Spatial Behaviour and Linguistic Representation

Local and global reference frames in survey knowledge acquired from navigation and maps Tobias Meilinger Max Planck Institute Tübingen, Germany It has been a long standing question in spatial navigation research whether navigators represent their environment within a single reference frame (i.e. coordinate system) or within multiple local interconnected representations. Most results suggest that route navigation is based on multiple local representations. For survey navigation most theories assume that a common reference frame for all locations within an environment is formed immediately or eventually within long-term memory. However, we have shown that survey knowledge of a recently navigated virtual environment was based on local reference frames rather than on a single global reference frame. On the contrary, pointing within ones city of residence relied on a single, north oriented reference frame. We interpret this result in a way that navigators derived survey relations from map experiences. Experiments we conduct currently examine the processes relevant for survey tasks after learning from navigation and from maps. Wayfinding Strategies and Environmental Features Christoph Hölscher University of Freiburg, Germany In this presentation I will report on our research on wayfinding strategies in complex public buildings. Such strategies are highly adaptive to geometric and topological properties of the building, visibility of path choices, salient landmarks and signage. At the same time, cognitive factors, like preference for certain strategies, differences in spatial competence and familiarity with the setting contribute to path choices as well. I will discuss how behavioral data, verbal reports as well as formal modeling of spatial properties with Space Syntax can help untangle the contributing variables, and how real-world experiments and systematic layout variations in Virtual Reality can inform each other. Space Perception, Spatial Decision Making, and Wayfinding Jan Wiener Bournemouth University, UK A series of experiments investigating gaze behavior and decision making in the context of wayfinding are presented. Participants were presented with screenshots of choice points taken in large virtual environments. Each screenshot depicted two path options. In Experiment 1, participants had to decide between them in order to find an object hidden in the environment. In Experiment 2, participants were informed about which path option to take as if following a guided route. Subsequently they were presented with the same images in random order and had to indicate which path option they chose during initial exposure. In Experiment 1, we demonstrate (1.) that participants reliably chose the path option that featured the longer line of sight, and (2.) a robust gaze bias towards the eventually chosen path option option. In Experiment 2, systematic differences in gaze behavior towards the alternative path options between encoding and decoding were observed. Finally, we present tentative models of gaze behavior that are based on geometrical features of the depicted environments and that allow predicting fixation behavior during space perception. Results from this study shed light onto the control of visual attention during space perception and wayfinding decision making. “Sense of Direction” and “Orientierungssinn”: Meaning of a SelfConcept as Revealed by Relations to Spatial Strategies and Learning Stefan Münzer Universität des Saarlandes, Germany Self-report measures contribute substantively to an explanation of individual differences inenvironmental spatial cognition. A central self-concept is “sense of direction” (SOD). Self-reported SOD can predict spatial learning significantly. What does SOD mean and is SODcomparable to the German “Orientierungssinn”? The meaning of SOD and“Orientierungssinn” is reflected by the factor structure of self-report instruments that intend tomeasure these self-concepts together with associated strategies and competences. The Santa-Barbara Sense of Direction Scale (SBSOD, Hegarty et al., 2002) comprises a number ofdifferent spatial strategies and competences (including route learning, knowing about cardinaldirections, formation of a mental map) in one dimension (factor). In contrast, with a Germanversion of the SBSOD scale (Freiburg-SBSOD, Meilinger & Knauff, 2004) a one-dimensional solution could not be found. Likewise, the factor structure of a recentlydeveloped German questionnaire on spatial strategies (Fragebogen Räumliche Strategien,FRS, Münzer & Hölscher, subm.) reflects the distinction between egocentric and allocentricspatial references frames. “Orientierungssinn” was related to egocentric strategies but not tosurvey strategies. These findings seem to illustrate a difference between a broad SOD and amore narrow meaning of “Orientierungssinn”. However, this difference is discussed withrespect to methodological aspects, behavioral studies, and cultural differences in the man-made environment. ReferencesHegarty, M., Richardson, A. E., Montello, D. R., Lovelace, K., & Subbiah, I. (2002).Development of a self-report measure of environmental spatial ability. Intelligence,30, 425-447.Meilinger, T., & Knauff, M. (2004). FSBSOD: Freiburg Version of the Santa Barbara Senseof Direction Scale. Retrieved August 18, 2009, fromhttp://cognition.iig.uni-freiburg.de/research/online-experiments/fsbsod.pdfMünzer, S. & Hölscher, C. (Revision submitted). Entwicklung und Validierung einesFragebogens zu räumlicher Orientierung (Development and validation of a self-reportmeasure of spatial orientation). (Diagnostica) Language as a representation of spatial thought Emile van der ZeeUniversity of Lincoln, UK, andNico van de WegheGhent University, Belgium Languages differ in the way they encode motion (e.g. Talmy, 2000). When considering onbuilding