See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237784566

Spatial Cognition in Architectural Design Anticipating User Behavior, Layout

Legibility, and Route Instructions in the Planning Process

Article · January 2007

CITATIONS READS

0 1,145

3 authors, including:

Zafer Bilda

39 PUBLICATIONS   865 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Zafer Bilda on 02 January 2014.

The user has requested enhancement of the downloaded file.

Spatial Cognition in Architectural Design
Anticipating User Behavior, Layout Legibility, and Route Instructions
in the Planning Process

Thomas Barkowsky, Zafer Bilda, Christoph Hölscher,

Georg Vrachliotis (Eds.)

SFB/TR 8 Report No. 014-09/2007

Report Series of the Transregional Collaborative Research Center SFB/TR 8 Spatial Cognition
Universität Bremen / Universität Freiburg
Contact Address:

Dr. Thomas Barkowsky

SFB/TR 8 Tel +49-421-218-64233
Universität Bremen Fax +49-421-218-64239
P.O.Box 330 440 [email protected]
28334 Bremen, Germany www.sfbtr8.spatial-cognition.de

© 2007 SFB/TR 8 Spatial Cognition

 Proceeding of the Workshop

Spatial Cognition in Architectural Design

Anticipating User Behavior, Layout Legibility,
and Route Instructions in the Planning Process

held in conjunction with the international

Conference on Spatial Information Theory (COSIT'07)

Melbourne, Australia, 19 September 2007

Workshop Organizers

Thomas Barkowsky Christoph Hölscher

University of Bremen, Germany University of Freiburg, Germany

Zafer Bilda
Australasian CRC for Interaction Design, Creativity and Design Studios, UTS, Australia

Georg Vrachliotis
(ETH Zurich, Switzerland)

Advisory Board
Ellen Do (Georgia Tech, USA), Christian Freksa (University of Bremen, Germany),
John Gero (University of Sydney, Australia), Gabriela Goldschmidt (Technion, Israel),
Barbara Tversky (Stanford University, USA)
 Contents

Martin Brösamle & Christoph Hölscher (U Freiburg):

Architects seeing through the eyes of building users - A qualitative analysis of design cases

Ruth Conroy Dalton (UCL):

Isovist characteristics of stopping behaviour

Ning Gu (U Newcastle):
Digital design with style: characterizing digital architecture in 3D virtual environments for design
collaboration

Leman Figen Gül (U Newcastle):

A study of design collaboration in virtual environments: interacting with the external design
representation

Joao Pinelo (UCL):

What about people, do they matter? Insights on the impact of people's presence in the built
environment
 Architects Seeing through the Eyes of
Building Users?
A Qualitative Analysis of Design Cases

Martin Brösamle and Christoph Hölscher

University of Freiburg, Centre for Cognitive Science, Friedrichstr. 50, 79098 Freiburg,
Germany, {hoelsch, martinb}@cognition.uni-freiburg.de

Abstract. Spatial cognition and wayfinding research as well as design

cognition are well established as fields of research. It is nevertheless
largely unknown how architects reason when they try to integrate way-
finding-friendly factors into their designs. In two semi-structured inter-
views we asked architects to give critique on real-world example cases
as well as to solve predefined design tasks. The qualitative analysis fo-
cuses on perspective taking and other skills related to the anticipation of
users in the building. The main finding is that the anticipation of visual
access for single locations is done well but that multiple locations are
seldom considered. A consequence would be that anticipation of users’
perspective is restricted to episode-like cognitive walk-throughs.

1 Introduction
To orientate oneself is an indispensable human ability in most everyday activi-
ties. Although in most cases navigation succeeds without much difficulty every-
body has already made the experience of being disoriented or lost. Moreover,
disorientation is not limited to unknown and pathless territory but is a recur-
ring phenomenon of human existence. Our research in the field of human spatial
cognition investigates how people perceive, reason and memorize architectural
space. Towards navigation the question is what properties make built environ-
ment easy or difficult to navigate. More practically speaking: How can buildings
be designed such that they facilitate wayfinding and spatial orientation?
When asking for wayfinding-critical environmental properties, at the same
time, the architectural designer who to a large portion shapes this very envi-
ronment comes into play. Investigating how better designs should be implies
thinking about how it can be achieved. As cognitive scientists are interested in
relevant informations and skills a designer needs to achieve wayfinding-friendly
designs. Architectural design processes have been studied intensively in the De-
sign Cognition community. (See for example Akin [1] and Goel & Pirolli [2])
Some researchers touch wayfinding issues in architectural design: For exam-
ple, Weisman [3] identifies factors of architectural legibility, Arthur & Passini
?
I would like to thank Georg Vrachliotis as well as my interview partners: Thank you
for spending your time and having the patience for giving such informative answers.
[4] give a prescriptice model for systematic wayfinding design. It is nevertheless
largely unknown how architects reason when they try to integrate wayfinding-
friendly factors into their designs. One of the reasons might be that wayfinding
is mostly treated implicitly in the discipline of architecture.
Research in the field has to consider the possibility that architects are able
to create navigation-suitable designs although never thinking explicitly about
wayfinding. Goldschmidt [5] identifies both declarative and procedural knowl-
edge as important in the design process. To address implicit and explicit design
knowledge our investigation combines semi-structured interviews on the role of
wayfinding in architectural design with example cases and design tasks.

1.1 Explicit Aspects of Designing for Wayfinding

Explicit Wayfinding design issues subsume everything that designers do with

the intention to improve their designs in terms of navigability. Here, the present
paperfocuses on the anticipation of the users’ situation in the object (building
or building collection) to be designed. Perspective shift refers to perspectives in
the visual, true sense of the word and means the anticipation of a user’s direct
perception from her point of view.
Perspective taking refers to perspective in a metaphorical sense and empha-
sizes discrepancies between the designing architect as an expert of the building
and the user as a layperson, possibly new to the building. Arthur & Passini con-
ceptualize “wayfinding design” as the systematic structuring of the environment
as well as providing relevant information to the users such that they can effi-
ciently solve relevant navigation tasks. Perspective taking means the anticipation
of the user’s informational circumstances in the designed environment.

1.2 Implicit aspects of Designing for Wayfinding

In contrast to the explicit aspects discussed in the section before, implicit wayfind-
ing design knowledge subsumes all aspects in architectural design which have
wayfinding related consequences but are not considered as such by the designing
architect. As no connection to wayfinding is given by the designers themselves
the identification of these factors must be theoretical or based on empirical re-
search. Weismans four factors, visual access, signage, architectural differentiation
and floor plan complexity provide a relatively general theoretical base. Our own
empirical work and formal architectural analysis [6] provides us with profound
knowledge about wayfinding problems in one particular building wich is therefore
used as example Case 2 (see below).
Continuing previous work by Hölscher et al. [7] the present study not only
relies on pure interviews but also asks informants to investigate existing cases
with respect to navigation. In addition, informers of the present study actively
carry out design tasks. The open role of inplicit knowledge in wayfinding design
is the majour reason for employing critiquing and design tasks.
2 Method
In order to investigate relevant knowledge in wayfinding design, in the first part
of the interview we asked our informers about the architectural design process
in general and the role wayfinding plays in it. To address implicit wayfinding
relevant knowledge, we then provided two example cases. For each case our
informers were asked to examine the provided floor plans and to give some
critique with respect to wayfinding. For Case 1 they also completed an example
design task.

2.1 Example cases and design tasks

In the first case they were then asked to place a waiting area in the hall of a
hospital1 . After completing this design task, the location where the waiting area
has been placed in reality was revealed. Finally, they were asked to evaluate the
present state plan and anticipate possible navigation errors of a visitor coming
down the stairs and trying to find her way back to the main entrance area.
Case 2 is a bit more open since it does not include a specific design task.
We presented to them floor plans of the conference centre Heinrich Lübke Haus,
Günne, Germany. This case was chosen because we have already collected em-
pirical data regarding wayfinding in the building. (See [6])

2.2 Data Collection and Analysis

During the interviews videos were recorded in oder to capture drawing actions,
pointing gestures on the drawings as well as verbal utterances. Content-based
transcription was the first step of data analysis where all potentially relevant
utterances had to be transcribed such that their content was captured. An ut-
terance was considered as relevant if it refered to one or more of the following:
overall building structure, circulation system, signage, visitors of the building,
wayfinding and orientation issues, perspectives different from plan view, planning
decisions. In parallel to the transcription drawing elements and areas pointed to
via gestures were numbered in the drawing such that the transcription can refer
to such areas precisely.

