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

net/publication/322593493

The Hide and Seek of Workspace: Towards Human-Centric Sustainable

Architecture

Conference Paper · January 2018

DOI: 10.1145/3173574.3173649

CITATIONS READS
9 777

4 authors:

Hamed Alavi Himanshu Verma

Université de Fribourg École Polytechnique Fédérale de Lausanne
37 PUBLICATIONS   443 CITATIONS    22 PUBLICATIONS   196 CITATIONS   

SEE PROFILE SEE PROFILE

Jakub Mlynar Denis Lalanne

Charles University in Prague Université de Fribourg
37 PUBLICATIONS   29 CITATIONS    177 PUBLICATIONS   3,010 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Multimodal dialogue modeling View project

Turkish Audio-Visual Bipolar Disorder Corpus View project

All content following this page was uploaded by Jakub Mlynar on 19 January 2018.

The user has requested enhancement of the downloaded file.

 The Hide and Seek of Workspace:
Towards Human-Centric Sustainable Architecture
Hamed S Alavi Himanshu Verma Jakub Mlynář Denis Lalanne
Human-IST Institute Human-IST Institute Dept. of Social Sciences Human-IST Institute
University of Fribourg University of Fribourg University of Fribourg University of Fribourg
Switzerland Switzerland Switzerland Switzerland
[email protected] [email protected] [email protected] [email protected]