a semantic typology for motion encoding several elicitation tests are available(Levinson & Wilkins, 2006). For example, the Frog Story test makes it possible to determinehow manner, path, and path roles – such as goal and source are cross-linguistically encoded(Berman & Slobin, 1994). However, current tests make it difficult to systematicallyinvestigate particular motion components (e.g., only manner, or only path roles).We have developed a test, based on Qualitative Trajectory Calculus (Van de Weghe, 2004),which allows us to construct a semantic typology relating to only path roles (Bogaert et al,2008; Van der Zee et al, submitted a, b). The Path Role test makes it possible to consider bothlanguage production and language comprehension. So far we have collected data fromEnglish, Finnish, Dutch, Bulgarian and Norwegian. We are planning to extend our research(1) to include a larger set of languages, (2) to include a more diverse set of stimuli allowing usto study more parameters impacting on path role use, and (3) to consider distinctions in otherareas of qualitative reasoning to enrich the current approach. ReferencesBogaert, Peter, van der Zee, Emile, Maddens, Ruben, van de Weghe, Nico & De Maeyer,Philippe (2008). Cognitive and Linguistic Adequacy of the Qualitative TrajectoryCalculus. In N. van der Weghe, R. Billen, B. Kuijpers, & P. Bogaert (Eds.),Proceedings of the International Workshop “Moving Objects: From Natural to FormalLanguage” (in cooperation with GIScience 2008), Park City, Utah.Van der Zee, Emile, Bogaert, Peter, Martinez, Liliana, Eshuis, Rik, Vulchanova, Mila,Saarinen, Leena, Niemi, Jussi, Nikanne, Urpo, Maddens, Ruben & Van de Weghe,Nico (submitted). Spatial features in path descriptions: a comparison between English,Dutch, Finnish, and Norwegian. Manuscript.Van der Zee, Emile, Maddens, Ruben and van de Weghe, Nico (submitted). The two-waystreet of qualitative information representation and natural language research.Manuscript.Van de Weghe, N., (2004).Representing and Reasoning about Moving Objects: A Qualitative Approach, Ghent,Belgium, University Ghent, Faculty of Sciences, Department of Geography. Understanding situational use of spatial language for thegeneration of spoken navigation maps Karl RehrlSalzburg Research Forschungsgesellschaft mbH, Austria Geospatial Information Systems nowadays are widely coined by visual user interfaces. Beingoriginally designed for desktop interaction, mobile access to geospatial information affordsnew kinds of "minimal attention" interfaces. The metaphor of the "spoken map" has recentlygained attention as a convenient way of user interaction, not only for the visually impaired.One crucial challenge for generating spoken maps is to find linguistic concepts being alignedto a commonsense use of spatial language. A major question therefore is to understand howpeople verbally describe spatial scenes with an emphasis on the cognitive semantics encodedin such descriptions.This talk reports on results from the SemWay project, focusing on verbal navigation supportfor pedestrians as well as alpine ski tourers in snow-covered mountains. In two field studiesverbal descriptions of route choices at 113 decision points were collected from participantsbeing unfamiliar with the navigation environment. The focus of the linguistic analysis was onextracting re-occurring conceptual motion patterns, relating qualitative motion and directionconcepts to linguistic concepts representing physical reference entities. The studies contributeto a broader understanding of language use (specifically German language) in situations ofspatial decision, needed for cognitive ergonomic language generation in spoken maps.The talk discusses similarities as well as differences in language use by different participants.Conclusions from direct experience descriptions given while walking the route to summariesgiven at the end of each route are drawn. Among the questions arising from the talk are thefollowing: What is the influence of the environment on language use in the description ofspatial scenes? Does unfamiliarity with a spatial environment foster the use of basic levelconcepts for physical reference entities? Which differences in the use of linguistic conceptscan be found in direct experience descriptions and summaries? What is a valid, formalizedway to map linguistic on cognitive concepts? How could cognitive concepts independently oflinguistic concepts be represented? Cognitive Discourse Analysis: Analyzing Spatial Thought asRepresented in Language Thora TenbrinkUniversity of Bremen, Germany Cognitive Discourse Analysis as a method is specifically designed for analyzing cognitiveprocesses as represented in language. The method has been applied in recent years both togain a better understanding of mental representations of scenes and events, and to investigatethought processes involved in problem solving tasks. The specific novelty of the method liesin the identification of those linguistic structures and patterns that are specifically informativewith respect to conceptual rather than communicative phenomena. Particularly in the areaof problem solving, the triangulation with other types of evidence, such as behavioral data, isessential. In my talk I will focus on the spatial domain and outline the main features of themethodological approach, illustrated by a range of findings gained in cooperativeinterdisciplinary studies.The studies were conducted during three years of cooperation with Jan Wiener in the scope ofour joint project "Wayfinding strategies in behavior and language", funded by theVolkswagen Foundation. I will briefly sketch two prominent application areas of this tandemendeavor. First, our research on cognitive strategies in the Traveling Salesperson Problemhighlighted the complex interplay of cognitive processes involved in finding the shortest pathto a predefined set of