Content uploaded by Kazuchika Manabe
Author content
All content in this area was uploaded by Kazuchika Manabe
Content may be subject to copyright.
JOURNAL
OF
THE
EXPERIMENTAL
ANALYSIS
OF
BEHAVIOR
REAL-TIME
DETECTION
OF
ORIENTATION
DURING
NEGATIVE
BEHAVIORAL
CONTRAST
WITH
KEY
PECKING
AND
A
TURNING
RESPONSE
KAZUCHIKA
MANABE
MEISEI
UNIVERSITY,
TOKYO,
JAPAN
We
developed
a
video
system
for
real-time
detection
of
a
pigeon's
orientation
and
for
reinforcement
of
a
"turning
response."
Using
this
system,
negative
behavioral
contrast
was
found
across
key-peck
and
turning
responses.
In
addition,
turning
away
from
the
pecking
key
was
detected
by
the
system
just
after
presentation
of
the
negative
discriminative
stimulus
on
the
key.
The
results
suggest
that
avoidance
of
the
discriminative
stimulus
in
the
constant
component,
which
has
been
regarded
as
a
causal
factor
for
negative
contrast
(additivity
theory),
is
not
the
primary
factor
for
negative
behavioral
contrast
of
pigeons'
key
pecking,
but
may
account
for
negative
local
contrast.
Key
words:
automated
technology,
video
system,
negative
behavioral
contrast,
turning
response,
key
peck,
pigeons
Under
typical
two-component
multiple
schedules
of
reinforcement,
two
discriminative
stimuli
are
presented
successively.
During
pre-
sentation
of
one
discriminative
stimulus,
a
cer-
tain
response
is
reinforced
according
to
a
schedule
of
reinforcement.
During
presenta-
tion
of
the
other
stimulus,
the
same
response
is
reinforced
by
another
schedule
of
reinforce-
ment.
In
these
multiple
schedules,
when
the
rate
of
reinforcement
in
one
of
the
components
is
decreased,
the
response
rate
in
the
changed
component
decreases
and
the
response
rate
in
the
other
unchanged
(constant)
component
in-
creases.
This
inverse
relation
between
re-
sponse
rate
in
the
constant
component
and
rate
of
reinforcement
in
the
changed
component
is
called
positive
behavioral
contrast.
Conversely,
when
the
rate
of
reinforcement
in
one
of
the
components
is
increased,
the
response
rate
in
the
other
constant
component
decreases
and
is
termed
negative
behavioral
contrast.
In
the
study
of
behavioral
contrast,
two
questions
have
been
discussed.
First,
are
pos-
itive
and
negative
behavioral
contrast
con-
This
research
was
supported
by
a
Grant-in-Aid
for
Developmental
Scientific
Research
from
the
Ministry
of
Education,
Science,
and
Culture
of
Japan.
We
wish
to
thank
Kurashige
san
for
her
assistance
with
this
Experi-
ment
and
Enomoto
shi
for
his
valuable
advice
with
pro-
gramming.
Correspondence
and
reprint
requests
may
be
sent
to
Kazuchika
Manabe,
Department
of
Psychology
and
Pedagogy,
Faculty
of
Humanities
and
Social
Sciences,
Meisei
University,
Hino,
Tokyo
191,
Japan.
trolled
by
the
same
mechanism?
Second,
do
positive
and
negative
behavioral
contrast
occur
in
the
same
manner
for
topographically
dif-
ferent
responses?
Herrnstein's
(1970)
matching
law
implies
that
both
types
of
contrast
are
explained
by
one
mechanism:
changes
in
relative
rate
of
re-
inforcement
across
components.
Therefore,
positive
and
negative
behavioral
contrast
should
occur
in
the
same
manner
for
topographically
different
responses,
differing
only
in
the
mag-
nitude
of
contrast.
Staddon
(1982)
and
Hinson
and
Staddon
(1978)
stated
in
the
behavior-competition
the-
ory
that
both
types
of
contrast
are
caused
by
reallocation
of
interim
activities
(such
as
wheel
running
for
rats
and
locomotion
for
pigeons)
from
the
component
with
the
higher
rate
of
reinforcement
to
the
component
with
the
lower
rate
of
reinforcement.
