Social learning and imitation in dogs (Canis familiaris)

Eötvös Loránd University Faculty of Science
Doctoral School of Biology, head of the school: Dr. Anna Erdei
Doctoral Program of Ethology, head of the program: Dr. Ádám Miklósi
Social learning and imitation in dogs
(Canis familiaris)
Thesis of Doctoral Dissertation
Claudia Fugazza
Supervisor: Ádám Miklósi
Eötvös Loránd University,
Department of Ethology
1117 Budapest, Pázmány Péter sétány 1/c
2014
1
Introduction
Dogs have undergone selection for living in human groups through domestication and these
changes contributed to form a species with surprisingly complex social skills. Thus they
represent particularly interesting subjects for the study of hetero-specific social learning abilities.
While imitation is usually studied between individuals of the same species, there is strong
evidence that dogs can learn socially from both con- and heterospecifics demonstrators (e.g.,
Kubinyi et al. 2003). Topál et al. (2006) used the Do as I do paradigm and discovered that dogs
are able to match functionally their behaviour to actions shown by a human demonstrator. With
the Do as I Do procedure, dogs first learn to match their behaviour to a small set of familiar
actions with operant conditioning methods and later are able to generalize the ‘copying rule’ to
other actions and demonstrators, thus they can be tested for imitation in various situations.
Previous studies on imitation in dogs did not include the two-action procedure, which allows
controlling for other social learning processes such as stimulus enhancement and goal emulation.
Deferred imitation is the ability to encode, retain and retrieve a memory of an action and then to
use it as the basis to reproduce the demonstrated action after a delay (Klein and Meltzoff 1999).
In humans this cognitive ability is considered a hallmark of mental representation as it indicates
the emergence of the infant’s ability to form a mental representation of the model’s behaviour at
the time of demonstration and recall of that image after a retention interval (Barr et al. 1996).
From a cognitive perspective, evidence for deferred imitation excludes alternative explanations
of behavioural similarity between demonstrator and observer where the demonstration triggers a
similar behaviour in the observer at the same time or shortly after it, such as contagion and
response facilitation (Bandura 1969).
It is known that human infants’ deferred imitation performance decreases if their recall is tested
in a different context, especially with increasing delays (e.g., Barnat et al. 1996). These results
2
suggest that the context serves as a retrieval cue that helps recalling the demonstration. The
contextual cues that can serve this function can be the respective positions of the objects in
space (spatial cues) and/or the features of the objects used during demonstration (figurative
cues).
Despite evidence that dogs are predisposed to acquire information socially from humans and are
even able to copy the actions of a human demonstrator, formal dog-training methods have
traditionally relied only on individual learning (e.g. Lindsay 2005) and the use of social learning
in dog training received little attention from researchers. However it is possible that the
involvement in training procedures of dogs’ predisposition to learn socially from humans
improves the current training methods.
General aims
The aims of the research presented in this thesis are twofold. First we used the Do as I do
method to shed light on dogs’ ability of deferred imitation and on its memory of different
elements of human actions by the use of modified versions of the Do as I do paradigm combined
with the two-action procedure. Second, from a more applied perspective, we assessed the
efficiency of the use of social learning in the field of dog training.
3
Part 1: Deferred imitation in domestic dogs
This study investigates on the ability of deferred imitation of novel actions in dogs after
retention intervals of 1.5 minutes and on their memory of familiar actions after intervals ranging
from 0.40 to 10 minutes. The subjects were trained using the Do as I do method to match their
own behaviour to actions displayed by a human demonstrator. They were then trained to wait for
a short interval to elapse before they were allowed to display the previously demonstrated action.
