Why are people good at finding their way through an airport? Yet why is it so difficult to teach a robot cleaner to deal with a wastepaper basket that was not there yesterday? Why can people easily follow an instruction: "Turn at the hotel," and have difficulties with: "Turn after 134m"? To what extent is "Melbourne" similar to "Carlton"? And why is a Swiss tourist calling a hill what Australians are calling a mountain? This is what spatial information theory is about. Spatial information theory deals with human or agent decision making in geographic space, with communication about space (you may call it interoperability), and with the formal description of geographic space. It aims to provide the foundations for applications as diverse as the design and construction of Web 2.0, interoperable services, next-generation location-based services, transport management and information, ubiquitous spatial computing, virtual environments and robotics.

These questions and others were discussed at COSIT'07, the Eighth International Conference on Spatial Information Theory, which was held 19-23 September 2007 in Mount Eliza, Victoria, Australia. The COSIT conference series has been shaping and leading interdisciplinary research in theoretical GIScience for more than a decade, involving disciplines such as spatial information science, artificial intelligence, cognitive science, neuroscience, cognitive psychology, cognitive anthropology, linguistics and philosophy.

COSIT usually takes place in Europe or North America. This year researchers from the Asian-Pacific region (forming about one-third of the participants) had a chance to participate and share their ideas. This mixture of "old" and "new" faces worked very successfully for COSIT'07. Held at Melbourne Business School, Mount Eliza , COSIT'07 had more participants and a more competitive paper selection process than at any previous COSIT. The tranquil setting created a relaxed atmosphere for the intensive discussions between participants, discussions that went on even through the nights.

The first day held four international workshops. The topics covered included spatial cognition and architectural design, semantic similarity in geographic applications, distributed and mobile spatial computing, and social networks in geographic space. The next three days were structured into formal knowledge representation, ontologies and similarity (day 1), perception and cognitive mapping (day 2), and qualitative reasoning, navigation and spatial uncertainty (day 3). The last conference day is traditionally reserved for a doctoral colloquium at which talented doctoral students have the opportunity to present their research ideas to their peers, as well as to more senior researchers, frequently for the first time.

So, where is spatial information theory currently located, and where is it heading? Here are the bigger issues at play:

• Ontology and semantics promise to provide foundations for integration and interoperability - pressing issues in practice, and increasingly in the future.
• Learning about our environments by developing models and studying cognitive representations that deal with incomplete knowledge and imprecise observations. Progress here is immediately relevant for robotics, autonomous vehicles and supporting human navigation.
• Communication of spatial information to people using cognitive and linguistic concepts, including understanding the differences between languages and cultures. The applications that will profit are not the expert tools, but all the ubiquitous services supporting everyday decisions.

The feedback from youthful to distinguished researchers was enthusiastic. Prof. Andrew Frank from TU Vienna, the founder of the COSIT conference series, publicly declared this to be the "best COSIT ever." In 2009, COSIT will return to Europe.

Bruce Thompson, Director of Spatial Information Infrastructure, Department of Sustainability and Environment of the Victorian Government, after his welcome address to COSIT'07, with conference chair Stephan Winter. (Photograph by Matt Duckham).

Falko Schmid (U Bremen, Germany) explaining his poster and demo to Volker Paelke (left, U Hannover, Germany) and Bernhard Klingseisen (Curtin U, Australia) after winning the best poster award. (Photograph by Matt Duckham).