They
also
stated
that
contrast
occurs
in
the
same
manner
for
topo-
graphically
different
responses.
On
the
other
hand,
additivity
theory
asserts
that
positive
behavioral
contrast
is
caused
by
an
addition
of
responses
elicited
by
the
rein-
forcer
(such
as
pigeons'
key
pecks
when
the
reinforcer
is
grain)
to
the
operant
response
baseline
(Gamzu
&
Schwartz,
1973;
Rachlin,
1973;
Schwartz,
1975;
Schwartz
&
Gamzu,
1977).
In
comparison
to
positive
behavioral
contrast,
negative
behavioral
contrast
is
caused
by
avoidance
of
or
escape
from
the
discrimi-
native
stimulus
that
predicts
that
fewer
posi-
209
1992,
57,
209-218
NUMBER
2
(MARCH)
KAZUCHIKA
MANABE
tive
reinforcers
will
be
presented,
thereby
de-
creasing
the
response
rate
to
that
stimulus.
Therefore,
additivity
theory
predicts
that
sub-
jects
will
show
an
avoidance
response
to
the
discriminative
stimulus
in
the
constant
com-
ponent
when
the
rate
of
reinforcement
is
in-
creased
in
the
changed
component.
Further-
more,
additivity
theory
predicts
that
positive
behavioral
contrast
occurs
only
for
the
re-
sponses
that
can
be
elicited
by
the
reinforcer.
In
previous
examinations
of
the
first
ques-
tion,
it
has
been
found
that
conditions
that
produce
one
type
of
contrast
do
not
necessarily
produce
the other
type
of
contrast
(e.g.,
Ettin-
ger
&
McSweeney,
1981;
King
&
McSweeney,
1987;
Schwartz,
1975).
These
studies
suggest
that
one
mechanism
cannot
explain
both
types
of
contrast.
(As
mentioned
above,
additivity
theory
asserts
two
different
mechanisms
for
the
two
types
of
contrast.)
Experiments
have
been
performed
to
test
whether
positive
contrast
is
caused
by
an
addition
of
responses
elicited
by
the
reinforcer
to
an
operant
response
baseline
(Keller,
1974;
Manabe
&
Kawashima,
1982;
Schwartz,
1978;
Schwartz,
Hamilton,
&
Sil-
berberg,
1975;
Spealman,
1976).
The
results
of
this
research
suggest
that
the
addition
of
responses
elicited
by
the
reinforcer
to
the
op-
erant
response
baseline
produces
an
increase
in
only
the
initial
part
of
the
constant
com-
ponent,
a
process
termed
positive
local
contrast.
On
the
other
hand,
there
have
been
no
ex-
aminations
of
avoidance
of
the
discriminative
stimulus
in
the
constant
component
under
neg-
ative
behavioral
contrast.
The
present
exper-
iment
examined
whether
subjects
show
any
avoidance
of
or
escape
from
the
discriminative
stimulus
in
the
constant
component
when
the
rate
of
reinforcement
is
increased
in
the
changed
component.
Several
studies
have
examined
the
second
question.
They
suggested
that
positive
and
negative
contrast
do
not
occur
in
the
same
man-
ner
for
key-peck,
treadle-press,
and
bar-press
responses
in
pigeons
(Hemmes,
1973;
How-
ard,
1979;
McSweeney,
1978,
1982;
Mc-
Sweeney,
Dougan,
Higa,
&
Farmer,
1986;
Scull
&
Westbrook,
1973;
Westbrook,
1973).
However,
additional
studies
are
needed
to
clar-
ify
the
correlation
between
types
of
response
topography
and
types
of
contrast
with
re-
sponses
other
than
key
peck,
treadle
press,
and
bar
press.
To
this
end,
we
have
developed
a
system
that
can
detect
not
only
the
positions
but
also
the
orientation
of
subjects
in
real
time
(cf.
Pear,
Rector,
&
Legris,
1982).
