Quiet please, computers in contemplation!
Can machines think? Or more importantly, do we think that they think? That is the question a team of EU-funded researchers asked, and the answer appears to be 'yes', if the machine in question looks like a human. Their results raise even more questions for future researchers. The study, which was led by Dr Sören Krach and Professor Tilo Kircher from the Clinic for Psychiatry and Psychotherapy at RWTH Aachen University in Germany, is published in the July 9 issue of the online, open-access journal PLoS ONE. The team investigated how people interact with computers, and more specifically whether humans respond differently to computers that look more 'human'. The participants of the study played 'the prisoner's dilemma'. Developed by the RAND Corporation, this seminal game pits two people against each other. In the scenario, both players are arrested by the police and are interrogated separately. Each player has the opportunity to go free if they testify against their friend, who would go to jail for ten years. However, if both remain silent, both go to jail for six months. This scenario has been adapted over the years to play out a range of socio-political theories. In this study, participants were pitted against four different game partners: a regular computer notebook, a functionally designed Lego-robot, the anthropomorphic robot BARTHOC Jr. and a human. All game partners played the same sequence, a fact which was not, however, revealed to the participants. The researchers observed that the participants rated the anthropomorphic robot as being more competitive and less cooperative than the functional robot and the computer. The human partner and the anthropomorphic robot were not perceived differently with respect to competitiveness. They also noticed that the human players found the anthropomorphic robot to be more sympathetic and pleasant to interact with than the functional robot. For the participants, the very appearance of the anthropomorphic robot was valued as more human-like and wins against it evoked more positive feelings compared to winning against the functional robot or the computer. This study provides the first evidence that the degree of human-likeness of a corresponding 'person' influences perception, communication and behaviour. They were able to show that this modulation is linear; therefore, the more an agent or entity exhibits human-like features, the more people build a model of its 'mind'. The findings also allow the researchers to conclude that humans alter the way they behave based on how human looking a robot is. Furthermore, the less human looking a robot is, the more they expect the robot to act in ways that are 'alien'. What this suggests is that the more a robot looks human, the more humans expect the robot to act like a human. EU support for the research came from the COGNIRON ('Cognitive robot companion') project, which is financed by the 'Information society technologies' (IST) Thematic Area of the Sixth Framework Programme (FP6).
Countries
Germany