Perception & Action at a Distance
Brian Gaudino and Brian Prue
Skidmore College, 2006

Perception and Action at a Distance
By
Brian Michael Gaudino
Brian Christopher Prue
In partial fulfillment of the degree
Bachelor of Arts
Skidmore College
Psychology Department
Advisor
______________________
Flip Phillips
Department of Psychology & Neuroscience Program

Table of Contents
Page # Contents
1 1 – Exploring ‘Unnatural’ Modes of Perception and Action
6 1.1 – The Theory of Affordances
15 1.2 – Current Work
18 2 – Methods
29 3 – Results
34 4 – General Discussion

Perception and Action at a Distance
Brian Michael Gaudino
Brian Christopher Prue
Skidmore College
Teleoperation of robots and autonomous vehicles introduces an interesting
series of questions with respect to perception and action at a distance. While
the pragmatics of this problem has been considered in the human factors
domain, there is little consideration of an overall theory of perception and
action at a distance in the perceptual domain. Our work attempts to erect a
scaffolding for the development of such a theory. Classically, studies of
perception and action take place in the self, self, i.e., those where the
embodiment of the perceiver and actor are the same entity. Our work
considers the external, other and self, other perspectives (e.g., watching a
machine carrying out our action and watching _from_ the machine carrying
out the action). The framework is complicated by the fact that external, other
and self, other embodiments may have different action capabilities than the
self, self, and self, other embodiments may have additional sensor
mechanisms able to provide information not available in the usual self, self
sense. Our overall strategy consists of external, other and self, other
replication of classic self, self perception-action paradigms and investigation
the resulting shifts (or lack thereof) in performance. Obviously some types of
performance will have little or no difference when differently-embodied while
others should experience significant modification. From these results, we can
model and predict expected performance in alternative perception-action
embodiments. Here, we present results from affordance-based experiments
modeled on Warren & Wang (1987) as well as Kinsella-Shaw, Shaw & Turvey
(2005), along with their relevant implications for our proposed theoretical
framework.

1 Exploring ‘Unnatural’ Modes of Perception and Action
Consider the following scenario: you are a member of an urban search and rescue
team, and an office building has just collapsed. There is a great risk for further collapse,
and much of the rubble is too compacted for a human to navigate. Your only hope for
finding survivors is to employ a search and rescue robot. Now imagine that you are
piloting this robot through the rubble. You must negotiate turns, judge the passability of
gaps and crossable terrain and recognize potential structural hazards all in a completely
foreign environment.
Navigating through a new and complex environment can prove to be a daunting
spatial task even when one is physically in this new environment and has the ability to
interact with it through one’s own eyes and body. Evidently, the complexities of the task
are immensely compounded when one is forced to navigate this environment with
nothing but a camera and remote control. A search and rescue robot shares very few
physical similarities with its operator. It is smaller in every dimension, has a different
means of movement, and “sees” the world from a different perspective. In short, the
operator must navigate a new environment from the perspective of another. The operator
can perceive and act in the collapsed building, but is not physically there. They are
perceiving and acting from a perspective that is not their own.
This search and rescue situation is just one of many scenarios involving
perception and action in an ‘unnatural’ setting; one in which the perceiver and actor are
separate beings, or when the primary perceiver and actor are somehow distanced from the
self. The search and rescue operator can take an entirely new perspective by standing
Gaudino, BM, Prue BC Perception and Action at a Distance 1

directly over the robot and controlling it from this overhead, or God’s eye view. This
mode of operation incorporates the idea of the primary perceiver being disconnected from
the physical dimensions of the actor. Another scenario is as simple as riding as a
passenger in a car. In this case, the passenger is seeing the world directly through their
own eyes, yet are passive in the sense that they are not actively controlling the car.
All of these scenarios can be classified with a 3 x 3 table outlining the various
modes of perception and action (table 1).
Table 1: Modes of perception and action
The two axes represent how a particular entity is both controlled and how it is perceived.
When the locus of control is the self you are acting with your own body. If it is other,
something other than the self is acting, and if it is none, there is no direct control over the
action. When the perception of controlled entity is the self, you perceive through the eyes
of the entity in control. If it is mediated, the perspective of the perceiver is separate from
the entity in control. The mediated case is split into embedded and external. Embedded
is when the perceiver is attached to the entity in such a way that the perceiver can see the
Gaudino, BM, Prue BC Perception and Action at a Distance 2

dimensions of the entity. External refers to situations in which the perceiver is
completely disconnected from the controlled entity.
Several examples should illuminate the ways in which the modes of perception
and action converge. The self, self scenario is the natural setting in which you are the
entity in control and you are perceiving the world around you (figure 1). In the self, none
scenario you are also perceiving the world through your own eyes, but are inactive. An
example of embedded self would be a simple reaching task. In this instance you are the
entity in control, but the entity in control (your hand) is perceived through your eyes,
which can directly perceive the dimensions of the hand. Straightforward examples of
every other scenario are presented in figure one.
The following research is based on situations in which the locus of control is
other, in particular external, other and self, other (Figures 2 & 3, respectively). The
search and rescue scenario in which the operator is perceiving and acting through the
camera of the robot is an example of self, other. The controlled entity is separate from
the self (the robot), but is perceived through the “eyes” of the robot. The controlled
entity is the perceiver in this case, hence the self label. The situation in which the search
and rescue operator stands directly over the robot and controls it from a God’s eye view is
an external, other situation. Once again, the controlled entity is something other than the
self, but this time, the controlled entity is not the perceiver. The search and rescue
operator is using his own eyes to perceive the robot.
Gaudino, BM, Prue BC Perception and Action at a Distance 3

Figure 1: self, self perspective Figure 2: external, other perspective
Figure 3: self, other perspective
Surprisingly, nearly all of the perception and action research is limited to the self,
self scenario. Every other mode of perception and action lacks a coherent theoretical
structure. As previously stated the current research is interested only in the self, other
(robot’s eye view) and external, other (God’s eye view) scenarios. These areas are of
particular interest because modern technology has forced many professionals into
unnatural modes of perception and action. Such instances include the previously
mentioned use of search and rescue robots as well as endoscopic, laparoscopic and
arthroscopic surgery procedures and the operation of unmanned aerial vehicles. The
dearth of quantitative data in this field could certainly be inhibiting to the designers and
Gaudino, BM, Prue BC Perception and Action at a Distance 4

operators of these technologies. It is not known whether or not there are systematic
differences between the self, self mode and the two modes of interest; the self, other, and
external, other scenarios. In order to mitigate this problem, a theory of self, other and
external, other perception and action (perception and action at a distance) must be
established.
The best way to do this is to examine two factors: the quantitative data that has
lead to an established theory of the self, self mode and our basic knowledge of our
capabilities in the two modes of interest. First, let us address what we already know
about the self, other scenario. Clearly, there is not a major methodical difference between
this and the self, self scenario. If this were the case, it would be extremely difficult to
control a search and rescue robot, or play a first person shooter video game. Hence, there
is most likely no difference in the most basic of perceptual tasks (depth and color
perception, spatial frequency). If differences are going to be found, they will exist in
more complex perceptual tasks; ones involving interactions between the environment and
the perceiver. In other words, the realm of perception and action must be examined. It is
these more complex perceptual tasks that often require or partially depend on
proprioceptive feedback or self-knowledge of ones own body. When controlling a robot
with only a remote control and an image provided by a camera, these factors are either
absent, incomplete or distorted.
The second component of establishing a theory for these “new” modes of
perception and action is to examine the well-established paradigms of the self, self mode
of perception and action. Take for instance a theory as broad as Tolman’s cognitive maps
Gaudino, BM, Prue BC Perception and Action at a Distance 5