3 Results
3.1 Case 1
Examination phase: Both informers stated having analyzed the building from
the main entrances along the “main circulation”2 – “as if one would enter [it]
oneself”3 . Both made extensive use of circulation-related vocabulary, e.g. path,
1
Universitätsklinik (University Clinical centre) Freiburg, Germany
2
“Erschließung” (both Interviews)
3
“als wenn man sie selbst betreten würde” (Interview 2)
entrance, circulation, axis to state the most prominent ones. In the second in-
terview “direction” and “movement” were also frequently used. An other group
of concepts employed by both is reflected in concepts referring to the function
of different zones. In the first interview “zone” and “usage”4 were explicitly
mentioned.

Design phase: Our informers consistently analyze the situation in the hall (where
they had to place the waiting area) in terms of pathways and flows of visitors
and visualize their analysis on transparent paper on top of the plan view. (See
Fig. 1)

Fig. 1. Case 1: Ground floor plan showing the hall where the waiting area had to be
placed (left). Informer 2 analyzes main pathways and visitor flows in the hall (right).

For the actual design, basically three possible locations A, B, C are taken into
account (See Fig. 1) where B is the present solution in reality. While reasoning
about what solution to choose both informants frequently refered to the main
visitor flows in the hall. Both rejected B pointing out the “main axis” passing
through that area. Informer 2 referred to lack of outlook after rejecting some
possible solution (which is unfortunately unknown because the pointing gesture
is not visible in the video). The final decision is consistently made in favour of
an “island” solution at location C.

Anticipation of navigation difficulties caused by the actually realized solution Af-

ter revealing B as the location for the present solution the task was to anticipate
probable navigation errors for a person coming down the stairs and trying to
find the way back to the main entrance area. Indeed, they both considered the
blockage of the main axis as the central problem as well as they realize that
the visual contact to the central junction J is broken. However, they anticipate
quite different navigation errors which seems be caused by differences in the pre-
ceeding navigation episode. Namely: Informer 1 consideres position p1 as stop
4
“Nutzung”
location after coming down the stairs whereas Informer 2 considers position p2.
Informer 1 then concludes that the way back to the main entrance, orthogonal to
the main axis (down in the diagram), will be taken whereas Informer 2 decides
for the pathway along the main axis (to the right in the diagram).

3.2 Case 2
Again our informers examined the building – this time the geometry and zones
of usage where more prominent which seems to result from the building plan.5
Interestingly, the spontaneous conclusion with respect to navigation is con-
sistently that the building is clearly laid out. Only after mentioning empirical
results demonstrating serious navigation difficulties these are detected of our
informers as well. Furthermore, their conclusion differed in so far as they em-
phasized different usability hot spots. (see [6]) This, like in Case 1, seems to
depend on the navigation episode considered which was induced by the inter-
viewer. Whereas Informer 1 focuses on lack of visual contact to the main stairs in
the building – the task was to consider a visitor trying to find to the basement.
Informer 2 primarily finds the non-congruency of ground floor and basement –
he shifted attention to the basement after being asked to pay more attention to
wayfinding issues (without mentioning a specific example).

Fig. 2. Case 2: Informer 1 (left) recommended to remove several elements in the en-
trance hall and use glass walls in order to improve visual access to the central staircase.
Informer 2 (right) redesigned the circulation system in the basement such that it re-
sembles that in the ground floor more closely.

4 Discussion
Our findings suggest that architects are able to anticipate the situation of a user
in a particular location in the building. During evaluation of the present solu-
tion in Case 1 navigation difficulties and ego-centered views were anticipated
5
different areas of usage employ their own geometric typus, respectively.
well for a single location. However other potential difficulties with respect to
visual access from further nearby locations with different characteristics were
not spontaneously found. This is reflected in the fact that both informers ad-
equately anticipate visual access. However, depending on the assumed location
they come to different conclusions about potential errors. Our Informers seem
to have restricted their anticipation to single but relevant points.
Further support comes from the fact that our informers largely refer to circu-
lation either by egocentric but episode-like movement anecdotes or reason from
an allocentric view about axes, flows and paths which refer, however, to aggre-
gate movements. Especially during the pre-design analysis pathways and visitor
flows are considered more like location factors than as actual navigation relevant
circulation elements.
The reason for this “single-point anticipation” might be the high amount of
information to be processed. Lee et al. ([8]) for example mention cognitive limits
as one source of design errors. Design tasks involve large amounts of data and
designers therefore rely extensively on drawings as external representations. [9]
Following this interpretation, reasoning about wayfinding would be restricted to
the anticipation of prototypical navigation scenarios.
Further studies will therefore investigate the role of external representations
in the anticipation of visual access from multiple locations. More restricted design
cases with sophisticated visual structure will be employed. An other issue will be
the analysis of visual attention during such tasks – pimarily based on drawing
actions, pointing gestures already collected but also via eye tracking in future
studies.

References
1. Akin, O.: Psychology of architectural design. Pion Ltd., London (1986)
2. Goel, V., Pirolli, P.: The structure of design problem space. Cognitive Science 16
(1992) 395–429
3. Weisman, J.: Evaluating architectureal legibility. Environment and Behavior 13(2)
(März 1981) 189–204
4. Arthur, P., Passini, R.: Wayfinding: People, signs, and architecture. McGraw-Hill
Ryerson, Toronto (1992)
5. Goldschmidt, G.: Expert knowledge or creative spark? predicaments in design ed-
ucation. Paper presented at Design Research Symposium 6 (2003) Hosted by Cre-
ativity and Cognition Studios, University of Technology, Sydney, Australia.
6. Brösamle, M., Hölscher, C., Vrachliotis, G.: Multi-level complexity in terms of space
syntax: A case study. In: Proceedings of the 6th Space Syntax Symposion, Istanbul
(2007) (in press).
7. Hölscher, C., Vrachliotis, G., Brösamle, M.: Design cognition meets wayfinding
cognition: How architects support orientation in built environments. In: Poster
Proceedings of DCC Design Computing & Cognition ’06, Eindhoven, NL (2006)
8. Lee, G., Eastman, C., Zimring, C.: Avoiding design errors: A case study of redesign-
ing an architectural studio. Design Studies 24 (2003) 411–435
9. Goel, V.: Sketches of Thought. Massechusetts Institute of Technology (1995)
 Isovist Characteristics of Stopping Behaviour

Dr. Ruth Conroy Dalton

The Bartlett School of Graduate Studies, University College London,

Gower Street, London WC1E 6BT
[email protected]

Abstract. This paper presents a method for analyzing the isovist characeristics
of stopping behavior in a virtual simulation of the Tate Gallery. All spaces in
the virtual environment were filled with an array of points from which an
isovist was generated and whose attributes were calculated and stored. The
resultant ‘set’ of attributes was considered, in statistical terms, to be a
‘population’. A second series of isovists were computed, generated at the pause
point locations (where experiment-subjects paused whilst exploring the
simulation). Again, the attributes of these isovists were calculated and recorded
and were held to be a ‘sample’ of the wider ‘population’. Using the z-test and
central limit theorem, the sample of pause-point isovists was compared to the
population to determine how likely it was that the sample was generated
randomly. The results suggest that people are not pausing randomly; they are
pausing in locations providing maximum visual information.

Keywords: isovists, visibility graph analysis, VGA, wayfinding behaviour.

1 Introduction

Visual field analysis has proved a useful and robust tool in the arsenal of spatial
analysis techniques and has been effectively applied to the spatial analysis of
buildings and, to a lesser degree, urban neighborhoods. In particular, the type of
spatial representation termed the isovist, as coined by Benedikt [4], [5] has had
considerable utility and longevity. An isovist is simply a polygonal representation of a
two-dimensional slice through the potential visual field; it is usually constructed at
eye-height and parallel to the floor plane. However, the notion of the isovist is
strongly related to previous methods that attempted to represent or describe the visual
experience of an environment. Most significantly, it can be seen to be strongly related
to Gibson’s optic array [12], [13], a theoretic entity effectively constituting a three-
dimensional isovist. Another notable antecedent lies in attempts by Lynch to represent
the continuous, visual experience of routes through complex environments [18]
which, in turn, related to his conviction that functional neighborhoods should be
legible [19], namely should contain a number of imageable visual/spatial attributes. A
recent attempt to redefine Lynch’s work on legibility with methods of visual, spatial
and configurational analysis was recently undertaken by Conroy Dalton and Bafna [7]
which demonstrated relationships between isovist characteristics of urban
environments and a number of Lynch’s city elements (in particular, nodes, landmarks
and edges). Aligned work, on the classification of urban types, can be found in a
paper by Batty [8]. Other work on the experience of landscape has its origins in early
work by Appleton [2]. In more recent years, space syntax (a family of theories and
techniques examining the relationship between space and society) have employed
isovist analyses to quantify the visual experience of complex spatial environments
[14], [16], [17] and this work has been extended to take into account the
configurational properties of visual fields, that is to say, examining the set of visual
relationships between distinct locations. This work has been pioneered by Turner et
al. [21], [22], [23], however, such work on networked visual models can ultimately, in
turn, be related to early work by Braaksma [6].
From the references above, it can be seen that isovist analysis is becoming an
increasingly popular method for representing and quantifying the visual experience of
an environment. However, very little work has been done of the relationship between
visual field measures and associated behavior, particularly at the small-scale
resolution permitted by isovist analysis. Studies on the Tate Gallery [23] show a
correspondence with isovist measures. Equally, measures of stopping and engagement
at science museum exhibits were shown to correspond with isovist measures of the
location of particular exhibits in work by Peponis, Wineman and Conroy Dalton [20].
However, one of the most interesting studies demonstrating a clear correlating with
behavior, spatial judgment and isovist properties was demonstrated in a paper by
Wiener and Franz [24], where subjects were able to effectively locate regions of high
and low isovist area (corresponding to best ‘overview’ and ‘hiding’ places). The
objective of this paper was to determine whether there was a relationship between
people’s stopping behavior in a virtual environment and fine-scale visual field
analyses of those environments. One hypothesis being tested was that people pause or
hesitate more frequently when they are lost or disorientated and that an examination
of the locations of pause points could reveal locations where people have become lost.
An alternative hypoethesis was that people pause in locations that contain more visual
information to aid them in re-orientating themselves. In this case, people would not
necessarily pause in locations where they were lost, but rather in locations where they
were able to relocate or reorient themselves.

