Epstein, W. (1961). The known-size-apparent-distance hypothesis. American Journal of Psychology, 74, 333-346.
Epstein, W., & Franklin, S. (1965). Some conditons of the effect of
relative size on perceived relative distance. American Journal of
Psychology, 78, 466-470.
Epstein, W., Park, J., & Casey, A. (1961). The current status of the
size-distance hypotheses. Psychological Bulletin, 58, 491-514.
Flach, J. M. (1990). Control with an eye for perception: Precursors
to an active psychophysics. Ecological Psychology, 2, 83-111.
Flach, J. M., Allen, B. L., Brickman, B.J., & Hutton, R. J. B.
(1992). Dynamic occlusion: Active versus passive observers.
Insight: The Visual Performance Technical Group Newsletter,
14, 5-7.
Gibson, J. J. (1950). The perception of the visual world. Boston:
Houghton Mifflin.
Gibson, J. J. (1962). Observations on active touch. Psychological
Review, 69, 477-491.
Gibson, J. J. (1979). The ecological approach to perception. Boston: Houghton Mifflin.
Graham, C. H. (Ed.). (1965). Vision and visual perception. New
York: Wiley.
Hastorf, A. H. (1950). The influence of suggestion on the relationship between stimulus size and perceived distance. Journal of
Psychology, 29, 195-217.
Hershenson, M. (1989). Moon illusion as anomaly. In M. Hershenson (Ed.), The moon illusion (pp. 123-146). Hillsdale, NJ:
Erlbaum.
Hills, B. L. (1975). Some studies of movement perception, age, and
accidents (Transport and Road Research Laboratory [TRRL]
Report SR-137). Crowthorne, Berkshire, England: TRRL.
Hochberg, J. (1982). How big is a stimulus? In J. Beck (Ed.),
Organization and representation in perception (pp. 191-216).
Hillsdale, NJ: Erlbaum.
Ittelson, W. H. (195 la). Size as a cue to distance: Static localization. American Journal of Psychology, 64, 54-67.
Ittelson, W. H. (1951b). Size as a cue to distance: Radial motion.
American Journal of Psychology, 64, 188-202.
Kilpatrick, F. P., & Ittelson, W. H. (1951). Three demonstrations
involving the visual perception of movement. Journal ofExpermental Psychology, 42, 394-402.
Laurent, M., Dinh-Phung, R., & Ripoll, H. (1989). What visual
information is used by riders in jumping? Human Movement
Science, 8, 481-501.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

798

PATRICIA R. DeLUCIA AND RIK WARREN

Lee, D. N. (1974). Visual information during locomotion. In R. B.


MacLeod & H. L. Pick, Jr. (Eds.), Perception: Essays in honor
of James J. Gibson (pp. 250-267). Ithaca, NY: Cornell University Press.
Lee, D. N. (1980a). The optic flow field: The foundation of vision.
Philosophical Transactions of the Royal Society of London,
B-290, 169-179.
Lee, D. N. (1980b). Visuo-motor coordination in space-time. In G.
Stelmach & J. Requin (Eds.), Tutorials in motor behavior (pp.
281-295). Amsterdam: North-Holland.
Lee, D. N., Lishman, J. R., & Thomson, J. A. (1982). Visual
regulation of gait in long jumping. Journal of Experimental
Psychology: Human Perception and Performance, 8, 448-459.
Lee, D. N., Young, D. S., Reddish, P. E., Lough, S., & Clayton,
T. M. H. (1983). Visual timing in hitting an accelerating ball.
Quarterly Journal of Experimental Psychology, 35 A, 333-346.
Regan, D., & Hamstra, S. (1991). Dissociation of time-to-contact
(T) and the rate of increase in angular size (dO/dt). In P. J. Beek,
R. J. Bootsma, & P. C. W. van Wieringen (Eds.), Studies in
perception and action: Posters presented at the Vlth international conference on event perception and action (pp. 89-93).
Amsterdam: Rodopi.
Savelsbergh, G. J. P., Whiting, H. T. A., & Bootsma, R. J. (1991).
Grasping tau. Journal of Experimental Psychology: Human Perception and Performance, 17, 315-322.
Schiff, W., & Detwiler, M. (1979). Information used in judging
impending collision. Perception, 8, 647-658.
Schiff, W., & Oldak, R. (1990). Accuracy of judging time to
arrival: Effects of modality, trajectory, and gender. Journal of
Experimental Psychology: Human Perception and Performance,
16, 303-316.
Sedgwick, H. A. (1983). Environment-centered representation of
spatial layout: Available visual information from texture and
perspective. In A. Rosenthal & J. Beck (Eds.), Human and
machine vision (pp. 425-458). San Diego, CA: Academic Press.

Todd, J. T. (1981). Visual information about moving objects.


Journal of Experimental Psychology: Human Perception and
Performance, 7, 795-810.
Tresilian, J. R. (1991). Empirical and theoretical issues in the
perception of time to contact. Journal of Experimental Psychology: Human Perception and Performance, 17, 865-876.
Tresilian, J. R. (1993). Four questions of time to contact: A critical
examination of research on interceptive timing. Perception, 22,
653-680.
Tresilian, J. R. (1994). Approximate information sources and perceptual variables in interceptive timing. Journal of Experimental
Psychology: Human Perception and Performance, 20, 154-173.
Wang, Y., & Frost, B. J. (1992). Time to collision is signalled by
neurons in the nucleus rotundus of pigeons. Nature, 356, 236238.
Warren, R. (1990). Preliminary questions for the study of egomotion. In R. Warren & A. H. Wertheim (Eds.), Perception and
control of self-motion (pp. 3-32). Hillsdale, NJ: Erlbaum.
Warren, R., & McMillan, G. R. (1984). Altitude control using
action-demanding interactive displays: Toward an active psychophysics. Proceedings of the 1984 IMAGE III Conference (pp.
37-51). Phoenix, AZ: Image Society, Inc.
Warren, R., & Owen, D. H. (1982). Functional optical invariants:
A new methodology for aviation research. Aviation, Space, and
Environmental Medicine, 53, 977-983.
Warren, W. H., Jr., Young, D. S., & Lee, D. N. (1986). Visual
control of step length during running over irregular terrain.
Journal of Experimental Psychology: Human Perception and
Performance, 12, 259-266.

Received April 9, 1992


Revision received January 12, 1994
Accepted January 12, 1994

S-ar putea să vă placă și