The
real-time
detec-
tion
of
orientation
makes
it
possible
to
detect
activities
that
indicate
changes
in
orientation
of
the
subject
(e.g.,
turning
response,
loco-
motion,
etc.).
Thereby,
the
present
system
can
reinforce
these
responses
automatically.
In
ad-
dition,
the
present
system
can
detect
an
avoid-
ance
response
away
from
the
discriminative
stimulus
in
the
constant
component
under
con-
ditions
of
negative
behavioral
contrast.
In
this
paper,
we
describe
the
system
and
show
pigeons'
orientation
in
an
experiment
using
negative
behavioral
contrast
in
which
key-peck
responses
were
reinforced
in
one
component
of
a
multiple
schedule
and
either
key-peck
responses
or
turning
responses
were
reinforced
in
the
other
component.
METHOD
Subjects
Three
adult
male
homing
pigeons
were
maintained
at
80%
of
their
free-feeding
weights.
The
birds
had
free
access
to
grit
and
water
in
their
home
cages.
Apparatus
Experimental
chamber.
One
black-painted
experimental
chamber
(30
cm
by
31
cm
by
30
cm)
was
used.
A
response
key
was
mounted
on
one
wall
behind
a
hole
(2
cm
diameter)
at
a
height
of
20
cm
from
the
floor.
A
force
of
approximately
0.15
N
activated
a
microswitch
behind
this
key.
The
key
could
be
transillu-
minated
by
red
and
green
lights.
Ten
centi-
meters
beneath
the
key
were
two
openings
(5
cm
by
5
cm)
to
food
hoppers
that
contained
grain.
The
reinforcer
was
a
3-s
period
of
access
to
grain.
The
left
food
hopper
(the
center
of
the
opening
was
7.5
cm
from
the
left
wall)
was
used
for
the
turning
responses,
and
the
right
hopper
(7.5
cm
from
the
right
wall)
was
used
for
key-peck
responses
(see
Figure
1).
The
ceiling
was
a
black-painted
rough
net.
Detection
of
orientation.
The
pigeons'
ori-
entations
were
detected
by
the
following
method.
Pigeons
wore
a
harness
with
two
white
ping-pong
balls
attached.
The
harness
did
not
prevent
most
activities.
The
ping-pong
balls
were
positioned
on
the
neck
and
the
tail
(see
Figure
1).
The
back
half
of
the
ping-pong
ball
210
REAL-TIME
DETECTION
OF
ORIENTATION
Fig.
1.
Top
view
of
a
subject
in
the
chamber.
The
two
circles
indicate
ping-pong
balls.
The
ceiling
was
a
rough
net
painted
black.
on
the
neck
was
painted
black.
A
videocamera
was
situated
70
cm
above
the
center
of
the
ceiling
on
a
tripod.
The
top-view
image
was
translated
to
an
X-Y
image
tracker.
The
X-Y
image
tracker
calculated
the
X-Y
coordinates
of
the
two
ping-pong
balls
(neck
and
tail),
and
transmitted
the
two
X-Y
coordinates
to
a
mi-
crocomputer
once
every
0.2
s.
The
microcom-
puter
recorded
the
two
X-Y
coordinates
and
calculated
the
angle
of
orientation.
In
this
cal-
culation,
the
arc-tangent
was
calculated
based
on
the
distance
between
X
and
Y
coordinates
of
the
neck
and
tail;
the
arc-tangent
was
then
translated
into
the
angle
of
orientation.
The
turning
response
was
defined
as
a
change
from
the
area
00
through
+45°
to
the
area
00
through
-45°
passing
through
+180°,
and
vice
versa
(see
Figure
2).
At
00
the
subject
oriented
its
body
directly
toward
the
front
panel,
and
+
1800
indicates
orientation
to
the
rear
panel.
A
plus
value
indicates
that
the
subject
turned
its
body
to
the
left
side,
and
a
minus
value
indicates
turns
to
the
right
side.