(1948). Based on this theory, much research has been conducted on how we store spatial
knowledge, and the navigational techniques we use. If a particular study concluded that
we depend on landmark-based navigation to find shortcuts in the self, self mode, the
methods of the study can then be duplicated using a one of the new modes of interest. A
small chunk of the theory can then be established regardless of whether the results are the
same or different to those of the self, self scenario. If they are the same, then we know
that we navigate shortcuts in the same manner regardless of perspective. If they are
different, then we know that we navigate shortcuts from the new mode of interest in a
fundamentally different way than we do in the self, self mode.
In order to determine the most efficient means of creating this theory, the most
meaningful self, self perception and action paradigms must first be examined.
1.1 The Theory of Affordances
Gibson’s theory of affordances is the cornerstone of the ecological theory of
perception and action. An affordance refers to an opportunity offered by the environment
to an organism with certain biomechanical limits (Chemero, 2003; Kinsella-Shaw, Shaw
& Turvey, 1992; Gibson, 1986; Proffitt, Stephanucci, Banton & Epstein, 2003;
Soffrengen, Yang & Bardy, 2005; Wagman & Taylor, 2005; Warren & Whang, 1987;
Witt, Proffitt & Epstein, 2005). Hence, an affordance is dependent on both the qualities
of an environment or object being acted upon as well as the biomechanical limitations of
the actor (Chemero, 2003; Kinsella-Shaw et al., 1992; Gibson, 1986; Proffitt et al., 2003;
Soffrengen et al., 2005; Wagman & Taylor, 2005; Warren & Whang, 1987; Witt et al.,
2005).
Gaudino, BM, Prue BC Perception and Action at a Distance 6

Take for instance, a sphere. What type of actions can this object afford to a
human being? In order to answer that question, one must describe the properties
possessed by the sphere as well as the capabilities of that person. If the sphere has a
diameter smaller than the length of the actor’s hand, it may very well afford
“graspability”. However, it may possess a slippery surface or an extremely high density
that would prevent the affordance of “graspability”. All of the properties of an object or
environment must be considered when determining what it affords to a certain organism
(Chemero, 2003; Gibson, 1986). If only diameter were considered, one might
fallaciously assume that any sphere the size of a basketball affords them “graspability”
with one hand. However, texture plays a major role in “graspability” when the object is
as large as a basketball. The rubbery texture of a basketball is extremely conducive to
“graspability”, whereas a smooth metallic surface most likely is not.
The biomechanics of the actor must be considered as well (Chemero, 2003;
Kinsella-Shaw et al., 1992; Gibson, 1986; Proffitt et al., 2003; Soffrengen et al., 2005;
Wagman & Taylor, 2005; Warren, 1984; Warren & Whang, 1987; Witt et al., 2005). The
size of a professional basketball player’s hand certainly permits “graspability” of a
basketball, yet the hand of a child is much to small take hold of the ball in an identical
manner. Clearly, affordances are dependent on two factors (an organism and an
environment), which in turn, are dependent on each other.
When considering the affordance of an object or environment, one must keep in
mind that affordances are in a sense limitless. A basketball does not only afford
“graspability”, it can be acted on in myriad ways. It has the potential to be thrown,
Gaudino, BM, Prue BC Perception and Action at a Distance 7

caught, bounced, sat on, and a whole host of other actions, performed by a whole host of
organisms. In essence, an object as simple as a basketball can afford a limitless number
of actions to the world of potential actors (Chemero, 2003; Gibson, 1986)
When considering one particular affordance that an object or environment
provides for one particular actor, limitations do exist. For example, when holding all the
properties of a sphere constant, apart from diameter, a point will exist when the actor can
no longer grasp the sphere with one hand. This transition state where an object no longer
affords a particular action is referred to as a critical point (Warren, 1984; Warren &
Whang, 1987). There can also be a point when a certain object or environment is ideal
for carrying out a particular action. This is referred to as an optimal point (Warren, 1984;
Warren & Whang, 1987). A sphere the size of a baseball is certainly more optimal for
throwing than a sphere the size of a basketball.
In order for organisms to interact with their respective environments, they must be
able to perceive the affordances offered to them. That is, the perception and action
system requires that the actions one makes are dependent on the perception of the
affordances offered by the environment (Chemero, 2003; Kinsella-Shaw et al., 1992;
Gibson, 1986; Proffitt et al., 2003; Soffrengen et al., 2005; Wagman & Taylor, 2005;
Warren, 1984; Warren & Whang, 1987; Witt et al., 2005). We must certainly be adept at
doing this, for survival depends on it. How then, do organisms perceive what an object
or environment affords them? Is this perception the same for all affordances? Is it
dependent on self-knowledge, and if so, do we store this self-knowledge in a geometric
manner or are we continually adapting to our environment?
Gaudino, BM, Prue BC Perception and Action at a Distance 8

Many studies have examined these questions, and in doing so have given insight
into the self, self mode of perception and action systems. The results from these studies
can be extrapolated and applied to our developing self, other and external, other theories
in order to provide a quantitative means of conjecture for the results of this study.
It is widely agreed by Gibsonians that affordances are perceived in body-scaled
terms (Chemero, 2003; Kinsella-Shaw et al., 1992; Gibson, 1986; Proffitt et al., 2003;
Soffrengen et al., 2005; Wagman & Taylor, 2005; Warren, 1984; Warren & Whang, 1987;
Witt et al., 2005). The basic idea is that our ability to perceive an affordance depends on
the physiological constraints of the body part or parts involved in carrying out an action.
A way to demonstrate this is to create a biomechanical model for the critical point ratio of
the body part/s involved in the action with the dimensional limitations of the object being
acted on. Experimental data can then be compared to this to see if the affordance-based
action is perceived in body-scaled terms. In Warren’s (1984) study of stair climbability, a
critical point of .89 was determined for the ratio of lower leg length to riser height. A
biomechanical model predicted a ratio of .88 (Warren, 1984). This indicates that the
affordance of climbability is, in fact, dependent on our knowledge of the biological
constraints of our leg (Warren, 1984). An intrinsic knowledge of our own body aides in
perceiving the actions that objects in the environment afford.
Many other studies have examined the relationship between the biomechanics of
the body and the action-based affordances offered by the environment. Warren and
Whang examined the affordance of passability through an aperture (1987). The
dimensions of the aperture and the widest dimension of the body (shoulder width) are the
Gaudino, BM, Prue BC Perception and Action at a Distance 9

