Content uploaded by A. Charles Catania
Author content
All content in this area was uploaded by A. Charles Catania on Jun 25, 2014
Content may be subject to copyright.
JOURNAL
OF
THE
EXPERIMENTAL
ANALYSIS
OF
BEHAVIOR
PREFERENCE
FOR
FREE
CHOICE
OVER
FORCED
CHOICE
IN
PIGEONS
A.
CHARLES
CATANIA
AND
TERJE
SAGVOLDEN
UNIVERSITY
OF
MARYLAND
BALTIMORE
COUNTY
In
a
six-key
chamber
variable-interval
initial
links
of
concurrent-chain
schedules
operated
on
two
lower
white
keys.
Terminal
links
operated
on
four
upper
keys;
green
keys
were
correlated
with
fixed-interval
reinforcement
and
red
keys
with
extinction.
Free-choice
ter-
minal
links
arranged
three
green
keys
and
one
red
key;
forced-choice
terminal
links
ar-
ranged
one
green
key
and
three
red
keys.
Thus,
terminal
links
were
equivalent
in
number,
variety,
and
information
value
(in
bits)
of
the
keylights.
Preferences
(relative
initial-link
rates)
were
studied
both
with
location
of
the
odd
key
color
varying
over
successive
terminal
links
and
with
the
odd
color
fixed
at
key
locations
that
had
controlled
either
relatively
high
or
relatively
low
terminal-link
response
rates.
Free
choice
was
consistently
preferred
to
forced
choice.
Magnitude
of
preference
did
not
vary
systematically
with
terminal-link
response
rate
or
stimulus
control
by
green
and
red
keys.
The
origins
of
free-choice
prefer-
ence
could
be
ontogenic
or
phylogenic:
organisms
may
learn
that
momentarily
preferred
alternatives
are
more
often
available
in
free
than
in
forced
choice,
and
evolutionary
con-
tingencies
may
favor
the
survival
of
organisms
that
prefer
free
to
forced
choice.
Key
words:
concurrent
chain
schedules,
preference,
free
vs
forced
choice,
response
vari-
ability,
position
preference,
freedom,
key
peck,
pigeon
Philosophical
discussions
of
freedom
of
choice
often
involve
the
debate
over
free
will
and
determinism
(e.g.,
Berofsky,
1966;
Ente-
man,
1967).
But
free
choice
implies
the
avail-
ability
of
alternatives,
and
thus
the
concept
of
freedom
may
also
occasion
empirical
questions.
To
argue
that
organisms
value
freedom,
for
example,
is
to
argue
that
free
choices
will
be
preferred
to
forced
choices.
It
may
then
be
asked
whether
such
preferences
exist,
whether
they
are
limited
to
humans,
and
whether
they
are
products
of
ontogeny
or
phylogeny.
Research
supported
by
NSF
Grants
GB-43251
and
BNS76-09723
to
the
University
of
Maryland
Baltimore
County.
Terje
Sagvolden,
now
with
the
Institutes
of
Neurophysiology
and
of
Psychology
at
the
University
of
Oslo,
was
supported
during
the
research
by
a
post-
doctoral
fellowship
from
the
Norwegian
Research
Council
for
Science
and
the
Humanities.
We
are
in-
debted
to
Kenneth
Keller,
Howard
Rachlin,
and
Mur-
ray
Sidman
for
some
probing
questions.
We
also
thank
several
UMBC
students
who
assisted
in
the
research,
and
must
especially
mention
Robert
Kountz,
Marian
Colleen
Owens,
and
Virginia
von
Lossberg.
Some
of
the
present
data
were
presented
at
the
1975
meeting
of
the
Eastern
Psychological
Association.
Preparation
of
the
manuscript
was
supported
in
part
by
NIMH
Grant
MH-33086.
For
reprints,
write
A.
C.
Catania,
Depart-
ment
of
Psychology,
University
of
Maryland
Baltimore
County,
5401
Wilkens
Avenue,
Catonsville,
Maryland
21228.
An
organism
can
choose
among
alternatives
only
to
the
extent
that
stimuli
are
correlated
with
their
availability.
For
this
reason
the
dis-
tinction
between
free
and
forced
choice
will
typically
be
confounded
with
the
number,
variety,
and
information
value
of
stimuli.
An
earlier
concurrent-chain
study
(Catania,
1975)
demonstrated
preferences
for
free
over
forced
choice
in
pigeons,
but
controlled
for
stimulus
number,
variety,
and
information
only
indi-
rectly,
across
a
sequence
of
procedures.
The
present
research
more
directly
separated
free-
choice
preference
from
the
effects
of
these
other
variables.
Free-choice
and
forced-choice
conditions
were
arranged
on
four
pigeon
keys.
In
free
choice,
pecks
on
any
of
three
green
keys
pro-
duced
food
according
to
a
fixed-interval
rein-
forcement
schedule
while
pecks
on
the
remain-
ing
red
key
had
no
scheduled
consequences.
In
forced
choice,
pecks
on
only
a
single
green
key
produced
food
according
to
the
fixed-inter-
val
schedule
while
pecks
on
the
three
remain-
ing
red
keys
had
no
scheduled
consequences.
Preferences
for
free
over
forced
choice
were
obtained
even
with
the
two
conditions
equated
for
stimulus
number
(four
lit
keys),
stimulus
variety
(one
odd
color
among
four),
and
tra-
ditional
measures
of
information
(two
bits
of
77
1980,
34,
77-86
NUMBER
I
(JULY)
A.