The dogs were then tested for deferred imitation of the demonstrated behaviour in various
conditions, including a modified version of the two-action procedure in which the same subjects
observed different actions displayed on the same object in different tests. A control for Clever
Hans effect and a condition without demonstration were also included. Dogs were typically able
to reproduce familiar actions after intervals as long as 10 minutes, even if distracted by different
activities during the retention interval and were able to match their behaviour to the
demonstration of a novel action after a delay of 1 minute. The performance of dogs in the twoaction procedure reveals that dogs are able to imitate the demonstrated actions after retention
intervals of 1.5 minutes, even if different actions are shown on one object. The ability to encode
an action and recall it after a delay implies the presence of the ability of deferred imitation in
dogs. They form a mental representation of the demonstration and use it as the basis to perform
a matching action at the time of recall, which is based on some form declarative memory.
4
Part 2: Dogs’ spatial bias affects their imitative performance in a Do as I do task
This study investigates on what information dogs encode and recall preferentially in a Do as I do
task with an interval between demonstration and recall (deferred imitation), between spatial
information (i.e., the location of demonstration) and figurative information (i.e., the object
manipulated by the demonstrator). The subjects were repeatedly tested in several conditions
with modified versions of the Do as I do paradigm. The position of target objects on which
actions were demonstrated were displaced following demonstration, so that, when the ‘Do it!’
command was given, dogs could only match either the original position of demonstration or the
object, but not both. We combined the Do as I do paradigm with the two-action procedure – i.e.,
two actions were shown on each object - to assess if, when the objects were displaced at the time
of retrieval, dogs could still recall and imitate the action. Dogs were able to match the action and
the object used by the demonstrator in conditions with only one target object displaced.
Nevertheless in conditions with two objects displaced, the majority of dogs matched only the
position (i.e. not the object nor the action).
Two-object conditions were repeated in which humans used ostensive communicative cues and
pointing gestures to draw the subjects’ attention to the object used by the demonstrator. In the latter
conditions dogs’ predisposition to follow human communicative gestures outweighed their spatial
bias, hence their object matching and imitative performance increased. Our results suggest that dogs
rely strongly on spatial information when imitating actions. Physical features of the object,
nevertheless, seem to be important retrieval cues that facilitate recalling the action.
5
Part 3: The efficiency of the Do as I do method and shaping / clicker training to train dogs
We compared the efficiency of the Do as I do training method, which relies on social learning,
with that of a training method that relies on individual learning (shaping/clicker training –
Skinner 1951) to teach dogs three different kinds of object-related actions. In order to control for
the comparability of the previous training experiences of our subjects, we tested experienced
dog-owner dyads that had previously achieved a certificate for either type of training (Do as I do
and shaping/clicker training). They were tested upon training three different novel actions:
simple, complex and sequences of two actions, in three separate sessions, using the training
method they were certified for. In each case the owners had 15 minutes for accomplishing the
task of training the dogs to perform the predetermined action. We used the latency of first
occurrence and the number of dyads that were successful within 15 minutes as measures of
training success. While we did not find a significant difference between the two training
methods with regard to simple actions, we found that subjects using the Do as I do method
outperformed those using shaping/clicker training in the case of complex actions and sequences
of two actions. This study is the first to formalize a method based on the Do as I do protocol for
training dogs and to assess its efficiency by comparing it with shaping/clicker training.
6
Conclusions
The presented research has lead to the following novel scientific discoveries:
•
Dogs possess the ability of deferred imitation. This evidence excludes alternative
explanations of behavioural similarity between demonstrator and observer in which the
demonstration triggers a similar behaviour almost simultaneously. Furthermore this
suggests that dogs form a mental representation of the demonstration and use it as the
basis to perform a similar behaviour after a delay.
•
The ability of dogs to recall a demonstration of an action after a delay without the
possibility to motor-practice on the action and to keep their mind active on the
demonstration by using contextual cues suggests that they may possess some form of
declarative memory.
•
In a deferred imitation task, when the contextual cues regarding spatial information (i.e.,
the location of demonstration) and figurative information (the features of the object
manipulated by the demonstrator) are conflicting at the time of recall, dogs show a
spatial bias that affects their imitative performance. However human ostensive cues can
outweigh dog’s preference to match the location of demonstration and restore their
ability to match the object and the action after a delay.