limiting factors that determine whether or not an aperture is passable. The critical point
for the transition from frontal walking to body rotation can be expressed by a ratio of
these two limitations (Warren & Whang, 1987). Warren and Whang referred to this as the
aperture to shoulder width ratio (A/S) (Warren & Whang, 1987). Participants making
judgments based only on static monocular information had a mean A/S of 1.16, whereas
participants making dynamic judgments had a statistically different mean A/S of 1.3
(Warren & Whang, 1987). The difference in aperture to shoulder width ratios between the
two conditions can be attributed the perceptual distortion from body sway (Warren &
Whang, 1987). That is, the body sway from walking makes one believe that they are
wider than when they are standing still (Warren & Whang, 1987). It is interesting to note
that the mental model of the actor, even on the static condition, does not match the
biomechanical model of shoulder width. This is a good example of an affordance in
which the mental model does not correspond to the biomechanical model.
Warren and Whang also determined that the participant’s eye height is crucial to
their self-knowledge of their own body size (1987). When eye height manipulations were
made, participants were deceived into believing that they were smaller than they actually
are resulting in a skewed A/S ratio (Warren & Whang, 1987).
Affordances are perceived based on body-scaled terms, but a network of our
bodily senses represents information about this self-knowledge (Chemero, 2003;
Kinsella-Shaw et al., 1992; Gibson, 1986; Proffitt et al., 2003; Soffrengen et al., 2005;
Wagman & Taylor, 2005; Warren, 1984; Warren & Whang, 1987; Witt et al., 2005). Our
bodies are constantly integrating all sorts of information presented by the environment
Gaudino, BM, Prue BC Perception and Action at a Distance 10

with our own self-knowledge in order to allow us to perceive and interact with the world.
Exactly how much of this information is needed to perceive the affordances of the
environment? The answer depends on the environment or object, and what you intend to
do with it. Hence, there is no particular body-sense that universally perceives every
affordance accurately. For instance, in their experiment on perceived walk-on-ability of
slopes, Kinsella-Shaw et al. found that participants were able to accurately perceive the
affordance in a binocular vision condition or a haptic condition (1992). This demonstrates
two bodily senses that can perceive an affordance with high precision and independently
from each other (Kinsella-Shaw et al., 1992).
Other experiments have examined how locomotion or even simple body
movements can help us perceive affordances (Oudejans, Michaels, Bakker, & Dolne,
1996; Stoffrengen et al., 2005). It is well known that movement helps with basic visual
perception. For instance, by moving ones head the slightest bit, occluded objects can be
identified and information about depth can be interpreted. One study examined the
importance of locomotion in catching a fly ball (Oudejans et al., 1996). Professional
baseball players could not accurately perceive the catchability of a fly ball when they
relied only on nonlocomotor body movement (body sway) (Oudejans et al., 1996).
However, in another study, participants could judge the maximum sitting height of a
surface while standing with normal shoes or while standing with ten cm blocks on their
feet (Stoffrengen et al., 2005). This demonstrates that one bodily sense or ability
(nonlocomotor body movement) is sufficient for perceiving one affordance (sitability) but
not another (catchability). This is of course due to the dynamics of the affordances.
Gaudino, BM, Prue BC Perception and Action at a Distance 11

Judging the catchability of a fly ball requires one to perceive something that cannot be
conveyed without locomotion.
Thus far the discussion of affordances has been limited to the interaction between
an actor and objects in the environment. Larger scale affordances must be considered as
well. Distance and space can both be perceived and acted upon all the time. In fact, the
ground beneath our feet is the foundation for most actions. In essence, the environment
as a whole does present us with some large-scale opportunities that can be interpreted as
affordances. Distance, for example has two perceptual factors: slant and extent (Proffitt
et al., 2003). It has been demonstrated, that we perceive the affordance of distance as
being longer when there is a considerable grade to ascend or descend or when under
physiological duress compared to gently graded slopes and when we are in good physical
standing (Proffitt et al., 2003).
There are certain times when an actor must perceive and act in an environment
with either limited or skewed self-knowledge. We often interact with the environment
with objects that extend the dimensions of our body (Higuchi, Takada, Matsuura &
Imanaka, 2004; Wagman et al., 2005; Witt et al., 2005). This is referred to as the person-
plus-object system. Examples include carry luggage or being confined to a wheelchair
(Higuchi et al., 2004; Wagman et al., 2005). The person-plus-object system can create
difficulties in accurately perceiving affordances. When confined to a wheelchair several
problems emerge. One’s eye height is now considerably lower, they no longer possess
body sway during locomotion, and their dimensions are slightly wider (Higuchi et al.,
2004). It should be no surprise that when Warren and Whang’s aperture study was
Gaudino, BM, Prue BC Perception and Action at a Distance 12

repeated with novice wheelchair users, the mean A/S ratio was .92 (Higuchi et al., 2004).
This indicates that participants believed that they could fit through certain apertures that
were actually impassable. Clearly, this is resultant of the limitations of being in a
wheelchair. In the original study the A/S ratio was greater than 1.0 partly because
participants took into consideration their own body sway (Warren & Whang, 1987).
Also, eye height was found to be crucial in judging one’s own shoulder width (Warren &
Whang, 1987). As stated above, a wheelchair prevents body sway, lowers eye height and
slightly widens the body beyond the shoulders. However difficult these factors become,
with practice we are able to adapt to the person-plus-object system. Expert wheelchair
users have an A/S ratio greater than 1.0 (Higuchi et al., 2004).
Another study examining the person-plus-object system for aperature passability
manipulated the information available to the participants while carrying a T-shaped object
that was wider than their shoulders (Wagman et al., 2005). In all instances, the aperture
to object (A/O) ratio was lower than 1.0, indicating a slight decrease in the accuracy of
perceiving the affordance (Wagman et al., 2005). However, it was found that participants
could perceive the affordance to a similar accuracy (still below 1.0) in haptic, visual and
dynamic touch conditions (Wagman et al., 2005). This indicates that although we are
slightly worse at perceiving the affordance of passability in the person-plus-object
system, we can still do it with limited information.
Instances exist that allow an altered self-knowledge to be advantageous to an
actor once they have adapted to the new dimensions. For instance, Witt et al., found that
when an object is just out of arms reach, but can be reached with the aid of a stick it
Gaudino, BM, Prue BC Perception and Action at a Distance 13

