Perception and Spatial Representation - Deliverables from the First Mandate of the Chair

While the second seven year mandate of the Canada Research Chair on Cognitive Geomatics is focussed on the relationships between identity, body and space, the first mandate was concerned with understanding our mental representations of space, as derived from perception and mental imagery, and with the development of tools and software that put this knowledge to use. Application areas for this work included rehabilitation, the performing arts, navigation, landscape design, and database design.

Nested perceptions of the world

A variety of researchers have studied how our perceptions of the world are organized as a function of scale. Several schemas that describe scaled perceptions exist. One of the most interesting is that put forward by Dan Montello in 1993 (Scale and multiple psychologies of space). Montello's is interesting in part because it synthesizes the work of several researchers, but also because it critically examines the different approaches. Within Montello's framework, roughly four spaces exist at different scales :
(a) Figural space
(b) Vista space
(c) Environmental space
(d) Geographic space
Montello describes Figural space as being the space of drawings and maps, representations of the world. Vista space is conceived of as the region that can be viewed from a single location. Environmental space is defined as the region accessible via displacement or navigation. Geographic space is the space that is too large to be visited. A fifth space,
(e) Cosmic space,
covers spaces that are not accessible on the Earth.

Other categorisations of spaces also exist, some of them very useful. Hence several researchers focus on what are called "table top spaces" to describe the spaces in which objects can be picked up and manipulated (for example, Andrew Frank). In Montello's scheme, Table top space may be viewed as intermediate between Figural and Vista space. Another approach distinguishes between "within body" space, "body space" and "near body space" (Three spaces of Spatial Cognition by B. Tversky et al.). These are also spaces that are smaller than Vista space. Tversky also points out the cognitive importance of barriers within a space.


Figure 1 : Local displacement space for a household in Sillery, Quebec City

Along with my collaborator-postdoc Dr. Isabelle Reginster, we found that to apply these theoretical ideas to a real application, it was necessary to subdivide Montello's Environmental Space into two different scale spaces, what we called the Local Displacement Space (see Figure 1) and the Extended Displacement Space (see Figure 2). Local Displacement Space dealt with the part of Environmental Space that can be accessed by foot, while the Extended Displacement Space accommodated the region accessed by car. We applied a time limit for displacement as a means to characterize the size of these spaces, and used the three scales (Vista, Local Displacement and Extended Displacement) as spatial units within which information was aggregated to infer perceptions for different households - perceptions of how many municipal services were accessible, of access to schools, and perceptions of how much green areas were to be found within the local environment. We showed how to determine the Vista, LDS and EDS spaces from satellite imagery, and we used the aggregated statistics to explore the relationships between scale, perception and house prices (see the publication Reginster and Edwards, 2001, for details). We tracked, in particular, the location of barriers in the space, both perceptual and navigational barriers.

At the time this work was undertaken, the displacements were tracked using phone interviews carried out with a variety of households in an Origin-Destination survey that had been conducted by colleagues in the Département d'Aménagament of Laval University. These days, the survey could be carried out much more cheaply using a portable GPS unit with a data logger.


Figure 2 : Extended displacement space for the household in Sillery, Quebec City. Note that the extended displacement space consists of corridors around each road used, and that frequency of travel along the road reinforces the intensity of that part of the displacement space

Hence we were able to implement these theoretical constructions of embedded spaces in a study on scaled perceptions of the local environment from the point of view of members of a household.

Rooms and gateways

A second study, undertaken with Dr. Gerard Ligozat and later with his daughter, Anne Laure Ligozat, focused on the development of formal representations of perceived space, especially outdoor, natural spaces. In this project, we were interested in developing a formal (i.e. mathematical) representation of perceived space and in implementing this representation on a computer. We were particularly interested in the fact that, within exterior environments, one may move a certain distance and yet still conclude that one was within the same place as before the move. What determines when we conclude that a change has occurred in our location?