ABSTRACT HCI that pursues the broadening of its scope over and above
This contribution exemplifies how the study of space percep- the realm of “artifacts” to the realm of “environments”. In the
tion and its impact on space-use behavior can inform sustain- vision of Ubicomp, this can be seen as a logical development,
able architecture. We describe our attempt to integrate the following the transition from virtual to physical-tangible.
methods of user research in an architectural project that was
Along with the scattered instances of HCI research in built
focused on optimization of space usage.
environments, framing attempts have also been proposed. A
In an office building, two large office rooms were refurbished recent example is the emerging notion of Human-Building
to provide desk-sharing opportunities through hot-desking. Interaction (HBI), introduced in CHI 2016 [3], which has
We studied the space-use behavior of 33 office workers over sought to provide a unifying framework that can be used to
eight weeks in those two rooms as well as their occasional compare and relate the converging research efforts from the
presence in ten other areas (cafeteria, atrium, meeting rooms, fields of architecture and HCI. The discourses on the definition
etc.). Quantitative and qualitative analyses were performed to and scope of HBI and the attempts to chart its landscape have
understand the nature and nuances of space occupancy at the resulted in a possible classification, in which two themes are
scope of the building and within the refurbished offices. While notable: physical and spatial [2, 3]. In the following, we
at the scope of building the patterns of movements between briefly describe these two themes and then specify how the
rooms were found to be related to the professional profile of work that we present in this paper contributes to the current
the users, at the scope of office the occupancy patterns were in- challenges of HBI research.
fluenced by the spatial design of workspaces. More precisely,
The physical theme accounts for the functional logic of the
certain visual attributes of a workspace, namely Visual Expo-
building – a shelter that guarantees the basic measures of ther-
sure and Visual Openness, could determine whether or not it
mal, visual, and acoustic comfort. Historically, in the research
was regularly used. In this paper, we describe our findings in
domain of architecture, comfort has been investigated as an
detail and discuss their implications for sustainable building
objective quality of building (e.g. [11, 16, 29, 37]). Recent
design.
works, nevertheless, have extended this view by highlighting
the role of human. They suggest that comfort can be studied
ACM Classification Keywords
and designed as interactive experiences with the built environ-
H.5.m. Information Interfaces and Presentation (e.g. HCI):
ment [4, 12]. The attempts to study smart automation systems
Miscellaneous
(e.g. smart thermostat [1]) are prevalent examples of research
contributions to the physical theme of HBI.
Author Keywords
Sustainable Architecture; Human-Building Interaction; Visual The spatial theme, the one upon which we wish to focus at-
Attributes of Space; Sustainability; Shared Offices. tention, corresponds to what Hillier and Hanson, in their 1984
book The Social Logic of Space, describe as the “meaning” of
INTRODUCTION the building. They argue that built environments inherently
In recent years, Human-Computer Interaction (HCI) has of- differ from artifacts (or assemblies of artifacts) because they
fered numerous and notable contributions to the study of create and order the empty volumes of space, and it is this
humans’ interactive experiences within buildings and urban very spatial ordering that shapes the occupants’ individual and
spaces. These contributions represent a growing branch of social experiences. “Spatiality”, in their view, is what com-
plicates the relation between the physical-material attributes
Permission to make digital or hard copies of all or part of this work for personal or exhibited by buildings and their meanings to the users. With
classroom use is granted without fee provided that copies are not made or distributed this stance, a transition from studying artifacts to studying
for profit or commercial advantage and that copies bear this notice and the full citation
on the first page. Copyrights for components of this work owned by others than the environments directly entails the development of new methods
author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or that can help investigate spatiality as their point of divergence.
republish, to post on servers or to redistribute to lists, requires prior specific permission The project we describe here contributes to this method devel-
and/or a fee. Request permissions from [email protected].
CHI 2018, April 21–26, 2018, Montreal, QC, Canada opment agenda.
© 2018 Copyright is held by the owner/author(s). Publication rights licensed to ACM.
ACM 978-1-4503-5620-6/18/04...$15.00
https://doi.org/10.1145/3173574.3173649
There is precedence for the use of space analysis tools (e.g. connected if they are mutually visible, or, in another version,
Space Syntax) in assessing the impact of alternative architec- if their isovists intersect.
tural layouts on human perception and behavior (as we briefly
review in the next section). The distinguishing component in Isovist and Space Syntax theoretical concepts, representations,
the method that we designed and examined, however, is the measuring techniques, and software tools have been capital-
integration of user studies throughout the process of the archi- ized on by researchers and practitioners in various contexts to
tectural project (with specific sustainability goals), as well as explore and foresee possible links between spatial structure
the system of data collection that provides multi-dimensional and user experience. The studied environments cover a wide
information about the users. In an office building, we ob- range, from cities and urban public spaces to public buildings,
served 33 office workers over eight weeks. Two pervasive offices, schools, and private homes. The targeted questions
also vary largely depending on the project at hand. In the
sensing systems were complemented by on-site observations
following, we review a selection of these works that address
and semi-formal interviews. At the scope of building, Blue-
two specific topics that are of considerable relevance to our
tooth beacon bracelets were used to monitor the occupants’
contribution.
space-use behavior between rooms, and at the scope of the
room, infrared cameras were employed to monitor how dif-
Interpersonal Interaction
ferent areas within the rooms were used. The visualization of
Spontaneous interactions between co-workers have been
the user data superimposed on the spatial layout of the rooms
shown to be constrained and guided by the spatial config-
generated hypotheses about why certain workspaces were not
uration of office areas. Hillier and Penn found correlations
used as expected, and then inferential statistics was applied
between the level of spatial integration of an area in the build-
to validate those hypotheses. In this paper, the data collec-
ing and the degree of (perceived) “usefulness” of the people
tion, analyses, results, and implications for sustainable space
who worked in that area [27, 28]. They suggested that this
design are described in detail. Before that, in the next two sec-
is due to the formation of impromptu conversations resulting
tions, we first briefly give an overview of the research domains
from frequent personal contacts. Conory extended these stud-
within which we offer a methodological contribution, and then
ies by investigating the visual configuration of the area while
describe the specific architectural project within which the ob-
removing the spatial integration variable. She found that the
jectives of our study are defined and our contextualized results
shape and size of the isovist at a work desk had an impact on
are utilized.
whether the person who worked at that desk was recruited into
conversations (an unpublished manuscript cited by [14]).
RELATED WORK
Our contribution falls at the intersection of HBI, particularly Visual Privacy
the research on human experiences in relation to architectural The perception of visual privacy was studied by Shach-Pinsly
design, with applications for sustainability. et al. [39] in relation to visibility characteristics of built envi-
ronment in residential neighborhoods. Based on the study of
3D isovists, they proposed a method of spatial analysis to eval-
Behavioural and Perceptual Studies in Architecture
uate the potential visual penetration into one’s privacy from
Following the discourses in experimental psychology that pro-
external spaces (street or other buildings). In another study,
pounded spatial structure as being fundamental in the experi-
