Marc Green, Ph. D.
Visual Expert Human Factors
Toronto, Ontario
Police officers sometimes shoot unarmed suspects in
error. The majority, about 71%1, occur at night under poor
visibility conditions. In one common scenario, a suspect suddenly
produces an object in his hand. The officer believes the object to be a
weapon and fires at the suspect. The officer then approaches only to see
that “gun” was actually something as innocuous as a driver’s license or
wallet.
The subsequent inquiry must judge the whether the
officer acted reasonably. One component of reasonableness is the
officer’s perception; should s/he have correctly recognized that the
object was not a gun? The visibility conditions, especially light
levels, usually become a central issue. Investigators often attempt to
assess visibility from witness statements such as “It was bright,” “I
could see three blocks down the street,” “There was enough light to
read my notebook,” etc.
In reality, these
statements reveal little about visibility conditions at the time of the
shooting. A realistic assessment requires an in-depth analysis of
specific situational factors that determined perception. Roughly
speaking, these factors can be divided into two categories, sensory and
cognitive. Sensory processing is the initial stage, where light enters
the eye and forms an image. When people speak of visibility conditions,
they are usually talking about light levels and other factors that
affect registration of the image information.
However, seeing involves much more than an image, and
visibility is not the only factor affecting the officer’s ability to
discriminate between a gun and an innocuous object. The quality of the
image and it contents provide only raw sensory information that the
viewer must next interpret. This requires use of memory and of other
innate cognition processes, most importantly expectation. Whether the
officer will see a wallet or a gun depends as much on his knowledge and
experiences as it does on the sensory factors.
Sensory Factors
Inquiries often
focus on light levels. They presume that if lighting level were
sufficient, then the officer should have correctly identified the
object. The mistake was then due to negligence or at least improper
care.
This logic is flawed because amount light is
only one sensory factor in determining visibility, and it is often a
relatively minor one. All things being equal, it is true that people see
better in brighter light. But things are seldom equal because many
situational factors are more important than overall light levels.
1.
People see contrast, not light. In a
real sense, people do not see light; instead, they see contrast, the
difference in light coming from
an object and it’s background. As the Illuminating Engineering Handbook
says, “Contrast detection is the basic task from which all other
visual behaviors are derived.” Contrast is the property that
allows people to detect shapes and contours and ultimately to recognize
objects. In most cases, contrast refers to brightness, but it can
sometimes refer to differences in color or texture.
Viewers see most objects in reflected light. A
source, such as a lamp, emits light that bounces off of a surface and
reaches the eye. To see contrast well, the there must usually be a
significant difference in reflectance between the object and background.
A piece of coal is highly visible against a white sheet because the coal
reflects very little light while the sheet reflects almost all of the
light. Further, the coal against a sheet will be highly visible even in
dim light. It is difficult to see in fog even in daylight because the
contrast is low. Contrast, not overall light level is often the key
factor in visibility.
The only important light is the light that reflects
off the object and it’s background because that is what determines
contrast. As a result, witness statements about general lighting
conditions are often irrelevant. A witness statement saying that he
could read his newspaper or see people walking a block away reveal
little about visibility of the object in the suspect’s hand. Moreover,
both object reflectance and background change with viewpoint, so a
witness standing at a different location may not be seeing what the
officer saw.
2.
The
location of the light is often more important than the amount of light.
Light location determines whether the contrast will be negative or
positive (Figure
2). Negative contrast (also called “silhouette contrast”) occurs then
there is a dark object on a bright background2. Newspapers,
for example, are printed in negative contrast. Positive contrast (also called “reverse
silhouette contrast”) occurs when there is a bright object on a dark
background.
The difference between the light location, and the
two types of contrast is often critical in determining object
recognition. Suppose the suspect is standing in front of a lighted
doorway, car headlight or a streetlight. Then the light is behind the
suspect, and the officer will see the object in negative contrast. The
object in the hand will appear only as a silhouette and its detail will
be invisible. Most importantly, objects of similar general shape will be
highly confusable. For example, a wallet and a 25 mm automatic pistol
have a similar square outline. An officer who sees a suspect’s hand
emerge from a pocket with an object having a square profile will have
difficulty determining whether it is a gun or a wallet. The straight
lines and the square edge are the most salient perceptions.
If the light is primarily in front of the suspect, it
will strike the object and reflect to the officer’s eye. S/he will have
a much better chance of seeing the fine details and discriminating one
object from another, i. e., a gun from a wallet. The more reflective the
object, the better the recognition. If the object in the hand is dark,
however, then positive contrast will be minimal regardless of light
position.