In
this
system,
tracking
errors
occurred
(a)
when
the
subject's
head
accidentally
covered
a
ping-pong
ball
on
the
neck,
(b)
when
the
angle
of
the
subject's
body
relative
to
the
floor
was
near
900
so
that
the
ping-pong
ball
on
the
neck
prevented
detection
of
the
one
on
the
tail,
and
(c)
when
either
of
two
points
(neck
and
tail)
moved
more
than
the
distance
of
about
176
cm
in
1
s.
When
any
of
these
errors
occurred,
the
microcomputer
FRONT
PANEL
+9
O9
5.
-9
0
±
1
8
0
°
Fig.
2.
The
definition
of
turning.
A
turning
response
occurred
when
the
angle
of
a
subject's
orientation
changed
from
the
area
(0°
through
+45°)
to
the
area
(00
through
-45°)
through
±180°,
or
vice
versa.
automatically
stopped
the
session,
beeped,
and
turned
off
the
keylight.
Without
delay,
an
ex-
perimenter
reset
the
X-Y
image
tracker
and
proceeded
with
the
session.
These
errors
oc-
curred
rarely.
Programs
for
experimental
con-
trol
and
data
collection
were
written
in
C
lan-
guage
(Microsoft®
C).
Procedure
After
all
birds
were
habituated
to
the
har-
ness,
they
were
trained
to
eat
grain
from
the
two
food
hoppers.
Then,
key-peck
and
turning
responses
were
shaped
by
successive
approx-
imation.
Subsequently,
each
subject
was
ex-
posed
to
the
following
A-B-A-C-A-B-A
se-
quence
(see
Table
1).
In
Condition
A,
the
key-peck
response
was
reinforced
on
a
vari-
able-interval
30-s
schedule
(VI
30)
under
one
component
(constant
component;
key
color
was
green
for
Subject
S1
and
red
for
Subjects
S2
and
S3);
no
responses
were
reinforced
under
the
other
component
(changed
component;
key
color
was
red
for
Subject
S1
and
green
for
Subjects
S2
and
S3).
In
Condition
B,
the
key-
peck
response
was
reinforced
under
a
VI
30
schedule
during
the
constant
component,
and
the
turning
response
was
reinforced
on
a
VI
30
schedule
during
the
changed
component.
In
Condition
C,
the
key-peck
response
was
re-
inforced
on
a
VI
30
schedule
during
both
com-
ponents.
To
prevent
chaining
of
key-peck
and
turning
responses,
a
changeover
delay
(COD)
of
3
s
was
used.
Thus,
a
key-peck
response
KAZUCHIKA
MANABE
Table
1
Experimental
conditions.
Experiments
were
conducted
from
top
to
bottom.
Subject
Sl
Subjects
S2
and
S3
Component
1
(green)
Component
2
(red)
Component
1
(green)
Component
2
(red)
Key
peck
Turn
Key
peck
Turn
Key
peck
Turn
Key
peck
Turn
A
VI
30a
EXTb
EXT
EXT
EXT
EXT
VI
30
EXT
B
VI
30
EXT
EXT
VI
30
EXT
VI
30
VI
30
EXT
A
VI
30
EXT
EXT
EXT
EXT
EXT
VI
30
EXT
C
VI
30
EXT
VI
30
EXT
VI
30
EXT
VI
30
EXT
A
VI
30
EXT
EXT
EXT
EXT
EXT
VI
30
EXT
B
VI
30
EXT
EXT
VI
30
EXT
VI
30
VI
30
EXT
A
VI
30
EXT
EXT
EXT
EXT
EXT
VI
30
EXT
a
Variable-interval
30-s
schedule.
bExtinction.
was
not
reinforced
during
the
first
3
s
following
the
first
peck
just
after
a
turning
response,
and
vice
versa.
All
subjects
received
two
sessions
daily,
7
days
per
week.
The
first
daily
session
began
at
about
7:00
a.m.,
and
the
last
daily
session
began
at
about
4:00
p.m.
Each
session
was
terminated
after
each
component
had
been
presented
20
times.
The
component
duration
was
30
s.
Components
were
alternated
ac-
cording
to
the
Gellerman
(1933)
series.
Each
condition
was
in
effect
for
15
sessions.