appears closer when the tool was used (2005). This also indicates a like between
affordances, effort, and how self-knowledge is stored (Witt et al., 2005).
The way in which we store self-knowledge can occur in one of two ways. Either
we store fixed, quantitative information about our action capabilities (geometric) or we
are adapting to our environment though exploratory movements (Chemero, 2003; Gibson,
1986; Proffitt et al., 2003; Soffrengen et al., 2005; Witt et al., 2005). Geometric storage
is similar to the path integration theory of navigation. Both ideas rest heavily on our
mind’s ability to constantly calculate complex algorithms to inform us about our body’s
ability to interact with the environment (Foo, P., Warren, W. H., Duchon, A. & Tarr, M. J.,
2005). Although some of this might exist, it appears that our self-knowledge is never as
mathematically accurate as a geometric model would predict (Chemero, 2003; Gibson,
1986; Proffitt et al., 2003; Soffrengen et al., 2005; Witt et al., 2005). This actually turns
out to be advantageous to the way we perceive and act.
The stick-reaching study demonstrates this point. In this case, we do not perceive
ourselves in geometric terms. Instead, our perception is slightly skewed in a manner to
assist in the action we intend to carry out (Witt et al., 2005). Without a stick, one
perceives an object to be further away than it actually is because the intended action
cannot be carried out (Witt et al., 2005). However, when the stick is wielded, the object
appears closer because the action of graspability can be obtained (Witt et al., 2005). This
indicates that a nongeometric model for affordances is in fact beneficial because it creates
a wider discernablilty between what actions are attainable and what actions are not.
Gaudino, BM, Prue BC Perception and Action at a Distance 14

The use of exploratory movements to build self-knowledge is made evident by the
fact that movement assists in the perception of affordances. As stated earlier, participants
were able to accurately judge the sitability of varying ledges even when ten cm blocks
were attached to their feet (Stroffregen et al., 2005). Participants were able to adapt to
their taller environment though exploratory movements. If we possessed a purely static
geometric self-knowledge, then it would prove extremely difficult, if not impossible, to
accurately perceive affordances when our body dimensions are skewed. In fact, it would
prove to be even more rigorous when we are perceiving and acting through a medium
completely different than ourselves. That is, perception and action through the self, other
mode would be impossible; which of course, is not the case.
1.2 Current Work
How can this previous research from self, self mode of perception and action be
applied to the undeveloped self, other and external, other theories. For one, we can look
at the results from the person-plus-object experiments. In the person-plus-object system,
we must create a new self-knowledge to adapt to our changed body dimensions. In a
sense, we are relying on a new self-knowledge to allow us to accurately perceive and act
in the world. This can be viewed as middle grounds between perceiving affordances with
our normal body dimensions and perceiving affordances from the self, other mode in
particular.
As the person-plus-object experiments have demonstrated, it takes sometime to
first adapt to one’s new dimensions, and even when this occurs, the adaptation is not
Gaudino, BM, Prue BC Perception and Action at a Distance 15

always as accurate. This can be attributed the lack of certain available senses as well as
varying factors to be considered in order to perceive an affordance. The wheelchair
aperture study demonstrates this well. Novice wheelchair riders had an A/S ratio of .92,
whereas expert wheelchair riders had an A/S ratio just over 1.0 (Higuchi et al., 2004).
Neither of these values was very close to the A/S ratio of walkers (1.3) (Warren &
Whang, 1987). This can be accredited to the fact that locomotion on feet is very different
from locomotion on wheels (Higuchi et al., 2004). Although the environment is the
same, different factors contribute to how it is perceived because the dimensions and
mechanics of the actor are different.
Now consider the self, other scenario. Here, the actor (robot) is completely
different from a human actor. Its dimensions are completely different, as are its means of
locomotion. Furthermore, we are dealing with two perceivers and two actors. The one
making the choices is not physically present in the environment being acted upon.
Finally, the human actor is limited to the sense of sight, and even this is perceived in an
unnatural manner.
Based on the dearth of perceptual information provided to the actor, it is amazing
that we have capabilities for the self, other mode of perception and action. Clearly, we
have this capability because of our ability to adapt to dynamic environments and dynamic
bodily dimensions and mechanics. If differences in affordance perception can be found
when one’s self-knowledge and biomechanics are slightly tweaked (as in the person-plus-
object systems), then differences may be found for the vastly distorted worlds of self,
other and external, other perception and action.
Gaudino, BM, Prue BC Perception and Action at a Distance 16

Because of the wealth of research and the relative ease of implementation, our
first two experiments were external, other and self, other replications of Warren and
Whang’s aperture studies. Experiment one will examine self, other aperture affordances
in a virtual environment using the Unreal Tournament engine. In order to teach
participants the shoulder width dimensions of the virtual character, a forced learning
phase must be completed first. With this participants will have a mental model of the
virtual character’s width, which can then be tested. A static aperture passability phase
will determine aperture to shoulder width ratios for a self, other mode and will also
determine how behavior in the learning phase corresponds to the mental model. This can
then be compared to Warren and Whang’s results to determine if fundamental differences
exist in aperture affordances from the self, self mode and the self, other mode.
Experiment two will also be examining the affordance of aperture passability. In
this case, participants will use the external, other and self, other modes in a telerobotics
environment to judge aperture passability. Just as in experiment one a learning phase will
be needed to assist participants in creating a mental model of the robot, which will then
be tested and compared to the behavior data and aperture to shoulder width data from
experiment one and the self, self experiments.
Finally, a new affordance paradigm will be introduced based on the walk-on-
ability of slopes study of Kinsella-Shaw, Shaw and Turvey. Participants will judge the
maximum climability of textured and untextured slopes from the external, other and self,
self modes using the robot from experiment two. Range of accuracy in judgments can be
Gaudino, BM, Prue BC Perception and Action at a Distance 17

compared to the self, self data in order to see if a systematic difference exists between
scenarios.
2 Methods
Experiment 1.A
Participants
Participants (six men, four women) were Skidmore College introduction to
psychology students, and acquaintances of the researchers. All participants signed
informed consents in accordance with the ethical standards of the American
Psychological Association.
Gaudino, BM, Prue BC Perception and Action at a Distance 18

