This is Jeff Hawkins on Brains

The man -- and inspiration -- behind the PalmPilot.

Main story: The Next Small Thing More: The Right Way to Make Mistakes

Where do people who design breakthrough products find their inspiration? That question goes to the heart of both creativity and entrepreneurial success. For Jeff Hawkins, the visionary behind the PalmPilot, the answer doesn't start with handheld computers. It starts with something more powerful than a computer: the human brain.

Hawkins loves the PalmPilot. And he is excited about the future of the fast-growing market that he created. But he is obsessed with the workings of the human brain. We asked him to explain his research into cognition - "brain stuff," as he calls it.

"My fascination with cognition and neurobiology - the physical basis of human intelligence - goes back almost 20 years. In fact, in January 1986, I abandoned my business career to pursue that interest full-time. I enrolled in a graduate-level biophysics program and began searching for a theoretical understanding of what our brains do and how they work.

"I had two big insights that eventually related to my work at Palm Computing. The first had to do with memory. People love to make comparisons between computers and the brain. But the brain is not at all like a digital computer. It has no processor, no software, no random-access memory. Your brain uses a type of memory called auto-associative memory. Auto-associative memory can generalize, fill in missing pieces of information, and work well with incomplete and even inaccurate data.

"Research in auto-associative memory has remained obscure to computer people because they haven't seen how to apply it. What I figured out was how to apply auto-associative memories to temporal data - data that varies over time. That's what the brain does. The 'pattern recognizer' used in Graffiti (the PalmPilot's handwriting-recognition software) is based on the mathematics used in auto-associative memory.

"My second insight involved understanding - that is, how we know something. 'Intelligent systems' don't just act; they anticipate. They make predictions about their environment. And auto-associative memories are good at helping us do that.

"My ultimate goal is to build a new industry around silicon-based, temporal, auto-associative memories. Products that incorporate these memories will understand the world much as you and I do."

Add New Comment

0 Comments