Yawning,
scratching, and protruded lips : differential
conditionability of natural acts in Pigtail
Monkeys
(Macaca
nemestrina)
Moutou Louboungou, James R Anderson
Most studies of learning abilities in
nonhurnan primates involve operant procedures in
which the subject executes motor acts on the
environement in order to obtain rewards.
Typically, the required response is biologically
neutral. For example, selectively displacing
only one of three geometric objects, or touching
one of an array of previously illuminated panels
do not form part of the animal's everyday
behavioral repertoire. Attempts at operant
conditioning of behavioral patterns found in the
natural repertoire are much more rare, although
a number of such studies using birds, fish and
rodents have been reported. Experiments aimed at
conditioning natural patterns are important for
testing the generality of laws of learning
formulated on the basis of traditional
conditioning paradigms and may be of interest
for the understanding of neuroanatomical
substrates of species typical behavior.
The present paper describes three experiments
in which we attempted to control the perfomance
of three naturally occurring behavioral patterns
in macaque rnonkeys, using food as a reinforcer.
The patterns studied were scratching. the
"protruded lips" facial expression, and yawning.
With regard to scratching. we attempted to
replicate and extend the findings from the only
previous study on operant conditioning of
scratching in a monkey. A facial expression was
studied to examine whether this type of
communicative act could be readily brought under
voluntary control. The classic view of nonhuman
primate facial expressions is that they are
directly linked to the individual's emotional
stat. The question posed in the present study
was whether facial expressions were "hard-wired"
responses quite resistant to external schedules
of reinforcement, or whether they ressemble
vocal responses. which can be operantly
conditioned using food reinforcement. Yawning
was included as a hypothesized intermediate
pattern in the event that we succeeded in
conditioning scratching but not the facial
expression.
Experiment 1 : scratching
Macaques scratch themselves with their hands
(the fingertips nails are rapidly and repeatedly
drawn across the fur at a particular body
location) or, less frequently, with their feet
(like a dog). While it naturraly occurs as a
response to itching, scratching also has
characteristics of "a displacement activity"
appearing in situations characterized by
frustration or tension. Adult, group-living
rhesus monkeys scratch more often in social
contextss than when alone, and scratch often in
close temporal association with a change of
belia\ior in social contexts. Recently Iversen
et al. ( 1984. Experiment 3) reported obtaining
a high rate and impressive variety of scratching
response in an adult male vervet monkey
(Cercopithecus aethiops) when the presentation
of food was made contingent upon scratching. In
the present experiment, we used a different
species of Old World monkey in a further
assessement of' the conditionability of
scratching.
The subject was a young adult male pigtail
macaque (Macaca nemestrina) ("Paul") living in a
large, indoor-outdoor cage with an adult female
and an infant conspecific. The latter two were
socially subordinate to Paul and did not
interfere with the experiment. Food (dried
pellets) and water were available ad libitum.
Paul had been a subject in one previous
experiment. inolving finding hidden food. but he
was otherwise experimentally naive.
Experimental sessions lasted 1 hr and were
conducted daily. Two baseline phases, each
consisting of three sessions, preceded the first
conditioning phase. During the first baseline
phase the experimenter simply sat in front of
the subject'scage and recorded the spontaneous
frequency of scratching. The same experimenter
carried out all of the tests reported here.
During the second baseline phase the
experimenter presented 20 thin slices of fresh
banana to the subject at regular intervals
during each session, independent of the
subject's behavior. Conditioning Phase 1 (four
sessions) consisted of saying the subject's name
and presenting a banana slice contingent upon a
bout of scratching by the subject. Continuous
reinforcement was used and the reward was given
by hand. A criterion of at least four scratching
movements per bout was set for an acceptable
scratching response, and all responses were
allowed to continue until they stopped
spontaneously. A written record was made of
every response.
Following, the establishment of the desired
response in Phase 1, two sessions were conducted
in wich reinforcement was made contingent upon
the subject changing the form of the response
from the preceding one. For example, if Paul
scratched his back with his right hand an
immediate repetition of this act went
unrewarded, whereas any other form of scratching
was f'ollowed by the presentation of another
banana slice.
Phase 3 involved rewarding only one
particular type of scratching. namely with the
foot. Following some unsuccessful conditioning
procedures not reported here, a final,
extinction phase was conducted. This consisted
of three sessions during Ahich no rewards were
given to the subject.
Results : The major results are
illustrated in Figure 1. Scratching was quickly
and reliably brought under the control of the
continuous reinforcement regime. During baseline
sessions, Paul scratched himself between 19 and
27 tinies per hour, and there was virtually no
effect of giving him food independent of his
behavior. In contrast, in the first session of
conditionning Phase 1, the rate of scratching
rose to over 100 responses per hour. Response
rates remained high throughout this conditioning
phase, peaking at 135 in session 3.