Apparatus
This experiment was conducted at the Eye, Brain and Vision Laboratory at
Skidmore College. A virtual world was created using UnrealEd version 3.0. The
program was run on a Macintosh G5 presented by a rear projection DLP screen. The
image was 160 cm x 142 cm. A Microsoft SideWinder Precision 2 joystick used “stats”
and “OLSstats” mutators to assist in recording data.
Virtual Environments
Three different virtual environments were created with the UnrealEd v.3.0 and
WOTgreal v.3.005. The first environment was a practice maze produced solely for the
purpose of acquainting the participants with the tasks controls. The maze consisted of
five rooms with dimensions of 256 x 512 x 512 unreal units connected clockwise by
apertures of 256 x 256 x 256, 256 x 256 x 256, 256 x 256 x 128, 256 x 256 x 64, 256 x
256 x 128 unreal units. A sixth room off to the side of the third room in the loop was
connected by two separate apertures of 256 x 32 x 256, 256 x 64 x 256 unreal units.
The second environment was a learning test designed to teach the participants to
learn the width of the unreal character. It consisted of 20 pairs of 512 x 512 x 512 unreal
unit rooms connected by varying aperture breaths (128, 56, 46, 44, 96, 64, 32, 42, 38, 30,
112, 80, 48, 50, 52, 54, 36, 28, 72, 88 unreal units). The height of the apertures was 512
unreal units and the width was 256 unreal units. The unreal character began the trial
centered in the back of one room (starting room), directly facing the 128 unreal unit-
breath aperture. Blue cylindrical health packs were located directly in front of the
aperture, directly to the character’s left in the corner of the room, and in the center of the
Gaudino, BM, Prue BC Perception and Action at a Distance 19

room connected by the aperture (back room). These health packs marked locations of
teleporters, and recorded the time the participant used a teleporter. The teleporters would
send the participant to a random room-pair with the possibility of repeating the same
room-pair.
The third environment was a test phase identical to the second environment with
several exceptions. A green “yes” box was located directly to the right of each aperture
in each room, and a red “no” box was located directly to the left of each aperture in each
room. The teleporters were located in the back left and right corners of the starting room,
directly to the characters left and right. “Yes” or “No” was selected by sliding the
character in the corresponding direction (left/right) and the teleporter would send the
participant to a randomly selected room, possibly the same. “Flak pack” pickups were
used, much like the blue health packs in the learning test environment, to record “yes/no”
decisions and time of decision.
Tasks
The purpose of the practice task was for the participant to learn the controls of the
joystick and program. They began in one of the six rooms and navigated around the
maze until they felt comfortable with the maze.
The purpose of the learning task was for the participant to learn the width of the
unreal character. They began in a starting room with an aperture of random width. The
unreal character could fit through half of the apertures. Participants had two options,
both of which result in being teleported to a new room. They could either attempt to go
directly through the aperture to the back room, where a teleporter was located, or make a
Gaudino, BM, Prue BC Perception and Action at a Distance 20

sharp left turn to the top left corner of the starting room where another teleporter was
located. Once they crossed over a teleporter, they were transported to a new, random
starting room.
A feedback system was set up so participants could better learn their widths. The
three blue cylinders acted as health packs. Each participant began with 100 health points.
Every time a wrong decision was made participants gained health, and every time a
correct decision was made, participants lost health. Once their overall health reached
zero, the experiment ended.
Every time participants attempted to pass through an aperture they were forced to
cross over a health pack that added two health points to their overall health. This health
pack re-spawned every six seconds, so if a participant took longer than six seconds to
pass through an aperture, they were penalized with an additional two health points. If an
aperture was passable, the health pack in the back room subtracted three health points,
resulting in a net health loss of one. If an aperture was unpassable, participants were
forced to retreat to the health pack in the back left of the starting room. This health pack
subtracted one health point. Hence, if a participant attempted to pass through an
unpassable aperture, they would pick up two health points in front of the aperture and be
forced to collect the minus one health pack in the back corner, resulting in a net gain of
one health. However, if the participant knew that the aperture was unpassable, and went
directly to the minus one health pack in the left corner, they would end up with a net loss
of one health point for making a correct decision.
Gaudino, BM, Prue BC Perception and Action at a Distance 21

For every unpassable aperture, the health pack in the back left corner of the
starting room subtracted one health. Conversely, every passable aperture had a health
pack that added one health point to the participant’s overall health. This penalized
participants for incorrectly assuming that they cannot pass through a passable aperture.
The purpose of the testing task was for participants to demonstrate their
knowledge of the character’s width. Participants simply made judgments of whether or
not they could fit through the aperture that was presented to them on the screen. The
unreal character began in the exact same starting position as the learning task, yet was
only permitted to make lateral movements. Hence, participants were forced to make
passability judgments at a fixed length (512 unreal units) from the aperture. Once the
participant made a judgment, the experimenter moved the character to the appropriate
teleporter (“yes” or “no”) and the character was teleported to a random room. This was
repeated over 100 trials.
Figure 4: An aperture used in the learning and test phase of experiments 1.A and 1.B.
Gaudino, BM, Prue BC Perception and Action at a Distance 22

Procedure
Participants were seated 132 cm from the DLP screen with their eyes aligned with
the center of the screen. Their eye height was approximately 68 cm from the bottom of
the screen and 74 cm from the top of the screen. Participants were given the option of
placing a wooden board on their lap as a means of stabilizing the joystick.
Experimenters explained the controls of the joystick to the participants and the
purpose of the practice environment. Participants began by navigating the practice
environment and were encouraged to pass through as many apertures as possible to get a
feel for the joystick and program. When the participant felt comfortable with the
controls, the task was aborted.
In the learning phase of the experiment, participants were first briefed on the task
at hand. They were informed that they should attempt to go through as many apertures as
possible until they believe that they know what they can and cannot fit through. Once
they felt confident in their width they were informed to only attempt to pass through the
apertures that they knew they could fit through. When the participants’ health reached
zero, the learning phase was terminated.
In the test phase, participants were informed that they are to make “yes” or “no”
judgments of aperture passability in the same environment as the learning phase.
Experiment 1.B
Experiment 1.B was identical to experiment 1.A, with the exception of the
displays used. A Macintosh G5, 20-inch monitor was used in place of the DLP projector.
Participants sat one meter from the screen.
Gaudino, BM, Prue BC Perception and Action at a Distance 23

Experiment 2
Participants
Participants (seven men, three women) were Skidmore College introduction to
psychology students, and acquaintances of the researchers. All participants signed
informed consents in accordance with the ethical standards of the American
Psychological Association.
Materials and Apparatus
This experiment was conducted in the psychology and neuroscience laboratory at
Skidmore College. A Vex robot (27 cm x 33 cm) was constructed with a rotating camera
(21.5 cm height) mounted to the front of the robot (figure 5). The camera had a viewing
angle of 42 – 45 ° at 84 cm. The signal from the camera was picked up by a 2.4 GHz
Swann receiver and projected on a 14 inch Panasonic television.
Ten adjustable apertures were created by mounting 26 cm x 38 cm multicolored
plastic boards to 20 bricks. Fifteen different set widths were cut from three cm wide
wooden sticks and were bound together from longest to shortest. These widths were
exactly proportional to the 15 smallest aperture widths of the unreal environments. The
sticks were numbered from three to 18, with three being the smallest. Stick number ten
was the same width as the robot.
Gaudino, BM, Prue BC Perception and Action at a Distance 24