2 The Experiment

The experimental environment was a simulation of the Tate Gallery. This

experiment initially served as a pilot-study for a much larger study (based on seven
virtual worlds) into navigational behavior, published in full by Conroy [10]. Equally,
one of the associated experiments (from the larger study) has been further
documented in subsequent paper by Conroy Dalton [8].
There were 25 volunteer subjects who participated in this study; the male to female
ratio was 3:1. They had a diverse range of experience of virtual environments, ranging
from complete novices to those with a high degree of familiarity. The plan of the
virtual Tate Gallery is shown in figure 1. The location where the subjects began the
experiment is the bottom-centre entrance: the main entrance of the real gallery.
 Fig. 1. Plan of the Tate Gallery simulation

The virtual simulation was run on a Silicon Graphics computer with a head-
mounted display and a 3d mouse. The HMD provided the subject with an immersive,
stereo, full colour LCD visual display and a wide field of view; the horizontal field of
view of this headset was 105° with a vertical field of view of 41° and a 40° horizontal
overlap. The virtual Tate Gallery was initially modelled using the 2d/3d CAD
application MicroStation and then exported as a 3d DXF file. A scripted user event
programmed into the world calculated the position of the subject and the orientation
of their head ten times every second and saved it as a ASCII text log-file.

3 Measures of Isovist Arrays

The spatial layout of the Tate Gallery was analysed using the computer application
OmniVista [9]. OmniVista calculates two dimensional, planar isovists, parallel to the
ground plane, identical to Benedikt’s original methods of isovist calculation [4]. After
importing the building plan, a grid of points was generated throughout the
environment, flooding all navigable space. The horizontal and vertical spacing of the
grid points was set to four metres. Each one of these grid points was used as a
viewpoint from which a single isovist was generated and its attributes calculated and
stored. The geometric properties calculated for each isovist location are listed below
and the measures that were used by Benedikt in his paper [4] are marked with an
asterisk: area*, perimeter, area to perimeter ratio, circularity*, dispersion and absolute
dispersion, drift, maximum, mean and minimum radial length and standard deviation,
variance* and skewness* of the isovist radials. A brief explanation of how some of
the less-commonly used measures were calculated is included in the following
section.
Dispersion [10] is the difference between the values of the mean and the standard
deviation of the isovist’s radial lengths. This measure can take either a positive or
negative value. Drift [10] is an exceedingly interesting measure; it is the distance
between the location from which the isovist is generated and its ‘centre of gravity’.
The centre of gravity of an isovist is calculated as if the isovist were a polygonal
lamina of negligible but uniform thickness. This measure can only take positive
values. Drift will tend towards local minima at the centres of spaces and along
corridors. Similarities exist between areas of low drift and the axial breakup of spaces.
Maximum, mean and minimum radial lengths are calculated by measuring the
lengths of isovist radials at specified intervals (for example every one-degree). These
attributes are generated by calculating the maximum length of any radial (or the
longest line of sight), the mean length of all the radials (another measure of
‘spikiness’) and the minimum length of the radials (or the distance from the isovist
generating-point to its closest occluding surface). Standard deviation, variance and
skewness of radials are also a family of measures based upon the distribution of the
radial lengths of an isovist. Skewness is the third moment of the radials, a measure
used by Benedikt, who suggested that it is a good indicator of asymmetry of the
perimeter of an isovist polygon.
When the full set of attributes have been calculated for each isovist, the
relationship between every isovist viewpoint and every other isovist viewpoint may
be examined and a graph representation of intervisibility or the visibility graph is
constructed. There are two possible types of visibility connection, termed first and
second order relationships by Turner et al. [22]. The calculations performed by
OmniVista use a first order visibility relationship. Once the visibility graph has been
generated, it is used to develop a set of syntactic (as used in space syntax) measures.
The values calculated are connectivity, mean depth, radius 3 depth and total depth.

4 Results of the Pause Point Data Statistical Analyses

If the complete set of all possible isovists is held to be a population, then this
population can be compared to a sample of isovist properties calculated for each of
the pause point locations. There are many methods for comparing a sample to a
population to determine how representative that sample is of the whole population.
The two methods used in this paper are the central limit theorem and the z-test. The z-
test relies upon the population being approximately normal whereas the central limit
theorem can be applied to a completely random population.
In essence, the sample of isovist measures for each pause point location is
compared to the population of the equivalent measure for each grid isovist location.
Two values of z are calculated for each measure (one using the central theorem limit
and one using the z-test). If z is less than a specified value then there is a 95%
confidence that the sample could have been randomly drawn from the population. In
other words, subjects are pausing randomly and the visual and spatial layout of the
environment has no effect upon their stopping behaviour. If, however, the value of z is
greater than this amount, then it is unlikely that the sample of pause points was drawn
randomly from the population of the grid isovist locations.
According to the results of the tests, the attribute that is least likely to have been
randomly drawn from the population is drift. The mean value of drift for the
population is 6.86m whereas the mean of drift for the sample of pause points is
4.81m. Since drift tends to towards local minima at the centres of spaces (both rooms
and corridors) then it seems appropriate that people should be stopping in these
locations. The results also indicate that people were pausing in locations with a much
higher than average isovist perimeter value (651.11m compared to 457.66m) and
consequently a much smaller area/perimeter value (5.17m compared to 5.85m) and
locations which offered longer lines of sight (173.22m compared to 113.22m). All of
these results are related: locations which permit the viewer long lines of sight would
consequently have a larger than average perimeter and lower area to perimeter ratio.
In the Tate Gallery, the parts of the building whose isovists have a high perimeter
value are predominantly located at the ends of long visual axes (for example at the
entrance looking towards the main galleries). The areas of greatest isovist perimeter,
however, are mostly concentrated at the junctions of the major visual axes. These
locations appear to be ideal locations for subjects to stop; pausing at locations with a
longer than average line of sight is strategically sensible. Locations with an unusually
high isovist perimeter value, at the intersection of major visual axes, are locations
where a route choice decision needed to be made. Pausing at such a point to scrutinise
the environment would appear to be natural wayfinding behaviour. Reinforcing the
above results is the fact that people are also stopping in locations with much higher
than average radial standard deviation, variance and skewness, which also occur at the
intersections of the major visual axes throughout the gallery.
In terms of space syntax measures, it also appears that people are pausing
strategically. Subjects are stopping in locations with a higher than average isovist
connectivity (212.12 compared to 161.87) and are pausing in locations with a lower
than average mean depth (3.84 compared to 4.24). In summary, people are pausing in
locations that are highly connected and highly integrated in terms of the building as a
whole. They are pausing in locations where they are more likely to glean the
maximum information about the configuration of the building (both local and global
information).