spatial goals. I will outline how linguistic analysis supported theinterpretation of behavioral data in this area. Second, a joint study with Jan Wiener andChristoph Hölscher on route planning in familiar environments showed that routes areplanned and described differently depending on the situation. While actual route navigation(without planning ahead) is predominantly direction-based and characterized by incrementalperception-based optimization processes, in-advance route descriptions rely more on salientgraph-based structures, and they are affected by concerns of communicability. Differences in the production and comprehension of spatial terms Matthijs L. NoordzijUniversity of Twente, The Netherlands People from the Netherlands tend to favor relative spatial terms (to the left) over canonicalspatial terms (to the West). I will discuss experiments showing that Dutch participantsconsistently favor relative spatial terms despite different situations of a fictitious addressee(someone driving a car vs. someone studying a map), the scale of the described environment(town vs. country), and explicit priming of cardinal directions. Paradoxically, studies haveshown that Dutch people actually have more accurate spatial memories from texts withcanonical than relative spatial terms. This suggests that people can immediately recode non-preferred spatial terms in a preferred equivalent one and that this recoding results in spatialmemories that are very effective. Results from transferring this idea to an applied setting(describing census data to the blind) will be discussed. Representation and modelling frameworks Christophe ClaramuntNaval Academy Research Institute, France Mental maps, conceptual maps and computational representations Early models of Geographical Information Systems have been widely influenced byquantitative models and geometrical representations of space. Despite the interest of theseapproaches for cartographical applications, they don’t completely reflect the way humansperceive and describe their environment since they preferably store and process qualitativeinformation. The perception of space encompasses cognitive principles that favormemorization of the properties of an environment, and communication of its salient propertiesto an external referent using natural language. The research presented conceptualizes such aperception and applies it to the verbal description of a natural environment and by underliningthe salient entities that emerge from space, the spatial relations binding them, and the overallstructure of the environment. It combines a conceptual analysis of the sentences and terms,and a structural-based study of the spatial and rhythmic properties that emerge from suchdescriptions. The approach favours generation of a conceptual map using different schematicand spatial views, this acting as a support for a mapping towards computationalrepresentations. When logic meets constraints: qualitative spatial and temporalreasoning from a modeller's perspective Gérard LigozaUniversity of Paris-Sud, France The main motivation for developing qualitative spatial and temporal formalisms is torepresent and reason about qualitative knowledge about space and time, as opposed to metricknowledge. Qualitative knowledge is typically what is conveyed by natural language, and thelevel of qualitative information is the natural medium of interaction with computerizedsystems of information such as GIS. The standard approach to knowledge representation and reasoning (KRR) is through logicalsystems, either classical, or non standard, such as modal logics, non monotonic logics, orabductive logics. Designing a logical system means first defining its abstract language and aformal reasoning framework, then considering four basic issues: expressiveness, complexity,reasoning procedures (e.g. defining algorithms), and models (measuring how well aformalism relates to its intended, "real world" model). Qualitative spatial and temporal reasoning (QSTR) arose from a shift of perspective, namely,concentrating on languages where formulas are seen as constraint networks. A landmark isAllen's 1983 paper where Allen's version of interval calculus is defined in terms of temporalconstraint networks. This move allows the use of the machinery developed for constrainsatisfaction problems (CSP) notably of filtering algorithms. The following two decades witnessed the development of a host of constraint-basedformalisms, covering various aspects of space such as topology, direction, or qualitativedistance. A prominent part of the research activity has been devoted to problems ofcomplexity, using both syntax-based and topologically-based approaches. The first approachis based on restricting the languages, while the second uses the structural properties of spaceand time to identify specific subsets of constraints. The question of determining the models ofthe calculi has been another active and fruitful domain of investigation. As a consequence of the introduction of the constraint-based perspective, mathematics hasentered the field in many forms, including universal algebra (relation algebras and theirrepresentations, clones of operations), geometry and topology (characterization of tractablesubsets), lattice theory (models of the RCC calculi), mereology and mereotopology, andcategory theory. Modelling the processes at work in the use of natural language in relation to space impliesintegrating aspects of that use which are beyond space and geometry, such as functionality.This emphasizes the need for developing systematic approaches to the combination offormalisms (space and time, space and function, topology and distance, topology andorientation). This is likely to put the consideration of families of formalisms, rather thanindividual ones, to the foreground of the agenda. Modeling Geospatial Activities for Navigation Sabine TimpfAugsburg