C.
CATANIA
and
T.
SAGVOLDEN
information
to
specify
either
one
or
three
of
four
keys).
METHOD
Subjects
Four
male
White
Carneaux
pigeons,
about
1
yr-old
at
the
start,
were
maintained
at
about
80%
of
free-feeding
weights.
Each
pigeon's
key
pecking
had
been
established
in
a
two-key
autoshaping
procedure
in
which
one
key
was
red
and
the
other
green
(Fisher
&
Catania,
1977);
in
that
procedure
all
pigeons
pecked
red
keys
more
often
than
green.
Apparatus
Daily
sessions
were
conducted
in
a
six-key
pigeon
chamber
(Catania,
1975).
Figure
1
shows
the
key
configuration.
The
four
upper
keys
could
be
lit
red
or
green
and
the
two
lower
keys
could
be
lit
white
by
6-W
lamps.
The
keys
were
matched
to
operate
at
mini-
mum
forces
of
about
.14
N.
Each
peck
on
a
lit
key
produced
a
feedback
click
from
a
relay
behind
the
panel.
Dark-key
pecks
had
no
scheduled
consequences.
A
standard
Gerbrands
feeder
was
centered
beneath
the
keys.
During
grain
deliveries
the
feeder
was
lit
and
all
keylights
were
off.
Standard
electromechanical
equipment
in
an
adjoining
room
arranged
scheduling
and
recording.
Procedure
Figure
1
illustrates
the
concurrent-chains
procedure
(Autor,
1969;
Herrnstein,
1964).
In
initial
links
equal
and
independent
variable-
interval
(VI)
30-sec
schedules
were
arranged
concurrently
for
pecks
on
the
two
white
lower
keys;
the
four
upper
keys
were
dark.
The
15-
interval
VI
30-sec
schedules
were
constructed
according
to
Catania
and
Reynolds
(1968,
Ap-
pendix
II),
with
the
order
for
one
the
reverse
of
the
other
so
that the
two
schedules
could
not
become
synchronized.
The
timing
of
inter-
vals
stopped
for
a
given
initial-link
key
when
a
terminal
link
was
scheduled
for
the
next
peck
on
that
key,
and
it
stopped
for
both
keys
throughout
terminal
links
and
food
deliveries.
The
first
peck
after
a
changeover
from
one
initial-link
key
to
the
other
was
not
eligible
to
produce
a
terminal
link.
In
terminal
links
the
four
upper
keys
were
lit
and
the
lower
keys
were
dark.
Forced-choice
terminal
links
consisted
of
one
green
and
three
red
keys
(Figure
1,
left);
free-choice
ter-
minal
links
consisted
of
three
green
and
one
red
keys
(Figure
1,
right).
In
both
types
of
terminal
links,
pecks
on
any
green
key
were
reinforced
according
to
a
fixed-interval
(FI)
30-sec
schedule;
pecks
on
any
red
key
had
no
scheduled
consequences
(EXT).
The
terminal-
link
reinforcer
was
a
3-sec
food
delivery,
after
which
initial
links
were
reinstated.
Although
the
same
keys
lit
in
both
cases,
terminal
links
produced
by
the
left
and
right
initial-link
schedules
are
referred
to
respectively
as
left
and
right
terminal
links.
Establishing
the
concurrent-chain
perfor-
mance.
With
each
pigeon's
key
pecking
al-
ready
established
through
autoshaping,
the
procedures
began
with
three
sessions
in
which
all
six
keys
were
dark
during
the
timing
of
initial
links
and
free-choice
and
forced-choice
terminal
links
were
each
presented
indepen-
dently
of
responding
at
mean
intervals
of
30
sec.
In
terminal
links,
green-key
pecks
were
reinforced
according
to
an
FI
5-sec
schedule
and
red-key
pecks
had
no
scheduled
conse-
quences.
In
the
next
session
the
concurrent
initial-link
VI
schedules
were
introduced
on
Fig.
1.
Example
of
the
concurrent-chains
procedure.
During
initial
links
the
four
upper
keys
were
dark
and
the
two
lower
keys
were
white
(W).
According
to
independent
variable-interval
(VI)
schedules,
pecks
on
the
left
and
right
white
keys
produced
their
respec-
tive
and
mutually
exclusive
terminal
links.
During
ter-
minal
links
the
upper
keys
were
green
(G)
and
red
(R)
and
the
lower
keys
were
dark.
In
the
example
a
forced-
choice
terminal
link
is
shown
on
the
left,
and
a
free-
choice
terminal
link
on
the
right.
In
forced
choice
one
key
was
green
and
the
three
remaining
keys
were
red.
In
free
choice
three
keys
were
green
and
the
one
re-
maining
key
was
red.
In
both
cases
green-key
pecks
produced
food
according
to
a
fixed-interval
(FI)
sched-
ule,
and
red-key
pecks
had
no
scheduled
consequences.
After
each
food
delivery
initial-link
conditions
were
reinstated.
78
FREE-CHOICE
PREFERENCE
the
lower
keys,
lit
white.
The
left
initial-link
key
produced
forced-choice
terminal
links,
the
right
initial-link
key
produced
free-choice
terminal
links,
and
the
terminal-link
green-
key
schedule
was
lengthened
to
Fl
10-sec.
Over
subsequent
sessions
free
choice
and
forced
choice
were
occasionally
alternated
as
respective
left
and
right
terminal
links,
session
durations
and
reinforcer
durations
were
ad-
justed,
and
the
terminal-link
schedule
was
lengthened
from
Fl
10-sec
to
Fl
20-sec
and
then
to
Fl
30-sec.
