Today I found a lovely paper on ResearchGate exploring the social relationships and their representations between computer machinery and dogs. The authors (Judit Abdai, Anna Gergely, Eszter Petro, Jozef Topal and Adam Miklosi) present a methodology of presenting a dog with two plates of food (one with one sausage and one with six) letting them choose a plate. Their choice is influenced under four scenarios with humans and an Unfamiliar Moving Object (UMO): 1) a Helper UMO 2) helper UMO control, 3) non-helper UMO and 4) Human partner.
Authors (from left-right): Adam Miklosi, Anna Gergely, Jozef Topal & Eszter Petro
Their paper starts off exploring the relationships a dog has with machinery and humans, questioning whether a dogs social-cognitive ability comes from their domestication allowing flexible representations or that these are formed individually and not genetic. The paper then leads onto other dog-choice literature with UMO and humans proving that dogs like social features in UMOs (Gergely et al., 2013), humans can influence dogs choices through both behaviour (Prato-Previde et al., 2008) and voice even if the human was not familiar (Marshall-Pescini et al., 2012/2001). Literature and the above authors suggests that dogs are sensitive to humans communicative cues and behaviour, even if they mislead the dog. This question of associative behaviour is then lead on from humans to UMOs (in this case remote controlled cars) to see if they can also influence behaviour.
It was found that while humans can influence a dog’s choice, so can UMOs but only with a short period of positive interaction (see below). However, despite this the humans influence is always greater than the UMOs (helper and control).
This exploration of the versatile relationship between humans animals and machines is an important task within Animal Computer Interaction (ACI). In order to design machines that engage with, in this case are influential, to animals, we must mix both animal behaviour and computer science in a similar way to Human Computer Interaction (HCI) has mixed both psychology and machinery. We need to create a better understanding of animals to design machinery while synchronously designing to be understood by animals.
I had a few questions around this paper, particularly with behaviour, such as if behaviour is more important than embodiment with technology in regards to dog-computer interaction – is there a particular behaviour that UMOs should possess? Within their work they did use beeps, but this is not a familiar or ordinary sound a dog would hear. Instead a human voice might have been more familiar. Leading onto the aesthetics in the 1950s Harry Harlow did studies with monkeys, where monkeys choose the fur surrogate over the wire surrogate to spend a majority of their time on suspected due to their similarity to their real mothers. It would be interesting to find out if the UMO, in this case the remote controlled car, had more animal qualities if the dogs would show a greater preference towards the social cues.
Also while the author states that there is a connection between the UMO helping a dog and as such the dog recognized this as a social cue: but how are we able to correctly determine these cues as social and not a cause and effect type scenario similar to Pavlov dog experiment.
I did find one bit of the paper amusing, where the human tried to aid their dog in picking the right choice. I have found in a lot of my research that especially owners of pets will always try to help their beloved animal, and so often there is a need to deceive the owner into the studies true intentions in order to get unbiased analysis. This also opens up the question to if the humans’ feelings about the UMO influenced the dog.
As lots of my work has been around head tracking and I tried to find out, to no avail, how the authors tracked the dogs vision. This also lead me to think that although the authors mention an unintended human influence, dogs have been shown to follow humans gaze in similar experiment (Hare, 1999) so it could be possible that the dogs were following the human gaze.
Lastly due to associative learning it would be interesting to introduce different UMOs to see if this factor affected the results e.g. different remote controlled cars/drones or if the dog built up, as the authors call it, a social relationship, with only that particular drone.
With UMOs becoming more prevalent and diverse this work is becoming very prominent, however clearly there are still lots of questions to answer.
Written by Ilyena Hirskyj-Douglas 30th September 2015