We determined that either the neighborhood had changed, or the order of landmarks on the horizon had change (the latter is called the "panorama" in technical terms). Therefore, we set about to characterise a space in terms of its neighborhoods and panoramas. This led to the idea of "perceptually stable zones" and "zones of transition", which one may metaphorically associate with "rooms" and "gateways". We found that all outdoor spaces could be reconfigured as a set of "rooms" and "gateways", making them analagous to interior spaces. Visual barriers act, within such a viewpoint, as metaphorical "walls". The set of rooms and gateways forms a kind of dual or alternate representation to the set of neighborhoods and panoramas, and we found that one could infer the one from the other and vice versa, to some extent. This work was published in two papers (Ligozat and Edwards, 1999; Edwards and Ligozat, 2004).


Figure 3 : A fictional landscape created within the software prototype PERSEUS

A software prototype called PERSEUS was developped to showcase the model. The prototype divides space up in terms of what are often called "viewsheds", that is, areas of intervisibility, and then subdivides these areas in terms of panoramas, defined as regions in which the order of landmarks on the horizon is stable. To some extent, therefore, the maps produced depend on what objects are labelled as landmarks by the user. We have been able to generate maps of stable perceptual zones for both fictitious landscapes, but also for a study of the Plains of Abraham, the large park within Quebec City.


Figure 4 : Map of the perceptually stable zones for the three landmarks within the fictional landscape used by the PERSEUS prototype

Understanding near-body spaces as a function of disability

The work by Reginster and Edwards, and that by Edwards, Ligozat and Ligozat, constitute new material representations of space (i.e. maps) that incorporate understanding of our mental representations of space as derived from modern cognitive psychology. However, they handle vista spaces and larger regions.

In work aimed at supporting the movement of disabled users in the landscape, another postdoc, Pierre-Emmanual Michon, and a full time Research Professional, David Duguay, and I, developed a new kind of representation, this time a special kind of 3D map, aimed at representing near-body spaces.


Figure 5 : An part of the research centre in rehab in Quebec City, as portrayed within the CADMUS prototype software

For this work we drew on the concept of affordances as proposed by James J. Gibson in the 1950s. The idea is that objects permit certain kinds of functional use but not others - they are said to "afford" such uses. Hence a chair affords that one can sit on it, but not that one can eat it (unless it were a chocolate chair!). We implemented the concept of affordances in a 3D database (see Edwards, 2006, for a description of this process). Hence in our database, doors may afford opening via a "door handle" or a "push button" as in some hospitals. In addition, we matched the affordances of such objects to the physical capabilities of the user. Hence a "door handle" requires the ability to twist as well as a certain level of physical strength, whereas the push button requires a much lower level of strength. Using the combination of affordances and user profiling, we were able to generate maps that showed areas of different accessibility levels as a function of a users physical profile. In a second version of the prototype, which is called CADMUS, we also incorporated mental competencies as well as physical competencies.


Figure 6 : The same region as shown in Figure 5, but color coded in terms of accessibility for a given class of disabled user. Red means access is difficult, green that access is easy

Image schemas and performance design

The work on affordances and user profiling, although it led to the creation of a new kind of 3D map, could also be used to evaluate the effectiveness of particular environmental or building designs for different handicapped profiles. Likewise, the rooms-and-gateways representation of outdoor spaces could be used not only to understand an outdoor space, but also as a support for redesigning such a space.

Our interest in designing spaces extended into another arena, that of performance design (i.e. for the performing arts). Here, the understanding of space requires a connection to their emotional impact and not just their perceptual impact. A useful tool for capturing the relationship between space and emotion is found in image schemata.

Image schemata were unearthed by philosopher Mark Johnson in the early 1980s (see his book The Body in The Mind for a clear exposition of the concept), and their study and use matured under Johnson's collaborator with the linguist, George Lakoff (see Women, Fire and Dangerous Things : What Categories Reveal About the Mind) for this later work. They are basic images that are found to be common across most languages, and that are used to talk about abstract ideas. Common examples of image schemata include CONTAINER, PATH, CYCLE, LINK, ENABLEMENT, FORCE, BLOCKAGE, SPLIT, and COLLECTION. Later studies have found that image schemata also turn up in most forms of expression, including the visual arts, music, gesture and dance, sculpture, and cinema - as such, they constitute a powerful means of coordinating design that must serve our many different senses.