The
daily
sessions
of
these
conditions
are
combined
with
data
in
Figure
2.
During
these
sessions
the
only
key
failure
occurred
on
terminal-link
Key
4
for
Pigeon
10
in
Session
52.
Three
sessions
of
left
initial-
link
extinction
and
three
of
right
initial-link
extinction
(a
test
of
the
sensitivity
of
the
pro-.
cedure
to
reinforcement
variables)
preceded
subsequent
experimental
conditions.
Terminal-link
key
positions:
sequence
of
conditions.
In
most
sessions
the
position
of
the
odd-colored
key
varied
over
successive
ter-
minal
links.
With
the
upper
keys
numbered
from
left
to
right,
Figure
1
shows
green
lo-
cated
on
Key
3
in
forced
choice
(left)
and
red
located
on
Key
4
in
free
choice
(right).
The
location
of
the
odd
key
was
determined
by
a
sequence
of
four
alternatives
over
successive
terminal
links:
AABA
CACC
DDAA
CDBD
CBBD
DABB
DCDA
BCBA
CCBD.
The
arbi-
bitrary
correspondence
between
the
lettered
alternatives
and
the
four
key
positions
was
occasionally
changed
over
blocks
of
sessions.
Because
the
concurrent
initial-link
VI
sched-
ules
produced
an
irregular
alternation
of
free-
choice
and
forced-choice
terminal
links,
this
arrangement
generated
an
effectively
non-
repeating
sequence
of
locations
for
the
odd
key
in
each
terminal
link.
The
effects
of
terminal-link
response
rates
and
stimulus
control
on
initial-link
prefer-
ences
were
examined
by
fixing
the
location
of
the
odd
key.
For
most
pigeons
Key
1
main-
tained
the
highest
and
Key
3
the
lowest
green
response
rates
in
both
free
and
forced
choice
(Table
1).
In
one
procedure
Key
1
was
always
the
odd
green
key
in
forced
choice
and
the
odd
red
key
in
free
choice.
In
a
second
pro-
cedure
Key
3
was
always
the
odd
green
key
in
forced
choice
and
the
odd
red
key
in
free
choice.
The
arrangement
of
free
and
forced
choice
as
left
and
right
terminal
links
and
the
sessions
at
each
condition
are
combined
with
data
in
Figure
3.
Sessions
were
ordinarily
scheduled
for
25
min
of
initial
links
but
were
occasion-
ally
increased
or
decreased
in
steps
of
5
min
for
a
given
pigeon
to
maintain
80%
weights
while
minimizing
postsession
feeding.
The
shortest
session
scheduled
was
20
min;
the
longest
was
35
min.
Because
changes
in
terminal
links
had
rap-
idly
affected
relative
initial-link
response
rates
in
earlier
research
(Catania,
1975),
fixed
num-
bers
of
sessions
at
each
condition
were
origi-
nally
planned.
This
plan
was
discarded
pri-
marily
on
the
basis
of
occasional
key
failures
and
of
the
performance
of
Pigeon
10,
which
developed
long
initial-link
pauses
and
variable
responding
(cf.
standard
deviations
for
Pigeon
10
in
Figure
3).
All
key
failures
during
these
procedures
occurred
only
on
terminal-link
keys.
The
key
failures
were
detected
through
la-
tencies
from
the
end
of
the
terminal-link
Fl
to
the
reinforced
green-key
peck;
these
times
were
consistently
short
(mean
values
less
than
1
sec)
and
did
not
differ
systematically
across
free
and
forced
choice.
The
following
sum-
marizes
the
sessions
with
such
failures
(session
numbers
correspond
to
those
in
Figure
3):
Pigeon
3,
Sessions
110,
126,
206,
and
217;
Pi-
geon
10,
Sessions
122
and
217;
Pigeon
18,
Ses-
sions
171
and
217;
Pigeon
23,
Sessions
110,
122,
and
230.
The
key
failure
in
Session
217
was
an
intermittency
inadvertently
created
during
apparatus
testing
after
the
procedure
change
in
Session
216.
Changes
in
terminal
links
nevertheless
had
rapid
effects;
for
all
pigeons
the
mean
values
of
data
based
on
Sessions
2
to
6
of
each
condition
typically
dif-
fered
by
less
than
one
standard
deviation
from
those
based
on
the
last
five
sessions
of
that
condition
(cf.
Figure
3).
RESULTS
Data
from
the
sessions
in
which
terminal
links
were
increased
from
FI
10-sec
to
Fl
30-
sec
and
session
and
reinforcer
durations
were
adjusted
are
shown
in
Figure
2.
The
x-axis
is
scaled
so
that
a
shift
to
the
left
corresponds
to
increasing
preference
for
the
left
terminal
link
and
a
shift
to
the
right
corresponds
to
increasing
preference
for
the
right
terminal
link.
The
data
are
shown
as
filled
apex-left
triangles
for
free
choice
in
left
terminal
links
and
unfilled
apex-right
triangles
for
free
choice
79
A.
C.
CATANIA
and
T.
SAGVOLDEN
181
C)
1-8F
L/)
z
-32
,
-41
,
)
Lu...47
LI)
I.-
,
-841-
-1021-
I4
3G1R
vs
1G3R
0.7
0.5
D
1G3R
vs
3G1R
0.5
0.5
RELATIVE
INITIAL-LINK
RATE
L
Fig.
