Virtual robot Tchnology links body to numbers just like humans

A virtual robot has acquired cognitive ride common in people - further evidence that computers need bodies if they're ever going to think like us Read more: "Squishybots: Soft, bendy and more intelligent than everMovie Camera" One of the many curious habits of the human brain is that we tend to associate a small number with the left side of our body and large numbers with our law. Now a virtual robot embedded in a world of synthesis has acquired the bizarre. This is to help unravel the puzzle of how even very abstract concepts such as numbers can be embedded in our physical interactions with the world. "The numbers are so abstract and pure, they seemed detached," says Marek Rucinski of the University of Plymouth, UK, who presented the virtual robot at the conference of Cognitive Science in Boston in July. "However, each concept we have is somehow grounded in the real world. " The so-called SNARC (spatial numerical association of response codes) effect is well established: people respond more quickly to a number (by pressing a button, for example) with their left hand when the number is small and with their right hand when the number is large. Similarly, people who have brain damage that causes them to ignore the left side of his body show a preference for a greater number when asked to report to the middle of a numeric range. These number-space associations embody broader insight surprising that many other aspects of the camera thoughtMovie abstract are believed to have roots in our sensory and motor interactions, says Martin Fischer, who specializes in so-called embodied cognition at the University of Potsdam in Germany. "SNARC effect shows that the abstract knowledge, even returns directly to the sensory and motor experiences that we had when we acquired the knowledge." The accepted explanation for the SNARC effect is that when we in the West learn to count, we are more likely to be taught in a direction from left to right - on a table or perhaps a parent to point objects in a row. This way of learning sets up connections between a small number and the part of the brain that controls our left side, and vice versa, which persists into adulthood. There is also evidence from cultures that read from right to left as the association is reversed. To better understand how these links could be formed, Rucinski and his colleagues turned to a virtual version of a child - a numerical simulation of the humanoid robot iCub - and exposed to processes that could help these connections are formed in the people. iCub development began with a phase called "motor babbling" random virtual moving his arms and the look that human babies are supposed to use to become aware of their bodies. In a rough simulation of the architecture of our brains, the researchers set up three zones for spatial processing in the brain of the iCub: two for each arm and one in his eyes. Next the researchers learned to count iCub by presenting it with a stream of numbers from 1 to 15. To imitate the cultural prejudices of the West towards counting from left to right, low numbers were presented to the left and top right. Learning software that mimics how synapses form connections in the human brain while a small number associated with the left area of ​​the map the eye and the left arm - and vice versa for the right. iCub is ready to take the test SNARC. He was presented with a series of random odd and even numbers. In one case iCub press a button with his left hand when the number is odd and the right hand when he was still in a second instance, the buttons were reversed. Like humans who took the same test, iCub was faster at a time when the number is small and pressed the button with his left hand, and when the number was big and the button was on the right. In a second amended SNARC, the robot was presented with a number followed by an object, either right or left. By the time the robot has taken notice of the object was then recorded. It was faster at a time when the number was small and that the object was on the left, and the number was large and the subject was on the right. Rucinski can explain what is happening. In both cases, the connections established during motor babbling and learn to count meant that simply by appearing, each number has been automatically activating the brain areas associated with the space or the right or left side of body iCub. This makes the robot faster with this side of the body to accomplish the task - whether to hit a button in the conventional test or SNARC noticed an object in the modified test. "The model provides an explanation for many embodied-space interactions," says Rucinski. The results show that these associations can emerge from a combination of learning from left to right and how the human body is built. This does not mean that the thought process iCub mirror exactly what is happening in a human brain - but it does not mean that the association can result from two factors only those that both human beings are said to it. "The approach is very good complement to what we do in experimental psychology," says Fischer, adding that he looks forward to the day when working in cognitive robotics could make a prediction on cognition that is then tested in humans, rather than the other way.