Serendip is an independent site partnering with faculty at multiple colleges and universities around the world. Happy exploring!

On Intelligence

gflaherty's picture

            OnIntelligence: How A New Understanding Of The Brain Will Lead To The Creation ofTruly Intelligent Machines offers a new perspective on the inner workingsof the most complex human organ. Written by Jeff Hawkins, this book links together the science of thebrain and the logic of the world of computers.  Hawkins is founder of Palm Computing and Handsping, twocomputing companies which invented the Palm Pilot and the Treo, respectively.  Known for his ability to innovateintelligent technology, Hawkins is yearning for more. 

            Heemphasizes a long history of computer scientists striving to duplicate theintelligence observed in humans. From the 1940s to the 1980s Artificial Intelligence was the pioneeringfield of computers science.  Theend goal of the time was to mirror the system of intelligence is seen in thehuman brain.  However, this systemcompletely ignored the rules of biology and thus failed, in the eyes ofHawkins, to emulate human behavior and the brain’s unique manner of processingof information.  Hawkins alsomentioned the era of artificial neural networking from the 1970s through the1990s.  Highlighted by the ratherweak attempt to shadow the human brain, neural networking made somewhat of aneffort to understand the biological processes involved in humanintelligence.  However, ultimatelyboth of these systems failed because they failed to address human intelligence,as a biological process, as their primary concern.  Instead, both systems allowed computer science to be theprimary concern, and biological processing took the back seat.

            Hawkinsproposes a new way to approach the creation of intelligent machines.  Real Intelligence, a biologicallyderived study of neural science and neural physiology, lends itself to a moreprogressive and viable method of creating intelligence.  Hawkins argues that in order to betruly successful in the business of creating intelligence, one must firstunderstand the brain at the most detailed of levels.  To this end, Hawkins sets out to highlight the neocortex asthe region of the brain most involved in the processing and storing ofintelligence.

            Hawkinsdescribes the neocortex as the localization of everything that people think ofas intelligence: language, thinking, music, art; basically anything that aperson can talk about is stored in the neocortex.  The brain is a very complex organ in any animal and ismostly made up of neurons.  Theneocortex, meaning ‘new layer’, is an evolutionarily rather recent additiononly found in mammals.  Differentregions of the neocortex are involved in different parts of function.

            Hawkinscategorizes inputs to the brain into three main branches.  The auditory nerve, the optical nerveand the spinal chord; these are the three ways that information can get sent tothe brain.  He also emphasizes thefact that the brain lives in perpetual darkness; it cannot hear or see orfeel.  In other words, it onlyprocesses electrical impulses sent by these three sources and must decipherthese impulses into images, sounds and feelings.  These impulses, regardless of which of the three areas theyare coming from, all look essentially the same as the enter the brain.

            Hawkinsalso proposes that the neocortex’s end goal is predicting the future.  He reiterates the fact that this is notsome sort of psychic ability, but that the neocortex has very concreteexpectations about very ordinary events of the future.  And these predictions have the abilityto mix and match from the three inputs. An example of this would if one was to watch a person put their handstogether to clap.  They wouldexpect to see the hands meet and stop. As the hands stopped, they expect to hear a loud noise.  If they did not hear this noise or theysaw the hands move past each other, then the neocortex would have to scrambleto decipher these unanticipated inputs.

            Hawkinscalls this ability of the neocortex to default to other parts of the neocortexwhen it receives a unanticipated input, Hierarchical Connectivity.  He expects the lowest form of cells inthe neocortex to be able to decipher ordinary everyday inputs, but as inputsbecome more and more unexpected, the electrical impulses are sent to more andmore sophisticated and specific neocortex cells.  Hawkins believes that there are cells within each region ofthe neocortex that are more specific than others.  The region of the neocortex that deals with language mighthave many cells that are used to decipher the spoken word.  However there might be much morespecific cells designed to respond to a specific word or phrase. 

            Hawkinsmain goal in explaining the workings is to suggest that there is a commonalgorithm underlying everything that humans do.  And that figuring this out can be done through the study ofbiology.  He also proposes that functionsof the brain can be expressed mathematically.  These mathematical equations are likely to give rise to apromising future of Real Intelligence machines.



Paul Grobstein's picture

brains and computers: a common algorithm?

The interplay between "computer science" and "biological processing" is indeed an interesting one. Maybe there will eventually be agreeement that there is no "common algorithm underlying everything that humans do"?