COSIT 2026

Description: We live and navigate continuously between indoor and outdoor spaces, yet the research landscape is fragmented into separate cognitive strategies, computational models, and support systems for the indoor vs outdoor contexts. This session invites the COSIT community to reflect where the differences are truly fundamental, and where convergence across these strands is possible. The session is designed as a structured, discussion-driven workshop centered around interdisciplinary exchange between the cognitive/empirical and computational/technical groups.

We begin with a short warm-up that anchors participants in the indoor-outdoor relationship of the venue and the city of York. Through a brief pointing and recall task, participants will reflect on how they locate themselves relative to the building entrance, external landmarks, and global reference frames. A brief discussion will highlight the diversity of strategies and information sources that people rely on when thinking across the indoor-outdoor boundary.

Participants then self-select into small groups organized around two core positions: emphasizing either similarities or differences between indoor and outdoor spatial information. Within each position, groups are further structured by disciplinary focus, resulting in six discussion groups in total: (a) similarities in cognition and behaviour, (b) differences in cognition and behaviour, (c) similarities in computation and information digitisation, (d) differences in computation and information digitisation, (e) similarities in HCI aspects, (f) differences in HCI aspects.

Each group has two key tasks, supported by the pre-designed Miro Board: (1) develop a mind-map of keywords that describe related research streams in the indoor vs outdoor contexts (are they always the same? when not, what’s the reason for different names for the same concepts?) (2) develop arguments convincing the others to your groups’ designated position (e.g. for group a: “cognitive and behavioural processes across indoor/outdoor are similar and need not be separated”, for group d: “computation and information digitisation across indoor/outdoor has so different needs, that it cannot be integrated”.

We then bring groups into plenary confrontation. Each presents its core position, followed by moderated cross-discussion encouraging other groups to question the arguments. We aim at identifying points of convergence, but also differences, and open research gaps across cognition and computation. These will be summarised on shared Miro Boards by the organisers after the session, as an output preserved from the workshop.

We close by returning to the experiential reflection: participants leave with a brief pre-designed task to observe how their spatial strategies, and information sources, shift when exiting the conference venue.

COSIT is a unique community where multiple streams of spatial research meet: spatial behaviour and cognition, spatial databases and digital twins, building information models, human-computer interaction for spatial systems, and sensing and data acquisition. Yet even within shared topics (for example, wayfinding), work tends to specialise in either indoor or outdoor contexts, with limited crossover. Given that spatial behaviour is continuous, Spatial Information Theory research is a perfect framework background to reflect on this divide. This session is aiming to bridge the gap by questioning if this divide is grounded in a fundamental cognitive/technical differences across indoor and outdoor contexts, or merely a legacy of fragmented computational tools and separate academic disciplines. The depth and breadth of expertise across theoretical, empirical, and technological approaches present at COSIT makes it an ideal setting to explore where integration is feasible.

Description: The session will start with a 10-minute elaboration of the meanings of Spatial Human-Computer Interaction. It will continue with a panel discussion involving 3 panelists, each of whom is a representative of one of the three domains. Each panelist is asked to give an initial 5-minute statement about their point of view on how their topical focus relates to the other two. Subsequently, the organizers will guide the discussion among panelists. After 30-minutes of discussion among them, the session will be opened up to questions and comments from the audience.

COSIT has seen a tradition in bringing interdisciplinary scholars together to discuss emerging issues. Spatial Human-Computer Interaction as understood in the traditional sense (“HCI in 3 dimensions”) has seen increased interest in particular related to virtual and augmented reality techniques; its interpretation can further be extended to HCI that is happening in geographic space. Spatial information theory can inform interaction design and analysis for both interpretations. Spatial Cognition is facilitating the understanding of the reasons for interactions and where they occur.

Description: The UK’s Safer Streets Mission aims to reduce serious harm and increase public confidence in policing and criminal justice system. Integral to the mission is the ambition to half knife crime and violence against women and girls (VAWG) within a decade. The US have their own programmes that focus similar missions (e.g., Project Safe Neighourhoods, or regional examples such as LA’s Operation Safe Cities). The challenge faced by law officers and agencies is that many are also having their budgets reduced, requiring them to achieve more with much less resources. The adoption of technology is increasingly seen as a force-multiplier and a way to assist law officers and agencies in meeting their objectives. Our thematic session focuses on the important role that spatial information theory will play. Specifically, this thematic session will focus on how the COSIT community can come together to address the concept of safeability.

