The human brain has been identify as a massively parallel computer science machine . But just how powerful is it ? A late brain scan analysis is offering some unexpected results .
Neuroscientist Harris Georgiou from the National Kapodistrian University of Athens in Greece lately take it upon himself to number the turn of discrete “ CPU heart ” at work in the human brainiac as it do round-eyed tasks in a operative magnetic resonance imaging ( fMRI ) motorcar .
This is quite the task when you cogitate about it . The human mastermind comprise of about 100 billion neuron that each make up about 10,000 connections with their counterparts . By using an fRMI car , and by quantizing the brain into 3-D pixels called voxels ( each about five three-dimensional millimeters in sizing ) , Georgiou was capable to represent the complete activity of the brain using a storage-battery grid of 60 x 60 x 30 voxels . Through this method , he create a dataset of around 30 million data detail .
mentality activity was register as participants perform one of two simple tasks , one a visuo - motor task , the other a visual recognition chore . MIT ’s Technology Reviewdescribes the determination :
Although the analysis is complex , the final result is dim-witted to country . Georgiou says that independent component psychoanalysis reveals that about 50 main processes are at body of work in human brains performing the complex visuo - motor project of indicating the presence of green and cherry boxes . However , the brain use fewer processes when carrying out simple tasks , like visual recognition .
That ’s a fascinating termination that has significant implications for the mode computer scientists should design buffalo chip mean to mimic human performance . It implies that parallelism in the brain does not occur on the layer of single neurons but on a much higher morphologic and functional stage , and that there are about 50 of these .
Georgiou points out that a typical voxel corresponds to just about three million neurons , each with several thousand joining with its neighbors . However , the current state - of - the - art neuromorphic chips contain a million stilted neurons each with only 256 connection . What is clear from this work is that the parallelism that Georgiou has measure occurs on a much gravid scale than this . [ vehemence supply ]
So an artificial equivalent of a genius - like cognitive social structure may not require a massively parallel architecture at the level of exclusive neurons . rather , and as noted by Georgiou , one could be built using “ a in good order designed exercise set of circumscribed operation that track down in parallel on a much lower scale . ”
scan theentire articleat MIT ’s Technology Review , and check out the scientific studyhere .
Cognitive scienceFmriNeuroscience
Daily Newsletter
Get the best technical school , science , and cultivation news in your inbox day by day .
tidings from the hereafter , deliver to your nowadays .
You May Also Like