Beck [7] used Space Syntax methods to investigate visual
ence of environment [40], the empirical attempts to study the
privacy of indoor areas in justice court facilities. He mea-
perception of architectural spaces began to flourish already in
sured the visual exposure of workspaces to other co-workers
the 1960’s (e.g. [20, 21, 22, 23, 25, 8, 34, 36]).
and found negative correspondence to their user evaluation
The breakthrough, however, was brought about by the mathe- gathered through questionnaires.
matical ways of describing architectural spaces, intended to
It was also interesting for our work to consider the attributes
make them more susceptible to quantitative analysis. In a
of space that may influence social interaction (similar to the
highly influential work, Benedikt [9] proposed a formal de-
work of Conory), or the perception of visual privacy (similar
scription of environment based on “visibility” polygons. In his
to the work of Shach-Pinsly et al.). However, what sets our
formulation, for a given vantage point x, the set of all points
contribution apart from the reviewed works is the very objec-
that are visible from x create the “isovist at x”. An isovist
tive of our analysis, and consequently the approach that we
is identified by its vantage point, size (area and perimeter),
took. We started with a rich set of quantitative and qualitative
and shape (which can also be numerically presented through
user data and extracted certain patterns of space use. The
measures such as circularity and skewness). Benedikt sug-
primary objective of our space analysis was to understand why
gested that isovists can represent perceptual and cognitive
those patterns appeared. Such understanding contributed to
factors such as visual privacy and control, and thus studying
the knowledge about user experience within the built environ-
them may allow certain spatial behaviors to be explained and
ment, which is broadly recognized as being indispensable for
predicted. Pursuing a broader objective, the theory of Space
attempts to design sustainable buildings [35].
Syntax [26] describes buildings and urban spaces through
graph representations. Broadly speaking, the nodes are the ar-
eas that the spatial layout creates, and the edges correspond to HCI Methods in Sustainable Design
the physically immediate adjacencies. In another type of repre- Blevis’ manifesto recommending the inclusion of sustainabil-
sentation, Benedikt’s isovists are adopted: a pair of points are ity within the core semantics of interaction design [10] has
since been followed by noticeable growth in research on sus- different areas within a building are perceived by the inhabi-
tainable HCI. A multitude of sustainability studies spanning tants, as well as the affordances of distinct spaces in terms of
across varied orientations and genres have been found to potential activities they endorse. In this article, we present this
address two kinds of issues based on Mankoff’s classifica- sub-project from an HCI point of view, while describing our
tion [33]. First, sustainability in the product design process collaboration with the other stakeholders. The collaborators
(e.g. [31, 32]). Second, encouraging sustainable ways of liv- (project stakeholders) comprised of experts from the following
ing through the use of technologies that can extend the users’ groups: 1) Interior architects from atelier oı̈2 henceforth
awareness about the environmental impact of their decisions referred to as Studio A, 2) Building performance researchers
(e.g. [6, 24, 13]), or through the sensing of our environmental from Building 20503 henceforth referred to as Lab B, 3) Ur-
resources and the manner in which they are used (e.g. [30]). ban sociology researchers from LASUR4 henceforth referred to
This latter class, with its roots in the sensing literature, is rele- as Lab L, and 4) HCI researchers (the authors of this article).
vant for the positioning of our research work, and is also a sub-
Beginning with the formulation of low occupancy as an op-
domain where HCI and Ubicomp have contributed through
timization problem, the researchers from Lab B suggested
the emerging notion of pervasive sensing.
“intensification of space” as a potential solution to increasing
Pervasive sensing, grounded within the tenets of calm comput- the occupancy rate. Accommodating more people and activ-
ing, allows for the observance of individuals’ interactions and ities in existing built spaces (while preserving the comfort
experiences through the use of ambient and wearable sensing qualities) reduces the need for constructing and maintaining
devices. For example, Atallah and Yang [5] modeled the ac- new spaces. Intensified use of space has been widely investi-
tivities of resident patients through the use of ambient blob gated in the domain of sustainable architecture, and has been
sensors and Bluetooth wrist-bands for the purpose of detecting shown to be achievable in three possible ways: a) adding
abnormal behavior. Our contribution, similar to the work of more building layers, b) optimizing the use of each layer’s
Atallah and Yang, leverages on a combination of pervasive surface, and/or c) designing places that can allow multiple
sensing systems to assess the quality of space through the activities at different times by different people [41]. Adding
analysis of inhabitants’ space-use behavior at varying scales. more layers to the existing building was not possible, hence
Positioned within the post-occupancy phase of the building it was ruled out as an option, and limited our choices to the
life-cycle, the encompassing goal of our research is to argue latter two. Therefore, the project stakeholders decided to ex-
for the inclusion of occupants’ space usage behavior as one of amine a new work situation in which several research teams
the design parameters towards the realization of sustainable use only two office spaces in the Smart Living Lab build-
buildings. ing that would be specially designed to provide opportunities
for desk-sharing, while trying to satisfy the users’ work and
ARCHITECTURAL PROJECT AND STAKEHOLDERS comfort needs.
In this section, we describe the architectural project, its objec-
tives and the collaborative context within which our study was Hot-Desking
conceptualized and conducted. Three groups which already had offices in the Smart Living
Lab building volunteered to consider a new work configura-
A newly constructed building that currently accommodates tion. Previously, these groups were separated and used four
several research groups affiliated with three academic institu- offices with individually allocated workspaces. In the new
tions, as well as a few start-ups, was inaugurated in late 2015. arrangement, they would share two refurbished office spaces
This building also serves as a living lab and provides spaces with the other groups. The other spaces in the building, in-
and facilities to test design ideas and to conduct user-studies. cluding the cafeteria and meeting rooms remained unchanged.
It is considered a “prototype building” – the first cycle of an Studio A proposed a re-modeling of the two offices that would
iterative architectural project known as Smart Living Lab1 . allow for hot-desking (i.e. desks are not assigned to individu-
The long-term objective is to learn from the occupants’ expe- als, instead they are used on-demand and cleared by the user
riences within the current prototype, in order to inform the before leaving). Hot-desking situations can create potential for
design of the future office building whose construction is due multiple use of space, but has also been criticized for inducing
to start in the year 2020. detachment from the workplace, and causing disruption of
work dynamics [15]. In an attempt to avoid such negative
Sustainability Goal of the Project consequences, we (the HCI researchers) together with the so-
Prior to the study that we present in this paper, an analysis of ciologists from Lab L recommended to organize co-design
occupancy rate was conducted, which revealed less than ex- sessions with the potential users to: a) gather knowledge
pected occupation of the building (around 20%). This finding about their expectations and preferences, b) obtain any con-
established that the building use was far from being sustainable scious inhibitions related to sharing work places with others,
in terms of building material and energy consumption (heating and c) empower them to inclusively and actively play a part in
and ventilation). Consequently, a sub-project was instigated the organizational policy making. In the next section, we will
to comprehend the nature of such low space occupancy behav- describe the details of the co-designing of the two offices.
ior, and subsequently devise potential strategies to optimize
2 www.atelier-oi.ch
the utilization of space. This entailed understanding how the
3 building2050.epfl.ch
1 smartlivinglab.ch 4 lasur.epfl.ch
 (a) Connective Room (b) Calm Room