University, Germany One of my main research goals is to understand, model and support spatial behavior,especially navigation behavior using geospatial information. GI-Science and spatial cognitionresearch has in the past been much concerned with representations and structures. While thoseare important they are lacking in providing meaning and fostering understanding and supportof spatial behavior. In order to support behavior we need to look at human’s activities, whichprovide the context for human behavior. Spatially distributed activities produce the need fornavigation but also the need for information on navigation and behavior choices. In my talk Iwill explore the need for geographic activity models as a basis for supporting navigationbehavior and information on navigation choices. An Urban Knowledge Data Structure Kai-Florian RichterUniversity of Pittsburgh, USA, and University of Bremen, Germany There are some fundamental differences in how machines and humans communicate aboutspace, which becomes especially apparent in our daily lives when using navigation services.This is to a large part rooted in the differences in spatial representations used by machines andhumans. I will focus on route directions as a specific form of spatial communication. In recent years,several approaches have been developed that contribute to bridging the gap between machineand human communication of these directions. Predominantly, these try to shift from metricdistances to references to environmental features (landmarks) in describing actions to beperformed. Another line of work investigates how to structure instructions to ease theirunderstanding. Both call for mechanisms that support these operations already on therepresentational level. In my talk I will illustrate what is required for such representations andpresent the Urban Knowledge Data Structure (UKDS), which is based on the Open LocationServices specification of the Open Geospatial Consortium (OGC). UKDS combines thecomputational efficiency of a network representation of space with formal specifications ofcognitive principles of human direction giving. Interrelations and combinations John BatemanUniversity of Bremen, Germany Relating Spatial Language, Representation and Behaviour Our work on extracting the linguistic semantic commitments of spatial language has led us toa two-level approach to semantic interpretation and generation that provides particularsolutions in the area of spatial language. More specifically, we adopt a primarilygrammatically-motivated abstraction covering the spatial commitments of decontextualisedlanguage and then relate this to situated language use in a separate step. The grammatically-motivated abstraction is represented in terms of a linguistically-motivated ontology. Thisallows us to explore the process of contextualisation as a case of inter-ontology mapping. Thetwo-level architecture supports a decoupling of qualitative treatments of spatial context orbehavioural simulation on the one hand, and abstract linguistic spatial configurations on theother. In the presentation, I will briefly set out this approach to handling spatial language,suggest examples of where it shows itself to be necessary, and discuss open problems andresearch directions where empirical investigation needs to be taken further. Strategic choices in spatial cognition Marios AvraamidesUniversity of Cyprus Much of the research in spatial cognition focuses on examining how various factors influencespatial performance in an attempt to understand how spatial information is encoded andmaintained in spatial memory. For example, scientists in this field routinely manipulatevariables such as the number of objects in a layout, the arrangement of the objects, thepresence of extrinsic cues, the viewpoint from which a layout is learned, the modality inwhich the layout is experienced, the location of testing, the response mode and so on. Basedon the pattern of findings, inferences about the organizational structure of spatial memories(i.e., what reference frames are involved in maintaining information in spatial memory) andits properties (e.g., if and how memories can be updated during observer movement) aretypically drawn.In this presentation I will discuss the hypothesis that by manipulating these variables,scientists are merely biasing the strategies that participants use to encode spatial informationand to use it when executing tasks. Findings from previous studies, as well as studiesconducted in my lab, that are compatible with this hypothesis will be presented and discussed. Do language based representations support spatial performance? Hanspeter A. MallotUniversity of Tübingen, Germany Recent theories of embodied cognition suggest that verbal representations are given meaningby “simulation” processes performed in non-verbal, sensori-motor representations. Thistheory seems to suggest that non-verbal representation alone would be sufficient to generatefunctional behaviour. The role of verbal representations would be limited to communicativeand dialogic uses. In a series of recent experiments, we have tested the (alternative) hypothesis that language-based representations do support spatial behaviour even in monologic situations. Subjectswere trained to find goals in a virtual environment comprising an iterated Y-maze with clearlynameable landmark information at each decision point. In three conditions, subjects wereeither allowed to explore and learn on their own (“baseline condition”), or were instructed ateach place to select from a list either a suitable name of a place (“word condition”) or an iconsummarizing the gist of the landmarks presented at that place (“icon condition”). Theexperiment was performed both for a route and for a map learning task. The talk will discusspreliminary results obtained from this experiment. Rainer MalakaUniversity of Bremen, Germany