|
Dogged
Pursuit of Human Affection
Our choice of dogs is far from accidental. Dogs represent
a fascinating model of a computational system.
|
Our
choice of dogs is far from accidental. Dogs represent a fascinating
model of a computational system. While perhaps not possessing
the full cognitive richness that we associate with humans,
dogs nonetheless do a remarkable job of learning what they
ought to learn and doing what they ought to do so as to exploit
the hugely successful adaptive niche of "man's best friend."
Not only do dogs exploit this niche, but also they do so in
such a way that we often find ourselves seeing the very best
of human qualities in the behavior of dogs. And they do this
despite having virtually no understanding of human language
beyond the use of words as cues, little apparent ability to
learn more than proximate causality, and our limited ability
to understand their internal state.
 |
 |
|
These
animated wolves have a social hierarchy similar to that
real wolf packs.
|
|
To
put their success in perspective, there are perhaps 400 million
dogs in the world (Coppinger and Coppinger 2001). By contrast,
there are only 400 thousand wolves despite their having a
20% greater brain-to-body-weight ratio than dogs. While many
of these 400 million dogs undoubtedly fend for themselves,
the fact remains that humans devote considerable resources
to the welfare of dogs. Americans alone spend over $30 billion
a year on pet food, supplies, veterinary care and other services.
Yet, for all of the cost, one third of dog-owners consider
themselves closer to their dog than to any family member,
and the most popular place for a pet dog to sleep is on its
master's bed. While these statistics may say as much about
us as they do about dogs, the fact remains that dogs have
hit on a highly successful evolutionary strategy.
One third of dog-owners consider themselves closer
to their dog than to any family member.
|
One
element of dogs' success is their ability to fit into the
social structure of a human household, and use it to their
advantage. The social learning of which dogs are capable may
not be as sophisticated as that observed in wolf packs, but
it is sufficient for their needs. Through their interactions
with members of their human family, they appear to learn their
relative "place" in the social order with respect to access
to resources. Implicit here is the seeming ability to associate
relevant qualities with different individuals (e.g., this
person feeds me, this person doesn't.) They also act as if
they learn contexts that signal an important interaction with
their human companions (e.g., supper-time, getting ready to
go for a walk, the human arriving home at the end of the day,
etc.) Similarly, they act as if they pick up on cues that
signal our internal state and respond appropriately. Stories
abound of the dog that "knows when its master is sad, and
comes over and lies down at the master's feet as if to provide
comfort."
 |
|
|
Alan howls to train his virtual wolf.
|
|
Much
of this social learning rests on their ability to learn apparent
proximate causality, especially when the immediate consequences
of their actions are motivationally significant to them. Indeed,
this is the basis of our ability to train dogs with comparative
ease (although we may be overstating the ease given the fact
that there are almost 600 books available from Amazon.com
on dog training.) To put this in perspective, a dog can be
taught to rollover on command in less than 100 repetitions
(Pryor 1999, Wilkes 1995). This is no mean feat given that
this requires them to perform some level of motor learning,
to associate the action of rolling-over with the subsequent
appearance of a treat, and to learn that the association is
only valid in the context of a unique acoustic or gestural
pattern. In the case of aversive stimuli, dogs are sometimes
capable of learning from a single example.
Finally, part of the dog's success in fitting into our social
structure is undoubtedly due to the tendency of dogs to provide
consistent and seemingly easy-to-interpret cues (e.g., ear
position, posture, tail movement and position and vocalizations
such as whines and growls) that help us explain and predict
their behavior. Not only do such cues suggest an internal
state that is explicable in human terms, but they also have
the effect of eliciting strong and perhaps innate responses
in us.
-
- - - - - - - - - - -
3
pages: | 1 | 2 | 3
|
|