Figure 1: Snapshots of the two refurbished (hot-desking) offices (14×9×2.5 m3 ).

CO-DESIGNING OFFICES tic privacy because the latter could be easily achieved with
We organized two participatory design sessions in tandem headphones.
with the interior architects from Studio A. The first session
We presented our analysis of the first co-design session to the
was intended to gather the inhabitants’ preferences for their
other stakeholders and discussed the possibilities and practical
future workplace and understand the issues relevant to their
constraints for the new design. A couple of weeks later, the
current offices. The second session, set two weeks apart, was
interior architects consolidated the observations from the par-
planned to validate the blueprints for a future re-design, which
ticipatory session, and proposed two different layouts for the
were based on the assimilated user preferences from the first
two rooms. One office was designed to provide opportunities
session.
for spontaneous collaborations, and the other one focused on
Twelve participants affiliated with the three research groups addressing the concern of visual privacy. In the rest of this
that had volunteered, took part in the first session. The par- paper, we will refer to the former as Connective and the latter
ticipants were randomly split into four groups. The groups as the Calm room. Please note that this labeling was not used
were provided with 18 cards – each containing an adjective during the study, it is created solely in the interest of clarity
corresponding to a potential positive quality attributed to an for this paper.
office (silent, comfortable, luminous, etc.). In the first ten
In the second session, the interior architects from Studio A
minutes, each group was asked to collaboratively select up to
presented the new designs to the future participants and col-
five adjectives (either from the cards or their own suggestions)
lected their feedback. Given the users’ feedback, the initial
as the qualities they might attribute to their preferred offices.
layouts were majorly modified, and then, through internal dis-
Besides the five chosen qualities, the groups were asked to cussions among the stakeholders, minor modifications were
consider two more adjectives: “sustainable” (both in terms of applied in several iterations. Finally, upon being approved
energy and space) and “flexible” (adapting to changing work by the leaders of the participating groups, the two rooms
dynamics and accommodating new colleagues). Next, the were re-modeled with the new furniture (stationary, lighting,
group members used their set of adjectives to brainstorm and plants, separators, etc.). Figure 1 shows the Connective and
materialize their notion of the future workplace by sketching Calm rooms just before inauguration. The Connective room
their ideas on paper for 30 minutes. Following the design (Figure 1a) was equipped with a large (central) shared table,
phase, each group took turns to present their sketches to the additional workspaces along the window, and labeled stor-
other groups, and answer their questions or criticisms. Finally, age spaces. The Connective room while affording informal
we opened the floor to discussions about the participants’ work collaboration was different from the meeting rooms in the
practices and their concerns with the current and future setups. building that are equipped and designated for formal meetings.
For the purpose of analysis, we audio-recorded the entire Figure 1b provides a glimpse into the Calm room with four
session. central workspaces with separators. Besides, the Calm room
also incorporated two visually isolated sofas, ideal for reading
We observed an overlap among the set of adjectives that were activities (behind the labeled storage space in this picture).
chosen by the groups. Collaboration and privacy (visual and
acoustic) were the two (contrasting) aspects that were unan- USER STUDY
imously chosen by all four groups. These two aspects were In this section, we illustrate the eight-week process during
also reflected in the sketches. The groups partitioned the office which we studied the participants’ space-use behavior within
space into collaboration zones which can facilitate opportunis- the building, and particularly in relation to the attributes of the
tic conversations, and isolated zones which afford for focused new designs. The study of space use was motivated by the goal
individual work as well as subtly indicating an individual’s of the project towards optimized use of space, entailing an
wish not to be disturbed. Participants’ illustrations of isolated understanding of how different workspaces in the refurbished
zones were preferentially biased for visual rather than acous- rooms were used considering the broader knowledge of move-
ment patterns at the scope of the building. We used pervasive
sensing methods to monitor the inhabitants’ presence in differ-
ent spaces at short time intervals. As part of the data collection,
we also conducted on-site observation based on the conception
of sociology as a “natural observational science” [38], which
is grounded in ethnomethodology [18, 19] and its detailed
methods of observation and self-reflection [17].

Participants
33 participants (10 females, 23 males) affiliated with three
research groups participated in the user study. The partici-
pants held varied professional responsibilities (6 professors,
3 project managers, 9 architects, 15 researchers including
post-docs, PhD students, and master students).