Lakoff and Johnson developed a theory that image schemata are formed during early childhood by a process of binding embodied actions to word concepts. Within this framework, therefore, image schemata are linked to emotional responses, albeit in a manner that is itself rather complex and likely to vary from one individual to another. Nevertheless, artistic design uses image schemata, often unconsciously and intuitively.

In a study callaborative study carried out in 2005, Marie Louise Bourbeau, a mezzo-soprano soloist, and I used image schemata to design a performance of Claudio Monteverdi's opera fragment, Arianna. We showed that image schemata, when used consciously and explicitly, constitute a powerful tool for performance design, for delivering an experience to an audience. In a sense, image schemata allow us to design experiences directly rather than just their progenitors, the objects or events that produce experiences. More details of this work can be found in our paper on the subject (Edwards and Bourbeau, 2005), and the results of the design can be viewed on youTube.


Ariadne Emerging Video Clip

Cognitive Design of Assistive Technologies

In addition to the work on designing maps and spaces, we have been interesting in designing tools that facilitate the navigation and movement within spaces, not just mapping tools. The first significant effort in this direction has been undertaken by a Ph.D. student, Mr. Reda Yaagoubi. The idea is to use what we know about how people represent spaces mentally to assist in the navigation of the blind.

Modern geomatics technology that is useful in this context is, of course, the GPS receiver. However, all current GPS devices rely on the visualization of a map to provide what one might call situational awareness. Instructions on where to go might be provided by a computerized voice, but the devices rely on the visual availability of a map to let people know where they are and the location of objects and landmarks in their immediate environment. Without such landmarks, directional instructions are useless. For the blind, this is a problem - situation awareness is lost many times over the course of a day, and direction that are given without situation awareness may be less than useless.

We are therefore using information about how people, in particular people without recourse to sight, store and maintain mental representations of their immediate surroundings. The tool we are developing seeks to use natural strategies to help individuals update their local mental representation in such a way that a GPS directional instructions may become meaningful. The design process is quite challenging, because it requires that one understands both the cognitive processes and representations in operation and that these inform the engineering and technical principles that must be used to develop a particular form of technology. A paper has been submitted to a journal describing this work (Yaagoubi and Edwards, 2007).

We also undertook behavioral experiments that tested the ability of blind subjects to understand and manipulate mental representations of space. A paper has been submitted describing this work as well (Eardley, A., G. Edwards, F. Malouin, P.-E. Michon and J. Kennedy, 2007). We found that a certain group of people without sight (those born with sight but who lost it a year or so after birth) actually perform better than the sighted at certain tasks involving spatial reasoning on their mental representations. Those born blind from birth with no neurological complications had similar competency as the sighted. Only those born blind with neurological complications performed significantly worse than the sighted on these manipulation tasks.

Resonant Installations - Designing the Immersive Experience for Maximum Impact

The work on image schemata in support of performance design was aimed at connecting performance spaces to their emotional impact. Although image schemata were found to be a powerful tool for design, their connection to emotional response was weaker than we would have liked. In an attempt to develop a stronger connection to emotional response, we (Marie Louise Bourbeau and myself) investigated the use of devices that enhance our awareness of our own bodies during performance.

We began this work by focusing on the act of breathing, perhaps the most important aspect of body-awareness because it is the source of ongoing life. So many studies and body-training disciplines are all based on the act of taking a breath, including all of modern athletics, but also all the performance arts. Marie Louise Bourbeau is a specialist in breath training for singers and dancers, so this choice made double sense as a first target.

We developed an installation, called Incarnatus, that sought to create a new relationship between the participating audience and classical lyrical music. Using one of Mahler's lieder (the Schildwache Nachtlied, or Soldier's Nightsong), sung in German and based on a traditional folk verse, we developed an instatallation that culminated in the use of a device we call the "co-breather". This is a cushion that breathes at the same time as the singer, while the participant is listening to the music sung by the same singer. Far from being experienced as an imposition, participants adapted their breathing to that provided by the co-breather within seconds, and many participants reported a near-ecstatic connection with the music, completely unexpected.


