ArticlePDF Available

Abstract

Cats were reared in a light-tight box in which the only source of illumination was a 9-musec strobe flash every 2 sec. This allowed them to experience visual form but they did not experience visual movement. Receptive-field properties of single units in area 17 of the visual cortex of cats reared in stroboscopic illumination (strobe-reared) were compared with properties of units in area 17 of normally reared cats. In strobe-reared cats both direction selectivity and orientation selectivity were greatly reduced relative to normally reared cats, and some units in the strobe-reared cats responded only to strobe flashes.
Proc.
Nat.
Acad.
Sci.
USA
Vol.
70,
No.
5,
pp.
1353-1354,
May
1973
Cats
Reared
in
Stroboscopic
Illumination:
Effects
on
Receptive
Fields
in
Visual
Cortex
(developmental
physiology/deprivation
effects/vision)
M.
CYNADER*,
N.
BERMANt,
AND
A.
HEINJ
Department
of
Psychology,
Massachusetts
Institute
of
Technology,
Cambridge,
Mass.
02139
Communicated
by
Eliot
Stellar,
March
2,
1973
ABSTRACT
Cats
were
reared
in
a
light-tight
box
in
which
the
only
source
of
illumination
was
a
9-psec
strobe
flash
every
2
sec.
This
allowed
them
to
experience
visual
form
but
they
did
not
experience
visual
movement.
Receptive-field
properties
of
single
units
in
area
17
of
the
visual
cortex
of
cats
reared
in
stroboscopic
illumination
(strobe-reared)
were
compared
with
properties
of
units
in
area
17
of
normally
reared
cats.
In
strobe-reared
cats
both
direction
selectivity
and
orientation
selectivity
were
greatly
reduced
relative
to
normally
reared
cats,
and
some
units
in
the
strobe-reared
cats
responded
only
to
strobe
flashes.
A
fundamental
property
of
visual
stimuli
is
their
constant
motion
across
the
retina.
The
importance
of
stimulus
motion
for
perception,
which
has
been
demonstrated
by
experiments
with
stabilized
images
(1),
suggested
that
motion
deprivation
should
have
a
substantial
effect
on
the
development
of
the
visual
system.
To
examine
this
suggestion,
two
kittens
were
reared
from
birth
for
6
months
in
an
enclosure
in
which
the
only
source
of
illumination
was
a
strobe
light
(strobe-reared).
A
9-/Asec
flash
and
a
2-sec
interflash
interval
provided
virtually
stopped
images
on
the
retinae.
The
receptive-field
organization
of
single
cells
in
the
striate
cortex
was
studied
when
the
kittens
were
removed
from
the
enclosure
at
6
months
of
age.
We
recorded
from
cats
under
conditions
of
reversible
intuba-
tion
(2-4).
The
responses
of
98
units
in
two
cats
studied
during
a
total
of
five
recording
sessions
were
compared
with
those
of
248
units
in
the
visual
cortex
of
nine
normally
reared
cats.
Three
aspects
of
receptive-field
organization
in
striate
cortex
cells
were
examined:
(i)
direction
selectivity,
(ii)
orientation
selectivity,
and
(iii)
responses
to
diffuse
strobe
flashes.
Columnar
organization
was
not
studied
systematically.
Preferential
firing
to
certain
directions
of
movement
was
the
criterion
for
direction
selectivity.
A
direction
selective
unit
was
judged
to
be
orientation-selective
if
(a)
it
responded
well
to
moving
slits,
bars,
or
edges
and
poorly
to
moving
spots,
or
(b)
it
responded
differentially
to
flashed
slits
of
various
orienta-
tions
confined
entirely
within
the
activating
region
of
the
receptive
field,
or
(c)
its
direction
tuning
for
movement
was
narrower
for
long
slits
or
edges
than
for
spots.
A
nondirec-
tional
unit
was
judged
to
be
orientation-selective
if
it
re-
sponded
differentially
to
moving
slits
of
certain
orientations.
*
Present
address:
Max-Planck
Institut
fur
psychiatrie,
Munich,
Germany.
t
Present
Address:
Department
of
Anatomy,
University
of
Pennsylvania,
Philadelphia,
Pa.
