Hanna Wirman, Anna Zamansky
You can discover more about a person in an hour of play than in a year of conversation. —Richard Lindgard
Beyond individual cases and uncontrolled settings, animals (by which we mean non-human animals) have been interacting with human technology since at least the 1930s, when Skinner's behavioral experiments required animals to interact with interfaces of operant chambers and when animals were made to wear tracking devices in conservation research. By the 1970s, primates in zoos were being given computer technology as a form of environmental enrichment. Today, the emerging discipline of animal-computer interaction (ACI) promotes a fundamentally user-centric approach—in terms of both methodology and theory—in the design of technology for animals, largely informed by the conventional wisdom of various subfields of HCI.
In the spirit of user-centric interaction design, animals' playful interactions with technology are drawing increasing attention within the ACI community. Designing animal play stems from the objective of various animal-welfare initiatives to enrich the environments of non-humans in captivity, as well as from an interest in attaining closer ties with companion species. In behavioral science, play has long been identified as a possible welfare indicator because it often disappears when animals are under stress and is thought to engender a pleasurable emotional experience. Therefore, playful interactions have the potential to both indicate and improve captive-animal well-being in homes, zoos, and beyond. Playful interactions may provide cognitive enrichment and physical exercise, as well as create stronger bonds between humans and animals. Moreover, as animals may perceive technological devices in ways completely different from humans, understanding these differences has the potential to lead to exciting new insights into the nature of play itself.
Behavioral research distinguishes between three main types of animal play: social play (with other animals of the same or different species), locomotor-rotational play (alone or in company), and object play (interacting with inanimate objects). While several ACI projects have addressed the design of technology for mediating or creating bonds between humans and animals, our main focus here is on the direct interactions of an animal with a device, or object play. In the context of ACI, our object is a technological device, which is placed in the animal's environment.
In what follows, we briefly review some examples of recent playful ACI projects in terms of their respective users, devices, and environments, as well as the evaluation methods used.
Technologies for Non-Human Animal Play
Several playful ACI studies have centered around dogs, who are perhaps the most studied and available subjects of study in this context. Recent work [1,2] has explored digital games for dogs. The device in the latter study was a standard tablet. The dog interacted with the tablet by trying to catch a moving virtual object on the screen. The device in the former study was designed specially for the experiment. It consisted of two audio speakers and corresponding buttons to be pressed by dogs' paws, and an electrically controlled dog-treat dispenser positioned midway between them. Each speaker played a sound in turn and the dog was supposed to press the button corresponding to it; if pressed correctly, food was automatically provided by the dispenser. Both studies were conducted in the dog's home. In the second study, researchers used so-called ethograms to evaluate the state of the animal, while in the first study they conducted daily measurements of cortisol stress hormone concentrations in the participants' saliva. The results of the first study indicate that digital interactive games can lessen physiological and behavioral stress responses in dogs, depending on a dog's personality, while the second study warns of possible over-stimulation and even aggression triggered by digital play in some dogs.
Other pets studied in playful ACI include cats  and hamsters . The device of the former was again a standard tablet, allowing humans and animals to play together simultaneously. The evaluation method included a behavioral analysis of video observations using video-coding software. The device was a flexible floor, on which the hamster was placed and presented moving food that it physically chased and caught. In a virtual game space, the human avatar was coupled to the physical food while being chased by the hamster's avatar. The evaluation method was based on standardized body condition scoring. After being allowed to voluntarily play for one hour per weekday over a period of six weeks, the hamster's health and fitness had improved.
Playful interactions may provide cognitive enrichment and physical exercise, as well as create stronger bonds between humans and animals.
Another study  is an example of a project in which the animal's environment is part of a captive setting, which makes evaluation particularly challenging. In the project, several game types and technologies were provided to Bornean orangutans at a wildlife rescue center in Indonesia. These experiments were aimed at recognizing preferences, affordances, and playing styles characteristic to orangutan play. One of the findings was that co-design approaches and the extension of orangutans' existing play practices by digital means are preferable to introducing unfamiliar interfaces and affordances.
Toward Shared Vocabulary (And Theory)
The design and analysis of playful ACIs as a scientific topic remains in its infancy. It calls for multidisciplinary approaches from ethology, interaction design, user-computer interaction, and media/game studies. These communities, however, use radically different research methods and theoretical frameworks, as well as overlapping terminologies. Establishing a shared discourse is potentially challenging, raising the need for a unifying conceptual ground.