2.
Relative
initial-link
response
rates
during
preliminary
conditions.
Arithmetic
means
over
the
last
five
sessions
with
free
choice
in
left
terminal
links
are
shown
by
filled
apex-left
triangles;
those
with
free
choice
in
right
terminal
links
are
shown
by
unfilled
apex-right
triangles.
The
x-axis
scale
is
arranged
so
that
a
shift
to
the
left
corresponds
to
an
increasing
preference
for
left
terminal
links.
Terminal
links
were
lengthened
from
FI
10-sec
through
Fl
20-sec
to
Fl
30-sec
over
successive
pairs
of
conditions,
during
which
reinforcer
and
session
durations
were
also
adjusted.
Pigeon
18
missed
several
sessions
of
the
first
condition.
Over
these
conditions
mean
initial-
link
response
rates
(resp/min,
both
keys)
and
changeover
rates
(changeovers/min,
one
direction)
for
each
pigeon
were,
respectively:
Pigeon
3,
37.6/5.8;
Pigeon
10,
39.7/6.1;
Pigeon
18,
30.2/5.3;
and
Pigeon
23,
40.5/5.7.
3G1R-
three
green
keys
and
one
red
key
(free
choice);
IG3R-one
green
key
and
three
red
keys
(forced
choice);
SD-
standard
deviation.
in
right
terminal
links.
Thus,
shifts
in
relative
initial-link
response
rate
that
follow
the
di-
rections
in
which
the
triangles
point
corre-
spond
to
changes
in
preference
that
follow
free
choice.
The
data
suggest
a
baseline
prefer-
ence
(equal
terminal
links)
of
about
.6
for
Pigeon
3,
about
.5
for
Pigeons
10
and
18,
and
about
.45
for
Pigeon
23
(cf.
Catania,
1975,
Figure
4).
The
evidence
for
free-choice
prefer-
ence
is
that,
except
for
the
early
conditions
for
Pigeon
18,
each
shift
to
left
or
right
of
the
free-choice
terminal
link
produced
a
cor-
responding
shift
in
preference
(seventeen
out
of
the
nineteen
possible
cases
in
Figure
2).
For
Pigeons
3
and
23
the
magnitude
of
this
shift
increased
over
successive
conditions.
For
Pi-
geon
10
it
decreased;
over
these
sessions
this
pigeon's
initial-
and
terminal-link
response
rates
decreased
and
variability
increased
(data
in
Figure
2
are
based
on
the
last five
sessions
of
each
condition
only).
For
Pigeon
18
the
shift
in
preference
emerged
only
in
the
last
two
conditions
and
was
small
relative
to
the
variability
of
relative
initial-link
response
rates;
for
this
pigeon
initial-
and
terminal-link
response
rates
increased
and
variability
de-
creased
over
these
sessions.
Figure
3
shows
relative
initial-link
response
rates
from
the
sequence
of
conditions
with
varied
or
fixed
positions
of
the
green
and
red
terminal-link
keys.
For
all
four
pigeons
pref-
erences
shifted
from
right
to
left
when
free
choice
was
switched
from
right
to
left
terminal
links
(after
Sessions
116
and
216),
and
from
left
to
right
when
free
choice
was
switched
from
left
to
right
terminal
links
(after
Session
159).
Compared
to
this
main
effect,
manipulat-
ing
the
position
of
the
odd
terminal-link
keys
had
relatively
small
effects
that
were
incon-
sistent
from
pigeon
to
pigeon.
The
two
largest
effects
(at
Sessions
143
through
147
for
Pigeon
18
and
Sessions
201
through
205
for
Pigeon
0.5
0.3
9
I
I
I a
H
I
I I
0
I
I
I
I)
I
I
I
80
I
I
-
-
9
I
v
-
-
I
9
9
-
-
I
5
I
I
I
I
FREE-CHOICE
PREFERENCE
3
10
-hm-->
4
lll.
e--l'
4
--4
"I..
CHOICE:
LEFT
RIGHT
i
D.
DO
.7
.6
.5
18
23
,-
,
,
IN
.K
POSITION:
VARIED
AP.
FIXED
*OEO
A
X
a
a
a
X
A
.5
.5
Meanl
__
(N=4)
|,D
~~~~~~
.1.
4
W410,
FORCED:
HIGHRATE
KEY
*
0
LOW-RATE
KEY
*
o
a
a
N
.
I
.k
A
.
a
.5
.6
.5
A
RELATIVE
INITIAL-LINK
RATE
(iL
)
Fig.
3.
Relative
initial-link
response
rates
over
the
last
five
sessions
of
each
condition,
represented
as
in
Fig.
2.
Standard
deviations
(SD)
are
shown
only
in
one
direction
(toward
baseline)
to
reduce
crowding.
Free-choice
ter-
minal
links
included
three
green
keys
and
one
red
key
(3G1R);
forced-choice
links
included
one
green
and
three
red
keys
(IG3R).
The
positions
of
red
and
green
in
successive
terminal
links
were
either
varied
or
fixed.
When
fixed,
the
odd
key
was
Key
1
with
forced-choice
responding
restricted
to
a
formerly
high-rate
key,
or
Key
3
with
forced-
choice
responding
restricted
to
a
formerly
low-rate
key.
During
these
sessions
mean
initial-link
response
rates
(resp/min,
both
keys)
and
changeover
rates
(changeovers/min,
one
direction)
for
each
pigeon
were
respectively:
Pigeon
3,
63.3/11.5;
Pigeon
10,
19.4/2.9;
Pigeon
18,
45.5/8.8;
and
Pigeon
23,
39.8/6.7.