Technical Setup
Two pervasive sensing approaches were used to record the
occupancy information. At the scale of the building, we used
zoning as a method to gather data about participants’ presence
in different areas. At the scale of the room, indoor localization Figure 2: The pervasive sensing setup we employed in our
was used to record the precise position of the participants user study. A Motion Capture (MoCap) infrared camera is
within the Connective and Calm rooms. displayed on the top, together with the data-logger used for
zoning. The bottom image shows a Beacon enabled bracelet
Macroscopic Localization (Zoning) and the marker badge for the MoCap system.
Twelve zones including the Connective and Calm rooms, five
meeting rooms, a cafeteria, an atrium, and three other open-
plan offices were identified as the areas that the participants bracelets and the marker badges rendered them similar to the
might visit, and hence chosen as the zones to be monitored. conventional fitness trackers and the university badges, and
Each zone was equipped with a data-logger5 (a Raspberry they were not perceived to be bothering or hindering activities
Pi 2 Model B), programmed to receive and record Bluetooth (as stated in the semi-formal interviews).
packets. Next, we asked the participants to wear a bracelet con-
taining a Bluetooth Beacon (EMBC01 manufactured by EM Procedure
Microelectronics, Switzerland6 ) as shown in Figure 2. These Our study spanned over eight consecutive weeks, during which
bracelets were configured to transmit a signal (along with a we collected bracelet data for the whole duration, and MoCap
unique identifier) every second, which was then received by data for last four weeks. We limited the use of MoCap sys-
a data-logger in close proximity, corresponding to the room tem, as initially we suspected that the extended use of marker
that was occupied by the participant. The data-loggers also badges could be annoying for some of the participants. The
registered the timestamp value corresponding to each received following list summarizes the crucial steps we took during our
signal along with the RSSI (Received Signal Strength Indica- user study:
tor) value. The RSSI information was later used to filter out
false positives resulting from the redundant reception of the Week 0: We organized an information session to brief the
same signal by one or multiple data-loggers. participants about the purpose of the study and the planned
procedure. First, we demonstrated the functioning of the
Microscopic Localization pervasive sensing setup. The participants then received their
Optitrack’s7 camera-based Motion Capture (MoCap) system assigned bracelets along with the usage instructions. We
comprising of 20 Infrared cameras (Prime 13 and 13W) was did not collect any data during this week.
used to locate the users within the Connective and Calm rooms. A substantial part of this session was dedicated to addressing
We provided a unique marker to each participant, which was the privacy and data-protection concerns. Our intention was
meant to be worn around the neck similar to conference badges to build trust with the participants, through a transparent
(see Figure 2). The MoCap system recorded the occupants’ elaboration of how their data will be used and the steps
location at 30 frames per second. undertaken to ensure the protection of the sensitive data.
These steps include: a) anonymizing and aliasing the
Our choice of these two approaches was driven by the need for collected data including the semi-structured interviews and
a complete understanding of inhabitants’ space-use behavior. on-site observations; b) encrypting the data before storing;
Besides, the relatively high cost difference between the two c) assuring the participants that their data will not be shared
approaches made MoCap a reasonable choice only at the scale with anyone including their employers; and d) offering them
of the room. Furthermore, the form-factor associated with the a choice to preemptively withdraw from the study, or request
5 Acquired from Starnberger Innovation & Technologie, Austria deletion of their data. The aforementioned details were
(starnberger.at) also presented to the participants in the form of a printed
6 www.emmicroelectronic.com/ “Informed Consent Document”, which upon agreement the
7 www.optitrack.com/ participant and the experimenters signed. Furthermore, we
 believe that the co-design sessions organized earlier helped Principal Component
us in gaining the participants’ confidence, as they created a Input Variables Avg. Usage First Second
sense of active involvement. (in mins.)
Week 1: We started the monitoring of participants’ presence Connective Room 37.80 0.661 -0.653
in the 12 zones within the building through the recording of Calm Room 35.18 0.604 0.595
bracelet data. Cafeteria 3.45 0.658 -0.214
Week 5: The MoCap system was deployed over the weekend Atrium 0.55 0.344 0.387
and the participants received their assigned marker badges Meeting Rooms 12.75 0.485 -
at the start of the week. Other Offices 35.94 0.375 0.360
Week 8+1: After the conclusion of the study, we interviewed Anchorage - 0.934 -
the participants in a semi-structured manner about their
experiences with the new furniture layout, desk-sharing, Table 1: The statistically significant correlation values between
and social interactions. the process variables and the first two principal components.
The values which were not significant were removed from
ANALYSES AND RESULTS this table. In addition, the average usage of different spaces,
In the context of office buildings, comprehending the way computed per person per day, is shown in minutes.
inhabitants use different spaces can furnish situated insights
about the relation between the ascribed functions and design of within the data in a reduced dimensional space; b) combine the
space, with the perception and exhibited behavior of occupants. various temporal variables in order to reveal their relationship
Manifestations of such knowledge can empower architects and to one another; and c) extract the differences in the inhabitants’
designers in making informed design choices about factors space-use behavior that may not be otherwise plainly visible.
that may inhibit sustainable usage of these spaces. In addition,
The first five principal components collectively explained
in the evaluation of the Connective and Calm rooms across the
90.46 % of the variance in the data. However, only the first two
dimensions of space-use intensification and user satisfaction,
principal components (accounting for 53.03 % of variance)
and the manner in which these two spaces are used, cannot
demonstrated the emergence of meaningful patterns of time
be studied in isolation. In this section, we first explain the
spent in different spaces. Table 1 summarizes this relationship
analysis that we performed at the scale of the building; then
through the correlation values between the input variables and
we outline the observed differences between the Connective
the principal components.
and Calm rooms; and finally, we narrow down our focus to
the analysis of workspaces within the the two rooms. We The first principal component was observed to be strongly
will then discuss the findings of these three parts together and correlated with the inhabitants’ tendency for anchorage. The
demonstrate how they translate to design implications, in the inhabitants’ score (i.e. the transformed coordinates in the new
next section. dimensional space) along the first principal component could
represent the strength of anchorage, hence transforming the
Analysis of Space Use within the Building first principal component into a continuous measure of inhabi-
We computed the time spent by the inhabitants within the 12 tants’ tendency to spend time in one kind of space rather than
monitored zones per day, and combined them into variables the others. This finding conforms with the distinct space-use
corresponding to the time spent in 1) meeting rooms, 2) cor- patterns identified by Verma et al. [42]: “Mutables” (corre-
ridors and atrium, 3) cafeteria, 4) Calm room, 5) Connective sponding to low anchorage) and “Habituals” (corresponding
room, and 6) other offices. In addition, we calculated the vari- to high anchorage). According to Verma and colleagues, mu-
ability in the amount of time spent in these different spaces, tables’ work practices encompass frequent coordination with
represented by the standard deviation. A high value of vari- different groups and projects which results in their extensive
ability may refer to the extended use of one kind of space use of the meeting rooms, and relatively weaker adherence
than others, and could be understood as an indicator of the to their office. On the other hand, habituals spend a signifi-
inhabitants’ tendency for Anchorage to a zone. This could be cant amount of their time on focused individual work in their
attributed to the inhabitants’ daily work schedules (meetings, offices, with seldom formal meetings. Amongst our partici-
etc.), affinity to specific tools required to accomplish tasks, or pants, the professors and project managers exhibited space-use
a disposition for one kind of space. Moreover, anchorage is a behavior that corresponds to the definition of mutables; the
dynamic quality of inhabitants’ working experience and style, architects and researchers, on the other hand, were classified
as it might change from one day to another depending on the as habituals. Subsequently, the analysis of variance (ANOVA)
tasks at hand and the social context. demonstrated that the habituals (architects and researchers)
had a significantly higher anchorage as compared to the mu-
As the next step towards incorporating the aforementioned
tables (project managers and professors) (F(1, 255.9)=8.37,
variables into a cohesive analysis that can accentuate the time p=.004).
spent in varied spaces in relation to the Calm and Connective
rooms, and to examine the similarities in the inhabitants’ space- Moreover, the correlation values corresponding to the time
use behavior, we performed the Principal Component Analysis spent in the Calm and Connective rooms with the second