RESULTS
Figure
3
shows
the
mean
response
rate
(re-
sponses
per
minute)
for
key
pecking
and
turn-
ing
for
the
last
five
sessions
of
all
conditions.
Turning
rates
during
the
changed
component
increased
when
turning
was
reinforced
(Con-
dition
B).
Similarly,
key-peck
rates
during
the
changed
component
increased
when
pecking
was
reinforced
in
the
changed
component
(Condition
C).
On
the
other
hand,
key-peck
rates
during
the
constant
component
decreased
OPECK
(CONSTANT)
*PECK((CHANGED)
LU
150
z
2~~z5i
LU
100
a.
Q
z
LU
o
-
LU
LL
z
LU
0i
ATURN
(CONSTANT)
ATURN
(CHANGED)
A
I
A
A
I
A
A
1
A
k
C
A
B
A
C
O
N
D
I
T
I
O
N
S
Fig.
3.
The
mean
frequencies
of
the
last
five
sessions
for
key-peck
and
turning
responses
under
seven
conditions.
Circles
indicate
pecking
response,
triangles
turning
response,
open
symbols
responses
in
the
constant
component,
and
filled
symbols
responses
in
the
changed
component.
212
REAL-TIME
DETECTION
OF
ORIENTATION
when
either
key-peck
or
turning
responses
were
reinforced
in
the
changed
component
for
Sub-
jects
Si
and
S3.
That
is,
negative
behavioral
contrast
occurred
for
2
subjects;
Subject
S2
showed
no
schedule
interaction
across
com-
ponents.
Analysis
of
Orientation
Figure
4
illustrates
subjects'
angle
of
ori-
entation
and
cumulative
records
of
key-peck
responses
in
the
19th
and
20th
components
of
the
last
session
of
each
condition.
In
Figure
4,
a
curve
from
00
to
00
through
±
1800
indicates
a
full
turn.
If
the
subject
kept
a
constant
angle
of
orientation,
the
line
was
parallel
to
the
x
axis.
Under
components
in
which
only
key-
peck
responses
were
reinforced,
the
angles
of
orientation
were
near
00,
indicating
that
the
subject
oriented
its
body
to
the
pecking
key.
On
the
other
hand,
under
components
in
which
turning
responses
were
reinforced,
the
angles
varied
from
0°
to
±1800
for
Subjects
S1
and
S2.
For
Subject
S3,
the
angles
varied
from
about
+450
to
about
-450
through
±
1800.
This
indicates
that
Subject
S3
turned
economically.
Typical
locomotion,
in
which
the
subject
walks
about
in
no
particular
orientation,
was
found
for
Subject
S2
during
the
changed
component
(extinction)
of
the
seventh
condition.
The
line
oscillated
about
+
900.
Figure
5
shows
the
mean
absolute
degrees
of
angle
under
the
constant
component
when
preceded
by
the
changed
component
and
those
under
the
changed
component
when
preceded
by
the
constant
component.
The
data
are
based
on
the
last
three
sessions.
In
Condition
A
(first,
third,
fifth,
and
seventh
conditions)
in
the
changed
component
(extinction),
the
mean
ab-
solute
degree
of
angle
abruptly
increased
in
every
case
and
gradually
decreased,
except
for
Subject
S2
in
the
third
and
seventh
conditions.
These
large
angles
of
orientation
in
the
initial
part
of
the
extinction
components
indicate
that
the
subjects
turned
away
from
the
pecking
key
just
after
presentation
of
the
negative
discrim-
inative
stimulus.
These
turning
responses
are
also
found
in
Figure
4
(see
extinction
com-
ponent
under
Condition
A).
In
the
component
in
which
turning
responses
were
reinforced,
the
mean
angles
were
large
and
the
curves are
parallel
to
the
x
axis,
indicating
continued
turning
throughout
the
component.
Although
the
rate
of
key
pecking
during
the
constant
component
decreased
in
Condition
C
(see
Fig-
ure
3),
the
mean
absolute
angles
approximated
those
of
Condition
A.