Figure 5: Robot used in experiments two and three
Procedure
Before each participant arrived, the stimulus was set up in the psychology and
neuroscience laboratory at Skidmore College. A rectangular track was set up around the
perimeter of the room, with ten designated aperture positions (figure 6). The television
screen was set up in a corner of the room in such a way so that when looking at the screen
participants could not see the robot or the track. The width of each aperture corresponded
to the widths of sticks five through 15, excluding width ten. A random number generator
was used to arrange the aperture widths in a random order for each trial.
Upon arrival, participants were seated at the television screen and instructed not
to turn around until the experiment was over. They were briefed on the purpose of the
Gaudino, BM, Prue BC Perception and Action at a Distance 25

experiment and taught the controls of the robot. The robot was kept in a hidden location,
preventing the participants from seeing its size.
In order for participants to get acquainted with the controls of the robot, a practice
phase was completed. The robot was placed in the hallway, out of view of the
experimental setting. Participants were taught the controls and allowed to navigate the
robot through the hallway through the self, other mode of the television screen. Half of
the participants were taught how to move the camera as well as the robot. The other half
only learned how to move the robot. The camera had the ability to rotate left, right, up
and down. Those that were taught to rotate the camera were permitted to do so
throughout the entire experiment. Those that were taught only to navigate the robot were
not permitted to rotate the camera at any time in the experiment.
When participants felt comfortable with the controls, the camera was turned off,
and the robot was placed at the beginning of the rectangular track to prepare for the
learning phase. Participants were instructed to navigate the robot through the course. It
was explained that some apertures were passable, while others were not. They were
instructed to pass though the apertures they believed were passable, and to simply
navigate around the impassable apertures. Once again, they were reminded to only view
the television screen while navigating the robot.
As participants navigated the track, they were timed and their passability
judgments were recorded. Upon completion of the first trial, participants were told to
remain looking at the television screen. The experimenter then handed them the 15
wooden sticks of varying lengths. Participants estimated the width of the robot by
Gaudino, BM, Prue BC Perception and Action at a Distance 26

choosing which stick corresponded to their mental model. They were then asked to give
a confidence rating on a five point Likert scale. The experiment was repeated two more
times for each participant. The ten apertures were randomly rearranged after each trial.
After the learning phase, participants were tested on their mental models of the
robot’s width. A method of limits task was repeated six times for each participant. The
robot was set 2.5 meters from the first aperture and one of the experimenters adjusted the
aperture together and apart until the participant believed it matched the appropriate width
of the robot. At this point participants still had never seen the robot and were making
width judgments through the television screen.
After participants made their six judgments, their external, other mode of
perception and action was tested. The robot was set at the start of the course and
participants navigated the robots through the course as they followed behind. As usual,
participants were timed and instructed to attempt to pass through the apertures they felt
were passable.
Figure 6: experiment two room layout for learning course
Gaudino, BM, Prue BC Perception and Action at a Distance 27

Experiment Three
Participants
Participants (five men, one woman) were Skidmore College undergraduate
students. All participants signed informed consents in accordance with the ethical
standards of the American Psychological Association.
Materials and Apparatus
This experiment was conducted in the psychology and neuroscience laboratory at
Skidmore College. A foam core ramp 51 cm x 76 cm was covered on one side with
carpeted texture. The other, untextured side consisted of the blank white foam core with
10 cm x 5 cm black boxes drawn on it. The ramp was set up at the end of a hallway and
attached to a single pulley connected to the ceiling. The rope from the pulley extended
into the Eye Brain and Vision lab and allowed for adjusting the angle of the ramp without
directly viewing the ramp. The same television set up from experiment two was arranged
inside the Eye Brain and Vision Lab. This television was connected to the camera on the
robot used in experiment two.
Procedure
Participants sat approximately one meter from the television screen and the robot
was placed facing the ramp, centered 1.5 meters away. The textured side of the ramp was
presented first. Participants had seen the robot and were familiar with its dimensions and
operation. The ramp started in a prone position and a researcher in the lab lifted the ramp
until the participant believed that it was at the maximum angle that the robot could
traverse. Participants could only view the ramp through the television screen. In each
Gaudino, BM, Prue BC Perception and Action at a Distance 28

block, participants made six judgments of traversability, three where the ramp started
from a prone position and three where it started from a position perpendicular to the
ground.
After block one, the experimenter took the average of the six judgments and set
the slope to that angle. The participant was then allowed to drive the robot onto the ramp
to see if their estimate was accurate or not. This was repeated for blocks two and three.
Next participants were run using the untextured surface of the ramp in the self,
other mode. The procedure was identical to the textured condition.
Finally, participants’ judgments were tested in the external, other using only the
untextured side of the ramp. This time participants stood approximately four meters from
the ramp, and the robot remained 1.5 meters behind it. Judgments were recorded and
blocks were run in an identical manner as the self, other mode.
Gaudino, BM, Prue BC Perception and Action at a Distance 29

Figure 7: Robot in textured ramp experiment
3 Results
Experiments 1.A and 1.B
In the learning task for the small screen and big screen conditions, participants
learned the width of the robot to 99.9% after a mean of 151 trials and a 37 trial SD.
However, participants were at 90% learning after a mean of 27 trials.
In the test phase the aperture to shoulder width ratio vs. percent no judgments for
the big screen and large screen conditions exhibited nearly the same psychometric
function as the Warren and Whang self, self study (1987) (figure 8). The aperture to
shoulder width ratio at 50% no for the large and small screen was 1.15, which is no
different that that of Warren and Whang’s self, self static aperture to shoulder width ratio
of 1.16 (p > .05) (1987) (figure 8).
Gaudino, BM, Prue BC Perception and Action at a Distance 30

Figure 8: experiment one A/S ratio vs. percent no judgments
Experiment 2
In the learning task, participants learned to a mean of 77% after thirty trials in the
robot’s eye view condition. Based on the aperture to shoulder width ratio vs. percent no
judgments, participants were considerably more accurate in their passability judgments in
the God’s eye view condition than the robot’s eye view condition (p < .01) (figure 9).
Based on learning data alone, participants had a mean aperture to should width ratio of
1.0 at 50% no judgments.
In the test phase for the robot’s eye view condition, participants’ 50% no mean
aperture to shoulder width ratio was 1.13 with the stick judgments, or direct view, and
1.16 for the method of limits, or robot’s eye view (figure 10). There was no significant
difference between these two aperture to shoulder width ratios and that static aperture to
shoulder width ratio of Warren and Whang’s self, self study (p > .05) (1987).
Gaudino, BM, Prue BC Perception and Action at a Distance 31

Figure 9: God’s eye view and Robot’s eye view A/S ratio vs. percent no judgments
Figure 10: Robot’s eye view stick judgment and method of limits A/S ratios
Experiment 3
In the carpet textured condition, participants’ estimations ranged by 18.5° (figure
11). All but two participants underestimated the maximum traversabilty of the slope; the
largest underestimation being 17.5° (figure 8). One participant was perfect in the first
Gaudino, BM, Prue BC Perception and Action at a Distance 32

estimation and the other overestimated by 1° (figures 11 & 13). By the second block of
trials, participants’ estimations converged to a range of 12.5°, with all but two
participants underestimating the maximum traversability (figures 11 & 13). In the third
trial participants had a 9° range in estimations with four underestimations (figures 11 &
13).
In the no texture condition, participants’ trial one estimations ranged by 16°, with
all but one participant making underestimations (figures 12 & 13). In trial two
participants’ estimations ranged by 4.5°, with all but two participants making an
overestimation (figures 12 & 13). In trial three participants’ estimations ranged by 4°,
with all but one participant making underestimations (figures 12 & 13).
In the external other no texture and textured conditions participants’ estimations
ranged by 5°.
There was no significant difference between the self, other and external, other
conditions in the texture and no texture conditions (p > .05).
Gaudino, BM, Prue BC Perception and Action at a Distance 33