(PCA). Our initial choice of PCA as a data exploration tool was principal component exhibits orthogonality (referring to the
driven by the need to a) summarize the underlying structure opposite signs of the correlations), signifying the mutual ex-
clusivity (to a certain degree) in the usage of these two rooms. separated teams, had a positive effect on social convergence,
This might reveal that the occupants rarely alternated between mostly during lunchtime and coffee-breaks.
the Connective and Calm rooms during a day.
Analysis of Space Use within the Two Rooms
The inhabitants’ tendency for anchorage to a specific func- Although the collected data from the occupants’ Bluetooth
tional space and its relationship to the professional profile of bracelets was sufficient to distinguish their presence across
the participants were revealed in the PCA results. Manifesta- different rooms, this data was not accurate enough to specify
tion of this tendency might be supported by the divergent work their location within the rooms. With the MoCap setup, how-
responsibilities and the practices of the varying professional ever, we were able to trace the precise location of occupants
profiles, which in turn were reflected in their choice of either inside the two refurbished rooms.
the Calm or the Connective room.
The analysis that we present in this section is based on 20
episodes of MoCap recordings from the Connective and Calm
rooms (10 recordings in each room). The episodes were ex-
Analysis of Space Use between the Two Rooms
actly 2 hours long, and were selected randomly across different
In order to compare the specificities of the Connective and
times of the day (between 9:30 and 18:00) and different days
Calm rooms, we began by examining the difference in the time
of the week.
spent by the two profiles in the two rooms. We normalized
the data before the analysis by computing the standard score To be able to visually explore the space-use behavior within
corresponding to each participant for each working day (by the rooms, we developed a tool that animates the MoCap data
computing the ratio of the difference with the mean value overlaid on the rooms’ floor plan. Figure 3 shows a snapshot of
to the standard deviation). Analysis of Variance (ANOVA) this tool, in which the occupants’ locations are visualized with
showed that the mutables spent significantly higher proportion colored circles, the radius of which is directly proportional to
of their time in the Connective room as compared to the Calm the time spent at that location. Furthermore, the tool displays
room (F(1, 229.07)=5.47, p=0.02). On the contrary, the ha- the trace as occupants move with the progression of time. Be-
bituals spent more time in the Calm room as compared to the sides visualizing, the tool has provisions for fast-forwarding,
Connective room, however this difference was not found to skipping forward to a specific time, and customizing the visu-
be statistically significant (F(1, 737.21)=0.99, p>.1). These alization parameters.
findings exhibited that the way the different profiles used the
Connective and Calm rooms was in accordance with their While browsing through the MoCap animations, it appeared to
working practices and responsibilities. However, these find- us that in both rooms certain workspaces were relatively more
ings have to be interpreted carefully as they do not imply that popular than the others. These workspaces got occupied in
the two profiles used these two rooms in a mutually exclu- the morning before the others, and once cleared-up during the
sive manner. Instead, the findings outline a tendency of the day got re-occupied quickly. Seemingly, this was not about
participants to perceive one kind of space as relatively more personal preferences, but rather a general behavior exhibited
conducive to their work practices. across the occupants. In order to compare the popularity of
workspaces, we extended our visual analytics tool with a “heat-
Findings on the differences in usage patterns of the Connective map” function, which after drawing a grid over the room plan,
and Calm room can be further expanded by on-site observa- computes the occupancy of each cell (10×10 cm2 ) in the gird,
tions. Although the two rooms were designed to support and generates the heat-map for a given episode of MoCap
different sets of activities, participants used workspaces in data.
both rooms for individual focused work over extended peri-
ods of time. In principle, the Connective room afforded for Examining the heat-maps reconfirmed that, in both rooms,
the same purposes as the Calm room, but not the other way there was a clear disparity between workspaces in terms of oc-
around. Connective room proved more flexible and enabled
variety of activities: focused individual work as well as collab-
orative work (discussions and meetings). It was not unusual to
observe people in the Connective room working individually
while other occupants were having meetings and discussions
nearby in the same room. On the other hand, we observed
a distinct tendency to speak more quietly – if at all – in the
Calm room, but only when there were other people present
and visibly absorbed in their work. Whispering as a distinctive
speech practice was observed only in the Calm room, not in
the Connective room. People did occasionally engage in short
small-talk exchanges in the Calm room, but the Connective
room appeared to be more inviting for acts of informal so-
cial interaction (e.g. during routine activities such as packing
up before leaving). On-site observations and semi-formal in-
terviews with participants also indicated that the intensified Figure 3: A Snapshot of the visualization tool we developed
space configuration, bringing together people from previously to animate the MoCap data. Each detected occupant is repre-
sented with a circle whose radius increases with the duration
of stay.
cupancy rate. By matching the occupancy heat-maps with our We found evidence that suggests a positive answer to Q1 only
on-site observations and considering the participants’ concerns for the Calm room, and a positive answer to Q2 only for
for visual privacy and interpersonal interaction (mentioned dur- the Connective room. In other words, in the Calm room the
ing the co-design session), we came up with two conjectures: workspaces with low visual exposure, and in the Connective
1) the occupants often avoided the workspaces where their room the workspaces with high visual openness were preferred.
computer screen would have been visible from a large area In the remainder of this section, we present the details of the
behind them; 2) the occupants often chose to sit and work in quantitative analysis that produced these answers.
the areas where they could have a wider view and higher visual
control of the room. To structure a more precise discussion of Our attempt to answer Q1 and Q2 is organized in six steps,
these two conjectures, we introduce and formally define the applied to each room:
following notions8 :
1. From the MoCap data we extracted the Availability of each
• Visual Exposure of a workspace is a scalar value that mea- workspace (seat) during each episode E, computed as the ac-
sures the volume of the area behind the workspace from cumulation of time intervals when the seat was not occupied.
which its center is visible. This can be described as the 3D We refer to the average availability of Si as Availability(Si ).
isovist of the computer screen position, removing the part High average availability implies a collective negative per-
that rests behind it (Figure 4, blue isovist). Visual exposure ception of the seat, and lack of interest for using it.
corresponds to the potential of being observable by others, 2. For each pair of seats (Si , S j ), we applied a paired t-test to
which has been shown to contribute to the perception of compare their availability across different episodes.
visual privacy [9, 39, 7].
3. The results of the t-tests were arranged in a matrix, in which
• Visual Openness of a workspace is a scalar value that mea- item (i, j) summarizes the results of comparing Si and S j ,
sures the volume of the area that is visible to the occupant of including the p-value and the difference of means. Figure 5a
that workspace. This can be described as the 3D isovist of and 5d show the resulting matrix for the Calm and Connec-
the occupant’s position, removing the part that rests behind tive rooms, respectively. For a cell at (i, j), the fill-color is
her (Figure 4, red isovist). Visual openness may relate to
blue if Availability(Si ) < Availability(S j ) and gray other-
the sense of visual control, and integration into the social
wise. The higher the difference of means the more saturated
fabric of space [27].
the color. The p-values are mentioned inside the cells and
Using these definitions, the two aforementioned conjectures the ones less than 0.05 are framed in red.
translate to the following questions: 4. For each seat, visual exposure V E and visual openness VO
Question 1: Did visual exposure have an impact on the occu- are measured. To do so, we made a 3D model of the room
pants’ choice of workplace? including the visual obstacles that are higher than the desks’
Question 2: Did visual openness have an impact on the occu- height, and computed the area of isovists according to the
pants’ choice of workplace? aforementioned definitions of visual exposure and visual
openness.
8 These notions were previously defined in the urban context [39].
5. We rearranged the columns (and similarly rows) in the ma-
However, our definition is uniquely fitted to the indoor context. trix, sorted them by visual exposure of the seats, such that
for a matrix item (i, j) if i > j then V E(Si ) > V E(S j ). Fig-
ures 5b and 5e visualize the results for the Calm and Con-
nective rooms respectively.
6. We rearranged the columns (and similarly rows) in the ma-
trix, sorted them by visual openness of the seats, such that
for a matrix item (i, j) if i > j then VO(Si ) > VO(S j ). Fig-
ures 5c and 5f visualize the results for the Calm and Con-
nective rooms respectively.

