The use of virtual characters to practice communication skills has clear advantages. That was the main conclusion of the International Symposium on Software for Practicing Communication Skills, which took place on 16 November in Utrecht. At the symposium, four researchers spoke about their views on practicing communication skills with the help of virtual characters. The symposium was organised in the context of the festive transfer of the Virtual Patient.
Tibor Bosse (VU Amsterdam) kicked off the symposium with an overview of opportunities and challenges in computer-based communication training. Such training programmes are often used to practice communication when emotions are involved, and where taking an appropriate communication style is very important. Application areas are diverse: from doctors who use it to practice bad news conversations, to the Dutch Ministry of Defence, where it is used in cultural awareness training for soldiers.
While it is not possible to completely replace roleplay, there are many upsides to computer-based training, such as lower costs, adaptability – even at runtime – and objectivity: every practice session is exactly the same. Much of the learning effect comes from the responses of the virtual character to the player – for example, if a shopkeeper responds agressively to a robber, things may go very wrong – but explicit feedback and reflection afterwards may be used as well. In the future, virtual touch may even be added: what if that robber actually slaps you? Although those scenarios are quite extreme, they may help users to practice reacting under stress.
Raja Lala (Utrecht University) gave an extensive overview of scenario development tools and their properties. Scenarios have essentially the same structure, and are generally represented in directed graphs, where each statement from the computer is followed by a choice from the player. Lala also presented an XML language for modelling communication scenarios, and a configurable authoring tool, which can be accessed via the Communicate! website.
Jeroen Linssen (Twente University) spoke about different types of virtual environments for training police officers, based on real police interviews with actors. He presented a training programme with cartoon characters, which incorporates "thought bubbles" from the characters to present direct, explicit feedback to the user on their current action. The training gets more difficult as the user gets more competent. Another simulation works with virtual reality, which has the advantage that officers report feeling very immersed in the environment.
Linssen gave some suggestions for further research, such as role reversal, where players play their antagonists to get better insight. He suggested the use of big data to reuse player behaviour to generate new scenarios. Setting standards for evaluating virtual agents will help to make it easier and clearer to determine how believable or social a virtual agent is.
Stephen Chapman (Keele University) spoke about virtual training scenarios from the point of view of medical training. In his experience, simulations with actual actors give a complete hands-on experience, but they are difficult to standardise. Whereas the behaviour of actors may differ between cases, "technology doesn't get tired", Chapman said.
He shared his experiences with the use of virtual reality caves, where groups of students practice patient care inside a simulation. He has observed his students acting engaged, relaxed, happy, and learning. Evaluations show that virtual reality environments have a clear effect on learning, with students scoring 8% higher on both retention and understanding than a control group who used traditional training methods. Another evaluation shows that pharmacists that use a virtual training environment experience a significant increase in their ability to conduct pharmaceutical plans and in their clinical knowledge.
In a panel discussion, Utrecht University researchers Zerrin Yumak and Ronald Poppe reflected on the symposium and shared their views on the challenges for software for practicing communication skills. Yumak said she would be interested in research into incorporating computer animation and artificial intelligence, and the use of data analysis to automate behaviour without too much authoring effort. Poppe advocated for further research into the transfer of skills from the virtual to the real environment.