19104
$
Address
reprint
requests
to
this
author.
Table
1
compares
the
responses
of
units
in
the
strobe-reared
cats
and
normally
reared
cats
with
regard
to
these
parameters.
In
the
normal
cat,
virtually
all
units exhibited
orientation
selectivity
and
most
were
direction-selective
as
well.
Few
units
responded
to
diffuse
strobe
flashes.
In
contrast,
the
fraction
of
units
exhibiting
orientation
or
direction
selectivity
was
se-
verely
reduced
in
strobe-reared
cats,
while
more
than
half
of
all
units
tested
responded
to
diffuse
strobe
flashes.
Some-units
in
strobe-reared
cats
could
only
be
activated
by
the
strobe
flashes.
Almost
all
units
in
normal
cats
could
be
classified
in
one
of
two
groups:
simple
or
complex
(5,
6).
Units
in
the
strobe-reared
cats
could
not
easily
be
identified
as
simple
or
complex;
however,
they
could
be
classified
as
direction-
selective
or
nondirectional.
The
responses
of
a
representative
unit
from
each
group
are
illustrated
in
Fig.
1.
The
nondirectional
group
comprised
62%
of
the
units
in
strobe-reared
cats.
These
units
did
not
reveal
orientation,
direction
selectivity,
or
preference
for
slits
or
edges
over
spots.
Presentation
of
flashing
spots
elicited
only
"on"
responses
throughout
the
activating
region,
and
annuli
revealed
a
weak
"off"
region
surrounding
the
strong
"on"
center.
These
units,
distinguished
from
lateral
geniculate
nucleus
afferents
in
that
they
have
cell-type
spikes,
were
sometimes
activated
bi-
nocularly
and
often
lacked
spontaneous
activity.
They
re-
semble
the
simple
cells
in
the
striate
cortex
of
the
normal
cat
in
that
both
have
receptive
fields,
which
can
be
mapped
into
discrete
"on"
and
"off"
zones
with
flashing
stimuli.
In
addition
to
these
units,
we
encountered
four
cells
that
were
orientation-
selective
and had
strong
"on"
regions
with
weak
"off"
flanks.
It
is
possible
that
the
concentrically
organized
units
in
strobe-
TABLE
1.
Comparison
of
receptive-field
properties
in
strobe-
reared
and
normally
reared
cats
Normally-
Strobe
reared
reared
Direction
selective
187
(83%)
32
(38%)
Nondirectional
43
(17%)
52
(62%)
Total
tested
230
84
Orientation
selective
241
(97%)
12
(15%)
Nonoriented
7
(3%)
73
(85%)
Total
tested
248
85
Response
only
to
strobe
0
13
Total
strobe
responsive
12
(10%)
32
(53%)
Strobe
unresponsive
109
(90%)
28
(47%)
Total
tested
121
60
1353
1354
Physiology:
Cynader
et
al.
/
z
,~A
%d
ItA
-
_,L
O
111111
UNIT
FIG.
1.
Histograms
illustrating
the
sum
of
16
sweeps
and
a
sample
unit
record
showing
the
responses
on
one
sweep
of
two
units
in
strobe-reared
cats
to
slits
moved
in
various
directions
across
their
receptive
fields.
The
unit
illustrated
(left)
displays
neither
orientation
nor
direction-selectivity,
responding
well
to
slits
moving
in
any
of
the
eight
directions
illustrated.
Receptive-
field
size:
30;
stimulus
velocity:
40/sec.;
slit
size:
1/20
X
10°;
stimulus
cycle:
4
sec.
The
unit
illustrated
(right)
displays
a
broadly
tuned
direction
selectivity,
responding
vigorously
to
the
three
or
eight
directions
of
movement
shown.
Receptive
field
side:
50;
stimulus
velocity:
70/sec.;
slit
size:
1/20
X
100;
stimulus
cycle:
4
sec.
reared
cats
are
simple
cells
that
have
either
failed
to
de-
velop
orientation
selectivity
(7)
or
have
lost
an
innate
orienta-
tion
selectivity
(8).
A
second
group
of
cells
comprise
32%
of
the
units
encoun-
tered
in
strobe-reared
cats.