Figure 11: Difference from threshold for carpet texture condition
Figure 12: Difference from threshold for no texture condition
Figure 13: Difference from threshold for texture and no texture conditions
Gaudino, BM, Prue BC Perception and Action at a Distance 34

4 General Discussion
Within each experiment, the relationship between behavior, mental model and the
veridical world was of particular interest. Because of the participants detached viewing
and acting perspectives, particularly in the self, other mode, it proved challenging for all
of these factors to coincide. This is of no surprise because it is rare for mental model and
reality to match up in a direct one-to-one ratio in the self, self perception and action
studies. Warren and Whang’s aperture to shoulder width ratio of 1.16 is a perfect
example of this (1987). Participants’ mental models of their shoulder widths were larger
than the actual dimensions of their own bodies, or the reality of their width (Warren &
Whang, 1987). Self, self perception and action studies do not generally examine the
behavior of participants because these particular studies are examining the relationship
Gaudino, BM, Prue BC Perception and Action at a Distance 35

between biomechanics and perception of one’s own body in a particular environment.
Because the external, other and self, other scenarios are unnatural, the affordances offered
by these modes must be learned. Hence, the behavior from these learning conditions can
be compared to mental models and reality to examine the relationship between the three
factors in the external, other and self, other modes of perception and action.
In experiment one, we found that there was no difference between the visual cues
provided by the large, lifelike screen and the small screen; indicating that external field of
view is not an issue. The large screen encompasses participants’ field of view and is
more successful at embodying the participant in the task, thus making it more true to the
more artificial view on the small screen. This difference is of negligible importance in
the perception of aperture affordances in the self, other mode. The results of the static
condition in Warren and Whang overlap almost perfectly with those observed in
experiment one. Warren and Whang calculated an aperture to shoulder width ratio of
1.16, which is statistically the equivalent to the aperture to shoulder width ratio of 1.15 in
our self, other virtual environment (1987). This also indicates that regardless of the
disconnected view of the operator, participants acquired the exact same trend in mental
model to real world as participants in the self, self studies.
The learning condition of experiment one is of particular interest. After a mean of
151 trials, participants were 99.9% accurate at judging the passability of the each
aperture. That is, their behavior matched the veridical world. However, in the testing
phase participants overestimated their widths in the same way as Warren and Whang’s
participants (1987). This indicates that participants’ behavior matches the veridical
Gaudino, BM, Prue BC Perception and Action at a Distance 36

world, yet their mental model does not. Perhaps the static nature of the test phase could
account for this anomaly. In the learning phase, participants were allowed to control the
unreal character, and often made exploratory movements before judging the passability of
the aperture at hand. However, in the test phase the experimenter had full control over
the character and only made lateral movements after participants’ judgments were made.
Hence, participants were forced to make judgments from a static position. Perhaps if
participants were permitted to control the unreal character and make the exploratory
movements they used in the learning task, their mental model might match up to their
behavior and the veridical world.
Experiment two resulted in similar findings to experiment one in the self, other
condition with regard to the aperture to shoulder width ratio. Interestingly enough, the
behavior of participants was extremely consistent with their mental models. The stick
measurement was a heavy behavioral based representation of the mental model, whereas
the method of limits was a measure of participants’ mental models of the self, other
perspective. This also indicates that the mental model acquired by participants through
the medium of a television screen (method of limits) translates to their mental model in
the real world (stick estimates). The aperture to shoulder width ratios calculated for the
stick-based model measurement (1.13) did not differ significantly from the mental model
based representation found in the method of limits test (1.16).
The learning data in experiment two evokes several interesting implications. In
this experiment, participants were unable to learn as well as participants in experiment
one because of the time constraints involved in setting up and implementing the task. In
Gaudino, BM, Prue BC Perception and Action at a Distance 37

this experiment participants were only trained to 77% accuracy in judgments of apertures;
whereas participants were trained to 99.9% in experiment one. Despite this major
difference, participants still acquired the same mental model as experiment one. This
could be resultant of the difficulty of the learning task. Participants often knew that they
could make it through a particular aperture, but the mechanical constraints of controlling
the robot prevented this. As a result, participants were often forced to go around a gate,
when their actions dictated that they most likely could have made it through. In essence,
the mechanical constraints forced participants to make decision errors.
Although the aperture to shoulder width ratio was consistent across the two
experiments and Warren and Whang’s self, self experiment, there still might be
differences in aperture affordances in the self, self and self, other modes. This
experiment only examined static aperture judgments, whereas Warren and Whang
examined both static and dynamic (1987). When participants were allowed to walk
towards the aperture and make a passability judgment, Warren and Whang found that the
aperture to shoulder width ratio increases significantly to 1.3 (1987). This increase was
attributed to the increase in perceived width due to body sway (Warren & Whang, 1987).
Experiment one and two used robots that experienced no body sway. Hence in a self,
other dynamic condition, it would be expected that the aperture to shoulder width ratio
would be the same than the static condition. The behavioral data from experiment one
might indicate that aperture to shoulder width ratios actually improve in a dynamic
setting. As mentioned earlier, participants were trained to 99.9% accuracy in learning
phase passability judgments where they were permitted to make exploratory movements.
Gaudino, BM, Prue BC Perception and Action at a Distance 38

This high degree of accuracy might indicate that behavior, mental model and the veridical
world all match with one another in a dynamic setting.
Another important implication of the self, other aperture affordances is the fact
that “normal” body dimensions and abilities are not a problem. Warren and Whang found
that the aperture to shoulder width ratio depends on the eye height of participants (1987).
In the unreal condition, the dimensions of the characters eye height and shoulder width
were in an unnatural one-to-one ratio. In other words, the character’s dimensions
mirrored that of a square, whereas a normal human is heavily favored in the dimensions
of “tall” rectangle. Furthermore, the camera angles in both experiments were
approximately 45°, which is drastically narrower than the natural human 180° field of
view.
Because the aperture to shoulder width ratio was consistent between experiment
one and two, the virtual simulation used in experiment one appears to be a viable
alternative to the more costly and time involved approach taken in experiment two.
Although experiment two is more real to the realm of search and rescue, computer
simulation programs, such as unreal tournament, can also be used for future research into
the self, other scenario. This certainly provides insight into ways of training operators for
self, other tasks such as certain and rescue operations and endoscopic surgery.
The external, other data from experiment two demonstrated a difference in the
way participants approached the task. It appears that participants focused less on an
internalized mental model of that robot because they had a direct view of the
characteristics pertaining to the task. As a result, they were significantly better at judging
Gaudino, BM, Prue BC Perception and Action at a Distance 39

