Content uploaded by Anthony Estey
Author content
All content in this area was uploaded by Anthony Estey on Jun 27, 2015
Content may be subject to copyright.
Catalyst: Seeing Through the Eyes of a Cat
Jeremy Long Anthony Estey David Bartle
University of Victoria
Sven Olsen Amy Gooch
Abstract
We have implemented a simulation of the cat visual system that is
based on the research of neuroscientists [Shuurmans and Zrenner
1981; Mouat 1985; Govardovskii et al. 2000]. The purpose of our
simulation is to illustrate four of the major differences between the
cat and human visual systems, and to map those contrasts into a
space that can be readily observed by humans. Three of the four
effects are achieved using fragment shaders, and can be applied
as post processes to a game engine or to a live video feed in real
time, provided that enough information about the scene is avail-
able. We also developed Catalyst, an educational game that teaches
the player about how their vision differs from that of a cat. The
tasks included in Catalyst require the player to switch between the
two visual systems, thus emphasizing the contrasts between them.
1 Simulating the Cat Visual System
1.1 Color Transformation
Color vision is the ability to distinguish objects based on the wave-
lengths of light that they reflect. Cones are the primary receptors
associated with color vision, although rod receptors can play a role
under low light viewing conditions. The model of cat vision that
we have implemented is intended only to simulate cone-dominated
color vision, as the interaction between cones and rods is complex.
We base our model of cat color vision on research done by neuro-
scientists that indicates that the cat visual system includes two cone
classes with peaks at 450 nm and 556 nm [Mouat 1985; Shuurmans
and Zrenner 1981]. This model of vision implies that cats would
only be able to see combinations of two color primaries. Some re-
searchers have suggested the presence of a third cone class within
the cat visual system [Guenther and Zrenner 1993], but they be-
lieve this cone class would only be present in a tiny number of reti-
nal ganglion cells, limiting its ability to mediate color vision. We
adopted the two cone model of cat vision, as it is well suited to il-
lustrate the trouble cats would have in distinguishing certain colors.
Some researchers have suggested that the pigment in cone photore-
ceptors has relatively invariant properties across different species,
allowing the curve of a cone receptor to be defined simply by spec-
ifying its peak absorption point. Several different formulations of
the universal pigment template have been proposed that take differ-
ent data sets and visual systems into consideration [Lamb 1995].
We used the alpha-bands proposed by Govardovskii et al. [2000]
to create response curves for the cat cones. Figure 1 shows how
the human cone response curves compare to the cat cone response
curves. Based on these curves, we were able to derive a 3x3 matrix
transform that takes in a signal recorded by the human cones and
translates it into coordinates within the cat visual system, which we
then display within the space that can be observed by the human vi-
sual system. We built a fragment shader that can perform this color
transformation in real time.
Figure 1: Response curves for cone photoreceptors in the human
visual system (left) and the cat visual system (right).
1.2 Luminance Transformation
Cats have superior night vision to humans. This is partially due
to an abundance of rod sensors, which dominate the visual system
under scotopic conditions. Cats also have a tapetum lucidum, which
is a reflective layer that reflects light that passes through the retina
back into the eye. The characteristics described above give the cat
a minimum light detection threshold up to seven times lower than
that of humans. This enhances vision under low light conditions,
but it appears to reduce vision in cases where light is abundant.
We have modeled the relationship between the human visual system
and the cat visual system’s ability to register light using a logarith-
mic function. Figure 2 shows that the slope of the resulting curve
approximates the cat’s ability to function at low light levels, while
also simulating the manner in which the responses from their light
receptors plateau when light is more abundant. This function was
built into a fragment shader that can be run in real time.
Figure 2: The cat luminance transformation function.
1.3 Distance Blur Transformation
Visual acuity describes the sharpness of the images that are pro-
duced by the visual system. It is estimated that the visual acuity of
cats is between 4 and 10 times worse than humans. We approxi-
mated this difference using a depth blur shader that applies an in-
creasingly powerful blur as distance from the viewpoint increases.
1.4 Field of View Transformation
Cats are thought to have a wider field of view than humans. This
gives cats more expansive peripheral vision. We simulate this char-
acteristic by changing the projection matrix used by the virtual cam-
era from the standard 45 degree viewing angle to a 70 degree field
of view when rendering the scene from a cat’s perspective.
2 Catalyst
Our purpose in creating a simulation of the cat visual system is to
educate people about the differences between their visual systems
and the cat visual system. To this end, we decided to implement
an educational game aimed at younger players that uses our sim-
ulation to convey some of the important characteristics of the cat
visual system. Running our simulation on a virtual scene also sim-
plifies the process of providing the information for the depth and
luminance transformations.
Catalyst is a first person puzzle game where the user must switch
between controlling a human and a cat in order to make their way
past a series of obstacles. When controlling the cat, the player sees
the world through our cat visual system simulation. The puzzles
that the player must overcome are based on color discrimination,
low light vision, and field of view, and are specifically designed
to require numerous switches between the two visual modes, thus
emphasizing the differences between them. The video submitted
with this project shows an example playthrough of Catalyst.
References
GOVARDOV SKII , V. I., FY HRQU IST, N ., RE UTER , T. O. M., K UZMIN, D . G., AND DONNER, K. 2000. In search of
the visual pigment template. Visual Neuroscience 17, 4, 509–528.
GUENTHER, E., AN D ZRENNER , E. 1993. The spectral sensitivityof dark- and light-adapted cat retinal ganglion cells.
Journal of Neuroscience 13, 1543–1550.
LAMB, T. 1995. Photoreceptor spectral sensitivities: Common shape in the long-wavelength region. Vision Research35,
22, 3083–3091.
MOUAT, G. S. V. 1985. Photoreceptor inputs to cat lateral geniculate nucleus cells. Experimental Brain Research59, 2,
242–248.
SHUURMANS, R. P., AND ZR ENN ER, E. 1981. Responses of the blue sensitive cone system from the visual cortex and
the arterially perfused eye in cat and monkey. Vision Research21, 1611–1615.