They
showed
broadly
tuned
direction
selectivity
when
tested
with
spots
and
responded
well
to
moving
slits
over
a
wide
range
of
orientations
(Fig.
1,
right).
These
units
typically
did
not
meet
the
first
two
criteria
for
orientation
selectivity
described
above.
About
25%
of
them
met
the
third
criterion
by
exhibiting
a
somewhat
narrower
direction
tuning
when
tested
with
moving
slits
than
with
mov-
ing
spots.
Tests
with
flashing
stimuli
revealed
similarities
be-
tween
direction-selective
cells
of
strobe-reared
animals
and
complex
cells
in
the
normal
cat.
Most
units
in
both
groups
gave
"on"
and
"off"
responses
intermingled
throughout
the
receptive
field,
although
in
direction-selective
units
in
the
strobe-reared
cats
the
"on"
responses
were
typically
stronger.
In
some
units
of
the
strobe-reared
cats,
"off"
responses
were
absent.
Tests
with
moving
spots
and
slits
indicated
further
similarities
with
the
normal
cat.
Many
complex
cells
in
the
normal
cat
gave
surprisingly
strong
responses
to
small
moving
spots,
as
also
shown
recently
by
Palmer,
Rosenquist,
and
Sprague
(9).
The
complex
cells
in
normal
cats,
and
those
units
in
strobe-reared
cats
which
show
direction-selectiv-
ity,
display
comparable
breadth
of
tuning
for
direction-
selectivity.
These
features
suggest
a
resemblance
between
direction-selective
cells
in
the
strobe-reared
cats
and
complex
cells
in
the
normal
cat.
Although
most
of
these
direction-
selective
units
in
the
strobe-reared
cats
showed
substantially
broadened
tuning
for
orientation,
a
few
were
indistinguishable
from
complex
cells
in
the
normal
striate
cortex.
The
data
presented
here
show
that
strobe
rearing
augments
responses
of
striate
cortex
units
to
strobe
flashes
and
alters
the
normal
balance
between
"on"
and
"off"
responses
to
flashed
stimuli.
The
percentage
of
units
having
direction
and
orientation
selectivity
is
considerably
reduced.
Raising
kittens
in
an
environment
where
a
strobe
light
provides
the
only
source
of
illumination
can
substantially
alter
the
basic
char-
acteristics
of
the
visual
cortex.
This
research
was
supported
by
the
National
Institutes
of
Health
and
the
Sloan
Foundation.
1.
Ditchburn,
R.
W.
&
Ginsborg,
B.
L.
(1952)
Nature
170,
36-
37.
2.
Cynader,
M.
&
Berman,
N.
(1972)
J.
Neurophysiol.
35,
187-
201.
3.
Berman, N.
&
Cynader,
M.
(1972)
J.
Physiol.
224,
363-389.
4.
Wickelgren,
B.
G.
(1971)
Science
173,
69-72.
5.
Hubel,
D.
H.
&
Wiesel,
T.
N.
(1962)
J.
Physiol.
100,
106-
154.
6.
Pettigrew,
J.
D.,
Nikara,
T.
&
Bishop,
P.
0.
(1968)
Exp.
Brain
Res.
6,
373-390.
7.
Barlow,
H.
B.
&
Pettigrew,
J.
D.
(1971)
J.
Physiwl.
218,
98P-
loop.
8.
Hubel,
D.
H.
&
Wiesel,
T.
N.
(1963)
J.
Neurophysiol.
26,
994-1002.
9.
Palmer,
L.,
Rosenquist,
A.
C.
&
Sprague,
J.
M.
(1973)
Corticothalamic
projections
and
sensoriomtor
activities,
eds.
Friggesi,
T.
L.,
Rinvik
E.
&
Yahr,
M.
D.
(Raven
Press,
New
York),
in
press.
DIRECTIONAL
NON-DIRECTIONAL
LiA-
"I\
I
.|
T
--
2<
4,
Proc.
Nat.
Acad.
Sci.
USA
70
(1973)
... Concretely, we choose to focus on dynamic objects, which we define as entities that are capable of moving independently in the world. Independent object motion is a strong grouping cue, which has been shown to drive object learning in animal perception [14,52]. In computer vision, there exists a long line of works on motion segmentation that automatically separate moving objects from the background based on optical flow [7,15,35,43,43,64]. ...