5 Conclusion

This paper concludes that subjects do not appear to be pausing randomly in the
virtual Tate Gallery. People are pausing in locations offering strategic visual
properties, locations that afford unusually long lines of sight and large isovist areas.
These are also the kinds of locations where the isovists are highly integrated as well
as spatially connected. Lastly, it seems that people are pausing in locations far from
any occluding surfaces such as internal walls (which would limit available
environmental information). It may be summarised that the subjects are being
exceedingly strategic in terms of where they stop to survey the worlds. They pause
only in locations offering maximum visual, local/global information, reducing the
necessity to pause more frequently. People’s navigational tactics can therefore be seen
to be both strategic and maximally efficient.
References

1. G. L. Allen, A. W. Siegel, and R. R. Rosinski, The Role of Perceptual Context in

Structuring Spatial Knowledge, Journal of Experimental Psychology: Human Learning and
Memory, vol. 4, pp. 617-630, 1978.
2. J. Appleton, The Experience of Landscape, 1996 ed. New York: John Wiley, 1975.
3. M. Batty, Exploring Isovist Fields: Space and Shape in Architectural and Urban
Morphology, Environment and Planning B: Planning and Design, vol. 28, pp. 123-150,
4. M. L. Benedikt, To take hold of space: isovists and isovist fields, Environment and Planning
B, vol. 6, pp. 47-65, 1979.
5. M. L. Benedikt and C. Burnham, Perceiving architectural space: from optic arrays to
Isovists, in Persistance and Change, W. H. Warren and R. E. Shaw, Eds. Hillsdale, NY:
Lawrence Erlbaum, 1985, pp. 103-114.
6. J. P. Braaksma and W. J. Cook, Human Orientation in Transportation Terminals,
Transportation Engineering Journal of ASCE, Proceedings of the American Society of Civil
Engineers, vol. 106, pp. 189-203, 1980.
7. R. Conroy Dalton and S. Bafna, The Syntactical Image of the City: A Reciprocal Definition
of Spatial Elements and Spatial Syntaxes, presented at Space Syntax - IV International
Symposium, London, 2003.
8. R. Conroy Dalton, The Secret is to Follow Your Nose: Route Path Selection and Angularity,
Environment and Behavior, vol. 35, pp. 107-131, 2003.
9. R. Conroy and N. Dalton, OmniVista: An Application for Isovist Field and Path Analysis,
presented at Space Syntax - III International Symposium, Atlanta, Georgia, 2001.
10. R. A. Conroy, Spatial Navigation in Immersive Virtual Environments, in Department of
Architecture. London: University College London, 2001, pp. 249.
11. N. M. Dalton and R. A. Conroy, OmniVista, 1.0 ed. London, 1999.
12. J. J. Gibson, The Perception of the Visual World. Boston, MA: Houghton-Mifflin, 1950.
13. J. J. Gibson, The Ecological Approach to Visual Perception. Boston, Houghton-Mifflin, 1979.
14. J. Hanson, Decoding Homes and Houses, 1999.
15. D. Henry, Spatial Perception in Virtual Environments: Evaluating an Architectural
Application, in Human Interface Technology Laboratory. Seattle: U. of Washington, 1992.
16. B. Hillier and J. Hanson, The Social Logic of Space, 1993, reprint, paperback edition ed.
Cambridge: Cambridge University Press, 1984.
17. B. Hillier, Space is the Machine, First, hardback ed. London: Cambridge University Press,
1996.
18. K. Lynch, The City as Environment, Scientific American, vol. 213, pp. 209-214, 1965.
19. K. Lynch, The Image of the City, Twenty-third printing, 1994 ed. Cambridge, MA: MIT
Press, 1960.
20.Peponis, J., Dalton, R., Wineman, J. & Dalton, N.: Measuring the effects of layout on
visitors’ spatial behaviors in open plan exhibition settings. Environment and Planning B:
Planning and Design, 31 (2004) 453-473
21. A. Turner, Analysing the Visual Dynamics of Spatial Morphology, Environment and
Planning B: Planning and Design, vol. 30, pp. 657-676, 2003.
22. A. Turner, M. Doxa, D. O'Sullivan, and A. Penn, From Isovists to Visibility Graphs: A
Methodology for the Analysis of Architectural Space, Environment and Planning B:
Planning and Design, vol. 28, 2001.
23. A. Turner and A. Penn, Making Isovists Syntactic: Isovist Integration Analysis, presented at
2nd International Symposium on Space Syntax, Universidad de Brasil, Brasilia, Brazil,
1999.
23. J. M. Wiener and G. Franz, Isovists as a Means to Predict Spatial Experience and Behavior,
presented at Spatial Cognition 2004, Frauenchiemsee, Germany, 2004.
 Digital Design with Style:
Characterising Digital Architecture
in 3D Virtual Environments for Design Collaboration

Ning Gu

School of Architecture and Built Environment

University of Newcastle, Callaghan 2308, Australia
[email protected]

Abstract. 3D virtual environments (3DVE) have the potential to make a major

impact on global design teams by supporting distant design collaboration. To
maximise the effect of 3DVE as a design tool for collaboration, it is important
to explore alternatives other than 3D geometrics for representing digital
architecture in 3DVE in order to better support the variety and complexity of
design languages that are commonly required for exploring architectural
designs during collaboration. To achieve such aim, one important step is to gain
a formal understanding of 3DVE as a design tool and current designs in 3DVE.
This paper presents the stylistic characterisations of digital architecture in
3DVE based on various design examples. The findings provide a base to
develop and integrate alternative design elements other than 3D geometrics in
3DVE for supporting sophisticated design collaboration.

Keywords: Digital architecture; 3D virtual environments; design collaboration;

stylistic characterisation.

1. Introduction

The increasing globalisation requires architecture and design firms to collaborate

across time zones and geographical locations. This has post new demands on the
design industry in terms of additional time and financial inputs for relocating human
and design resources. 3D virtual environments (3DVE) have the potential to make a
major impact on global design teams by supporting distant design collaboration.
Collaborative 3DVE provide a design platform where the design team can be co-
located without physically being there. However, similar to many other computer
aided design (CAD) modelling tools, design representations in traditional 3DVE also
focus on the representations of 3D geometric forms. This low level of representation
is inadequate to support sophisticated design collaboration where the team develop,
communicate and externalise ideas that are far more complex than 3D geometrics.
To maximise the effect of 3DVE as a design tool for remote team collaboration, it
is important to address the issue of design representation of digital models in 3DVE.
A research project has set out to explore alternatives for representing digital
2 Digital Design with Style

architecture in 3DVE in order to better support the variety and complexity of design
languages that are commonly required for exploring architectural designs during
collaboration, for example, layout, spatial adjacency, volume, closure, openness,
orientation and way finding. To achieve such aim, one important step is to gain a
formal understanding of 3DVE as a design tool and current designs in 3DVE. This
paper presents the stylistic characterisations of digital architecture in 3DVE based on
various design examples. The findings provide a base to develop and integrate
alternative design elements other than 3D geometrics in 3DVE for supporting
sophisticated design collaboration.

2. Stylistic Characterisations of Digital Architecture in 3DVE

To characterise art and design we usually refer to the concept of style [1]. In general,
styles can be understood as conventions or agreements that are used to recognise
similarities or differences among design instances for design studies and design
practices. A specific style is exemplified when several design instances “each create a
similar impression” [2]. Depending on the actual design domain, the style may be
further described with a set of formal properties such as shape, colour, arrangement,
texture, size and orientation [3]. The study of style is mainly concerned with the
characterisation of these properties. Architectural design has a long association with
style. The charm and complexity of different architectural styles lies in the reflections
of their cultural references, geographical references, historical references and
references to individual artistries. Digital architecture on the other hand has less-
developed design theories and principles, and its design examples are comparatively
limited. The rest of the section therefore aims at categorising the characterisations that
distinguish different designs of digital architecture in 3DVE by observing selected
current examples. We call these characterisations stylistic characterisations of digital
architecture.
The view of 3DVE as functional places that support an extended range of activities
online provides a common ground for designing digital architecture. This common
ground highlights two key issues: activities and metaphor. Firstly, 3DVE exist for
certain purposes supporting various professional activities. Secondly, their designs
apply the metaphor of architecture. Based on this understanding, designing digital
architecture in 3DVE can be divided into the following four phases: (1) To layout
virtual places/areas: each virtual place/area has a purpose that accommodates certain
intended activities. (2) To configure the virtual places/areas: each virtual place/area is
then configured with certain objects, which provides visual boundaries of the
place/area and visual cues for supporting the intended activities. (3) To specify
navigation methods: navigation in 3DVE can be facilitated to consider the use of way
finding aids or hyperlinks for assisting people’s movements among different virtual
places/areas. (4) To establish interactions: in general this is a process of ascribing
behaviours to selected objects in each virtual place/area so that people can interact
with the virtual place/area and with each other. Based on the outcomes of each design
phase, we characterise the designs of digital architecture in terms of space design and
 Digital Design with Style 3

visualisation (layout and object design), navigation and interaction. These are the
three inseparable parts that provide an integral “impression” of 3DVE.

