Even though we lost Trivers, we will have a great speaker and a super-interesting topic for next week - word-learning and cognition in dogs! I already sent the two papers that Andrew Olney wants you to read for his talk by e-mail and they are also on the UMdrive.
Yuna Jhang will present the Pillei & Reid paper and Chia-Cheng Lee will present the Kis et al Paper.
There are lots of video clips on the net that you can watch to see Chaser, the border collie at work - she is awesome!
This one is about the actual experimental conditions:
http://www.youtube.com/watch?
Have fun!
Uli
Although I haven’t have a dog before, I have heard lots of stories that dogs and humans can be very good friends because they seem to know you. Kis et al’s (2012) paper further confirms these previous stories that dogs are really sensitive to their caregivers’ cues. Like children learning joint attention (i.e., sharing attention, intention, and affect) to learn words, dogs also learn joint attention with caregivers to learn the tasks. What is different from infants and dogs is that dogs are so influenced by human cues so it is not truly a-not-b errors mentioned in the infant development literature (i.e., infants made such errors because of limited working memory or inhibitory control). Nevertheless, in Pilley and Reid’s (2011) paper, the dog Chase appears to know the real meanings of words without being interfered by human cues.
ReplyDeleteConnecting to the readings last week, Orwen, Amoss, and Rendall (2011) did not mention dogs’ behaviors and focused more on non-human primates, but at the end of the article, they discussed similar phenomena in captive animals. Some human-like vocalizations were heard from these captive animals but not those in the wild. After reading the two articles, my question is that why it is important in the topic of language evolution? Dogs’ behaviors in the two articles are definitely a kind of communication and they show behaviors similar to human infants. However, do dogs also play a role in the evolution of language? For example, in Orwen et al’s (in press) article, the authors mentioned smiles and laughs are good indicators of future cooperation in human relationships, so can we say that the intimate relationship between human and dogs is a kind of connection and thus cooperation in the course of evolution?
I've heard before that border collies have the same cognitive (if cognitive is the right word) ability as a 3-year-old child. I'm not sure if that's actually true or not, but from personal experience and also from what I read in these articles, I suppose it is possible.
ReplyDeleteI had a border collie for several years and she was certainly very smart and seemed to understand commands with a great deal more precision than other dogs I'd had before.
The first paper we read about the "A-not-B error in adult pet dogs" was interesting in that it showed how sensitive dogs are to ostensive cuing. One of the observations made in the paper about their sensitivity to such cuing is that it is motivated by the dogs' desire to please the person cuing, which is why they will follow the cue even when it is misguiding (p. 4).
The second paper about the border collie Chase, discussed the extent to which dogs (Chase in particular, in this case) have a "referential understanding of nouns". The relevancy of this study, especially in reference to the overall focus of the course, is that the referential understanding of nouns as was found in Chase was something that is normally attributed to children.
The Kis et al (2012) paper was quite interesting. The work of Piaget has been at the forefront of cognitive and developmental literature for several years. Piaget makes note that the primary reason why the A-not-B error is so robust in infants is because infants (typically 10 months are younger) are lacking the schema of object permanence--understanding that objects continue to exist even after they become hidden. Conversely, other theorists have also provided other explanations for this finding. More recent explanations have suggested that the infants are unable to inhibit a previously rewarded motor response, perhaps reflecting the immaturity of the prefrontal cortex, or that the error is due to limitations on working (short-term) memory. In addition, others have suggested that infants are unable to switch their attention from location A to location B, presumably also due to underdeveloped executive functioning in the prefrontal cortex. In retrospect, all these proposed theoretical implications are quite plausible; however, I wonder which theory can be upheld by the findings in Kis et al (2012). The findings in the paper are in line with the view that dogs’ perseverative search bias may stem from their propensity to follow human social cues rather than from inhibitory control problems or interference effects in working memory. I wonder if conducting a study with fMRI or other devices to capture neural activity in dogs would provide further support for this theory. However, keeping a dog still in a fMRI scanner would be quite a challenge.
ReplyDeleteI'm trying to wrap my head around the broader implications of these studies. So, dogs have a tendency to make A-not-B errors. This A-not-B error highlights canine tendency to make (unwittingly)sub-optimal decisions in order to maintain obedience towards their leader or owner, in this case. They are given a task, and instead of using their own previous knowledge as input for completing this task they rely on cues of their owners.
ReplyDeleteInfants make the same A-not-B errors. While infants are able to map new words onto objects per their parents' labeling, they continue to rely heavily on the cues of their caregivers, rather than prior knowledge.
So my question is, at what point do infants rely more heavily on prior knowledge for decision making than they do from cues of their care givers. When do infants, if at all, acquire skepticism?
Pilley
ReplyDeleteLike most dogs owners, I have used certain word combinations to train my dogs. With that in mind. I have often wondered how my dogs know what I want when I say it. As purely anecdotal evidence, I have experimented in similar ways by saying substituting gibberish words, but while using the same vocal inflection—figuring what I said was far less important than how I said it. Unsurprisingly, I found that my dogs would get excited by keys words like “Go get your toy” or “Let’s go for a walk.” When I substituted the fake words with similar inflections and sounds, they initially got equally excited. However, over time as they got more familiar with the fake words, they no longer reacted to the fake words but only with the real words. Little did I know that by using the command of “get your toy” with fake words, this was actually evidence of Goodwin’s exclusionary learning through inferential reasoning. Unlike Chaser’s trainer, however, I have not been able to teach my dogs to identify 1000 proper-noun objects. One aspect not discussed in the paper is whether or not they were able to determine if dogs, like humans, have a critical period in which they learn a huge part of their possible vocabulary and possibly level off in their learning. They did cite that after three years they no longer had the time to train Chaser and that there was no decrease in speed of learning, however, I wonder if there was any discernable critical period in place. Most surprising to me was that Chaser was able to learn the various nomenclature of the larger group of toys and the subgroup of the various types of toys themselves. As someone who has often discussed novelty in language acquisition, I was surprised that Chaser didn’t choose an unfamiliar item merely because of its novelty.