•
The Do as I Do method has also proven proficient in the applied field of dog training.
This method is more efficient for teaching dogs complex object-related actions,
compared to shaping/clicker training, a method that relies on individual learning.
7
References
Bandura A (1969) Social learning theory of identificatory processes. Goslin DA (ed)
Handbook of socialization theory and research. Rand-McNally, Chicago, pp. 213–262
Barnat SB, Klein PJ, Meltzoff AN (1996) Deferred imitation across changes in context and
object: memory and generalization in 14-month-old infants. Infant Behav Dev 19:241-151
Barr R, Dowden A, Hayne H (1996) Developmental changes in deferred imitation by 6- to 24month-old infants. Infant Behav Dev 19:159-170
Klein PJ, Meltzoff AN (1999) Long-term memory, forgetting and deferred imitation in 12month-old infants. Developmental Science 2:102-113
Kubinyi E, Miklósi Á, Topál J, Csányi V (2003) Dogs (Canis familiaris) learn from their
owners via observation in a manipulation task. J Comp Psychol 117:156–165
Lindsay, SR (2005) Handbook of Applied Dog Behavior and Training, Procedures and
Protocols. Ed. Blackwell, Oxford, pp. 13-27
Skinner, BF (1951) How to teach animals. Sci Am 185, 26-29
Topál J, Byrne R, Miklósi Á, Csányi V (2006) Reproducing human actions and action
sequences: “Do as I Do!” in a dog. Anim Cogn 9:355-367, DOI:10.1007/s10071-006-00516
Publications included in the dissertation
Fugazza C, Miklósi Á (2013) Deferred imitation and declarative memory in dogs. Animal
Cognition. DOI:10.1007/s10071-013-0656-5
Fugazza C, Miklósi Á (2014) Should old dog trainers learn new tricks? The efficiency of the
Do as I do method and shaping / clicker training method to train dogs. Appl Anim Behav
Sci. DOI 10.1016/j.applanim.2014.01.009
8
Fugazza, C., Pogány Á., Miklósi Á. (manuscript under preparation) Spatial bias affects
imitative performances of dogs in Do as I do tasks
Conference presentations / posters
Colloque SFECA (Societé Française pour L’Etude du Comportement Animal): L’imitation
différée chez le chien. C. Fugazza & A. Miklósi (oral presentation), 9-10 May 2012, Lyon
Canine Science Forum: Deferred imitation in dogs. C. Fugazza & A. Miklósi (poster), 25-27
July 2012, Barcelona
Conference SOFIVET: Deferred imitation in dogs. C. Fugazza & A. Miklósi (poster), 28
September 2012, Chioggia
International Congress IRSEA (Institut de Recherce en Sémiochimie et Ethologie Appliquée):
Deferred imitation in dogs. C. Fugazza & A. Miklósi (oral presentation), 15-16 November
2012, Apt
Ethology Symposium Wroclaw University: Social learning in dogs (plenary talk), 9 May 2013,
Wroclaw
CompCog Closing Conference: Old wine in new bottles: Does it matter how dogs learn?
(poster), 4-5 July 2013, Vienna
ConAmore: Comparative Perspectives on Autobiographical Memory: Do as I …did! Dogs’
memory and deferred imitation of human actions (poster) 18-19 June 2014, Aarhus
Canine Science Forum: The effectiveness of Do as I do and shaping for training dogs. C.
Fugazza & A. Miklósi (poster) 15-17 July 2014
Canine Science Forum: Do as I did! Dogs’ memory and deferred imitation of human actions.
C. Fugazza & A. Miklósi (oral presentation) 15-17 July 2014
9
Other publications
Fugazza C, Miklósi A (2014) Measuring the Behaviour of Dogs: An Ethological Approach. In
Horowitz A. Domestic Dog Cognition and Behaviour, The scientific study of Canis
familiaris. Springer Berlin Heidelberg, DOI 10.1007/978-3-642-53994-7_8
10
`