[In this case CASA can stand for “Cars Are Social Actors”! -Matthew]
Workshop on the attribution of cognitive abilities to vehicles
(in conjunction with The 6th International Conference on Automotive User Interfaces and Interactive Vehicular Applications)
WEBSITE
IMPORTANT DATES
Paper submission deadline: July 20 (extended)
Notification of acceptance: July 27
Camera ready version due: Aug. 03
Workshop date: Sept. 1
BACKGROUND
Humans tend to modulate their behaviour based on beliefs about the agent they interact with (Branigan et al., 2011), including cognitive abilities. This remains true when interacting with robots (Vollmer et al., 2013; Kopp, 2010) and interactive vehicles (Thill et al., 2014). Present-day and future cars arguably offer skills and abilities that one would normally ascribe to intelligent agents, either contributing to the driving task in many ways (e.g. through adaptive cruise control or congestion assistance), or even taking it over entirely. It can reasonably be expected that humans either are already treating cars as intelligent agents or will do so in the near future.
It is important, in this context, to highlight the distinction between perceived (or attributed) and actual intelligence: the vehicle may well be operating using sophisticated and complicated algorithms but if their effects (or even existence) are not perceivable by the driver, they may not contribute to the driver’s beliefs about the vehicle’s intelligence. On the other hand, the human tendency to project intentions and cognitive abilities on even simple shapes has long been known (Heider and Simmel, 1944), which may influence expectations in vehicle behaviour: drivers, for instance, tend to expect near-perfect performance from automated systems (Ma, 2005). A mismatch between expected and actual abilities — which can be lead to an inappropriate level of trust in the system (Helldin et al., 2013; Riveiro et al., 2014) — may lead to an inability of the drivers to accurately identify the limits of the automated system (Goodrich et al., 1999), with consequences for road safety.
Overall, it is therefore not just the actual abilities of the vehicle (and how to convey these to the driver) that matter; it is also the driver’s perception of the vehicle as an artificial, possibly intelligent, agent. It is this perception of a vehicle, along with the challenges and opportunities it offers for human-vehicle interaction and vehicle UX (user experience) design, that we are interested in exploring in the present workshop.
TOPICS OF INTEREST
The goal of the workshop is to further the theoretical understanding of the ways in which the driver’s perception (or attribution) of a vehicle’s cognitive abilities should be explicitly considered in vehicle UX design. Specific topics of interest to the present workshop include (without being limited to):
- Characterisation of attributed abilities. E.g.: what type of cognitive abilities do humans attribute to artificial agents? What are the contributing factors to such attributions?
- Relationship between perceived and actual abilities. E.g.: what relation (if any) exists between actual abilities and attributed abilities?
- Relationship between perceived abilities and trust. E.g.: how do perceived abilities affect trust, for instance in function of the degree to which they match with actual abilities?
- Relationship between perceived abilities and UX. E.g.: to what degree do user interface elements, or design decisions (rather than actual abilities), affect perceived abilities?
- Metrics, methods, and tools for measuring perceived abilities. E.g.: How do we effectively determine what abilities a driver attributes to a vehicle? How can these be measured and quantified?
- Incorporating perceived abilities in the system design process. E.g.: How should a system design process take into account the abilities that a driver might attribute to a system? Should it encourage attribution of particular abilities, while discouraging others?
- Perception of non-safety critical systems (such as navigation assistants or systems to encourage eco-friendly driving behaviour). E.g.: do non-safety critical systems influence the overall perception of the vehicle’s ability, even though they are not necessary for the driving task?
These topics are naturally intertwined, and we encourage contributions that address several of these at once. In particular, we expect all contributions to discuss possible consequences for vehicle UX design. We particularly welcome submissions beyond the automotive domain in the strictest sense. Any work on human/machine interaction (such as cognitive robotics) will be considered relevant as long as it can make a clear statement of relevance to the automotive domain.
SCHEDULE OUTLINE
We expect to allocate 20-minute slots to each paper (12 minutes of presentation; 8 minutes of discussion). The workshop will conclude with a roundtable discussion, whose purpose it is to distill a unified take-home message from the papers and discussions of the workshop. It will discuss, in particular, the direction future research is heading, and identify the most important areas for current and near-future development.
SUBMISSION
Papers may be up to 6 pages using the CHI Extended Abstract format. Position papers are welcome. See the webpage for the submission process details. Please submit by July 10.
Papers are peer-reviewed by at least two peers. Decisions of acceptance are made by the organisers based on reviewer feedback. Accepted papers will be published in the adjunct proceedings of Automotive UI 2015, to be published on the conference website. Depending on the quality of publications and the outcome of the roundtable discussions, we will consider a journal special issue for which extended papers will be invited.
ATTENDANCE
The workshop takes place on September 1, from 13.30 to 17.30. There are no additional attendance fees, but registration for the main conference is mandatory to attend the workshop. A video projector will be provided; please get in touch if you require additional material.
ORGANISERS
Serge Thill, University of Skövde (serge.thill@his.se)
Azra Habibovic, Viktoria Swedish ICT (azra.habibovic@viktoria.se)
Maria Riveiro, University of Skövde (maria.riveiro@his.se)