3. Lighting affects perception of outline shape
and detail differently. As light level decreases, visibility falls
much more rapidly for fine detail than for general shape. It can still
be possible to clearly discern and outline but difficult see the
specific pieces or components. For an officer who views the object in
the suspect’s hand, there might well be enough contrast to see the
rectangular shape of a gun/wallet, but there may not be enough contrast
to see the fine detail that discriminates a gun from a wallet.
4. Perception of general shape is faster.
People mentally process the general shape of an object faster than the
fine details. Everyone has had the experience of “recognizing” a
familiar person and then starting to wave or say hello, but then
realizing that it was actually a stranger with the same general build,
hair, etc. This occurs because the visual system transmits information
about larger areas faster from the eye to the brain. Roughly speaking,
viewers make initial recognition after seeing the general outline and
before the details can catch up.
An officer who must make a potentially life-and-death
decision reacts quickly. He is likely to make a decision based on the
visual information that is first perceived, which will usually be the
general outline and shape. If the light is in front of the suspect then
the officer will still see the general shape first may be forced to make
a rapid decision before the objects details become visible. If the light
is behind the suspect, of course, then the general shape is the only
information available to decide whether it is a gun or is a wallet.
5.
Motion impairs perception. If an object
is moving as you click the shutter on a camera, the resulting image may
blur. If the suspect is moving his hand, then the object he is holding
will produce a similar effect in the eye. The image is degraded,
lowering visibility and impairing recognition. Movement most degrades
fine details, the very information needed to distinguish a gun from a
wallet.
6.
Light adaptation impairs perception. The
eye sees contrast best when it is adapted to the same light level as the
current scene. For example, a person walking into a movie theater during
the day will at first see little more than blackness. After a few
moments, the theater becomes brighter and objects become visible. This
occurs because the eye adjusts sensitivity to the prevailing light level
through the process of light adaptation. Visibility loss is greater
when the difference between adapting level and viewing level is high and
when there has been little time to adapt.
An officer’s contrast perception will be reduced if
s/he has spent time looking at a brighter area prior to the time when
the object must be recognized. Even time spent viewing a streetlight or
writing on a sheet of white paper (which is highly reflective) could
reduce raise adaptation level and impair visibility of subsequently
viewed objects.
While these six factors are important in most
situations, there are others that may be important in some specific
conditions. One is glare from a light source that far brighter than the
ambient level. As any driver knows, for example, a passing car’s
headlights can lowered contrast perception and may also destroy dark
adaptation and lower visibility for several minutes. Another is
“masking,” which occurs when a patterned background lies behind the
object. The viewer will have more difficulty seeing the object clearly
amid the clutter. Finally, there are a wide variety of other possible
factors, weather, age, viewer eyesight and spectacles, etc., that may
also be relevant in some situations.
Cognitive Factors
Perception is the process of applying stored
knowledge to the sensory input and forming an interpretation. Once the
image is sensed, the officer must decide whether it represents a gun or
some other object based on what he knows and what he has learned through
experience.
This is a problem-solving task that humans address by
using innate cognitive strategies. In psychology they are called
“heuristics,” general rules of thumb, or “biases.” The term “bias” is
not meant in the negative sense of everyday language. In cognitive
psychology, a “bias” can be a reasoning strategy that aids in quickly
arriving at a correct answer. In critical situations, it is dangerous to
engage in the slow process of conscious reasoning. Instead, the human
mind has evolved heuristics and biases to speed the interpretation and
to make sense of a complex situation.
Here are some of the major cognitive heuristics and
biases. It is important to remember that these are innate and “normal,”
in the sense that virtually all humans rely on them heavily. Further,
they operate unconsciously, outside of normal awareness.
1.
Experience creates expectation People
become skilled at their jobs through learning and development of
expectation. Beginners act is a “controlled” mode, where they must stop
and consciously think about every decision. Their performance is slow
and inefficient and usually errorful. With learning, people switch to a
fast and “automatic” mode where responses seem to occur without
conscious thought or decision. Part of the shift occurs because the
skilled person learns what to expect, and can act without the need for
conscious decision-making in order to recognize objects.
While it is often said, “see is believing,” it is
equally true that “believing is seeing.” Officers who have previously
been in situations where a suspect was armed, for example, have more
reason to believe that future suspects will be armed and are far more
likely to expect to see a gun. However, the experience need not be
first-hand. If the station house talk dwells in instances of suspects
with weapons, officers shot by suspects, etc., this can also sway
expectation in favor of interpreting indistinct objects as guns.
2. The
“availability heuristic” primes perception.
People must use information stored in
memory to make decisions. In a critical situation, they will most likely
use the first information that comes to mind. One factor that causes
quick recall is that the information was recently acquired. In general,
the most recently learned information is the most available in memory
and is likely to have the strongest influence on decision-making and on
interpretation of low visibility objects. There are also a large number
of other cognitive factors that can affect memory3.