Preprint
This paper studies the problem of object discovery -- separating objects from the background without manual labels. Existing approaches utilize appearance cues, such as color, texture, and location, to group pixels into object-like regions. However, by relying on appearance alone, these methods fail to separate objects from the background in cluttered scenes. This is a fundamental limitation since the definition of an object is inherently ambiguous and context-dependent. To resolve this ambiguity, we choose to focus on dynamic objects -- entities that can move independently in the world. We then scale the recent auto-encoder based frameworks for unsupervised object discovery from toy synthetic images to complex real-world scenes. To this end, we simplify their architecture, and augment the resulting model with a weak learning signal from general motion segmentation algorithms. Our experiments demonstrate that, despite only capturing a small subset of the objects that move, this signal is enough to generalize to segment both moving and static instances of dynamic objects. We show that our model scales to a newly collected, photo-realistic synthetic dataset with street driving scenarios. Additionally, we leverage ground truth segmentation and flow annotations in this dataset for thorough ablation and evaluation. Finally, our experiments on the real-world KITTI benchmark demonstrate that the proposed approach outperforms both heuristic- and learning-based methods by capitalizing on motion cues.
... Fifty years ago, scientists used to rear animals in altered environments to understand the relation between ecological sensory stimulations, behavior and brain functional networks, and tease apart innate from environmental factors. For instance, kittens reared in a world made only of horizontal and vertical stripes lacked sensitivity to oblique orientations [6], while those reared in stroboscopic illumination could not develop proper motion-selective responses [7]. Now, similar causal links between environmental features and behavioral properties can easily be established in deep neural networks trained in virtual environments. ...
Article
Similarities and differences between deep learning models and primate vision have been the focus of recent research. Audition is comparatively less-studied. A new report describes the emergence of human-like auditory perception in a deep neural network, and suggests a promising way to relate perceptual behaviour to specific aspects of the environment.
Chapter
This chapter is concemed with the relationship of some simple visual perceptual capacities to the development of specific nervous system components and processes. The paralle1s that have been discovered between sensory experience and nervous system anatomy and physiology yield insights not only into the development of perception but, more basically, into its neural basis. Thus, this brief review is as much about development as about the neural basis of visual experience. Because almost any biological or psychological data related to sensory systems is relevant to the basic theme of this chapter, some limits have been arbitrarily set to permit the development of several specific ideas. Thus, many possibly relevant areas will not be incIuded here: (1) the development of nonvisual perception; (2) the embryological development of the nervous system; (3) the development of language; (4) physiological development not directly related to sensory behavior; and (5) the development of sensory integration. Numerous recent and excellent reviews and symposia are available on some of these subjects. In particular, the following sources are highly recommended: general CNS development (Jacobson, 1978; Reinis & Goldman, 1980; Gottlieb, 1974); the development ofthe nonvisual components ofthe nervous system (Aslin, Alberts, & Petersen, 1981; Gottlieb, 1976); the development of vision (Freeman, 1979; Gottlieb, 1976; Rosinski, 1977; Lund, 1978; Aslin et al., Vol. 2, 1981). These sources represent easily available, integrated sources that provide both background material and an excellent entrance point to the vast experimentalliterature.
Chapter
Changes in brain function having detrimental behavioral consequences may be produced by factors other than those associated with trauma or disease. In this chapter we consider recent evidence which indicates that isolation from normal sensory experience is one of these factors. Our goal is to outline some of the conditions of visual deprivation which may result in physiological and behavioral aberrations and to discuss several hypotheses concerning the mechanisms of these effects. Additionally, and in substantially greater detail, we review investigations of recovery of function following deprivation.
Chapter
The topic of “brain and human behavior” leads us into problems as vast as our entire field, as difficult, and as unfinished. Concern with human behavior may be as old as humankind, but the attempt to understand ourselves by means of experimental science is young: My father still listened to Wundt, my grandfather could have listened to Helmholtz. For only a few generations have we seen truly systematic efforts at relating aspects of behavior to aspects of brain function. The ultimate coalescence of psychology and neurology that my own teacher, Lashley, demanded (1941a) is still a long way off. Why then try to survey where we are, if we are so very much at the start?