2.1. Space Design and Visualisation

First of all, space design and visualisation of 3DVE is affected by the use of the
architectural metaphor. For example, in Figure 1, the image on the left shows a virtual
campus design that strictly applies the metaphor of a campus, one in which visitors
can find familiar references to their experiences in a physical campus. The image on
the right is a virtual museum design that applies a more abstract metaphor and
introduces different visual elements in addition to the conventional element that can
be found in a physical museum. Visitors in this virtual museum need to explore and
learn about the semantics of the abstract metaphor in order to fully absorb the
environments.

Fig. 1. Two examples from the AW educational universe (from left to right):
Wec3D virtual campus and TCWF virtual museum.

The style of the applied architectural metaphor also affects the visualisation of
3DVE. In Figure 2, the image on the left is the reconstruction of a village, depicting
the style of Van Gogh’s paintings, whilst the image on the right shows the simulation
of a rather modern building.

Fig. 2. Two examples from the AW educational universe (from left to right):
VanGogh world and VLearn e-learning centre.

Finally, different uses of forms and layouts can also change space design and
visualisation in 3DVE. For example, in Figure 3 the design shown on the left uses
rectangles, one of the geometric primitives, as the basic design element. In contrast,
4 Digital Design with Style

the design shown on the right uses more organic forms. In Figure 4, the design on the
left has its layout expanded vertically, following a spiral curve, while the one on the
right has the layout expanded horizontally along a floating path.

Fig. 3. Two examples from the AW University of Sydney universe,

designed by students (form left to right): a virtual gallery and an information centre.

Fig. 4. Two examples from the AW University of Sydney universe,

designed by students (form left to right): a virtual studio and a virtual gallery.

2.2 Navigation
Navigation in 3DVE has been studied with direct reference to way finding aids in the
physical world [4, 5]. There are at least two kinds of way finding aids that can be
integrated into 3DVE from the physical world: (1) The use of spatial elements; for
example, paths, openings, hallways, stairs, intersections, landmarks, maps, signs and
so on. (2) The use of social elements; for example, the assistance gained from guides
(softbots) or other occupants.
Besides these way finding aids originating from the physical world, 3DVE also
have their unique forms of navigation where virtual places/areas are hyper-linked.
Most 3DVE allow people to move directly between any two locations using
hyperlinks. For example, in Figure 5, the design shown on the left has its areas
spatially adjacent to each other. Therefore, people can travel from one area to another
by following the relevant paths, signs and openings. The image in the middle is an
interactive map for a virtual gallery. This map appears at several key locations of the
gallery. Visitors can mouse-click on areas that are marked with numbers to teleport
directly to the indicated locations in the virtual gallery. The image on the right
captures a snapshot of a conversational softbot implemented with Active Worlds
 Digital Design with Style 5

(www.activeworlds.com). This softbot is able to guide the visitors in the virtual

environment by responding to their general enquires.

Fig. 5. Three examples from the AW University of Sydney universe:

(left) a bird’s view of CRC world; (middle) an interactive map used in a virtual gallery
designed by students; (right) a snapshot of a conversational softbot.

2.3 Interaction

Interactions in 3DVE are realised by triggering the behaviours of relevant objects.

Designers can ascribe behaviours to selected virtual objects and carefully arrange the
ways to trigger them in 3DVE to create very interactive designs. For example, Figure
6 illustrates three different states of a virtual studio showing how the studio can
respond differently to the existence of its visitors by selectively showing/hiding parts
of the design accordingly.

Fig. 6. A virtual studio designed by students at the University of Sydney.

3. Conclusion

The paper has discussed how digital architecture in 3DVE can be characterised in
terms of space design and visualisation, navigation and interactions. Some of these
characterisations can be similar to the qualities of styles in the physical world due to
the use of the architectural metaphor. However, digital architecture also has many
unique stylistic characterisations that are different from its physical counterparts,
especially in terms of navigation and interaction.
6 Digital Design with Style

By understanding the stylistic characterisations of digital architecture, it is then

possible to develop design formalisms such as grammars [6] which can be applied to
design and implement digital architecture that capture these spatial, visual and
interactive qualities in 3DVE. In an ongoing study, we have developed a Generative
Design Grammar [7, 8] that describes a specific style of digital architecture. Applied
by computational agents, it is possible to dynamically design and implement digital
architecture that captures this particular style in 3DVE, on behalf of the designers as
needed. Generative Design Grammar uses four sets of design rules: layout rules,
object placement rules, navigation rules and interactions rules to define the design
conventions of space design and visualisation, navigation and interaction for digital
architecture. The use of design rules has shown potentials in describing and
addressing complex architectural design problems such as spatial adjacency,
orientation, way finding and interactive method that are beyond the simple operations
of 3D geometrics. For the future study, we will further explore the use of Generative
Design Grammar as a design tool in 3DVE and the use of design rules for
implementing and integrating stylistic design elements in 3DVE for supporting
sophisticated design collaboration.

Acknowledgement

This research is supported by an Early Career Researcher Grant funded by the

University of Newcastle, Australia.

References

1. Schapiro, M.: Style, Theory and Philosophy of Art: Style, Artist and Society: Selected
Papers. George Braziller, New York 1994 (1953) 51-102.
2. Stiny, G. and Mitchell, W.J.: The Palladian Grammar. Environment and Planning B 5 (1978)
5-18.
3. Kirsch, R.A.: Using Computers to Describe Style. American Indian Rock Art 22 (1998) 153-
160.
4. Vinson, N.G.: Design Guidelines for Landmarks to Support Navigation in Virtual
Environments. Proceedings of CHI’99. Pittsburgh (1999) 278-285.
5. Darken, R.P. and Sibert J.L.: Way Finding Strategies and Behaviours in Large Virtual
Worlds. Proceedings of ACM SIGCHI’96. ACM, New York (1996) 142-149.
6. Stiny G. and Gips J.: Shape Grammars and the Generative Specification of Painting and
Sculpture. Proceedings of Information Processing 71. North Holland, Amsterdam (1972)
1460-1465.
7. Gu, N. and Maher, M.L.: Dynamic Designs of Virtual Worlds Using Generative Design
Agents. Proceedings of CAAD Futures 2005. Springer, the Netherlands (2005) 239-248.
8. Gu, N. and Maher, M.L.: A Grammar for the Dynamic Design of Virtual Architecture Using
Rational Agents. International Journal of Architectural Computing 4(1) (2003) 489-501.
 A study of design collaboration in virtual
environments: interacting with the external design
representation

Leman Figen Gül1

1
School of Architecture & Built Environment, The University of Newcastle,
2308 Callaghan, Australia
[email protected]

Abstract. Recent developments in virtual environments have the potential to

bring significant changes in the way that design-related professionals
collaborate and design. Designing and collaboration has been changing with the
introduction of these new design environments. The aim of the paper is to
identify similarities and differences of interaction with the design representation
between co-located sketching and remote designing in different virtual
environments. Our results show that the type of representation has an effect on
design behaviour and on the externalisation of design ideas.

Keywords: design collaboration, external design representation, spatial and

visual perception,

1 Introduction

Recent developments in virtual environments have the potential to bring significant

changes in the way which design-related professionals collaborate and design.
Although there is a variety of research on collaborative design that examines the way
architects design and collaborate using traditional and/or digital media [1, 2], it is still
not clear how different types of design environments affect architects’ interaction
with the design representation. The aim of the paper is to identify similarities and
differences of interaction with the external design representation between co-located
sketching and remote designing in different virtual environments. In this paper, we
present two designers collaborating over four different design environments and
report a comparison of those environments with face-to-face (FTF) collaboration,
using protocol analysis [3].
The focus of the paper is on the design protocols (verbal and visual) in a collaborative
design context, based on the assumption that interaction between team members
requires effective communication through discourse and visual representations in the
design environments. Collaborative designing involves communication and working
together in order to jointly establish design goals, search through design problem
spaces, determine design constraints and construct a design solution [4].
 In design research, external design representations (sketches, models, diagrams)
are recognized as having important roles in the design process [see 5 for the details,
6]. In particular, sketches comprise possible design solutions and also seem to be
essential for recognizing conflicts and possibilities [see 7 for more detail on the
experiment setting] as well as for revising and refining ideas, generating concepts and
facilitating problem solving [5, 8]. Designers sketching in the early stages of a design
process and the role of sketching in designing have been examined [borrowed from 9]
showing the different roles sketches play in visuo-spatial reasoning and their
connection to different aspects of the design process.

2 Studying design collaboration

In order to understand the changes in designers’ behaviour, there is a need to have

baseline data that characterises collaborative design activity without the technology,
that is, co-located sketching. Then there is a need to gather the data to characterise
and identify the changes that occur in design collaboration when the different
technologies are introduced. With these ideas in mind, a series of experiments were
conducted to investigate the impact of technology on design collaboration. The
empirical data forms two groups: baseline study (co-located sketching) and
comparison study (remote designing) [see 5, for more detail6]

2.1 Baseline study

In the baseline study, the aim is to understand the nature of the collaborative design
process when the designers are using traditional materials: pen, paper, scale, etc. and
without the digital systems for designing and communication, as shown in Fig. 1.