Kis
I experience the A-not-B error every time I look for my car keys; however, I did not realize my dogs were cursed with the same problem. I do not have many questions regarding these two papers because their questions and purpose are well defined. As a musician, my only question is regarding their methods of data collection and various formulas used to synthesize the data. These questions, are less to do with the specific papers than with a general question of data collection.
I see Chaser’s word learning process as a net product of both biological and cultural (social) interactions. Pilley and Reid (2010)’s experiments probably could not be successful if doing without the social bond/interactions between the experimenter/trainer and Chaser. It is not to say that Chaser does not have the ability to learn words, but only through the facilitating and scaffolding provided by humans in addition to Chaser’s sociality that she is able to, just like the facilitatory effect of parent’s social interactions with children.
ReplyDeleteThere are two reasons that I am skeptical about whether the results of their study will be analogous to children’s study, and whether the procedures will be replicable to children’s study.
First, Although Chaser’s recognition of objects and categorization is quite impressive, it seems that there is an inevitable bias toward Chaser in the experimental procedures. For example, the category labels (i.e. ‘toy,’ ‘Frisbee,’ and ‘ball’) used by experimenters in experiment 3 are based on either the “playfulness” of an object or some observable characteristics that Chaser can easily see. Children, on the other hand, may not start from categorizing words according to object playfulness, or object shapes, etc.
Second, it seems that the authors praised Chaser’s performance on recognizing an object both by its category label and its proper name. Yet in reality most of the children do not do that right away in the process of their vocabulary learning. Children start from having an awareness of mutual exclusivity, especially in the beginning of learning nominals. Since they have not developed their nominal terms hierarchically, one item can only have one name, or one label. I am wondering how the authors are going to respond this.
The rate of acquisition, according to the authors, was not maximally emphasized, because they would prefer having higher accuracy in reaching objects.
I really like authors’ second experiment, in which they tested Chaser’s free recombinability of commands and names of object. Also, it would be even more interesting to see how dogs segment these verbal utterances, by prosodic cues or by each word’s meaning?
Pilley and Reid’s conclusion that Chaser understands the referential meaning of words may sound very convincing, but is not. No doubt their empirical findings are very interesting and indeed admirable, yet their conclusions are pretty misleading. This is because they fail to distinguish between learning the association between words and objects/actions and the referential meaning that words have for children when they start talking.
ReplyDeletePilley and Reid use experiment 2 to argue that “Chaser understood reference” (p. 9). To be precise, what the experiment demonstrates is only that Chaser has learnt to associate words to objects and actions independently of each other. Since Chaser has been trained to respond with determined actions to determined commands and to respond with actions towards determined objects when determined nouns are pronounced, when a new combination of command and noun is presented she knows which action she has to perform and with regard to which object. Indeed, there is no reason why we should have expected Chaser not to be able to do so: if she has learnt the two associations separately (command with action / noun with object) there is no reason why she should not perform the two associations together, given that the two associations are not in contradiction with each other. So, if reference means “association”, then it’s clear that words are referred to objects and actions independently. The problem is that this kind of reference has very little to do with language. Indeed, when we say that words refer to things, we mean that they “present” things. Let me give you an example. When I go to the bar with friends, we usually talk about “soccer”; this word presents to each of us the world of soccer, each of us has soccer in mind, can express his or her own opinion and can talk about it (perhaps trying to demonstrate that he or she knows more than the others); through words we share reference to the same facts about soccer and we can keep informing (or contradicting) each other about the same things for hours. Thus, words can present objects to us as mere objects of our consideration, implying no action to perform with regard to them. This is what children start doing when they start talking: for example, they listen to stories and they tell their own stories.
On the contrary, when Chaser hears a word, she is never presented with a mere object of her possible considerations, but she is always stimulated to an action relative to the object. Even if she can combine a word with different actions, the word never presents “the object in itself”, but functions as the specification of a feature of the commanded action: the noun specifies the object toward which the commanded action has to directed. In “Take the Lamb”, the word “Lamb” specifies that the appropriate response is to take the lamb, not the Lips or ABC. In conclusion, for Chaser words are signals for appropriate actions and do not present things (even if words may evoke “mental images”, these mental images are not used by Chaser as objects of consideration, as objects about which she can say “it is so and so”). Therefore, for Chaser words do not refer in the sense words refer for people who can talk. The only possible demonstration that animals have referential symbols would be that they started talking (perhaps by means of some sort of keyboard that they can use).
At this point it would be possible to explain why also the conclusion that Pilley and Reid draw from experiment 3 – that words, for Chaser, may represent categories – is misleading. This would imply the explanation of what it means for us to refer to universal objects such as “human beings in general”, “dogs in general”, “toys in general”, etc. Since I don’t want to bother people too much, I won’t talk about this but also a discussion about this topic would be very useful for me.