This method of analysis, which is based on comparing the pair

of seats, reflects the way one routinely chooses a workspace
by comparing the available seats given their ordering based on
salient qualities.
Upon sorting the workspaces in the two rooms based on their
visual exposure (see Figures 5b and 5e), we observed an accu-
Figure 4: Given a 3D model of the rooms, visual exposure and mulation of blue cells in the top-right corner of Figure 5b for
visual openness of each workspace were measured. The visual the Calm room. This accumulation suggests that, in the Calm
exposure of a workspace (shown in blue) was computed as room, visual exposure was a defining attribute for the way the
the size of the 3D isovist of the screen position, removing the various seats were occupied. Particularly, the seats with higher
part that rests behind it, and visual openness (shown in red) is visual exposure exhibited higher availability, signifying that
the size of the 3D isovist of the occupant’s position, removing they were often avoided and remained available. This finding
the part that rests behind her. This figure illustrates the visual is an affirmative answer to Q1 for the Calm room, whereas the
exposure and openness of one of the workspaces in the Calm
room.
 (a) Calm room (b) Calm room (c) Calm room
seats not sorted seats sorted by visual exposure seats sorted by visual openness

(d) Connective room (e) Connective room (f) Connective room

seats not sorted seats sorted by visual exposure seats sorted by visual openness

Figure 5: Subfigures a and d illustrate the relative average availability of seats in the Calm and Connective rooms, respectively.
For a cell at (i, j), the fill-color is blue if Availability(Si ) < Availability(S j ) and gray otherwise. The higher the difference of
means the more saturated the color. The p-values are mentioned inside the cells. The matrices in Subfigures b and c are the
rearrangements of the matrix in Subfigure a. In b the seats are sorted based on their visual exposure, and in c based on visual
openness. Similarly, Subfigures e and f are the rearrangements of the matrix in Subfigure d. In e the seats are sorted based on
their visual exposure, and in f based on visual openness.
lack of any such clustering in Figure 5e provides evidence that 2. The outcome of PCA uncovered the participants’ tendency
visual exposure did not influence the choice of a workspace for anchoring themselves to a specific office during a work
in the Connective room. Similar analysis regarding the visual day, as if choosing their daily headquarters, and this was
openness revealed an opposite trend for the Connective room specifically visible in the context of Connective and Calm
as shown by the clustering of blue cells in the bottom-left rooms.
corner of Figure 5f. In other words, the seats with high visual 3. Related to the previously described observations of small
openness were occupied often and for longer duration in the talk and sharing coffee/lunch breaks, space intensification
Connective room, and hence providing a positive answer for seemed to have a positive effect on social convergence of
Q2. However, this was not observed to be the case for the previously separated teams.
Calm room (see Figure 5c). 4. Comparing the Connective and the Calm rooms across the
varied activities they endorsed, we observed a relative versa-
Summary of Results
tility on the part of Connective room for supporting different
To conclude this section, we summarize the main findings: kinds of activities (from focused work to collaborations). In
1. The analysis of space-use patterns at the scale of building contrast, Calm room only fostered focused individual work.
showed a correlation between the occupants’ professional 5. The answers to Q1 and Q2 also suggest that in the Connec-
profiles and their daily routine of occupying different rooms tive room the workspaces with low visual openness were
and transitioning between them. At the scale of office, how- less popular, while in the Calm room, the workspaces with
ever, according to the answers to Q1 and Q2, the occupancy high visual exposure were less popular.
of workspaces seemed to be determined by visual exposure
and visual openness – spatial attributes of the offices.
DESIGN IMPLICATIONS AND DISCUSSION can shape the diverging perception of occupants and their
In this section, we discuss the presented findings and explain behavioral patterns, both in short-term and especially in the
how they were interpreted in the context of the architectural long-term. In the Calm room, for instance, four workspaces
project and in reference to its objectives. provided maximal visual privacy; consequently they became
outstandingly popular for focused work in high privacy; this,
User Evaluation of the Office Design in turn, created a social atmosphere that tightened the type of
Recalling that the Calm room was initially intended to address acceptable behavior in the calm room towards quiet, focused
the occupants’ concern for visual privacy, and the Connective work; and thus constructed a sustained collective perception
room was designed to foster collaboration, the answer to Q1 of the room. In other words, such cluster of workspaces in one
and Q2 can be interpreted as confirmatory of the architectural room that hold an extreme degree of a spatial attribute create
works. The fact that the most popular workspaces in the Calm a gravitational force guiding the occupants’ perception of the
room were the ones with lowest visual exposure may suggest room and consequently constraining the occupancy of other
that the Calm room succeeded to attract the occupants who workspaces in the same room.
were looking for a private workspace. Similarly, the fact that In this perspective, the following five steps can be carried out
in the Connective room the most occupied workspaces were to evaluate and improve a spatial layout prior to its imple-
the ones that offered highest visual openness may suggest that mentation: a) identifying the future users’ concerns through
the Connective room could appeal to the occupants’ interest an early user study, b) foraging spatial attributes that corre-
in opportunistic social interaction. The latter is reinforced spond to the users’ concerns; these make a coordinate system
by our observations and the interviews, confirming that the such as the one shown in Figure 6, c) placing the designed
Connective room hosted many episodes of inter- and intra- workspaces along the coordinates, d) detecting the zones of
group interaction. In contrast, the same findings highlight the maximal spatial quality, and e) alternating the design such that
flaws of the architectural works: certain workspaces remained the workspaces outside the detected clusters move towards
under-used, and the answers to Q1 and Q2 suggests that this the clusters that are populated by the workspaces in the same
was partly due to the design of the space and an alternative room.
layout could have prevented such cases of low occupancy.

Strategies for Space Optimization CONCLUDING REMARKS

The other direct implication of the answers to Q1 and Q2 is
Integrating HCI Approaches in Architectural Projects
that the Calm and Connective rooms require different sustain-
ability strategies. Concretely, the future plan of the Calm room Our contributions as HCI researchers have complemented and
will eliminate some of the highly visually-exposed workspaces guided the architectural project objectives and procedures at
and try to enhance some of the others in terms of visual pri- different stages of the project. In addition, our involvement
vacy. In the Connective room, however, visual privacy does with both the inhabitants and stakeholders coalesced their
not seem to be a salient factor, and the modifications will be respective perceptions and expectations towards the establish-
rather towards enhancing elements that concern with visual ment of a homogeneous framework for dialogue. With the
integration. Figure 6 unveils a generalizable explanation for stakeholders, we collaboratively framed the problem related
such distinction through the notion of Zone of Maximal Spa- to the less than optimal use of spaces, followed by the iden-
tial Quality. In Figure 6, the (schematic) positioning of all tification of intensification as a potential solution. Besides,
the workspaces along the axes of visual exposure and visual the outcomes of co-design sessions and the user evaluation of
openness, uncovers two distinct clusters at the extreme end refurbished offices were subjected to participatory analyses
of the two dimensions. We call these clusters the zones of and discussions with the stakeholders. Furthermore, our role
maximal spatial quality and argue that in a general case they in engaging the occupants to collectively devise strategies for
the sustainable use of space manifested in the redesign of the
Connective and Calm rooms, followed by their evaluation by
the occupants. We experienced that it is only through such
continual confluent measures that the human factors can be
profoundly assimilated and acknowledged as an essential com-
ponent in the processes of problem formulation and solution
seeking in sustainable architecture.