What is a playful ACI, how can we recognize it when we see it, and what are the guiding principles that should inform the design and evaluation of such interactions? To the best of our knowledge, no widely accepted definitions of playful ACI are yet available. One paper proposes a general definition of an "intelligent playful environment" as "an animal-centered ecosystem with intelligent capabilities which is able to learn from the animals' behaviors and interactions, using the acquired knowledge to adapt itself to the context, creating engaging playful activities which do not necessarily need human mediation to evolve" . However, this definition addresses only the device, leaving out the animal. Moreover, "animal's interaction" (with the device/environment) and "engaging playful activity" still require clarification.
We propose another step toward a general framework for describing, analyzing, and comparing playful ACI interactions, grounded in concepts from animal science and game studies. Within this framework, we attempt to reconcile various notions and elements of play interactions from existing playful ACI research.
Animal Play, A Working Definition
Play interaction is one of the most natural ways animals gain new experiences and include new objects—as well as other individuals—in their perceptual world. Play is seen as particularly important for young mammals in several areas: practicing skills they will use as they grow (such as hunting and fighting behaviors), reinforcing social bonds, establishing hierarchies (by understanding their own physical limits and those of others), and exercising. Gordon Burghardt lists five essential criteria of play behavior that are now used as a standard in behavioral science to describe animal play :
- It is not "fully functional" (meaning it includes elements "that do not contribute to current survival").
- It is "autotelic," that is, self-rewarding (and so related to experiencing pleasure).
- It differs in structure and/or timing from the adult, "serious" form of the behavior.
- It is performed repeatedly but not stereotypically.
- It is initiated when the animal is in a "relaxed field" (by this is meant there are no immediate threats to the animal's fitness).
While game design for animals is at the stage of individual pilot projects and proofs of concept, it is meaningful to instead compare the above characterization with various definitions found in the literature that scrutinizes human play. Such theories have long served in the field of game studies to provide a theoretical basis for analyzing as well as designing play experiences. Roger Caillois, for instance, defines play as "a voluntary attempt to overcome unnecessary obstacles" and characterizes it as free, uncertain, unproductive, and separated by its own time and space. For Johan Huizinga, play is something non-serious, yet utterly absorbing, standing outside ordinary life. Nearly a hundred years ago, C. J. Ducasse defined play by its autotelic characteristic; just like in Burghardt's postulation, Ducasse sees that play is intrinsically motivating. We consider such obvious overlaps in theory a viable starting point for a general framework of animal play.
Toward A Framework for Playful ACI
Informed by Burghardt's definition of animal play and concepts on human play mentioned earlier, our starting point is the concept of behavior, which in animal science is considered a reaction to stimuli in the animal's environment. Another part of our framework is a technological device, which is placed in the animal's environment. In reaction to input, the device produces output in the form of sensory stimuli, which are part of the overall stimuli of the environment. The animal displays behaviors, which are responses to the environmental stimuli, some of which (device-oriented ones) trigger new stimuli.
Some of these behaviors are functional (the exact classification is species-specific) and some are non-functional. A subset of non-functional behaviors may be device-oriented (e.g., touching or smelling the screen). The environmental stimuli also affect the animal's state (both physiological and emotional), some parameters of which can be measured. Among other things, they may threaten or distract the animal, taking it out of its comfort zone.
In a playful ACI iteration, the device's stimuli trigger a set of behaviors from an animal, some of which are device-oriented and again produce new digital output, and vice versa. Playful ACI is therefore a sequence of such iterations, which has the following characteristics:
- The stimuli produced by the device are not externally rewarding for the animal (so its device-oriented behavior is self-rewarding).
- The stimuli produced by the device positively change the state of the animal (reducing stress and making it feel happiness, thereby potentially improving its welfare.)
- The behaviors that are not device-oriented are not triggered by environmental stimuli that threaten or distract the animal (the animal is in a "relaxed field").
We illustrate these concepts using an adaptation of a classical human factors interpretation of the human-machine interface from  in which a loop is created between a user and a machine via interacting with an interface. Figure 1 shows how we adapt this idea to our setting.
Playful ACI can therefore be described as the loop in the diagram created between an animal and a device: While the animal is in its comfort zone, it is exhibiting not fully functional behavior toward the device, which in turn produces sensory stimuli that affect the animal's state and behaviors.
