Earlier this year in October Annika Geurtsen, Maarten H. Lamers and Marcel Schaaf wrote a Springer journal article for Entertainment Computing (ICEC 2015) on a case study to see if ‘Interactive Digital Gameplay Can Lower Stress Hormone Levels in Home Alone Dogs’.
This exploratory work focuses on measuring cortisol levels in dogs, which has been shown to be an indicator of stress or excitement. The cortisol level is collected via saliva of the dog(s). Three dogs cortisol levels are measured before and after an isolated period of playing with an interactive button game over 15 days with and 15 days without the game.
These where then constructed into ethograms that showed that through the combined data that interactive games can lower the cortisol levels of home alone dogs. This paper aligns with previous, as well as my own, research on animal welfare through informatics by exploring how technology can benefit our pet dogs which are sometimes left in isolation. While games designed to combat these ‘home alone dogs’ are not intended to replace owner-dog interaction they aim to provide a mental stimulus, reducing stress and increasing enjoyment through interactivity.
I really enjoyed the concept mentioned within the paper of the change from human to animal focused technology through and together with ACI. This alteration of the center of animal technologies seems to be appearing in a lot of recent ACI work creating a cultural shift towards animal-attentive designs. This paper also brought out part of a bigger debate: ‘what is a game?’ and further questions I had for the authors. To find out answers to these and other questions, I asked the authors a few in-depth questions split into three types: Emotions, Study Design and lastly Study Findings.
Your paper is primarily about stress management in dogs while being left alone. Do you think games in general relieve stress and in comparison how do they compare to their human owner being their?
When I started this research for my graduation it actually took me quite some time to come up with the idea of using games to manage stress in home alone dogs. Other options I considered were crafting a device that would allow dogs (whether conditioned or not) to experience the amount of time passing while the owner is gone (it’s been proven in humans that the waiting is experienced as less dreadful when knowing the amount of time you have to wait (Osuna, E. E. (1985) “The psychological cost of waiting”, Journal of mathematical psychology, Vol. 29, p.p. 82 – 105)). And simulating the owner, which was based on research in humans where simulated pets such as PARO (Parorobots, 2003) are used to comfort them. Simulating a human whilst taking into consideration the cognition and perception of dogs has many drawbacks however (such as a possible uncanny valley).
The idea of using games came to existence both due to research into the (possible) effects of play and from first-hand experience with my own dog (and the many dogs my parents had). Our hypothesis was that the experience that results from playing can be so immersive that a dog might even ‘forget’ that the owner is not there and that way can be (emotionally) lifted out of the experienced anxiety resulting from feeling isolated. Though making a comparison between the dog being home alone with the game and being home with the owner is outside the scope of this research, I do think most dogs prefer being with their owners. They are a social species after all. This is actually the reason that social isolation in dogs is a big problem. Humans can’t always take their dogs with them.
That said, it is important to realize that our study is not about developing the optimal game to counter separation-anxiety related stress, but to investigate whether or not such an approach could work. Now that we have demonstrated it could lower stress-response, the next step is to find the dimensions or parameters of interactive gaming that optimally reduce stress.
You also mentioned that games could ‘enable (the canine) to become more (emotionally) independent from the human (owner)’. How do you think that games could reduce this independence? I agree with you that certain games could lower stress, but how would they enable the dog to gain independence?
The best way to describe what we mean with (emotional) independence is to make a comparison between the way we treat dogs with the way we treat cats. A cat is quite independent: he/she is (usually) free to go in and out of the house when he/she wishes. Their habitat isn’t bound to the walls of the homes of their owners. This means cats have quite some freedom to express their natural behaviour (such as hunting mice or trying to scare other cats from their territory). Now take a look at how we treat dogs. They are locked up inside the house till someone takes them out on a leash (at least in the Netherlands you need a leash almost everywhere). Most of them get told what to do by humans (e.g. when a dog has to stay on his/her spot because you have visitors over). I would call that dependent, because the quality of life for a dog is much more dependent on the owner than it is for a cat.
The easiest way to overcome the issue of separation anxiety in dogs is to always take them with you. Our society doesn’t allow that however, so we had to think of other options. What if we could make the world of a dog bigger than the four walls we usually keep them in by using digital media? Eventually I came up with the idea of creating a multiplayer game for dogs that are home alone. This way dogs would be empowered to form (digital) social networks again, the way they did before they were domesticated. Unfortunately the timeframe in which I had to set up my research demanded that I simplified it, which is why in this research the dog only interacts with the game. Apart from that, I do think the immersion any form of (game)play can offer, has the potential to make the quality of life of a dog less dependent of the presence of (and with that the choices made by) the owner.
I agree with you there, its a lovely idea to take a digital formation of socliability into ACI. Enjoyment in animals is a hard emotion to measure, if at all possible from a biological viewpoint, I myself have always relied on a combination of owner reports, behavioral experts and physiology. It makes it harder even still to measure enjoyment vs. conditioning as you do always not know the motivation behind the animal’s actions. In your paper, Metazoa Ludens states that the ‘hamsters enjoyed playing the game’. Do you think the dogs enjoyed your game, and if so, how did you measure this?
I agree with you completely on this point. I even think that it is difficult to objectively measure emotions in any animal. You could ask how someone (a human) feels and they could describe it to you, but who can ensure you the description is correct and whether or not talking about the emotion hasn’t already altered it? Next to that there is usually a bias in self reflection and in the perception of others. For instance we think it is good practice to involve the owners of the animals in the research. This however does not guarantee a complete objective measure of the animal’s behaviour.