23)
occurred
with
responding
forced
to
the
high-rate
green
key,
but
they
were
in
opposite
directions;
the
magnitude
of
free-choice
pref-
erence
decreased
for
Pigeon
18
and
increased
for
Pigeon
23.
No
consistent
effects
of
key
position
are
evident
in
the
mean
data
on
the
right
in
Figure
3.
The
selection
of
the
fixed
positions
of
the
odd
terminal-link
keys
was
based
on
the
termi-
nal-link
performances
summarized
in
Table
1.
In
general,
differences
in
responding
as
a
function
of
position
were
more
evident
in
free
than
in
forced choice
and
depended
primarily
on
the
relative
frequencies
with
which
pigeons
initiated
pecking
on
different
keys
when
they
were
concurrently
available;
occasional
change-
overs
among
green
keys
occurred
in
free
choice,
but
responding
was
more
often
restricted
to
a
single
key during
any
one
presentation
of
a
terminal
link.
Across
the
six
conditions
with
green
on
Key
1
in
free
choice,
that
position
maintained
the
highest
response
rate
for
Pi-
geon
3
and
was
consistently
one
of
the
two
highest-rate
positions
for
the
remaining
pi-
geons
except
for
the
last
condition
with
Pi-
geon
23.
Across
the
six
conditions
with
green
on
Key
3
in
free
choice,
that
position
was
consistently
one
of
the
two
lowest-rate
posi-
tions
for
all
pigeons,
again
except
for
the
last
condition
for
Pigeon
23.
Thus,
with
positions
fixed
so
that
Key
1
was
always
red
in
free
choice
and
green
in
forced
choice,
the
high-
rate
green-key
position
was
restricted
to
forced
choice;
with
positions
fixed
so
that
Key
3
was
always
red
in
free
choice
and
green
in
forced
choice,
the
high-rate
green-key
position
was
restricted
to
free
choice,
and
a
key
position
that
might
have
been
correlated
with
a
lower
109-116[
-140
-147
V)-153
0-159
L/)
)
-183
LI)
-194
-205
-216
-240
81
A.
C.
CATANIA
and
T.
SAGVOLDEN
Table
1
Percentage
of
total
green-key
(FI)
pecks
at
each
location
in
each
terminal
link.
For
both
free-choice-
left
and
free-choice-right,
arithmetic
means
over
the
last
five
sessions
are
shown
for
the
first
three
conditions,
and
arithmetic
means
over
the
first
five
sessions
are
shown
for
the
fourth
condition.
R
-
red,
G
-
green.
Pigeon:
3
10
18
23
Keys:
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
FREE-CHOICE
TERMINAL
LINKS
Free-choice-lk
Varied
position,
3GlR-lG3R
39
25
19
18
50
24
12
4
42
26
31
2
35
35
13
17
Fixed
position,
RGGG-GRRR
-
81
3
16
-
72
3
25
-
98
2
0
-
53
12
35
Fixed
position,
GGRG-RRGR
70
9
-
21
70 6
-
24
60
15
-
25
37
55
-
8
Varied
position,
3GlR-lG3R
61
24
10
4
56
29
7
9
48
23 25
4
32
37
19
12
Free-choice-right
Varied
position,
lG3R-3GlR
46
30
11
13
39
44
11
5
35 35
18
13
36 36
17
12
Fixed
position,
RRGR-GGRG
56
24
-
19
72
16
-
13
88
5
-
7
24 64
-
12
Fixed
position,
GRRR-RGGG
-
96
3
1
-
82
6
16
-
81
6
13
-
66
16
17
Varied
position,
lG3R-3GlR
51
30
13
5
34
35 26
5
60
24
16
1
24
33
31
13
FORCED-CHOICE
TERMINAL
LINKS
Free-choice-lke
Varied
position,
3GlR-lG3R
28
24
27
22
23
23
29
24
25
26
29 20
30
27
22 22
Fixed
position,
RGGG-GRRR
100
- -
-
100
- -
-
100
-
- -
100
---
Fixed
position,
GGRG-RRGR
- -
100
-
-
-
100
-
-
-
100
-
- -
100
-
Varied
position,
3GlR-lG3R
33
34
35
18
25
26
25
24
30 38
18
24
31
25
22
22
Free-choice-right
Varied
position,
lG3R-3GlR
22
26
21
31
21
32
21
27
16
25
24
35
24
26
17
32
Fixed
position,
RRGR-GGRG
- -
100
-
- -
100
- -
-
100
-
-
-
100
-
Fixed
position,
GRRR-RGGG
100
-
-
-
100
-
-
-
100
-
-
-
100
---
Varied
position,
lG3R-3GlR
29
21
23 27 25
30
23
22
23
25
28
24 26
27
23
24
Note:
In
the
respective
free-choice
and
forced-choice
terminal
links,
overall
green-key
response
rates
(resp/min)
were:
Pigeon
3,
38.6
and
42.5;
Pigeon
10,
64.8
and
66.3;
Pigeon
18,
20.4
and
17.1;
and
Pigeon
23,
72.9
and
82.0.
ratio
of
responses
to
reinforcers
by
virtue
of
its
lower
response
rate
was
restricted
to
forced
choice.
Another
property
of
terminal-link
respond-
ing
perhaps
relevant
to
preference
was
stim-
ulus
control
by
the
green
and
red
keys.