For an officer going out on patrol, the most
available information may originate in the pre-patrol briefing. For
example, the officers may be reminded to keep an eye out for a dangerous
suspect of a specific description or that there have been reports of
violent crimes an specific area. When the officer later must make a
split-second decision, s/he draws on memory in order interpret the
ambiguous visual information. S/he is especially likely to perceive a
gun if the suspect fits the general description of the dangerous suspect
or is located is the area where violent crimes occur.
3.
“Confirmation Bias” strengthens preconceived beliefs.
Confirmation bias refers to the tendency
to seek information that supports and already held belief and to avoid
or dismiss evidence that contradicts it. It is perhaps the most powerful
cognitive bias, and has been found to be a source of error and accident
across a wide spectrum of human activities.
If an officer believes that a person is likely to be
a suspect (he matches a description, is located in a high crime area,
acting strangely, etc.) s/he will likely form an opinion that the person
is potentially dangerous. Once this belief takes hold, the officer will
look for evidence to support the belief and ignore, avoid, or
rationalize away evidence that contradicts it. When the person appears
with an object in his hand, it will be interpreted as a gun. This
conclusion confirms the initial belief that the suspect is dangerous.
4. Payoffs affect decision-making. A person
who must decide under uncertain conditions,
such as low visibility, faces the quandary that is depicted
schematically in the Table I. S/he must decide whether a “signal” is
present or whether there is only “noise.” (The terms “signal” and
“noise” arose from the “Signal Detection Theory,” a mathematical
analysis originally used as a model of radar operators – how did they
decide whether there was an enemy plane on the screen or merely
background noise.) For an officer who sees the object in the viewer’s
hand, the “signal” is a gun and the noise is an innocuous object.
The situation can
produce one of four possible outcomes. The officer viewer will be
correct if he says “yes” when the signal is there and it really is
(“hit”) or says “no” and it really isn’t (“correct rejection.”) There
are also two kinds of errors. The officer can say “yes” when there is no
gun (“false alarm”) or “no” when there is a gun (“miss.”)
The officer must consider the payoff consequences of
each response. Of course, the officer will attempt to be correct, but,
since the situation is uncertain, s/he must also consider the
consequences of an error. If s/he makes a correct “yes response” he may
save his own life while a correct “correct rejection” leaves the
situation as status quo. There are two types of error possible. If s/he
makes a false alarm (says gun when there is no gun), he shoots an
unarmed and possibly innocent suspect, may face legal action, etc. If he
makes a miss (does not shoot when there is a gun), s/he, a fellow
officer or bystander may end up wounded or even dead.
The officers have no way to assign objective values
to the outcomes of the “yes” and “no” responses. Is it worse to shoot an
unarmed suspect than to have a fellow officer shot? Is there any worse
outcome than being shot and killed? If expectation, availability and
confirmation effects are operating, then the sum of the mental
processing is likely to weigh toward seeing the object as a gun.
Conclusion
Hindsight, as they
say, has 20-20 vision. After the outcome of an event is already known,
it is difficult to put oneself in the place of the people who must
decide on action before the event. The officer confronting a suspect at
night has a difficult decision because he does not have the luxury that
subsequent investigators have of knowing with certainty whether or not
the object is a gun. Instead, s/he must act based on uncertain visual
information. To resolve the uncertainty, the officer will draw on the
innate cognitive processes that all humans have evolved to help us make
quick decision under stress. These heuristics and biases function
automatically because there is no time for conscious decision making.
There is only time to react.
The reasonableness of action cannot be assessed by
generalities about lighting conditions or post hoc analysis based on
what is known now, after then event, rather than what was then, before
the event. Proper analysis requires a detailed reconstruction of the
visual factors described above and an assessment of the cognitive
process operating at the time. The task of reconstructing the past can
be difficult and investigators are prone to their own set of cognitive
biases, as I describe elsewhere4. Ultimately, the
reasonableness of any behavior can only be understood by the physical
constraints of the situation and by the likely operation normal human
perception and cognition.
When a police
shooting occur, inquiries frequently draw on expert in police
procedures, etc. However, a scientific and human factors analysis of the
situation may be more revealing, more objective and more informative.
When a police officer misperceives an object as a gun, it is a
perceptual error, not a police error.
References
1Aveni, T. (2003).
Police Marksmanship Under Fire: Paradox And Promise, Law & Order,
August.
2Green, M. (2002). “The Invisible
Pedestrian,” Occupational Health & Safety Canada, pp44-50,
October, 2002.
Green, M. (2004). Eyewitness memory is unreliable.
http://www.visualexpert.com/Resources/eyewitnessmemory.html
4 Green, M. (2003). Skewed View: Accident
Investigation, ”Occupational Health & Safety Canada, June.
©2004 The Police Policy Studies Council. All rights reserved.
|