Event 2: Multiplayer Perceptual crossing Test

Multiplayer Perceptual crossing Test

In recent years, researchers in social cognition have become aware that many social phenomena cannot be understood simply by investigating offline situations and explaining them on the basis of individual capacities of the agents in the interaction.

There are processes of dynamic emergence that only arise In online situations, when two or more individuals are engaged in a real-time interaction. Auvray’s perceptual crossing paradigm (Auvray, Lenay and Stewart, 2006) provides the most basic conditions for studying the genuine factors involved in recognising others in social interactions. In this shared virtual environment, pairs of blindfolded participants are placed in separate rooms and interact in a common virtual one-dimensional perceptual space. Each participant moves a cursor along a line using a computer mouse and receives a signal when encountering something on the line. This signal might mean the presence of the other participant but also a displaced “shadow image” of the partner, that behaves  in a strictly identical way in shape and movement terms. Surprisingly, previous experience on this has shown that we have a great ability to discriminate live interactions from those interactions from recorded motions that are identical to the live interaction but cannot react contingently. Simulation models of this task (Di Paolo et al., 2008) and variations (Iizuka et al., 2002) have been also investigated showing clear patterns.

In the first quarter of 2013 in Zaragoza, a citizen science experiment, based on the Auvray’ perceptual crossing paradigm, will be carried out by the Advanced Computer Graphics Group (GIGA) at the Aragón Institute of Engineering Research (I3A), together with the Fundación Ibercivis. The program “Living Science” of the local government of Aragon, that supports the dissemination of science among lay-audience, will coordinate logistic matters. It will be the largest study regarding real-time perceptual crossing to date. About 1,000 high school students (17–18 years old) from different high schools from Zaragoza will participate interacting electronically in real time via a perceptual crossing interface specifically created for the experiment that will be accessible through the computers available in their respective schools.

 
The best results will be used to encode the populations of a genetic algorithm, with which train a bot to solve the perceptual crossing task and prove whether it could fool a human in a perceptual crossing Turing test. Those games between those students with greater fitness will be selected and recombined to form new populations in the next iterations of the algorithm. Finally, further statistical analysis from the experimental data will be performed in order to infer the different strategies taken by the players to detect if they are interacting with either a human or a machine.
 

Collaborations

The event will be organized in collaboration with
 
                                 

Aragón Institute of Engineering Research   /    University of Zaragoza

            

 Ibercivis Foundation                                 Government of Aragon