Only
one
red
key
was
present
during
free
choice,
whereas
three
red
keys
were
present
during
forced
choice.
In
the
absence
of
stimulus
con-
trol,
the
probability
of
pecking
any
red
(EXT)
key
was
.75
in
forced
choice
but
only
.25
in
free
choice.
Red-key
responding
is
summarized
in
Table
2,
which
shows
that
under
most
con-
ditions
less
red-key
pecking
occurred
in
free
choice
(one
red
key)
than
in
forced
choice
(three
red
keys).
In
seven
instances,
however,
Table
2.
Rate
of
pecking
(resp/min)
on
red
(EXT)
keys
in
free-choice
and
forced-choice
terminal
links.
Data
are
arithmetic
means
over
the
last
five
sessions
of
each
condition.
R
red;
G
-
green.
Pigeon:
3
10
18
23
Terminal-link
choices:.
Free
Forced
Free
Forced
Free
Forced
Free
Forced
FREE-CHOICE-L.aT
Varied
position,
3GlR-lG3R
.01
.06
.09
1.13
.02 .12
.01
.03
Fixed
position,
RGGG-GRRR
.02
.02
.46
.24
.11
.04
.08
.00
Fixed
position,
GGRG-RRGR
.00
.17
.00
2.25
.00 .00 .00
.07
Varied
position,
3GlR-1G3R
.01
.10
.34
2.36
.00
.01
.00 .04
FREE-CHOICE-RIGHT
Varied
position,
lG3R-3GlR
.03
.12
.26
1.09
.00 .00
.30
.03
Fixed
position,
RRGR-GGRG
.03
.12
.03
2.93
.07
.00
.10
.45
Fixed
position,
GRRR-RGGG
.00
.10
.14
.10
.00
.01
.22
.03
Varied
position,
lG3R-3GlR
.00
.14
.16
.46
.00
.00
.02
.14
82
FREE-CHOICE
PREFERENCE
I
LL
z
-J
z
LUJ
LU
p
p
A
A
I
a
A
a
p
a
I
0O
Q4
0
*
*
I
0.4
0
~~~~~
~~~0.4
0
0.4
0.8
RELATIVE
TERMINAL-LINK
RATE
(i)
Fig.
4.
Scatterplot
of
relative
initial-link
and
terminal-link
response
rates,
from
sequences
of
sessions
correspond-
ing
to
Figures
2
and
3.
Diagonals
show
equal
relative
initial-link
and
terminal-link
rates.
this
relation
was
reversed;
five
of
these
rever-
sals
occurred
when
the
red
key
was
fixed
at
the
high-rate
(Key
1)
position
in
free
choice.
These
reversals
were
not
systematically
related
to
changes
in
preference.
A
scatterplot
of
initial-link
and
terminal-
link
relative
response
rates
is
shown
in
Figure
4.
No
systematic
relation
is
evident,
although
these
two
variables
varied
only
over
relatively
limited
ranges.
The
findings
are
consistent
with
those
obtained
in
other
research
(Cata-
nia,
1975,
Figure
9;
in
press,
Figure
6).
DISCUSSION
This
study
systematically
replicated
pigeons'
preferences
for
free
over
forced
choice
within
concurrent-chain
schedules.
The
preference
was
obtained
with
terminal-link
stimuli
that
were
equivalent
in
number,
variety,
and
bits
of
information.
Furthermore,
it
was
not
highly
correlated
with
terminal-link
position
prefer-
ences,
stimulus
control
by
terminal-link
stim-
uli,
or
relative
rates
of
terminal-link
respond-
ing.
The
finding
that
initial-link
preferences
83
A.
C.
CATANIA
and
T.
SAGVOLDEN
were
not
systematically
related
to
terminal-link
response
rates
(Figure
4)
is
consistent
with
other
studies
of
determinants
of
preference
in
concurrent
chains
(e.g.,
Moore
&
Fantino,
1975;
Neuringer,
1969).
The
present
study
used
concurrent-chain
schedules
to
examine
preference.
These
pro-
cedures
separate
preference
for
different
con-
ditions
from
the
contingencies
that
maintain
responding
in
those
conditions.
In
these
con-
current
chains,
preferences
are
observed
in
ini-
tial
links
whereas
the
contingencies
operate
in
terminal
links.
In
simple
concurrent
schedules,
however,
relative
rates
are
not
necessarily
equivalent
to
preferences.
For
example,
if
variable-ratio
(VR)
reinforcement
maintained
higher
response
rates
than
the
concurrent
differential
reinforcement
of
low
rate
(DRL),
it
would
be
inappropriate
to
assume
a
prefer-
ence
for
VR
over
DRL
responding.
Although
pairs
of
concurrent
schedules
can
be
arranged
concurrently
(e.g.,
as
in
Menlove,
Moffitt,
&
Shimp,
1973),
the
contingencies
controlling
separate
responses
and
the
changeovers
among
them
may
override
the
relative
magnitudes
of
the
reinforcers
correlated
with
each
schedule.
A
relevant
example
is
provided
by
Leigland
(1979).
Pigeons
could
change
over
between
a
single
lit
key
and
two
lit
keys
by
pecking
on
a
third
(changeover)
key.
Independent
con-
current
VI
schedules
of
food
reinforcement
operated
for
pecks
on
the
single
lit
key
and
for
pecks
on
the
two
lit
keys.
Responses
that
occurred
within
2
sec
of
a
changeover
from
one
schedule
to
the
other
could
not
be
rein-
forced
(changeover
delay
or
COD
2-sec:
Herrn-
stein,
1961a).