In the case of our research we judged whether dogs enjoyed the game by the amount of interest they showed in it. All dogs had the option to stay away from the game or to approach it. We based this judgment on the assumption that animals do not engage willfully and voluntarily in an interaction if they don’t want to. I think not all dogs necessarily enjoyed the game, due to several reasons (such as personality and breed). One of our subjects for instance didn’t show any interest in the game at all. She rather spend her time sleeping. Another interacted with the game nearly 90% of the time she was alone with it. (I recently even received a report from the owner of the subject stating that the subject uses her paws a lot more to interact with inane objects.)
However, as always in scientific research, you gain insight into complex questions (e.g. do dogs enjoy the game) by looking at related effects one-by-one. Although we are highly interested into whether canines would enjoy the game, or better yet, what types of games they would enjoy most, we limited ourselves in this study to simply investigating the hormonal response in combination with ethogram data. As such, our work cannot independently answer the question of enjoyment, but makes a small step towards such understanding.
That did make me smile about the dog using her paw to try and work other objects! In my work I also use the ideaof non-participation as a form of consent towards an activitiy. It is a hard area to approach in ACI as the animals can not 100% confirm or distain their emotions. I find myself wondering though if an animal would take part in an activity it did not enjoy, with a negative connotations, for the car-crash effect, where it is not something you enjoy but you can not help but look or take part. I think this is something all researchers in ACI face aswell as something that can not be confirmed.
My biggest question around your work is the idea that the button interaction is a game. It can be a complex task of defining what exactly a game is, often with certain elements and attributes needed to identify a game. Your game doesn’t appear to have varying levels, require creativity, elements of competition, memory tasks, strategy or other elements commonly attributed to games in order to take part (Prensku, 2001). So I was wondering what attributes, in your design, do you think classifies this system as a game?
This question is quite philosophical, and outside the scope of our personal interest. It is an interesting question, and one that people should investigate, just not us. Our idea is simple, we do an experiment, describe our observations and draw tentative conclusions. Hopefully this will inspire others to delve into related questions, such as “does it work for cats?”, “and what about goldfish?”, or “what is a game?”. Our works was set out as an exploratory study into the possibility of reducing stress-response. Whether or not the system should be considered a “game” or not is interesting, but not something we research.
You designed your machines to ‘hypothesized reactions of canines’. A lot of my work is focused about designing with dogs, especially my paper ‘Doggy Ladder of Participation DLOP’ (BHCI, 2015) which aims to empower dogs through including them within the interaction design process. This enables the dogs’ preferences to be derived rather than guessed through the humanized perspective. Why did you choose to design what you thought the dog wanted rather than include the dog, through things such as body language consultation? I believe the button option for a dog isn’t in dogs’ ordinary behaviour and that is maybe why the highest interactions came from a dog which had previously been trained with a button-type system (mentioned in section 6.4 Game Oriented Behaviour) who was able cognitively to understand the interaction. I know for this reason Robinson (2014) designs through tug toys as they are something a dog innately knows how to use within their affordance.
We did spend time watching various dogs play (with each other, with humans and with objects) and reading relevant literature, but this work was undertaken as a 30 EC graduation project, not a four-year PhD study. Therefore, there are strict time-limitations imposed by the university. For example, the total project (including conception, lit review, method selection, execution, analysis, thesis writing) should be completed within 5 months. This simply prevents us from studying all possible aspects, and forces us to make choices that accommodate the strict deadlines for graduation theses.
I know that some dogs have habitual behaviour, such as my dog loves to grolw when playing with my partner, not because he is annoyed or frustrated as growling is seen as, but so that my partner will growl back which they both seem to oddly enjoy and will both initiate. This may be construed in view of standard behaviour as like I said aggressive but we as his owner know his true intentions holding valuable knowledge about the dog. So I was wondering if the dog owners were involved with the construction of ethograms for the dogs’ behaviour?
The dog owners were not involved with the construction of the definitions. However they were consulted during the analyzation of the video data to indeed ensure we weren’t overlooking any subject-specific behaviors.
In Table 1 (the game steps diagram) there is a 20min wait before the dog can have each interaction with the game, why is this?
This is to give the dog a break from interacting with the device. From our short study into play in dogs we found that dogs usually take a break during play in order to cool down. We wanted to avoid mentally overloading the subjects.
When conducting studies withdogs I also give them breaks. I think this would be an intresting area to research, I know Patricia Pons has looked into developing Playful interactive guidelines for conducting research in this area and this would be a valuable method to add. In your study you use cortisol concentration analysis to measure a dog’s level of stress and/or excitement and within your game you gave the dog users a treat for pressing the button. Do you think the ingredients within the dog food (e.g. sugar/ refined flour) could have changed the dog’s hormones enough to affect the cortisol levels?
There is certainly a chance of the food influencing the results. It’s a good question, and one that deserves to be followed up. In the case of the subjects in this study we can state that subject Tommie and subject Isa didn’t consume any treats during the experimental condition.
You seemed in your section about game oriented behaviour to attribute a positive relationship between the present knowledge of trained behaviours (you mention flyball, agility, obedience and doggy dancing training) and their use of the system. Do you think that the more a dog is involved in these training and obedience activities the more it will be involved within the game system?
I think the answer to this question also lies in the question you asked earlier: “I believe the button option for a dog isn’t in dogs’ ordinary behaviour and that is maybe why the highest interactions came from a dog which had previously been trained with a button-type system (mentioned in section 6.4 Game Oriented Behaviour) who was able cognitively to understand the interaction”. It is a skill set that in the case of this interaction would have the potential to attribute to a successful interaction with the game.
I would like to thank the authors for sharing this interesting conversation with me, in particular Annika. To find out more about their work below are the links to their chapter and the full springer book
Link to Springer paper: http://link.springer.com/chapter/10.1007%2F978-3-319-24589-8_18