passability of apertures compared to the self, self and self, other conditions. The external,
other mode was drastically different from the self, self and self, other modes, which
themselves were significantly similar across experiment one and two. As such, it can be
hypothesized that participants used a different mechanism to judge the affordance of
passability in the external, other mode compared to the self, self and self, other modes.
Additionally, due to the similarities of the self, self and self, other approach leads one to
believe that the interalization of the sensory information is managed in a similar way in
the self, self and self, other modes.
Experiment three addressed a different paradigm, the traversability of slopes with
and without texture gradient information. In the no textured self, other condition,
participants decreased variance and overestimated the maximum traversable slope from
block one to block two. Block three demonstrated a consistent shift towards a more
conservative estimation where all participants underestimated the maximum traversable
slope. However all participants were within five degrees of threshold, matching Shaw et
al.’s results (1992). The external, other no textured conditions did not differ significantly
from the self, other no textured conditions.
The textured condition demonstrated a slightly larger range from threshold of nine
degrees. However, if two of the participants’ data is omitted, the range decreases to the
five degree range. One of these two participants followed a completely different pattern
than the rest of the participants. In trial one, this participant estimated the maximum
traversability perfectly, and in trial two an overestimation of two degrees was made.
Every other participant had made an underestimation before trial three. Because it is
Gaudino, BM, Prue BC Perception and Action at a Distance 40

impossible to tell the degree in which one has overestimated the slope, this participant
might have believed a drastic overestimation was made in trial two and thus drastically
underestimated trial three. If an underestimation had been made in trial one or two this
participant would have been able to make an estimate based on converging
overestimations and underestimations. The other participant that contributed to a
widening range drastically underestimated the traversability in trial one and two; hence,
although improvements were made, it was difficult for this participant to approach zero
degrees.
The external, other results for the textured condition demonstrated a mean range
within five degrees of threshold, indicating that this mode of perception and action is no
different than the self, other conditions.
Based on the results from all three experiments and the self, self experiments of
interest, it appears that there is consistency between the way people perceive slope and
aperture affordances in the self, self and self, other scenarios. Perhaps this can be
attributed to the fact that in both instances participants are “embodied” in the actor. This
is obvious in the self, self mode, but can also be true for the self, other. Although a
search and rescue operator is not actually inside the robot, they are able to take on a
particular perspective as if they were looking through the eyes of the robot. In a sense,
this is a form of embodiment. The external, other mode is one in which the participants
were detached from the robot completely. Instead of embodying the robot, the operator
takes a “God’s eye view” of the robot. This elevated perspective can assist greatly in
certain tasks such as the passability of apertures because the operator has a direct view of
Gaudino, BM, Prue BC Perception and Action at a Distance 41

the dimensions of the robot. Other tasks, like the traversability of slopes, are not
mitigated by the external, other mode because the dimensions of the robot do not directly
assist the operator. That is, the angle of the treads does not assist the operator in
determining traversability because the robot has the ability to climb slopes of a steeper
angle than the treads. Also, texture becomes a major factor in determining this
affordance. Different textures afford different degrees of traversability. Hence, simply
seeing the dimensions of the robot will not assist in climbing a slope because the angle of
traversability is a dynamic quantity. In the case of passability, the external, other mode
assists greatly because this affordance directly relates to the dimension of width and
remains completely static.
In order to determine the implications of a dynamic test phase in the self, other
aperture passability tasks, this experiment should be repeated. As stated earlier, it is
hypothesized that participants will perform significantly better than participants in the
self, self experiments because of the lack of body sway. Furthermore, other affordance-
based tasks could be tested in the self, other and external, other modes to continue to
build these theories of perception and action.
References
Chemero, A. (2003). An outline of the theory of affordances. Ecological Psychology,
15(2), 181-195.
Foo, P., Warren, W. H., Duchon, A. & Tarr, M. J. (2005). Do humans integrate routes
into a cognitive map? Map-versus landmark-based navigation of novel shortcuts.
Journal of Experimental Psychology: Learning. Memory, and Cognition, 31(2),
195-215.
Gibson, J. (1986). The Theory of Affordances. In R.E. Shaw, W. M. Mace & M. E.
Turvey (Eds.), The Ecological Approach to Visual Perception (pp. 127 – 143).
Hillsdale, New Jersey: Lawrence Erlbaum Associates, Inc.
Higuchi, T., Takada, H., Matsuura, Y., & Imanaka, K. (2004). Visual estimation of
Gaudino, BM, Prue BC Perception and Action at a Distance 42

spatial requirements for locomotion in novice wheelchair users. Journal of
Experimental Psychology: Applied, 10, 55-66.
Janzen, G., Schade, M., Katz, S & Herrmann, T. (2001). Strategies for detour finding in
a virtual maze: the role of the visual perspective. Journal of Environmental
Psychology, 21, 149-163.
Kinsella-Shaw, J. M., Shaw, B. & Turvey, M. T. (1992). Perceiving “walk-on-able”
slopes. Ecological Psychology, 4(4), 223-239.
Ouddejans, R. R. D., Michaels, C. F., Bakker, F. C., & Dolne, M. (1996) The relevance
of action in perceiving affordances: Perception of catchableness of fly balls.
Journal of Experimental Psychology: Human Perception and Performance, 22,
879-891.
Proffitt, D. R., Stefanucci, J. Banton, T. & Epstein, W. (2003). The role of effort in
perceiving distance. Psychological Science, 14(2), 106-112.
Stoffrengen, T. A., Yang, C. & Bardy, B. G. (2005). Affordance judgments and
nonlocomotor body movement. Ecological Psychology, 17(2), 75-104.
Tolman, E. C. (1948). Cognitive maps in rats and men. The Psychological Review,55(4),
189-208.
Turvey, M. T., Fitzpatrick, P. (1993). Commentary: Development of perception-action
systems and general principles of pattern formation. Child Development, 64,
1175-1190.
Wagman, J. B. & Taylor, K. R. (2005). Perceiving affordances for aperature crossing for
the person-plus-object system. Ecological Psychology, 17(2), 105-130.Warren, W.
H. (1984). Perceiving affordances: Visual guidance of stair climbing.
Journal of Experimental Psychology: Human Perception and Performance, 10(5),
683-703.
Warren, W. H. & Whang, S. (1987). Visual guidance of walking through aperatures:
body-scaled information for affordances. Journal of Experimental Psychology:
Human Perception and Performance, 13(3), 371-383.
Witt, J. K., Proffitt, D. R. & Epstein, W. (2005). Tool use affects perceived distance, but
only when you intend to use it. Journal of Experimental Psychology: Human
Perception and Performance, 31(5), 880-888.
Gaudino, BM, Prue BC Perception and Action at a Distance 43