Fig. 1 The baseline study, face-to-face design session

In the baseline study (face-to-face design session, FTF), two architects are
collaborating on a design task in co-located sketching in which they are working
together around a table and develop a design representation using traditional materials
(pen-paper). The designer’s actions and communications are captured in the Digital
Video Recording (DVR) system [10].
2.2 Comparison study

In the comparison study, the aim is to compare three collaborative design sessions
with the baseline study (FTF): (1) FTF and remote sketching (RS), (2) FTF and 3D
virtual world (3D), and (3) FTF and 3D virtual world with sketching (3DS). Fig. 2
shows the DVR views of the three experiments in which the architects are
collaborating within the technology. The designers were located in the same room
with a panel between them to simulate high bandwidth audio communication. The
same architects were given different tasks of similar complexity in each setting.
Similar to the baseline study, their actions and communication were recorded in the
DVR system.

a b c

Fig. 2 The Comparison study, two designers collaborating within three different collaborative
virtual environments: (a) RS-Groupboard, (b) 3D-Active Worlds, (c) 3DS-DesignWorld

In the remote sketching (RS) session, the architects used a shared whiteboard
application (Groupboard) and digital pen interfaces (Mimio and SmartBoard), as
illustrated in Fig. 2a. Mimio and SmartBoard were digital touch systems allowing the
designers to use the digital pen as a mouse and to write in digital ink on the screen.
In the 3D virtual world (3D) session, the architects design in Active Worlds using a
typical desktop system with mouse, keyboard and a monitor, as shown in Fig. 2b.
In the 3D virtual world with sketching (3DS) session, the architects used a prototype
system, DesignWorld1 which included a 3D virtual world augmented with a number
of web-based communication and design tools, as shown in Fig. 2c (see [7], for
details of the experiments, training sessions and DesignWorld).

3 Method

Protocol analysis, which was first adopted by Eastman (1968) to study design
cognition, has been used as a research technique. We examine the protocols by using
a coding scheme, interaction with the design representation, which includes the
verbalisation and the visualisation of the external design representations (verbal-
visual design protocols) to communicate the design ideas with her/himself and/or to
others. The interaction with the design representation coding scheme captures: (1)
how architects create the external design representation, (2) how they approach
construction of the design representation, (3) how they use visual information and

1
This prototype was developed as part of a CRC Construction for Innovation project.
how they inspect/interact with the interface/tools, given materials and the
representation, and (4) what visuo-spatial features of the representation they focus on
while they are developing the design solution, as shown in Table 1.

Table 1 The first level of the interaction with the design representation coding scheme
DESCRIPTIONS Types of data
Interaction with design representation Verbal & Visual
Realisation Looks at discussions and actions about concretisations of design ideas: (1)
Realisation action: create – write – continue – delete and (2) Realisation
process: modelling – describe – decision (adapted from [7, 11]).
Agents Actions Looks at actions that are related to designers’ engagements with the
surrounding space: onTools – onElements - gesture
Perceptual focus Looks at discussions and actions that are related to visual features/form
articulation and spatial relationships of the design elements: spatial
relationships: alignment, arrangement, egocentric, allocentric -
object/entity: size, form, surface.
Design Space Looks at discussions that are related to dimensions of design space: 2D-3D
Representation Looks at actions that are related to the types of representation used: 2D-3D
Mode (adapted from [12])
Collaboration Looks at actions that are related to shared representation activities: meeting
mode – individual (borrowed from [13]).

The design protocols are divided into smaller units. This process is called
segmentation. The data of the study consist of a continuous stream of video and audio
that has two sources, the designers 1 (Greg) and designer 2 (Lee). Similar to Maher et
al. [7], each design session is segmented twice: (1) reflecting Greg’s design actions
and intentions, and (2) reflecting Lee’s design actions and intentions. Consequently,
the two major segmentation rules, which are the utterances-based segmentation
method [11] and the actions-and-intentions based segmentation method [14], are
combined in this study.

4 Results

The duration percentages of each action category are examined to measure the
similarities and differences of designers’ behaviour in each design session. Fig. 3
shows the duration percentages of the designers’ realisation actions comparing the
baseline study with the virtual environments. As shown in Fig. 3a, in the baseline
study, the duration percentages of the create and the write actions are higher, when
the designers spent time on writing down the areas and listing the requirements, and
drew the design solution. In the RS session, there is an overall increase in the duration
percentages of the realisation actions, in that the duration percentages of the create
action are higher, compared to the baseline study, as shown in Fig. 3a. The 3D and the
3DS sessions show different trends of the realisation actions, compared to the baseline
study. The duration percentages of the continue element action is significantly high,
followed by the create element and the write action categories in the 3D modelling
environments, as shown in Fig. 3b and c. This also consisted of a cycle of actions
such as move/rotate/transfer/group, etc., as pointed out by Maher et al. [7]. Thus the
“continue” action consists of a series of actions that require a continuing attention on
the particular object.
FTF 3D c - Rea lis atio n ac tio n s F T F - 3 D S FTF 3D S
a - Rea lis a tio n ac tio n s F T F - RS FTF RS b - Rea lis atio n a c tio n s F T F - 3 D

40% 40% 40%

35% 35% 35%

duration percentages per code

duration percentages per code

duration percentages per code

30% 30% 30%

25% 25% 25%

20% 20% 20%

15% 15% 15%

10% 10% 10%

5% 5% 5%

0% 0% 0%
create continue delete w rite create continue delete w rite create continue delete w rite
actions actio ns actio ns

Fig. 3 The duration percentages of the realisation actions: (a) FTF and RS, (b) FTF and 3D, and
(c) FTF and 3DS

Fig. 4 shows the duration percentages of the perceptual focus actions of the designers
comparing the baseline study with the virtual environments. The duration percentages
of the object/entity action are higher in the baseline study, as shown in Fig. 4a. The
RS session shows a similar trend, with a drop in the duration percentages, as shown in
Fig. 4a. In the 3D and 3DS sessions, there is an increase in the duration percentages of
the spatial relationships actions, compared to the baseline study, as illustrated in Fig.
4b and c.
a- Perceptual focus FTF - RS FTF RS b- Perceptual focus FTF - 3D FTF 3D c- Perceptual focus FTF - 3DS FTF 3DS

50% 50% 50%

duration percentages per code

40% 40% 40%

duration percentages per code

duration percentages per code

30% 30% 30%

20% 20% 20%

10% 10% 10%

0% 0%
0%
object spatial object spatial
object spatial actions
actions actions

Fig. 4 The duration percentages of the perceptual focus actions: (a) FTF and RS, (b) FTF and
3D, and (c) FTF and 3DS
The durations of the spatial relationships actions are investigated in order to
understand what kind of spatial features the designers focused on. Table 1 shows the
duration percentages of the spatial relationships actions in all the design sessions. In
the baseline study, the time spent on the spatial features of the design representation
was limited (the highest percentage is 7.2%, for the arrangement actions), as shown in
Table 1. We observed that the architects focused more on the alignment action
(spatial adjacency of the design objects) in the 3DS sessions (30.9%), and focused
more on the arrangement action (how the objects should come together) in the 3D
session (12%). The duration percentages of the allocentric referencing are high in the
baseline study. There is an increase in the duration percentages of the allocentric and
the egocentric actions in the 3D virtual worlds, compared to the baseline study, as
shown in Table . The highest percentages of the actions are shaded in grey, as shown
in Table .
Table 1. The durations of the perceptual focus on the spatial relationships action (second and
percentages of the total elapsed time)

Sec - % FTF RS 3D 3DS

alignment 105.44 2.9% 95.07 2.6% 742 20.6% 1113.54 30.9%
arrangement 258.24 7.2% 129.26 3.6% 444.47 12.3% 339.03 9.4%
egocentric 18.47 0.5% 110.3 3.1% 94 2.6% 81.4 2.3%
allocentric 170.57 4.7% 133.17 3.7% 280.54 7.8% 206.3 5.7%

4 Conclusions

First, we conclude that the two sketching sessions show a similar trend, that the
realisation actions of the design representation are based on the “create” and the
“write” actions. In contrast, in 3D modelling environments, the realisation actions of
the design model were based on the “continue” action. Second, analysis of the
protocol shows that the type of presentation has an effect on designers’ perceptual
focus on the spatial properties of the design solution: (1) the designers focused more
on the visual features of the design, which are size, form, colour and materials, while
sketching, and (2) the designers focused on the spatial relationship of the design
objects, which are spatial adjacency, arrangements, position, etc., while 3D
modelling. Third, our analysis showed that there was an increase in the designers’
referencing (egocentric and allocentric) in 3D virtual worlds, and they tended to
position themselves outside the design representation in sketching. In contrast, some
actions that are related to interaction with the surrounding design space were different
between the both sketching environments. For example, while they were engaging
more with the problem space and the solution space in the baseline study, the same
designers engaged more with the tools and interface of the applications in the RS
session. Different interaction behaviour was also observed in both 3D modelling
environments. For example, in the 3D session, the designers engaged more with the
tools and interface of the applications. On the other hand, in the 3DS session, similar
to the baseline study, the designers engaged more with the visual analysis of the
design solution whereby they inspected the model by flying over and walking through
it.