Chapter
The paradigm that we shall use in this chapter is the influence of sensory restriction on the development of the physiology of the visual and visuo-motor system. Conclusions drawn from this work are pertinent to the understanding this work are pertinent to the understanding of neural mechanisms operant in the normal animal. We have used a quantitative approach which is particularly interesting in that it provides, in mechanistic terms, a description of neuronal deficits in deprived animals. Before proceeding to an overview of our work, we shall situate the selective rearing paradigmboth historically and in the context of recent advances in the study of the epigenetic processes governing neural development.
Chapter
A better understanding of the function of sensory neurons has been achieved in the last decade by the determination of the stimulus parameters which are critical for triggering these neurons. Numerous experiments, following the pioneering studies by Hubel and Wiesel (10, 11) have clearly demonstrated that the neurons in the primary visual cortex of the cat (and of the monkey) are preferentially activated by rather complex visual stimuli, usually edges, slits or bars of particular and precise orientation, moving across their receptive field. The orientation of the edge which maximally activates each neuron is different from one cell to the next, but every orientation is equally represented. The majority of these neurons are binocularly driven, but are differentially influenced by the two eyes (2, 11, 16, 17).
Chapter
An der Bedeutung des Zentralnervensystems für die Speicherung angeborener und erworbener Informationen besteht heute kein Zweifel. Der Mensch hat, verglichen mit anderen Tieren gleicher Größe, ein beachtlich großes Hirngewicht entwickelt. Das absolute Gehirngewicht des Menschen wird von manchen Tierarten erreicht und zum Teil überboten; vernünftiger ist es wohl, einen Quotienten Hirngewicht: Körpermasse zu betrachten, wobei dann die bezüglich geistiger Leistungen ungewöhnliche Stellung des Homo sapiens erkennbar wird.
Chapter
Is the perception of form innate? Since the analysis of the question by Locke in his Essay Concerning Human Understanding (1690) and by Berkeley in his Essay Towards a New Theory of Vision (1709), it has been widely assumed that the perception of primitive sensations such as color or light intensity is innate, but that form perception is acquired slowly through visual experience. It was recognized, first, that the retinal image represents a two-dimensional projection of a three-dimensional world, and hence could supply information from which the third dimension had to be inferred. Secondly, it was understood that any object of stimulation had a nonunique relationship to any retinal image, since conditions of illumination, distance of the object, and the orientation of the receptor organ (e.g., eye and head movements) were constantly changing. It was thought likely that considerable experience was required to discover the underlying invariances inherent in what William James called the booming, buzzing confusion of stimulation projected onto the proximal receptor surface. The perception of a line, given a projected linear array of light dots projected onto the retina, was for empiricists not a simple, innately given competence. Rather, they believed an organism had to experience those light dots simultaneously so that associations could form between the neural representatives of those light dots. Empiricism has retained its theoretical vigor, especially following Hebb’S neurophysiological formalization of its principles (1949). The present chapter, Blakemore’S (this volume), and Haith’S (this volume) document the continuing wealth of behavioral and neurophysiological research which Empiricism has stimulated.
Article
Many cells in the intermediate and deep gray layers of the superior colliculus of the cat respond to both auditory and visual stimuli. These cells have similar receptive fields for both modalities and are directionally selective for both modalities, requiring stimuli moving laterally away from the animal. Perhaps cells that integrate auditory and visual information participate in the control of orienting and following responses to stimuli of both modalities.