A
critical
comparison,
examined
between
groups
of
pigeons
rather
than
within
individual
pigeons,
was
between
two
con-
tingencies
arranged
for
pecks
on
the
two
lit
keys.
In
one
case
a
peck
on
either
key
was
eligible
to
produce
any
scheduled
VI
rein-
forcers.
In
the
other
a
peck
on
only
one
of
the
two
keys
was
eligible
to
produce
scheduled
re-
inforcers;
the
eligible
key
varied
from
one
reinforcer
to
the
next,
but
no
discriminative
stimuli
were
correlated
with
its
position.
Consider
now
how
these
contingencies
might
make
the
relative
rate
of
pecking
on
the
two
keys
higher
when
reinforcers
are
scheduled
for
only
one
of
them
than
when
they
are
scheduled
for
either.
Suppose
responding
on
one
of
the
keys
continues
past
the
changeover
delay
without
producing
a
reinforcer.
In
the
first
procedure
this
occurs
either
when
no
re-
inforcer
has
yet
been
scheduled
or
when
a
reinforcer
already
scheduled
has
been
as-
signed
to
the
other
key.
In
the
second
pro-
cedure
this
occurs
only
when
no
reinforcer
has
yet
been
scheduled,
because
a
peck
on
either
key
produces
any
scheduled
reinforcers.
Thus,
the
former
contingencies
are
likely
to
gener-
ate
a
few
pecks
on
the
other
key
before
a
changeover
to
the
single-key
schedule,
whereas
an
immediate
changeover
to
the
single-key
schedule
will
be
favored
by
the
latter
contin-
gencies.
Those
few
additional
pecks
on
the
other
key
would
make
relative
response
rates
maintained
by
the
two-key
schedule
higher
in
the
first
than
in
the
second
procedure.
These
properties
of
the
Leigland
experiment
were
not
subjected
to
an
experimental
analysis,
but
it
is
sufficient
to
note
the
different
contingen-
cies
on
changeovers
in
the
two
procedures.
The
advantage
of
concurrent-chain
schedules
is
that
they
separate
preference
from
the
con-
tingencies
that
distribute
responses
among
the
alternatives.
Either
or
both
of
Leigland's
two-key
pro-
cedures
might
be
regarded
as
free-choice
con-
ditions.
The
question
is
an
experimental
one.
Once
free-choice
preference
has
been
demon-
strated,
further
research
may
be
devoted
to
refining
its
definition
by
exploring
the
bound-
aries
of
the
preference.
For
example,
prior
research
showed
that
EXT
keys,
whether
lit
or
dark,
do
not
constitute
choices
(else
every
condition
in
the
present
apparatus
would
in-
volve
six
free-choice
alternatives).
The
present
research
does
not
distinguish
free
choice
as
the
availability
of
independent
discriminated
operants
from
free
choice
as
the
opportunity
for
more
variable
responding
(e.g.,
Eckerman
&
Lanson,
1969;
Herrnstein,
1961b;
Schoen-
feld,
Harris,
ge
Farmer,
1966).
Would
a
three-
fold
increase
in
the
area
of
a
green
key
be
equivalent
to
an
increase
from
one
to
three
green
keys?
The
preference
for
free
choice
over
forced
choice
may
have
ontogenic
or
phylogenic
ori-
gins
(cf.
Skinner,
1966,
1975).
One
difference
here
between
free
and
forced
choice
was
that
there
were
other
green
keys
to
peck
if
a
green
key
became
inoperable
in
free
choice,
whereas
a
green-key
failure
in
forced
choice
left
the
pigeon
with
no
other
green
key
to
switch
to.
84
FREE-CHOICE
PREFERENCE
85
Free-choice
preferences
were
obtained,
how-
ever,
even
before
key
failures
exposed
the
pi-
geons
to
these
different
free-choice
and
forced-
choice
contingencies,
and
those
key
failures
that
did
occur
were
not
followed
by
systematic
changes
in
free-choice
preference.
If
preferences
among
several
alternatives
(e.g.,
those
for
particular
key
positions)
change
from
time
to
time,
a
free-choice
preference
may
develop
as
the
organism
learns
that
mo-
mentarily
preferred
alternatives
are
more
likely
to
be
available
in
free
choice
than
in
forced
choice.
But
opposing
position
preference
to
free-choice
preference
by
fixing
the
positions
of
terminal-link
keys
in
the
present
study
did
not
have
much
effect.
It
might
be
argued
that
this
manipulation
would
have
been
effective
if
it
had
been
maintained
for
an
extended
number
of
sessions.
Even
if
that
happened,
however,
it
would
be
inappropriate
to
attrib-
ute
a
preference
that
develops
rapidly
to
a
different
preference
that
develops
only
slowly.
Whatever
the
status
of
these
arguments
for
an
ontogenic
origin
of
free-choice
preference,
the
possibility
of
a
phylogenic
origin
must
not
be
excluded.
Organisms
that
prefer
the
avail-
ability
of
different
responses
or
opportunities
for
more
variable
behavior
may
have
evolu-
tionary
advantages
(e.g.,
Kavanau,
1969,
p.
268;
Roeder,
1975;
Rozin
&
Kalat,
1971).
For
example,
given
that
food
supplies
sometimes
may
be
lost
to
competitors
or
may
disappear
in
other
ways,
an
organism
that
chooses
patches
of
the
environment
in
which
two
or
more
food
supplies
are
available
will
prob-
ably
have
a
survival
advantage
over
one
that
chooses
patches
of
the
environment
contain-
ing
only
a
single
food
supply.