ACKNOWLEDGEMENTS
We would like to thank all our colleagues in the Smart
Living Lab project who contributed to the “User Experience
Study II”. The anonymous reviewers deserve special apprecia-
tion for their valuable comments and suggestions. To Klaus
Starnberger, Lukas Obletter, Bruno Vuillemin, Christophe Gol-
liard, Claude-Alain Jacot, Raphaël Tuor, and Agnes Lisowska
Figure 6: The circles and rectangles represent the workspaces we extend our sincere gratitude for their timely assistance.
in the Calm and Connective room positioned along the axes Finally, our special thanks to all the participants of our study.
of visual exposure and visual openness. The saturation of the
fill-color corresponds to relative occupancy.
REFERENCES Human Factors in Computing Systems. ACM,
1. Alper T. Alan, Mike Shann, Enrico Costanza, 3841–3846.
Sarvapali D. Ramchurn, and Sven Seuken. 2016. It is Too
15. Einar De Croon, Judith Sluiter, P Paul Kuijer, and
Hot: An In-Situ Study of Three Designs for Heating. In
Monique Frings-Dresen. 2005. The effect of office
Proc. of CHI’16. ACM, 5262–5273.
concepts on worker health and performance: a systematic
2. Hamed S. Alavi, Elizabeth Churchill, David Kirck, Julien review of the literature. Ergonomics 48, 2 (2005),
Nembrini, and Denis Lalanne. 2016a. Deconstructing 119–134.
Human-Building Interaction. Interactions (2016).
16. Povl Ole Fanger. 1970. Thermal Comfort. McGraw-Hill,
3. Hamed S. Alavi, Denis Lalanne, Julien Nembrini, New York.
Elizabeth Churchill, David Kirk, and Wendy Moncur.
2016b. Future of Human-Building Interaction. In CHI’16 17. David J. Francis and Stephen Hester. 2004. An Invitation
Extended Abstracts. ACM, 3408–3414. to Ethnomethodology: Language, Society and Interaction.
SAGE Publications, London, United Kingdom.
4. Hamed S. Alavi, Himanshu Verma, Michael Papinutto,
and Denis Lalanne. 2017. Comfort: A Coordinate of User 18. Harold Garfinkel. 1967. Studies in Ethnomethodology.
Experience in Interactive Built Environments. In Proc. of Prentice-Hall Inc., Englewood Cliffs, NJ.
Interact’17. 19. Harold Garfinkel. 2002. Ethnomethodology’s Program:
5. Louis Atallah and Guang-Zhong Yang. 2009. The use of Working Out Durkheim’s Aphorism. Rowman and
pervasive sensing for behaviour profiling - a survey. Littlefield, Lanham, MD.
Pervasive and Mobile Computing 5, 5 (2009), 447–464. 20. Tommy Gärling. 1969a. Studies in visual perception of
6. Magnus Bang, Carin Torstensson, and Cecilia Katzeff. architectural spaces and rooms: I. Judgment scales of
2006. The Powerhhouse: A persuasive computer game open and closed space. Scandinavian Journal of
designed to raise awareness of domestic energy Psychology 10, 1 (1969), 250–256.
consumption. In International Conference on Persuasive 21. Tommy Gärling. 1969b. Studies in visual perception of
Technology. Springer, 123–132. architectural spaces and rooms: II. Judgments of open
7. Mateus Paulo Beck. 2012. Visibility and Exposure in and closed space by category rating and magnitude
Workspaces. In Proceedings of the 9th International estimation. Scandinavian Journal of Psychology 10, 1
Space Syntax Symposium, 2012. Sejong University, Seoul, (1969), 257–268.
South Korea. 22. Tommy Gärling. 1970a. Studies in visual perception of
8. Franklin D. Becker. 1973. Study of Spatial Markers. architectural spaces and rooms III. Judgments of open
Journal of Personality and Social Psychology 26, 3 and closed space by category rating and magnitude
(1973), 439–445. estimation. Scandinavian Journal of Psychology 11, 1
(1970), 124–131.
9. Michael L. Benedikt. 1979. To take hold of space:
isovists and isovist fields. Environment and Planning B: 23. Tommy Gärling. 1970b. Studies in visual perception of
Planning and design 6, 1 (1979), 47–65. architectural spaces and rooms. IV. The relation of judged
depth to judged size of space under different viewing
10. Eli Blevis. 2007. Sustainable interaction design: conditions. Scandinavian Journal of Psychology 11, 3
invention & disposal, renewal & reuse. In Proc. of (1970), 133.
CHI’07. ACM, 503–512.
24. Anton Gustafsson and Magnus Gyllenswärd. 2005. The
11. Philomena M. Bluyssen. 2009. The indoor environment power-aware cord: energy awareness through ambient
handbook: How to make buildings healthy and information display. In CHI’05 Extended Abstracts.
comfortable. Earthscan. 400 pages. ACM, 1423–1426.
12. Adrian K. Clear, Janine Morley, Mike Hazas, Adrian
25. Scott C Hayward and Samuel S Franklin. 1974. Perceived
Friday, and Oliver Bates. 2013. Understanding adaptive
openness-enclosure of architectural space. Environment
thermal comfort: new directions for UbiComp. In
and Behavior 6, 1 (1974), 37–52.
Proceedings of the 2013 ACM international joint
conference on Pervasive and ubiquitous computing. 26. Bill Hillier and Julienne Hanson. 1989. The Social Logic
ACM, 113–122. of Space. Cambridge University Press.
13. Enrico Costanza, Benjamin Bedwell, Michael Jewell, 27. Bill Hillier and Alan Penn. 1991. Visible colleges:
James Colley, and Tom Rodden. 2016. ’A bit like British Structure and randomness in the place of discovery.
Weather, I suppose’ Design and Evaluation of the Science in Context 4, 1 (1991), 23–50.
Temperature Calendar. (2016).
28. Bill Hillier and Alan Penn. 1992. The social potential of
14. Sheep N. Dalton, Paul Marshall, and Ruth Conroy Dalton. buildings: Space and the innovative milieu in scientific
2010. Measuring environments for public displays: A research laboratories. In Proceedings of the Corporate
space syntax approach. In CHI’10 Extended Abstracts on Space and Architecture Conference.
 29. Ralph Galbraith Hopkinson. 1963. Architectural Physics: 36. Hirofumi Minami and Naoki Yoshida. 1993. The Hidden
Lighting. H.M. Stationery Off. Dimension of Space in School Environment: Implicit
Rules in the Use of Space. MERA 2, Sep. 93 (1993),
30. Alexander Ilic, Thorsten Staake, and Elgar Fleisch. 2009.
33–40.
Using sensor information to reduce the carbon footprint
of perishable goods. IEEE Pervasive Computing 8, 1 37. J. Fergus Nicol and Michael A. Humphreys. 2009. New
(2009), 22–29. standards for comfort and energy use in buildings.
Building Research & Information 37, 1 (2009), 68–73.
31. Ravi Jain and John Wullert II. 2002. Challenges:
environmental design for pervasive computing systems. 38. Harvey Sacks. 1989. Introduction. Human Studies 12, 3/4
In Proc. of MobiCom’02. ACM, 263–270. (1989), 211–215.
32. Azam Khan. 2011. Swimming upstream in sustainable 39. Dalit Shach-Pinsly, Dafna Fisher-Gewirtzman, and
design. interactions 18, 5 (2011), 12–14. Michael Burt. 2011. Visual exposure and visual openness:
an integrated approach and comparative evaluation.
33. Jennifer C. Mankoff, Eli Blevis, Alan Borning, Batya
Journal of Urban Design 16, 2 (2011), 233–256.
Friedman, Susan R. Fussell, Jay Hasbrouck, Allison
Woodruff, and Phoebe Sengers. 2007. Environmental 40. Cornelis Johannes Maria Van de Ven. 1976. Concerning
sustainability and interaction. In CHI’07 Extended the idea of space: the rise of a new fundamental in
Abstracts. ACM, 2121–2124. German architectural theory and the modern movements
until 1930. (1976).
34. Francis McAndrew, Richard M. Ryckman, William Horr,
and Robin Solomon. 1978. The Effects of Invader 41. Andy van den Dobbelsteen and Sebastiaan de Wilde.
Placement of Spatial Markers on Territorial Behavior in a 2004. Space use optimisation and sustainability -
College Population. The Journal of Social Psychology environmental assessment of space use concepts. Journal
104, 1 (1978), 149–150. of Environmental Management 73, 2 (2004), 81–89.
35. Isaac A. Meir, Yaakov Garb, Dixin Jiao, and Alex 42. Himanshu Verma, Hamed S. Alavi, and Denis Lalanne.
Cicelsky. 2009. Post-occupancy evaluation: an inevitable 2017. Studying Space Use: Bringing HCI Tools to
step toward sustainability. Advances in Building Energy Architectural Projects. In Proceedings of the 2017 CHI
Research 3, 1 (2009), 189–219. Conference on Human Factors in Computing Systems
(CHI ’17). ACM, New York, NY, USA, 3856–3866.

View publication stats