Acknowledgments. The empirical data that are used in this paper were collected for a
research project, “Team collaboration in high bandwidth virtual environments”, and
were provided by the Cooperative Research Centre for Construction Innovation (CRC
CI) (CRC study). The author wishes to thank the project partner companies and
researcher who worked in the project. Special thank also goes to the supervisor of this
study Prof. Mary Lou Maher whose contribution was invaluable.
References

1. Hennessy, S. and P. Murphy, The potential for collaborative problem solving in

design and technology. International journal of technology and design education,
1999. 9: p. 1-36.
2. Seitamaa-Hakkarainen, P., H. Lahti, H. Muukkonen, and K. Hakkarainen,
Collaborative designing in a networked learning environment, in Collaborative
Design, S.A.R. Scrivener, L.J. Ball, and A. Woocock, Editors. 2000, Springer-
Verlag: London. p. 411-420.
3. Akın, Ö. and C. Lin, Design protocol data and novel design decisions. Design
Studies, 1995. 16(2): p. 211-236.
4. Do, E.Y.-L., M.D. Gross, B. Neiman, and C. Zimring, Intentions in and relations
among design drawings. Design Studies, 2000. 21(5): p. 483-503.
5. Maher, M.L., Z. Bilda, and L.F. Gül. Impact of collaborative virtual environments on
design behaviour. in Design Computing and Cognition'06. 2006: Springer.
6. Maher, M.L., L.F. Gül, and Z. Bilda. Studying Design Behaviour in Collaborative
Virtual Environments. in Proceedings of 16th World Congress on Ergonomics
(IEA2006 Congress). 2006. Maastricht, Netherlands: Elsevier Ltd.
7. Maher, M.L., Z. Bilda, N. Gu, L.F. Gül, Y. Huang, M.J. Kim, M.L. Maher, D.
Marchant, and K. Namprempree, Collaborative Processes: Research Report on Use
of Virtual Environment. 2005, the University of Sydney: Sydney.
8. Gül, L.F. and M.L. Maher. The impact of virtual environments on design
collaboration. in 24th eCAADe Conference Proceedings, ISBN 0-9541183-5-9.
2006. Volos, Greece.
9. Atman, C.J. and K.M. Bursic, Verbal protocol analysis as a method to document
engineering student design processes. Journal of Engineering Education, 1998. 87(2):
p. 121-133.
10. Kvan, T., R. West, and A. Vera. Tools and channels of communication: Dealing with
the effects of computer mediation on design communication. in 1st International
Conference on Creative Collaboration in Virtual Communities. 1997. University of
Sydney.
11. Gabriel, G.C., Computer mediated collaborative design in architecture: the effects of
communication channels on collaborative design communication, in Architectural
and Design Science, Faculty of Architecture. 2000, University of Sydney: Sydney.
12. Gero, J.S. and T. McNeill, An approach to the analysis of design protocols. Design
Studies, 1998. 19: p. 21-61.
13. Maher, M., Z. Bilda, L.F. Gül, and D. Marchant. Studying design collaboration in
virtual worlds. in Studying Designers'05. 2005: Key Centre of Design Computing and
Cognition, University of Sydney.
14. McNeill, T., J.S. Gero, and J. Warren, Understanding Conceptual Electronic Design
Using Protocol Analysis. Research in Engineering Design, 1998. 10: p. 129-140.
 What about People, Do They Matter?
Insights on the Impact of People’s Presence in the
Built Environment

João Pinelo1

Bartlett School of Graduate Studies, University College

London.1-19 Torrington Place, Gower St., London WC1E 7HB
1
This paper reports work which, as part of the PhD Studies
of the author, is funded by the Portuguese Foundation For
Science and Technology - FCT
[email protected]

Abstract. The identification of spatial properties responsible for influencing

pedestrian navigation in non-familiar environments has been the focus of social
scientists’ research. Two findings from this were put together in this paper:
First, the fact that people seem to follow global properties of configuration
while navigating in complex buildings and urban environments; second, several
indications that the presence of other people in the urban environment might
influence the way an observer interprets and stores information about it. These
lead this research to focus on – the presence of people. A questionnaire was
made to investigate the impact of this to an observer in non-familiar urban
environments. Results suggest that people are seen as an affordance of the
environment, then acting as a feature considered on route choice. This finding
has implications both in spatial cognition and navigation processes, thus being
relevant to architecture in two different ways: conception/presentation and
building use.

Keywords: Environmental Cognition, Wayfinding, Space Syntax, Natural

Movement, Affordances, Awareness, Human Models, Complex Buildings.

1 Introduction

Pedestrian navigation in built environments has been a research focus of social

scientists for at least forty years. Trying to identify spatial properties responsible for
influencing this has been the main goal. Two findings from this line of research are
assembled and explored in this paper. First, some research findings have pointed to
the relevance of people in buildings and in the urban environment as possible
interference in the way an observer perceives and stores information about the
environment (Appleyard, 1970; Beaumont et al., 1984; Evans et al., 1982). Second,
Peponis suggested that people seem to follow global properties of configuration while
exploring a building (Peponis et al., 1990).
 The implications of these possibilities are far reaching and its exploration may help
to the comprehension of route choice behaviour in complex buildings and in the urban
environment, and potentially to the understanding of how individual behaviour creates
global patterns of movement. This means insights both to the Man-Environment and
Individual-Social paradigms. Also, considering the importance of people to the
understanding of environments, the space of graphics (Tversky, 2003) (namely
architectural and urban representations) may benefit from this knowledge. These lead
this research to focus on local properties of space which are simultaneously a
consequence of configuration, and which has been recognised as a relevant feature of
the urban environment – the presence of people.
Sociologists and psychologists give many possible explanations to the importance
of social interaction. These are suspected to be the cause of such importance given to
the presence of other people while navigating. However, it is not the intention of this
research to deeply investigate this, for now the aim is to check if this is the case. From
these considerations, three questions emerged, and a questionnaire was conducted in
order to answer them.

2 The Questionnaire

A questionnaire was made in order to explore the relationship between the

individual and other humans in non-familiar environments. The questions to answer
were: first, are people sensible to the presence of other people in the public space?
Second, in case they are, is it to the point that the density of people may have
influence on route decision? Third, what do people seem to associate to the presence
of different densities? These questions were addressed by showing three pictures of a
virtual street populated with different densities of people and ten questions were
asked.
The sample consists of 275 respondents of 29 countries. 168 female (61%) and 107
male (39%).

2.1 Results

The first question intended to answer: do people consider the same physical
(inanimate set) populated with different densities of people as different environments?
This question is based on the assumption that in case a difference is reported, being
the human presence the only difference, people not only actually see people, but get
from them enough information so as to consider the environment as different. The
answers clearly point towards the difference.
In question 6, people were asked to choose the street they would follow on the
context of a common exploratory situation. The text was: “Imagine you are just out of
the hotel in a city which you are visiting for the first time. You will start exploring the
surroundings. Which way would you take? Would you follow street a, b or c?”
Moreover, the results allowed people to say that “they have no idea”, and for those
who choose a street, the level of confidence was explored by giving to each street, the
possibility of choosing the expressions: “I think I would follow street…”, or “I am
sure I would follow street…”. In answer to this question 64% of the sample choose
street c (the most populated. Being a the less populated). From these, 57% showed a
limited level of confidence by choosing: I think, but 43% (28% of the total) showed
extra confidence by choosing: I am sure. 3% of the total answered that they had no
idea. The answer to this question stresses out the meaning of the previous one,
emphasizing that people are sensible to the presence of other people to the point that
they can feel the same physical set as different when populated by different densities
of people, but also that they show a clear preference for more crowded spaces (in non
familiar places).
This evidence becomes even more conspicuous when one looks at the 20% of the
respondents who choose I think street b (the second less populated). Because
cumulative answers to street a only account for 10%, the 20% I think street b is seen
as from people who were probably tempted to choose street c, but maybe found it too
crowded (especially due to the absence of I sure b). Even people who choose street a,
they are choosing it, so, they feel the difference. Anyway, the purpose of this question
was not to evaluate preferences in street density population, but simply to detect if
people’s density may have influence on route choice. It apparently does.
In questions 7, 8 and 9 people were asked to choose from one list of commercial
activities the ones they would expect to find in each street. This question tried to
investigate what people see in the environmental difference they report. One
possibility was thought to be that they could see more diversity in terms of
destinations, commercial types were used to test this. What is in stake is not to test
what kind of shop fit where. What really matters here is that overall scores show that
people identify more types of commercial activities in more populated streets. The
results show a clear tendency of people to associate more diversity of commerce to
more populated streets. Street a has a mean of 2.23 types of commerce, street b of
3.20 while street c has 4.42.
Despite all the evidence that people are sensitive to the presence of other people in the
urban environment, that it may influence route choice in non familiar places and that
they apparently associate more commercial diversity to more people, the reasons to
choose a certain street is not conclusive. In question 10, people were asked to rank
from 1 to 5 the features they took in account when they made the route choice in
question 6. This was the only question which imply that people bring to conscience
and justify a previous choice. The unambiguousness of choice showed in previous
answers is absent to the answer to what made them make that choice.