A
phylogenic
basis
of
free-choice
preference
does
not
imply
that
it
cannot
be
overridden
by
other
vari-
ables.
For
example,
a
forced-choice
preference
could
be
established
by
increasing
forced-
choice
but
not
free-choice
reinforcers
or
by
punishing
free-choice
but
not
forced-choice
responding.
Such
procedures,
however,
would
probably
involve
artificial
contingencies,
and
it
is
unlikely
that
they
could
be
maintained
for
as
long
a
time
as
the
natural
contingencies
that
generated
free-choice
preference
in
the
first
place.
If
this
is
so,
then
even
if
the
free-
choice
preference
were
suppressed
from
time
to
time,
its
suppression
would
be
only
tem-
porary.
REFERENCES
Autor,
S.
M.
The
strength
of
conditioned
reinforcers
as
a
function
of
frequency
and
probability
of
rein-
forcement.
In
D.
P.
Hendry
(Ed.),
Conditioned
rein-
forcement.
Homewood,
Ill.:
Dorsey,
1969.
Berofsky,
B.
(Ed.)
Free
will
and
determinism.
New
York:
Harper
&
Row,
1966.
Catania,
A.
C.
Freedom
and
knowledge:
An
experi-
mental
analysis
of
preference
in
pigeons.
Journal
of
the
Experimental
Analysis
of
Behavior,
1975,
24,
89-106.
Catania,
A.
C.
Freedom
of
choice:
A
behavioral
analy-
sis.
In
G.
H.
Bower
(Ed.),
The
psychology
of
learn-
ing
and
motivation
(Vol.
14).
New
York:
Academic
Press,
in
press.
Catania,
A.
C.,
&
Reynolds,
G.
S.
A
quantitative
analy-
sis
of
the
responding
maintained
by
interval
sched-
ules
of
reinforcement.
Journal
of
the
Experimental
Analysis
of
Behavior,
1968,
11,
327-383.
Eckerman,
D.
A.,
&
Lanson,
R.
N.
Variability
of
re-
sponse
location
for
pigeons
responding
under
con-
tinuous
reinforcement,
intermittent
reinforcement,
and
extinction.
Journal
of
the
Experimental
Analysis
of
Behavior,
1969,
12,
73-80.
Enteman,
W.
F.
(Ed.),
The
problem
of
free
will.
New
York:
Charles
Scribner's
Sons,
1967.
Fisher,
M.
A.,
&
Catania,
A.
C.
Autoshaping:
Relation
of
feeder
color
to
choice
of
key
color.
Bulletin
of
the
Psychonomic
Society,
1977,
9,
439-442.
Herrnstein,
R.
J.
Relative
and
absolute
strength
of
re-
sponse
as
a
function
of
frequency
of
reinforcement.
Journal
of
the
Experimental
Analysis
of
Behavior,
1961,
4,
267-272.
(a)
Herrnstein,
R.
J.
Stereotypy
and
intermittent
rein-
forcement.
Science,
1961,
133,
2067-2069.
(b)
Herrnstein,
R.
J.
Secondary
reinforcement
and
rate
of
primary
reinforcement.
Journal
of
the
Experimental
Analysis
of
Behavior,
1964,
7,
27-36.
Kavanau,
J.
L.
Behavior
of
captive
white-footed
mice.
In
E.
P.
Willems
&
H.
L.
Rausch
(Eds.),
Naturalistic
viewpoints
in
psychological
research.
New
York:
Holt,
Rinehart
&
Winston,
1969.
Leigland,
S.
M.
Deviations
from
matching
as
a
mea-
sure
of
preference
for
alternatives
in
pigeons.
Jour-
nal
of
the
Experimental
Analysis
of
Behavior,
1979,
32,
1-13.
Menlove,
R.
L.,
Moffitt,
M.,
&
Shimp,
C.
P.
Choice
be-
tween
concurrent
schedules.
Journal
of
the
Experi-
mental
Analysis
of
Behavior,
1973,
19,
331-334.
Moore,
J.,
&
Fantino,
E.
Choice
and
response
contin-
gencies.
Journal
of
the
Experimental
Analysis
of
Behavior,
1975,
23,
339-347.
Neuringer,
A.
Delayed
reinforcement
versus
reinforce-
ment
after
a
fixed
interval.
Journal
of
the
Experi-
mental
Analysis
of
Behavior,
1969,
12,
375-383.
Roeder,
K.
D.
Neural
factors
and
evitability
in
insect
behavior.
Journal
of
Experimental
Zoology,
1975,
194,
75-88.
Rozin,
P.,
&
Kalat,
J.
W.
Specific
hungers
and
poison
avoidance
as
adaptive
specializations
of
learning.
Psy-
chological
Review,
1971,
78,
459-486.
Schoenfeld,
W.
N.,
Harris,
A.
H.,
&
Farmer,
J.
Condi-
86
A.
C.
CATANIA
and
T.
SAGVOLDEN
tioning
response
variability.
Psychological
Reports,
1966,
19,
551-557.
Skinner,
B.
F.
The
phylogeny
and
ontogeny
of
behav-
ior.
Science,
1966,
153,
1205-1213.
Skinner,
B.
F.
The
shaping
of
phylogenic
behavior.
Journal
of
the
Experimental
Analysis
of
Behavior,
1975,
24,
117-120.
Received
October
15,
1979
Final
acceptance
February
14,
1980