Through Geospatial Innovation (the transformation of geospatial data into actionable insights using emerging technology), safeability measures and monitors how safe an environment is at a given time of day. Like concepts such as walkability and liveability, safeability is an experienced and perceived spatial phenomenon. It has two aspects: physical safety and perceived safety and can be considered for a particular individual, group or community. The concept is dynamic and will be considered as comprising two distinct phases: initial (i.e., designing environments where safeability is considered high based on what has been learnt from similar areas), proactive management (detecting and predicting individuals and groups whose safety has been compromised). Safeability should not just consider environmental features. It should also consider social aspects of an individual (i.e., how isolated they feel, how engaged they are with their community). The concept builds upon existing research into walkability and liveability but must be applicable to both include both urban and rural environments. The use of Geospatial Innovation should facilitate monitoring in real-time and predictive analysis.

The session will include a set of lightening sessions followed by a wider discussion session. The content of the discussion will be written up as a white paper that highlights the research that is already being undertaken, and the key research questions that remains unaddressed. Lightening talks will be included that address at least one of the following research questions:

• How can we leverage Geospatial Innovation to consider both perceived and physical components of safeability?
• How much information can be derived from GEO-INT (i.e., a subset of OSINT)?
• What safeability measures could be identified that do not rely on self-reporting but can be derived in real-time?
• How can spatial information theory ensure the concept of safeability can be applied in both rural and urban environments?
• How can the transparency and explainability of GeoAI solutions be considered?

Safeability is an experienced and perceived spatial phenomenon, which brings together many COSIT conference related topics. It also fits with COSIT 2026’s theme Cognition and Flows. Addressing safeability will require us to close the gap between our understanding of human spatial behaviour and aggregated flows, especially how cognition and behaviour influence the nature of societies. The application of the research (i.e., to assist law officers and agencies in meeting their objectives) is of increasing importance across the different countries that are represented within the COSIT communities. We will be inviting practitioners as well as researchers to partake in the session.

Description: The proposed thematic session will explore how today's emerging multi-modal AI systems represent and reason about geographic spaces and places. While these models are being increasingly applied to geospatial problems by industry, research, and even individuals, the implications for geographic theory, spatial representation, cognition, and ethics more broadly remain largely unexplored. While a lot of work has gone into understanding the accuracy (correctness, so to speak) of these systems, representation has received little attention. Put differently, while we understand the abilities of these systems to, for instance, infer topological relations, we do not know how distorted their latent geographic representations are and whether these distortions matter for everyday use. This is a substantial problem, as first research is showing that frontier models form strong geographic prototypes (called defaults). Therefore, we will focus on issues of bias, robustness, trust, and AI alignment but address them from a uniquely geographic perspective.

The session will intentionally avoid any lengthy presentations or even front-facing panels. Instead, we will focus on several short demos and exploratory experiments directly involving the audience (and prepared by our (12-14) Master's students that will be traveling with us). These experiments and their results will then be discussed in the form of an open, on-stage debate involving the entire audience, for instance, by bringing people to the stage. We hope to close the session by collectively reflecting on the observed system behavior, potential consequences, and research directions for future work. We will take notes and share all experiment results and discussion notes openly. Given the nature of experimentation/demos and discussions, we strongly favor 90min, but can do so in 45min as well.

The broader topic sits at the intersection of several core COSIT themes. Among them are geographic(spatial) theory, issues of representation (semantics), GeoAI, cognition, and ethics. We believe this mix makes the session uniquely suited for COSIT and ensures it is not limited to purely technically minded participants alone but also those interested in behavioral and cognitive geography. Most importantly, the topic is substantially broader than GeoAI alone.Our focus on representation aligns well with many classical COSIT papers of the past. Finally, the ethics perspective also seems a strong match for COSIT.