Figure
6
shows
the
percentages
of
time
that
the
subject
oriented
its
body
to
various
direc-
tions
in
the
last
session
of
the
third,
fourth,
and
fifth
conditions
(Conditions
B,
A,
and
C).
The
upper
half
of
the
vertical line
of
each
grid
indicates
the
area
from
-30°
through
+300.
The
lower
half
indicates
the
area
300
to
each
side
of
1800.
The
other
half
lines
also
indicate
the
areas
having
an
arc
of
600.
The
length
from
the
intersecting
point
describes
the
percentage
of
time
that
subjects
oriented
in
a
given
direc-
tion.
Intersection
at
the
end
of
a
line
indicates
that
the
subject
oriented
its
body
to
that
di-
rection
for
the
entire
session
(100%).
If
the
subject
oriented
to
each
area
for
the
same
amount
of
time,
the
figure
becomes
a
hexagon.
During
components
in
which
the
key-peck
re-
sponse
was
reinforced
(each
constant
compo-
nent
and
the
changed
component
of
Condition
C),
most
of
the
time
was
allocated
to
the
area
from
-30°
through
+30°.
On
the
other
hand,
time
was
allocated
to
various
orientations
in
the
component
in
which
turning
responses
were
reinforced
(changed
component
under
Con-
dition
B).
In
Condition
A,
the
percentages
of
time
allocated
to
the
area
from
-30°
through
+300
were
smaller
under
the
changed
com-
ponent,
in
which
responses
were
not
rein-
forced,
than
under
the
constant
component,
in
which
key
pecks
were
reinforced.
Under
the
changed
component
(extinction)
in
Condition
A,
Subject
S2
allocated
about
50%
of
the
time
to
the
left
side
from
+
300
through
+
1500.
Al-
though
the
rates
of
key
pecks
in
the
constant
component
for
Subjects
S1
and
S3
were
lower
in
Conditions
B
and
C
than
in
Condition
A,
the
time
allocated
to
the
area
from
-300
through
+300,
where
key
pecking
occurred,
was
longer
than
in
Condition A.
The
time
allocated
to
the
area
from
-30°
through
+
300
under
the
constant
component
in
Condition
B
for
both
subjects
was
shorter
than
in
Condi-
tion
A.
DISCUSSION
Using
the
present
video
system,
we
found
clear
negative
behavioral
contrast
of
key
peck-
ing
for
2
of
3
subjects
when
key-pecking
or
turning
responses
were
reinforced
in
the
changed
component.
This
extended
the
gen-
erality
of
negative
behavioral
contrast
for
key-
peck
responses.
However,
King
and
Mc-
213
214
S
1
KAZUCHIKA
MANABE
S
2
19th
component
20th
component
EXT
PECK-VI
+
180'
O-
I
go'
PECK-VI
TURN-VI
+180
PECK-VI
EXT
+180'
0.
-1801
PECK-VI
PECK-VI
+180'
PECK-VI
EXT
-180
__
-_
O
T
,
TURN-VI
PECK-VI
-180
60
19th
component
20th
component
EXT
EXT
PECK-VI
s_
PECK-VI
PECK-VI
PECK-VI
EXT
PECK-VI
TURN-VI
PECK-VI
EXT
0
30
60
19th
component
20th
component
EXT
PECK-VI
I
CONDITION
.AL
120
60
0
IB
120
60Z
0
o
0
-I,
AL
120
-
60
OC
120
C
,120
>
0
-
PECK-VI
EXT
<
|
11
|m
,--I
on
'IS'
120
PECK-VI
TURN-VI
PECK-VI
EXT
6t
0
30
60
0
30
60
I0
z
0
20
20
SUCCESSIVE
TIME
(S)
Fig.
4.
The
changes
in
angle
and
cumulative
records
for
key-peck
responses.
The
data
of
the
19th
and
20th
components
are
shown.
EXT
indicates
components
in
which
the
schedule
of
reinforcement
was
extinction.
PECK-VI
indicates
components
in
which
the
key-peck
response
was
reinforced
on
a
VI
30-s
schedule.
TURN-VI
indicates
components
in
which
the
turning
response
was
reinforced
on
a
VI
30-s
schedule.