Phase 2, in which the subject was required to
change the form of scratching from one response
to the next, began with an increase in the total
number of scratching bouts performed but a drop
in the number of rewards obtained. However, even
in the first session of this new phase Paul
obtained about four times more rewards (81) as
there were spontaneous bouts of scratching
during baseline sessions. The second session of
Phase 2 produced a substantial increase in the
number of rewards obtained, to 113. During
conditioning Phase 1, Paul had developed a
preference for scratching his right side or his
back with the right hand, and in the fast two
sessions of this "freestyle" phase 87% and 83%
of all responses were repetitions of the
response type just performed. The percentage of
repetitions in the two sessions of Phase 2 was
lower: 56% and 53% respectively.
Paul's reward-obtention rate remained high
during the two sessions of Phase 3, in which
scratching with the feet was selectively
reinforced. During the last two session of the
"free.style" Phase 1, 35% and 26% of all
scratching, responses were performed with a
foot. The corresponding values for the two
sessions of Phase 3 were 62% and 66% showing a
clear increase in the occurrence of the rewarded
form namely scratching with the feet.
Figure 2 shows the course of' scratching
during the extinction sessions. Spontaneous
recovery was evident at the start of each
session, but performance became progressively
weaker, until in the final two sessions there
were no responses at all after 20 min.
Experiment 2 : protruded lips
There have been several attempts to operantly
condition vocalization in nonhuman primates.
Initial success was limited, but it is now well
established that monkeys can control their
vocalization rate in order to obtain food
rewards. We are not aware of any previous
attempts to operantly condition another major
type of communicative act, namely facial
expressions. One expression in the repertoire of
the pigtail macaque is the protruded lips face,
described by van Hoff (1967) and Redican (1975)
and studied in some detail by Christopher and
Gelini (1977). This expression is displayed
mainly by adult males, often towards females in
sexual contexts. However. it is also observed in
a variety of nonsexual contexts. and may be
directed towards nonconspecifics, including
humans: in such cases the displaying individual
appears aroused (e.g.. during geeting). We chose
to study this particular expression because it
is a clearly definable act which usually lasts
several seconds.
The subjects were two adult male Macaca
nemestrina (Bernt and Charlie) living in
individual cages rileasuring 60 80 120 cm. Both
were experimentailly naive, and they were tested
separately. They were in visual contact with
other macaques in the colony room but not with
each other. Food (dried pellets) and water were
available at libitum.
As in Experiment 1, two baseline phases each
consisting of 3, 1 hr- long sessions were
conducted. During the first baseline phase all
spontaneous occurrences of the protruded lips
face were recorded by the experimenter, who
remained passive. The expression was recorded
whether it was directed towards other monkeys or
towards the experimenter. In the second baseline
phase, 20 thin slices of banana were given to
the subject during each session, independent of
the subject's behavior.
The conditioning phase followed the final
baseline. Bernt (three sessions) and Charlie
(six sessions) were rewarded with a slice of'
banana every time they displayed a protruded
lips face, regardless of the target to which the
expression was directed. Charlie then received
five induction sessions in which the
experimenter deliberately elicited the response
by talking softly, imitating the expression, or
tickling Charlie's chin. This procedure was also
tried with Bernt but was soon stopped because
this subject became very excited and aggressive
towards neighboring monkeys.
Two final continuous reinforcement sessions
were conducted with Charlie, in which all
spontaneous instances of the response were
rewarded with banana slices.
Results : The procedures involving
food reward failed to produce any increase in
the spontaneous production of the protruded lips
face. Typically, fewer than ten responses were
emitted during any session in which the
expression was not deliberately elicited by the
experimenter, with or withtout food reward. A
total of 294 protruded lips faces were elicited
from Charlie when the experimenter sought to
provoke them and all of these were rewarded with
food, but a return to conditions in which only
spontaneous occurrences were followed by reward
reduced to the rate of emission to baseline
levels.
Experiment 3:
yawning
Yawning has a particularly interesting place
in the behavioral repertoire of primates. On the
one hand. it is thought to be an innate response
which increases the oxygen supply to the brain
in a drowsy individual, though the evidence on
this point is equivocal (Deputte 1978). However,
its frequent appearance in social situations,
and the distinction between "directed" and
"undirected" forms indicates a role in
communication describes yawning, as "halfway
between a reflex and an expressive movement..."
(p. 203), and goes on to point out that yawns
can be produce volontarily by humans. The aim of
experiment 3 was to determine whetether macaques
could volontarily produce vawning for food
rewards.
The subjects were Bernt and Charlie fom
expermient 2. The six baseline sessions and the
first three session from conditioning phase of
experiments 2 provided baseline rates of
yawning. Conditioning procedures then differed
for the two subjects. In Session 1 of the
conditionning pahse, Bernt trained to yawn
through successi\e approximations. Every full
yawning response was rewarded with a slice of
banana and at first any mouth opening movement
was also reinforced. Progressive approximations
to the full yawning response were then required,
and by the end of the first session simply
opening the mouth wide went unrewarded. Charlic
was rewarded from the start only for complete,
naturally occurring yawns. Each subject thus
received five sessions in which food reward
could be obtained by yawning. These were
followed by two extinction sessions in which no
rewards were given, then by another continuous
reinforcement session, and then finally by three
extinction. sessions.
