Creativity in VR: Constraint versus Exploration

1 Current Work

The purpose of my research is to investigate creativity in virtual environ-
ments, taking an interdisciplinary approach. From the computing perspec-
tive, creativity is often cited as one of the ’grand goals’ of human achieve-
ment, and that systems in the ’new computing’ movement should all be
aiming to support (Shneiderman, 2002), while from the psychology perspec-
tive, creative design has been identified as one of the key challenges for
theories of human cognition (Simon, 1981).

Many different software packages do aim to support creativity, usually by
looking at specific domains and how to support those domains in a very task-
oriented manner. However, this is only really appropriate for clearly defined
problems, which is rarely true of most forms of creativity tasks, and most
software packages only support a relatively limited subset of possible tasks,
which may not be sufficient to allow creative ’leaps’. In order to support
creativity effectively, the cognitive processes involved must be understood,
both for the act of creation and for the type of system used. This requires
not only knowledge of the creative process and how it can be affected, but
also how the system used may affect cognitive processes.

As it is, the creative process is not currently well understood, and there
is still much scope for research in this area. Virtual reality (VR) offers a
very interesting vehicle for studying creativity: it opens up the possibility of
studying both the mental and physical domains that have traditionally been
explored by manipulating them in ways that may not otherwise be possible
while letting us study the theoretical gap between the two that VR creates.

1.1 Creativity in the Mental and Physical Domains

It has been widely suggested that a creative product is one which is both
original (or novel in some way) and practical (or fitting for its purpose)
(Finke, 1990) (Boden, 2004) (Warr & O’Neill, 2005). In terms of artistic
creativity, practicality might be assessed in terms of aesthetic values or emo-
tional properties. However, this is often highly sub jective. For this reason,
it was considered that the domain of creative design of tangible ’inventions’
may be most suitable for research at this time. Even for this though, studies
of how these creative inventions are built are contradictory and inconclusive.
One influential cognitive model of the creative process is the Geneplore
model (Finke, Ward, & Smith, 1992), which proposes that creative design
consists of two phases: generating ’preinventive structures’, then exploring
and interpreting them in an iterative process of focusing or expanding on
concepts, while taking the necessary product constraints into account in
order to ensure practicality. This model is advantageous as it has no in-
herent preconceptions or limitations about the type of creativity that it can
accommodate. The model was informed by mental synthesis experiments
conducted by Finke, in which participants generated ob jects from three 3D
shapes, to fit a category (Finke, 1990). He demonstrated that if partici-
pants generated structures before receiving categories then creativity scores
increased, and Finke argues that this ’preinventive phase’ forces higher levels
of exploration due to the structures being free of interpretive bias.

An experiment was run to provide a replication of Finke’s studies with
the addition of a physical synthesis task, where participants were given phys-
ical instantiations of the 3D shapes, arguing that the preinventive phase
should increase creativity irrespective of the type of ob ject generation, if
the Geneplore model is to hold (McKnight, Ormerod, Sas, & Dix, 2006).
However, the preinventive phase was not found to increase creativity in ei-
ther the mental or physical synthesis tasks, which is in contrast with Finke’s
results and throws the Geneplore model into question. Instead, the preinven-
tive phase was seen to cause an increase in the originality of the inventions,
at the cost of practicality. An alternate account for this effect would be that
the preinventive phase poses additional constraints on the task by forcing the
appearance of the invention, which means that participants are forced to be
more original in order to make their structures fit the designated category.
This may suggest that exploration is not in fact as beneficial to creativity
as previously assumed, and constraint may play a much larger role.

1.2 Supporting Creativity in Virtual Reality

Based on the results of the last experiment, it was planned that experiments
would then move to virtual environments, where the issue of constraint could
be further investigated. As a starting point, the constraints of physical laws
were considered. For example, the effects of gravity or solidity were never
imposed on the mental or physical synthesis tasks since participants were
told that shapes could be merged, but this would be hard to visualise when
presented with solid physical shapes, and may also be hard to imagine with
mental synthesis. However, in a virtual environment, this is easier to allow.
Pilot studies for the next phase of experiments were conducted using
environments built in Java3D and VRML, where the solidity and effect of
gravity on the shapes could be manipulated. While these studies appeared to
indicate an effect of the type of constraint, a number of technical issues were
discovered with using these systems. Therefore it was considered necessary
to investigate a more robust system for experimentation.

SecondLife ( is an online virtual community built
by Linden Labs, with the slogan ’Your World. Your Imagination’. Within
the environment, users (represented by avatars) can navigate a 3D virtual
world, interacting with each other in a manner similar to many popular
MMORPGs. However, a distinguishing feature of this environment is that
it is not, in itself, a game, although many games exist within it; there is no
overall ’goal’, just as in real life. Instead, users can choose to play games,
run businesses, socialise or just explore. The in-world currency of ’Linden
Dollars’ has a real economy, and can be converted to and from US dollars, so
virtual businesses may earn real incomes (approximately L$250 = $1 USD).
Currently, SecondLife has nearly 2 million users (as of December 2006).
The most relevant feature of SecondLife, however, is that all features of
the world are built by its users, and as such depends highly on supporting
and encouraging creativity. Players have the ability to create any form of
artifacts that they wish, such as buildings, vehicles, furniture or toys, which
can then have scripts added to them to allow animations or behaviours.
These artifacts are built using primitive 3D shapes (called ’prims’), such as
spheres, cones, cubes and cylinders, which can be manipulated in a variety
of ways (e.g. rotating, resizing, moving, cutting). The system uses the Ha-
vok physics engine, and users can make shapes ’physical’ (i.e. they can be
pushed and affected by gravity) or ’phantom’ (where they can pass through
other ob jects). They can choose from a variety of materials (e.g. wood,
metal, glass, rubber) or apply their own textures. This focus on creativ-
ity by synthesis of generic 3D shapes makes SecondLife an ideal platform
for studying creativity using our existing methods, while also giving the
opportunity to evaluate a system with a large existing userbase.

It is hoped that, once these experiments are run, it will be possible
to then investigate specific issues in more depth, with a view to building a
toolkit for supporting creativity effectively by allowing the most appropriate
levels of exploration or constraints.


Boden, M. (2004). The creative mind: Myths and mechanisms. London:
Finke, R. (1990). Creative imagery: Discoveries and inventions in visuali-
sation. Hillsdale, NJ: Erlbaum.
Finke, R., Ward, T., & Smith, S. (1992). Creative cognition theory, research
and application. Cambridge Massachusetts: The MIT Press.
McKnight, L., Ormerod, T. C., Sas, C., & Dix, A. (2006). As useful as a
bicycle to a fish: exploration versus constraint in creativity. In Pro-
ceedings of the cognitive science society conference. Montreal, Canada:
Cognitive Science Society.
Shneiderman, B. (2002). Leonardo’s laptop: Human needs and the new
computing technologies. Cambridge Massachusetts: The MIT Press.
Simon, H. A. (1981). The sciences of the artificial (2nd ed.). Cambridge
MA: MIT Press.
Warr, A., & O’Neill, E. (2005). Understanding design as a social creative
process. In Proceedings of the 5th conference on creativity & cognition
(p. 118-127). New York: ACM Press.