Having previously stated my aim to investigate emotional engagement through virtual pet design, I set out to design a virtual pet suitable for the undertaking.

Bates’ research on virtual agents sets out some basic requirements for the design:

  • The virtual pet must be able to display clearly recognisable, human-like emotions.
  • These emotions should be identifiable through facial expressions.
  • Emotions should also be expressed through body animations, following the 12 principles of animation developed by Disney’s original animators.

I also aim to draw on the history of virtual pets, beginning with Tamagotchi (1996),  to influence my design. Taking this into account added a couple more design requirements:

  • The virtual pet should be portable, and should ideally function on mobile phone and tablet devices.
  • The virtual pet behaviour should be persistent through time and react reliably to player input.

My response to these requirements is a cube-shaped feline creature I have named Catbox. I will cover the development of the Catbox design in detail here.

Catbox Concept 1-01

Initial Catbox sketch.


Early 3D model of Catbox.

Practical Considerations

Why a cat? Cats are a common pet with broad appeal. Cats feature regularly in modern media, from cartoons to online memes, and the rise of kawaii culture online has begun to define an iconic, codified way of drawing cats that is easy for people to parse, which I will go into in more detail shortly.

Cats also have uniquely expressive tails, a simple element that can be included in the design and animated to express emotion in the character.

Why a box? The cube shape of the character due to more practical considerations. Developing and animating a simple geometric shape requires a lot less time and resources than the development of a full animal shape. I can apply the principles of animation to a box-shape much faster and more easily, making for quicker prototyping and iteration.

The exploratory nature of the project concerns emotional recognition rather than realistic graphical fidelity, meaning a realistic representation of an animal is unnecessary. It makes sense to work with the simplest figure possible, providing that figure can fulfil all the design requirements listed above.

Keeping the character as visually simple as possible also aids with mobile development by keeping both the polygon count and animated joint count of the character relatively low. This will help ensure the virtual pet can operate smoothly on as wide a range of mobile devices as possible.

2D concept of the catbox design.

2D concept of the catbox design.

Facial Design

The key requirement when designing the face of the Catbox character was to develop a face that is appealing to as wide a range of people as possible, and that is capable of expressing a wide range of clearly identifiable emotions.

To understand what people find visually appealing about animal faces, I investigated the kindenschema or baby schema proposed by zoologist Konrad Lorenz in 1943. The baby schema attempts to identify why adult humans find certain sets of features, particularly those of baby animals, to be ‘cute’ or otherwise appealing.

Lorenz states that “humans feel affection for animals with juvenile features: large eyes, bulging craniums, retreating chins,” from this paper by Stephen Jay Gould, which concisely outlines Lorenz’s theory as it relates to the design development of Disney’s Mickey Mouse over time, as the character’s features trend towards those defined in the baby schema.

Further research in ethology (the science of animal behaviour, a practice developed by Lorenz) has shown that humans tend to react with positive bias to animal faces that show elements of the baby schema. People show a preference for faces with large eyes, big heads and small chins, whether that’s in humans, animals or virtual characters.

This is effectively illustrated by the Japanese idea of kawaii, roughly translating as ‘cute’ but with a more specific meaning in Japanese. Kawaii refers to a specific design of character, prevalent in eastern culture but increasingly common in the west, that heavily emphasises and iconifies the elements of Lorenz’s baby schema. Kawaii characters have large eyes, small or nonexistent noses, and small, simply rendered mouths.

Below are some characters from popular media that exemplify kawaii design:



Hello Kitty

Hello Kitty

Taiko Drum Master

Taiko Drum Master





Sushi GO

Sushi Go!

Kawaii characters also have a mostly-codified and easily identifiable set of facial expressions used to portray emotions. This lends itself well to the design of my virtual pet, as it allows me to draw on a pre-defined set of expressions that is already somewhat embedded into our culture, making it easier to quickly infer emotion with a visually simple character.

The catbox design complies with the standard sof the baby schema

The catbox design complies with the standards of the baby schema

Research into the effectiveness and popularity of kawaii design has shown “that viewing cute images narrows the breadth of attentional focus and reduces the global precedence effect in a subsequent task.” Essentially, people feel happier, more attentive and focused when viewing characters they find cute and appealing.

Catbox has a typical kawaii facial design, developed by referencing popular kawaii character designs and iterating through a series of sketches to arrive at an effective eyes/nose/mouth spacing solution.

Early sketches of facial designs for the cat character.

Early sketches of facial designs for the cat character.

Emotional expression

Convincingly portraying emotion requires tying the expressive features of the Catbox character to an emotional model suitable for interactive media.

PlutchikThe model I am using is Plutchik’s model of emotion or ‘wheel of emotion’, a representation of the psychoevolutionary theory of emotion developed by psychologist Robert Plutchik in 1980. The model places the range of human emotion across 4 axes, with emotional intensity ranging from mild on the outside of the diagram to intense at the centre.

Plutchik’s model is useful for interaction design because it portrays emotion as a navigable graph, which can easily broken down into a series of variables that can be read and controlled by a computer. By mapping a character’s emotional position on Plutchik’s model, a system can identify the character’s mood with a degree of accuracy and alter the character’s appearance and behaviour accordingly.

Plutchik’s model provides 8 core ‘driving’ emotions; ecstasy, vigilance, rage, loathing, grief, amazement, terror, and admiration. In order for my Catbox character to accurately portray a full range of emotion, I designed facial expressions to match each of these driving emotions.

Catbox's facial expression as specified by Plutchik's model.

Catbox’s facial expressions as specified by Plutchik’s model.

Each of these facial expressions are designed to be easily identifiable, and can also be scaled to show different extremes of emotion. By navigating Plutchik’s model in real-time and updating the character’s face accordingly, my virtual pet will be able to express a convincing range of emotions in response to a player’s input.


The Catbox expressions mapped onto Plutchik’s model.

With the visual design of my Catbox character now defined, the next step is to model the character in a form suitable for use in a game, and then rig and animate the character to provide the expressive body animation important to creating the ‘illusion of life’ as outlined in classic animation.