Article
1. The superior colliculus has been studied in Siamese and normal cats by recording the responses of single tectal units to visual stimuli.2. The retinotopic organization of the superior colliculus has been compared in the two breeds. In the normal cat, the contralateral half-field is represented in the central and caudal part of the colliculus, and a vertical strip of the ipsilateral half-field, 15-20 degrees wide, is represented at the anterior tip. The Siamese cat superior colliculus receives an abnormally large projection from the ipsilateral half-field so that units with visual receptive fields which extend as far as 40 degrees into the ipsilateral half-field can be found. The area of the tectal surface devoted to the representation of the ipsilateral half-field is about twice as large in Siamese cats as in normal cats. The enhanced representation of the ipsilateral half-field in Siamese cats is reflected in a displacement of the vertical meridian and the area centralis on the tectal surface.3. The area centralis in the Siamese cat is located at about the same point on the tectal surface as would be occupied by a point in the visual field about 6-7 degrees contralateral to the area centralis in the normal cat. The smallest receptive fields in both breeds are located near the area centralis. The size of the receptive field for a tectal unit seems to be determined by the retinal location of the receptive field and not by the absolute position of the unit on the tectal surface.4. The receptive-field characteristics of tectal units show many similarities in the two breeds. The receptive fields of individual units consist of activating regions flanked by suppressive surrounds. Units respond well to stimuli of different shapes and orientation provided they are moving. The optimum stimulus for a given unit can be much smaller than the size of the activating region. About two thirds of the units studied in both breeds show directional selectivity. Most of the units studied in normal cats can be activated by stimulation of either eye, while in Siamese cats, 80% of the units studied can be driven only by the contralateral eye. A few monocularly driven units with two separated receptive fields have been observed in Siamese cats.5. In the left tectum of both breeds, units respond well to left-to-right stimulus movement. The reverse situation obtains in the right tectum. In Siamese cats, units located at the anterior tip of the tectum with their receptive fields located in the visual half-field ipsilateral to the tectum under study respond better to stimulus movement toward the area centralis than away from it. The preferred direction for a tectal unit seems to be determined by its tectal location rather than by the location of its receptive field in the retina.6. Visual cortex lesions in both breeds increase the responsiveness of tectal units to flashing spots and almost entirely remove the directional selectivity exhibited by tectal units, although units with asymmetric surrounds are still found. In normal cats, the lesions change the ocular dominance distribution, skewing it more strongly toward the contralateral eye. In Siamese cats, the ocular dominance distribution remains unchanged after a visual cortex lesion.7. The squint commonly exhibited by Siamese cats is regarded as a compensation for the anomalous retinotectal topography. It is suggested that, in the absence of an adaptive modification, the anomalous retinotectal projection would lead to mislocalization in Siamese cats just as it does in frogs and hamsters whose retinotectal projection has been experimentally altered. The convergent strabismus which Siamese cats commonly exhibit may be a cure for the abnormal retinal projections rather than a disease.
Article
A quantitative study has been made of the responses to moving slit stimuli by single units in the cat striate cortex whose receptive fields lay within 5° of the visual axis. Special attention was given to finding the optimal stimulus parameters including slit width, length, orientation and speed. The analysis was largely based on averaged response vs. time histograms. Using the classification of simple and complex responses types, the units were further subdivided on the basis of the number of modes in the response and on the presence or absence of directional selectivity. Simple unimodal units with directional selectivity (SUDS) had the most specific stimulus requirements and nearly always had zero background activity. Complex units usually had a high level of background activity. SUDS units also showed a preference for horizontally- and vertically ****-orientated stimuli. Whenever the response survived reversal of contrast the directional selectivity remained independent of the change. Optimal stimulus speeds varied widely from unit to unit with a mean at 4°/sec: simple bimodal units and complex units tended to have higher optimal stimulus speeds and responded over a wider range of speeds than did simple unimodal units. While SUDS units with very small receptive fields tended to prefer slowly moving stimuli, in general there was no correlation between receptive field size and optimal stimulus speed.
Article
THE small movements of the eye which persist when a subject fixates (that is, tries to gaze steadily at a given target) have been studied by several workers1. We have endeavoured to deduce the movements of the image across the retina from a recent study of the rotations of the eyeball2. The movements of the retinal image have the effect of moving the boundary between two regions of differing brightness across the retinal pattern of receptors and may therefore play an important part in vision.
Corticothalamic projections and sensoriomtor activities
  • L Palmer
  • A C Rosenquist
  • J M Sprague
Palmer, L., Rosenquist, A. C. & Sprague, J. M. (1973) Corticothalamic projections and sensoriomtor activities, eds.
  • H B Barlow
  • J D Pettigrew
Barlow, H. B. & Pettigrew, J. D. (1971) J. Physiwl. 218, 98P-loop.