Cumulative
records
have
their
origin
in
the
bottom
left
corner
of
each
half-panel;
records
for
angle
of
orientation
usually
begin
near
the
0°
position
in
each
half-panel.
TURN-Vt
PECK-VI
TURN-VI
PECK-VI
:001
2A
L-
EXT
PECK-VI
13:
=,
I
z
0
I-
I-
z
w
0
Cl)
1-
m
w
Cl)
U-
0
w
.1
z
UL.
0
w
w
w
a
PECK-VI
PECK-VI
YUYIflL½
..
r--V
Sr-
3
v
D
REAL-TIME
DETECTION
OF
ORIENTATION
S
1
COMPONENT
1
COMPONENT
2
(CONSTANT)
(CHANGED)
80'
PECK-VI
EXT
90-
O0
-
AR.
9
9
90-
I
ou
7
PECK-VI
EXT
9011
0.I
I
f
180-
PECK-VI
TURN-VI
901
0.1
180-
PECK-VI
'
EXT
901,
0
10
20
S
2
COMPONENT
1
COMPONENT
2
(CHANGED)
(CONSTANT)
I
EXT
PECK-VI
TURN-VI
PECK-VI
EXT
PECK-VI
i
PECK-VI
]
PECK-VI
EXT
PECK-VI
TURN-VI
PECK-VI
.
E*XT
PECK-VI
.__________
I
10
20
30
0
10
20
0
10
20
30
S
3
COMPONENT
1
(CHANGED)
EXT
COMPONENT
2
(CONSTANT)
PECK-V
I
TURN-VI
PECK-VI
EXT
PECK-VI
PECK-VI
PECK-VI
EXT
PECK-VI
TURN-VI
PECK-VI
EXT
PECK-VI
0
10
20
0
10
20
30
SUCCESSIVE
TIME
(S)
Fig.
5.
The
mean
absolute
degrees
of
angular
deviation
during
the
constant
component
just
after
a
change
to
that
component
and
those
in
the
changed
component
following
a
constant
component.
The
data
are
shown
from
top
to
bottom
for
each
subject.
Sweeney
(1987)
failed
to
obtain
clear
negative
behavioral
contrast
for
the
key-peck
response
when
a
treadle-press
response
was
reinforced
under
the
changed
component.
There
are
several
possible
reasons
for
this
different
outcome.
First,
the
two
discrimina-
tive
stimuli
that
indicated
reinforcement
were
presented
in
different
positions
by
King
and
McSweeney
(1987).
In
this
study,
the
two
dis-
criminative
stimuli
were
presented
on
one
pecking
key.
In
most
experiments
in
which
negative
behavioral
contrast
is
found,
the
two
discriminative
stimuli
are
located
in
the
same
place.
Negative
behavioral
contrast
for
the
key-
peck
response
may
be
enhanced
by
a
condition
in
which
the
discriminative
stimuli
are
pre-
sented
in
the
same
position.
Second,
the
topography
of
the
response
re-
inforced
in
the
other
component
may
affect
negative
behavioral
contrast
for
the
key-peck
response.
For
example,
Scull
and
Westbrook
(1973)
suggested
that
one
requirement
for
pos-
itive
contrast
is
that
topographically
similar
behavior
be
required
in
both
components
of
the
multiple
schedule.
In
general,
negative
be-
havioral
contrast
may
be
determined
not
only
by
the
experimental
condition
but
also
by
the
type
of
responses.
Therefore,
to
test
the
gen-
I
PECK-VI
TURN-VI
0.
PECK-VI
EXT
0I
0.
z
0
z
Lu
0
(I)
3
co
n
(-
U
0
C,
z
0
Uf)
4c
z
4c
uJ
PECK-VI
PECK-VI
B
C
"
-
I
I
I
KAZUCHIKA
MANABE
CONDITION
B
A
C
S-30<-i30i
52
S3
Fig.
6.
The
percentage
of
time
that
the
subject
oriented
its
body
to
the
angle.