When designing a particular playful ACI interaction, the framework gives rise to the following checklist of questions to be considered:
Animal: What is the set of envisioned/possible device-oriented responses of the animal and how can they be measured? This can be answered via analyzing postures and constructing ethograms [2,3] or using sensors and other bio-logging devices (as proposed in ).
Device: What is the set of sensory stimuli produced by the device?
- In what way are they rewarding for the animal? In this context, interactions in  and , where food is offered to the animal during the interaction, can be compared with  and , where the animal tries to catch a moving object without any obvious reward (except satisfying a hunting instinct).
- What is their expected impact on the animal's state? What are the possible ways to measure and assess it? Ethnographic and ethological approaches can be used here as possible indications. A more objective measure is, for example, cortisol stress hormone concentrations measurements ( is the only study of which we are aware that uses cortisol measurement for evaluation of a playful ACI).
Environment: What are the sensory stimuli produced in the environment? How can they be reduced so as to bring the animal as close as possible to its comfort zone?
- What are the existing environmental factors contributing to the animal's stress level and to what extent can these be reduced?
- How do caretakers and everyday routines fit into the animal's perceived context? For example,  discusses the distractions present in a rescue center including weather conditions, feeding patterns, and volunteer presence.
- How is the device introduced to the animals? Introducing a new device to the animal's environment can create excitement, over-stimulation, and even aggression [2,5].
Summary and Future Work
We have presented a general framework grounded in concepts from animal science and (human) game studies. Using this framework, we have proposed a characterization of playful ACI that reconciles Burghardt's definition of animal play with various elements of play interactions from existing playful ACI research. It is our hope this will enable better exchange and comparisons between studies, and will serve as a starting point for recognizing and describing key aspects of playful ACI systems. With respect to playful interactions being repeated while avoiding stereotypical activities, earlier documentations of animal-computer play provide very little evidence on the duration or iteration of interaction. We therefore call for research that defines the duration and repetition of possible playful interactions.
Many playful ACIs are designed with the goal of strengthening the bond between human and non-human animals. To fully characterize interactions that involve a game between a human and an animal (e.g., Felino  or Metazoa Ludens ), the framework needs to be extended, incorporating the role of the human and his interaction with both animal and device. Further research is needed to understand the ways in which humans and animals use technology and games together and in parallel.
Finally, another point for future investigation concerns captive settings as environments of playful ACI (e.g., ). Such settings provide both "artificial" safety and departure from "natural" development from juvenile to adult age, while also allowing for free time, unlike "natural" environments in which play is a luxury performed at times of peace and full satisfaction. Domestic and zoo-bound animals, as well as captive animals living in rescue centers, do not follow a typical development pattern from juveniles to adults as seen in nature. They generally participate in play activities at all ages. It is therefore hard to identify whether an activity is a "non-serious," simplified version of later adult activity. Deeper comparisons between domestic animals and animals living in rescue centers seem to be an interesting area for further research.
1. Geurtsen, A., Lamers, M., and Schaaf, M. Interactive digital gameplay can lower stress hormone levels in home alone dogs—A case for animal welfare informatics. Entertainment Computing-ICEC 2015. Lecture Notes in Computer Science 9353 (2015), 238–251.
2. Baskin, S. and Zamansky, A. The player is chewing the tablet!: Towards a systematic analysis of user behavior in animal-computer interaction. Proc. of the 2015 Annual Symposium on Computer-Human Interaction in Play. ACM, 2015.
4. Cheok, A.D. et al. Metazoa Ludens: Mixed-reality interaction and play for small pets and humans. IEEE Transactions on Systems, Man and Cybernetics, Part A: Systems and Humans 41, 5 (2011), 876–891.
8. MacKenzie, I.S. Input devices and interaction techniques for advanced computing. In Virtual Environments and Advanced Interface Design. W. Barfield and T.A. Furness III, eds. Oxford Univ. Press, Oxford, UK, 1995, 437–470.
Hanna Wirman is a research assistant professor at the School of Design of the Hong Kong Polytechnic University, where she teaches game design. Her research focuses on marginal and critical ways of playing and making games, including animal play. She is the president of Chinese DiGRA and is on the executive board of DiGRA International. firstname.lastname@example.org
Anna Zamansky is a senior lecturer in the Information Systems Department of the University of Haifa. Her research interests include animal-computer interaction and the use of technology for animal welfare. email@example.com
Copyright held by authors. Publication rights licensed to ACM.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2016 ACM, Inc.