Incarnatus Video clip

Following the public presentation of the Incarnatus installation, we determined that the co-breather creates a rather paradoxical state that includes a heightened state of body awareness combined with a "loss of self" (as well as a stronger sense of identification with the music), a dropping of the barriers that define and protect the self.

We are now in the process of developing new installations that put such "body opening experiences" first, and follow it up with other body exploration processes as well as an integration and a closure phase. The installation that generates such a sequencing of experiences we call a "Resonant Installation", and several examples are presently under development (for more details about Resonant Installations, see the blogsite ResonantInstallations). These form the heart of the second mandate of the Canada Research Chair in Cognitive Geomatics.

Virtual and Mixed Reality Environments - Embodiement and Identity

Finally, during the final year of the first mandate of the Canada Research Chair in Cognitive Geomatics, we have begun a systematic investigation of the relationship between virtual worlds and embodied experience. This is another paradoxical study. At first site, virtual worlds would appear to be a perfect example of a "disembodied experience". This intuitive evaluation that many of us form at a distance is almost completely false, as it turns out. Virtual worlds generate highly embodied experiences. But our understanding of what embodiment is has changed, as a result of this work.

This work is supported by the ongoing discussions undertaken by the Embodied Research Group (ERG), an active group of researchers that meets every week online (on Second Life) to discuss our understanding of embodiment. For more information about the work of the group and the results of the discussions, see their blogsite, EmbodiedResearch. Within this context, it has become apparent that embodiment is "performative" rather than simply "physical". As a result, it is possible to develop a sense of embodiment within virtual worlds, even though our physical bodies are not directly engaged.

This results in a profound rethinking of what it means to be embodied, and even what constitutes "body awareness". In a recent discussion by the Embodied Research Group, it was noted that within Second Life we may actually develop a kind of "reflexive muscle" for an embodied functionality available within Second Life but not in our physical environments. An example is the use of a virtual camera to look at objects that would normally be "out of sight" of one's body. When the reflex develops, we find ourselves "trying" to use it in our physical bodies, and frustrated that we cannot. We have developed our virtual camera as a kind of phantom limb - the neurons still activate it, but there is no follow through to a muscular action.

Within virtual worlds, our identity is also multiplied, resulting in another rethinking of what it means to be a person. The multiplication of identity and the spatial redistribution of our sense of embodiment are two startling mutations in our sense of self that derive from an engaged presence within virtual worlds of such complexity.

This is of special interest for so-called "mixed reality" environments, that is, environments that combine parts of virtual worlds within our physical, material experience. Gaining understanding of the impacts of such mixed reality environments on our sense of self and our ability to act in the world has become a major source of study for the Canada Research Chair in Cogntive Geomatics, and this work will also form part of the second term mandate. A paper on this was recently presented (in French) at the Geocongrès International in Quebec City in October 2007.


Conclusions



Table I : Representations (R) and Tools (T) developed during the first seven year term of the Canada Research Chair in Cognitive Geomatics

Table I shows a summary of the innovations of the first seven year term of the Canada Research Chair in Cognitive Geomatics, in the area of Perceptions and Representations. At the scale of the body, we developed tools we call Resonant Installations. At the near body scale, we developed tools that use image schemata and representations based on the theory of affordances. At the scale of vista space, we developed a map representation that views all spaces, indoors and outdoors, in terms of conceptual rooms, barriers and gateways. At the scale of environmental spaces, we developed map representations in terms of local and extended displacement spaces that can be derived from satellite imagery (e.g. Google Earth). Together, these constitute an "end-to-end" collection of tools for representating, handling and manipulating the full range of perceptual spaces, and informing design processes focussed on the spaces themselves, their map representations, and tools and methodologies that facilitate their understanding. Applications presented include aids for the disabled, for the performing arts and museology, and aids for architecture and landscape design.