Each
axis
is
for
a
range
of
600.
erality
of
behavioral
contrast,
we
should
ex-
amine
a
larger
variety
of
responses.
The
pres-
ent
system
may
be
useful
as
a
tool
for
examining
behavioral
contrast
across
responses
that
con-
tain
changes
in
angle
and
position
of
the
body.
The
present
system
detected
an
initial
tem-
poral
increase
in
absolute
angle
of
orientation
just
after
the
change
from
the
VI
to
extinction.
Because
pigeons
cannot
peck
the
key
when
the
angle
is
large
(i.e.,
when
they
face
away
from
the
key),
this
phenomenon
may
be
a
major
factor
responsible
for
negative
local
contrast,
the
decrease
in
rate
of
response
that
is
observed
just
after
the
component
is
altered
from
a
rich
schedule
to
a
lean
schedule.
Hearst
and
Jen-
kins
(1974)
asserted,
according
to
their
sign-
tracking
hypothesis,
that
animals
will
tend
to
go
away
from
stimuli
that
predict
a
decrease
in
the
frequency
of
presentation
of
positive
reinforcers.
The
tendency
may
be
strongest
just
after
the
component
is
changed
from
VI
to
extinction.
If
this
is
true,
negative
local
con-
.I
REAL-TIME
DETECTION
OF
ORIENTATION
217
trast
may
result.
The
initial
turning
away
dur-
ing
the
extinction
component
should
be
ex-
amined
further.
Although
the
rate
of
key
pecking
during
the
constant
component
was
lower
in
Condition
C
than
in
Condition
A,
the
means
of
absolute
angular
deviation
during
the
constant
com-
ponent
under
those
two
conditions
were
almost
equal.
This
indicates
that
the
decrease
in
rate
of
key-peck
responses
during
the
constant
com-
ponent
in
Condition
C
might
not
have
been
caused
by
a
change
in
the
subject's
orientation
away
from
the
discriminative
stimulus
on
the
pecking
key
(cf.
Hearst
&
Jenkins,
1974).
The
decrease
in
rate
might
have
been
caused
by
inhibition
of
elicited
key-peck
response
in
Condition
C
or
by
the
other
factors
(cf.
Rach-
lin,
1973;
Schwartz,
1975).
An
analysis
of
the
percentage
of
time
that
subjects
orient
towards
a
particular
angle
may
produce
useful
data
for
analyzing
time
allo-
cation
in
some
experiments,
such
as
concurrent
schedules
in
which
subjects
peck
either
of
two
keys.
Because
subjects
must
approach
and
ori-
ent
towards
the
key
in
order
to
peck
the
key,
measures
of
angle
and
position
may
provide
a
more
accurate
indication
of
time
allocation.
In
previous
studies
of
operant
behavior,
only
a
few
types
of
responses
have
been
examined,
such
as
the
key-peck
response
for
pigeons
and
the
lever-press
response
for
rats
and
monkeys.
It
is
unclear
from
existing
data
whether
to-
pographically
different
responses
demonstrate
similar
phenomena
in
the
same
situation.
More
studies
are
needed
to
reexamine
various
well-
known
phenomena
with
more
types
of
re-
sponses.
The
automated
system
used
in
this
experiment
makes
such
studies
more
practical.
In
summary,
the
present
results
suggest
that
an
avoidance
response
to
the
discriminative
stimulus
in
the
constant
component
is
not
the
primary
factor
for
negative
contrast,
but
may
contribute
to
negative
local
contrast.
In
addi-
tion,
the
system
used
here
may
be
useful
as
a
tool
for
the
analysis
of
the
other
activities
that
microswitches
cannot
detect.
REFERENCES
Ettinger,
R.
H.,
&
McSweeney,
F.
K.
(1981).
Behav-
ioral
contrast
and
responding
during
multiple
food-
food,
food-water,
and
water-water
schedules.
Animal
Learning
&
Behavior,
9,
216-222.
Gamzu,
E.,
&
Schwartz,
B.
(1973).
The
maintenance
of
key
pecking
by
stimulus-contingent
and
response-
independent
food
presentation.
Journal
of
the
Experi-
mental