Results: As can be seen in figure 4,
in both subjects yawning was rapidly brought
under control using food reward. Baseline rates
were between 7 and 18 per hour, rising to
between 60 and 70 (Charlie) and 80 to 150
(Bernt) during the conditioning phases.
Withdrawal of contingent food presentation
rapidly brought yawning rates down to baseline
levels. The resumption of of high rates of
yawning during the reconditioning phase
interposed between the two extinction phases
frrther indicates that the reinforcement
procedures were effective in controlling the
yawning response.
Discussion
The results of the present experiments
indicate differential conditionability of the
three natural responses studied, with scratching
and yawning being quickly brought under control
using food as a reinforcer, but the facial
expression "protruded lips" being resistant to
conditioning using equivalent operant
procedures. Some implications of these findings
are discussed below.
The subject rewarded with food for scratching
quickly learned to emit this response in order
to obtain the reward. He also frequently varied
the form of scratching, and then subsequently
increased the rate of a normally relatively
infrequent form of scratching, namely with the
feet, when these were required by the
reinforcement regime. These results broadly
agree with those of Iversen et al. (1984), who
established that scratching could be operantly
conditioned in vervet monkeys. The present
results also lend support to the view that
scratching is not a rigid, stimulus-bound reflex
act in macaques. The fact that scratching can
easily be controlled, and the finding that
macaques frequently scratch around the time of a
change in activity in social contexts suggest
that scratching might be incorporated into the
repertoire as a communicative gesture.
The attempt to operantly condition the
protruded lips facial expression failed. Three
possible reasons for this failure are worth
mentioning briefly. First, it is conceivable
that in macaques the controlling neurological
mechanisms for facial expression are
predominantly subcortical to the extent that
voluntary control is minimal or absent. However,
such an explanation seems unlikely, given the
notable successes in operant conditioning of
vocalizations despite the evidence for
overwhelming subcortical control of vocalization
in nonhuman primates. Furthermore, there exist
descriptions of behavior in macaques which
suggest the purposeful use of' facial
expressions to obtain a goal. Bertrand (1976)
reports that a hand-reared pigtail macaque used
the protruded lips face to request to be let out
to urinate, and Chevalier Skolnikoff (1982)
describes adjustments of facial communicatory
acts in stumptail macaques according to the
orientation and responses of the recipient. For
these reasons we favor the second possible
reason for our failure to obtain a conditioned
facial expression, namely that the
expression-food contingency used in the present
study was not conducive to learning. Clearly,
not all responses are equally susceptible to
food reinforcement and it is conceivable that
some form of social reinforcement would be more
effective than food in attempts of condition
facial expressions, as has been recognized in
studies with human infants. The third possible
reason for our failure is related: the state of
arousal accompanying the facial expression,
especially experimenter-induced episodes, might
have been intense enough to interfere with the
establishment of an association between the act
and the reward by the subject. In conclusion,
the difficulty in conditioning the protruded
lips face may well be attributable to some
deficiency in the procedure used for this act.
In any case, the experiment highlights the type
of problems to be overcome in attempts to use
conditioning paradigins to study the question of
voluntary control of communication in nonhuman
primates.
The most striking finding in the present
study was the increase in yawning by the two
macaques rewarded with food for each yawn. Both
subjects appeared to behave in a fashi'on
similar to that described by Barbizet - (1958)
for humans making themselves yawn, i.e.,
inhaling deeply with the mouth open. Both
subjects were also seen to abandon sortie
attempted yawns, suggesting a deliberate effort
to trigger the response.
It is possible that the high number of
responses by the subject Bernt includes a
proportion of "pseudo-yawns,"' similar to
responses described in dogs by Konorski (1967).
Pseudoyawns in dogs consisted of simple opening
of the mouth. After the initial shaping trials
with Bernt, however, acceptable responses were
only those which contained all of the observable
motoric components of a yawn, including closing
the eyes and tilting the head back at the climax
of the act. Further, in contrast to the
successive approximation procedure used with
Bernit rewards were given to the second subject
only for full yawns from the start of the
conditioning phase. An experienced, independent
observer judged these yawns to be natural in
appearance.
Along with observations of phenomena such as
scratching in association with changes of social
activity, directed yawning and the suppression
of yawning in an adult male baboon with broken
canines, the present results suggest that
monkeys might well be capable of controlling the
production of certain basically preprogrammed
responses for social goals. A combination of
naturalistic and experimental studies will be
the most effective way to investigate these
capacities.
« It is
ironic that testosterone "the male sex hormone,"
is more closely associated with the yawning rate
than with the mounting or intromitting rates
» Charles Phoenix
Sexual
steroids
exert several effects on both central
dopaminergic and oxytocinergic systems by acting
either at the genomic or membrane level