3 Implications of the Findings to Environment and Behaviour

Theories
Consequences of these findings have an impact on established theories on urban
collective movement and individual behaviour; namely help to explain the links
between individual cognition and collective patterns and between man and
environment.
The author suggests in this paper that people are seen by other people as
affordances (Gibson, 1986). In this case, a very special kind because apart from
affordances as security, potential for interaction, for help, etc, some results of this
questionnaire (as the answer to question 5) suggest that we recognise ourselves in
others and recognise their actions as potentially our own and see ‘second hand’
affordances in their presence. In other words, the suggestion is that we see the
presence of other people as a sign of the existence of affordances; implying that some
kind of positive correlation exists between the two.
The fact that people seem to follow global patterns of configuration before they get
to know it, as suggested by Peponis (Peponis et al., 1990), may find here an
explanation. People do not know the configuration, but they may perceive signs of it
by comparing people’s density in different spaces. It is not clear if they read the
presence of people as configurational properties of the setting, or as local signs of
diversity, safety, or something else; however, what are these features (including
configurational properties) if not affordances?
Whatever the case, what are the implications of the fact that people recognise
differences in environments in function to the density of present people? And that, at
least to a certain point, they tend to choose the most populated ones?
The answer for this lies in the way one uses this information while exploring an
environment.

3.1 The Process of Getting to Know an Environment

The expression ‘familiar environment’ is often used in literature to refer to a space

or set of spaces which the individual knows. To refer to the opposite idea the term
‘non-familiar’ is used. However, it is often forgotten that an environment which we
refer to as being familiar implies that it has been non-familiar in the past. This calls
attention to the one-way process of getting to know an environment. If there is a
process, its understanding lies on the understanding of its rules. So, what are the
consequences of the application of the rule suggested by the questionnaire, that people
tend to be attracted by people? The recall of the process reshapes the question: what
are the implications of the systematic application of this rule? The author suggests a
look at it in through: the Individual-Social.

3.2 Implications to the Individual-Social Paradigm

If one considers for a moment that each urban environment is always new for each
one of us at some point in time, the situation portrayed by the presented experiment
will apply to each one of us in each environment that we know. The implications of
this suggest that we all get to know environments by using most populated spaces
first. These being the ones we first use, it is reasonable to accept they are going to be
the first ones to be stored in our memory. It is so logical that they are the ones we use
in the next day because they are what we know. The need or will to go to other places
will obviously widen our knowledge and allow us to store more information (although
the tendency when exploring new ways should be the same of using more populated
streets). This being the process of environmental knowledge acquirement and storage,
it is logical to expect that the more populated streets are more used, more present in
our journeys, and naturally more present in our mental environmental storage device.
One of the findings which deployed this research, that higher densities of people
around buildings reinforce the memory of those buildings (Appleyard, 1970; Evans et
al., 1982) reinforces this argument. It is then reasonable to expect the more populated
streets to be more important in our mental scheme not only for being more used, as for
being more populated. This shows a multiplier effect of people density and so its
importance.
Another phenomenon may join this in the cases where natural movement theory
(Hillier et al., 1993) applies. The most accessible streets being the most populated,
another multiplier effect takes place. This is of particular importance because, over
time, it may lead to people finding those streets useful in terms of access and this may,
again, reinforce their presence in our mental environmental storage device. Also, in
this case, the absence of exploring other routes will diminish the lack of options and
by doing that reinforce the use of the known ones, keeping them ‘central’ in our mind,
and life. Here a clear connection about the individual and the collective behaviour
emerges. In one person’s point of view a certain space is more attractive than another
and, as just explained, the use of it leads to repetition. This repetition over time creates
an individual pattern which could be irrelevant in terms of the collective pattern.
However, the individual pattern is deeply anchored in the presence of other people, or
in other words it is bound to the collective pattern. The principle seems to be based on
a cycle where people attract more people. Questionnaire results however also point to
a possible pervasive effect of this, suggesting that in a certain point people may find
some places too crowded and using instead the next less populated.
Haq also considers the hypothesis of a cycle where “Configuration creates
movement, which in turn promotes an understanding of the configurational properties.
This then contribute to more accurate movement and wayfinding.”(Haq et al., 2003).
This idea may well be right but it still does not explain how people are attracted to
configurationally relevant spaces in the first place. In the present paper the suggestion
is that configuration is ‘empowered’ by individual cognition which attracts people to
where other people already are. Haq also mentions the process of knowledge
acquirement suggesting that the local properties of connectivity is more meaningful
than integration in the early phase of contact with a new environment and that
afterwards individuals follow the space syntax measure of integration (Haq et al.,
2003). This may be the case, but to state that imply more direct analysis on the impact
of the measure of connectivity on people’s behaviour. However, this is an important
reminder which leads to the fact that rules may change within the process of learning,
claiming extra attention to the process, where knowledge seem to bound and influence
rules relative weight.
Another rule was identified in 2003. Dalton shows that people knowing the
direction they have to take to their destination, tend to follow it, choosing the streets
which imply the smaller deviation from it (Dalton, 2003). Note that the existence of
Dalton’s rule concerning directions is not a contradiction to the one suggested here.
Several rules may coexist, however further work is needed in order to understand their
relative weight.
 4 Conclusion

If an idea of configuration in its own right is built in our mind as we experience it,
many may be the environmental signs we read in order to build it. This paper shows
evidence that the presence of people may be one of them. Both by influencing the way
people move as well as one’s mental representations of environments, the presence of
people seem to be very powerful in influencing both the way space is perceived and
the way it is used. This being true to exploratory behaviour, is significant to the way
people use large spaces as complex building or urban environments, and its multiplier
effect should be considered in urban design strategies.
The suggestion is that we learn the environment by own experience but we rely on
others’ experiences in order to make it more effective.
Additionally, this evidence claims attention to the way architects and urban
designers populate drawings and models. If people are so influenced by the presence
of other people in environments, this means that they rely on the first to perceive the
latter, this may imply that the wrong use of human models in architectural
representations may be counter productive, but the adequate use may empower
understanding.
These conclusions suggest both the importance of visibility of populated spaces in
environments as ‘natural signage’, and the importance of the correct use of human
models in architectural representations, both in final representations and in conception
stages.

References

1. Appleyard, D. (1970) Styles and methods of structuring a city. Environment and Behavior, 2,
100-116.
2. Beaumont, P. B., Gray, J., Moore, G., and Robinson, B. (1984). orientation and wayfinding it
the tauranga Departmental Building. Proceedings, Environmental Design Research
Association Proceedings (
3. Dalton, R. C. (2003) The secret is to follow your nose - Route path selection and angularity.
Environment and Behavior, 35, 107-131.
4. Evans, G., Smith, C., and Pezdeck, K. (1982) Cognitive maps and urban form. Journal of the
American Planning Association, 48, 232-244.
5. Gibson, J J (1986), The ecological approach to visual perception, Hillsdale, N.J: L. Erlbaum.
6. Haq, S. and Zimring, C. (2003) Just down the road a piece - The development of topological
knowledge of building layouts. Environment and Behavior, 35, 132-160.
7. Hillier, B., Penn, A., Hanson, J., Grajewski, T., and Xu, J. (1993) Natural Movement - Or,
Configuration and Attraction in Urban Pedestrian Movement. Environment and Planning B:
Planning and Design - Pion, 20, 29-66.
8. Peponis, J., Zimring, C., and Choi, Y. K. (1990) Finding the Building in Wayfinding.
Environment and Behavior, 22, 555-590.
9. Tversky, B. (2003) Structures of mental spaces - How people think about space.
Environment and Behavior, 35, 66-80.

View publication stats