What is, or was, or yet again could be ‘Applied Neurodynamics’ ?
I was a student of the Mind via Philosophy and Psychology as an undergraduate, and I studied AI and Cognitive Science for a Masters. I was particularly interested in neural models of cognition. To apply my human factors and AI background I started my career in a faculty position leading what was then a cutting edge computer aided instruction laboratory (CAI, see also MOOCs for the modern web way ). It also gave me a chance to help change a lock-step education system that had frustrated me. After 4 years successfully building a CAI laboratory I became interested in the potential of microprocessors to radically individualize education. I went back to graduate school in San Diego briefly before entering industry to learn design engineering so I could build my own CAI workstation . After about 7 years and getting proficient at hardware design, Hopfield published his classic paper, and the connectionist PDP group at UCSD wrote their 2 volumes. That clicked with my undergraduate and graduate interest in neural modeling. Apples and PCs were now plentiful. So I did not need to design the CAI system any more. I used my hardware experience to warp my career over into hardware design for neurocomputers which were all the rage at the First International Joint Conference on Neural Networks in1987 in San Diego. I landed a job.
However, not long after I moved back to San Diego with a wife (x) and 4 kids, and just coming out of escrow on a new home purchase, I got laid off from that job where I was on a team that developed a neurocomputer. Jobs in my field were pretty scarce then. So I hung out a shingle and formed a consulting firm named to express my hunch about where neurocomputing was headed – neurodynamics. I had studied General Systems Theory and was reading about dynamics. While working in Dallas before coming here, we had a group there that met every month called the Metroplex Institute for Neural Dynamics (MIND). Hence, ‘Applied Neurodynamics’ or appliedneuro.com became the name. After various engineering contracts and some standards work for scalable neurocomputer communications, for the new IEEE Computational Intelligence Society, and for the IEEE SCI standard, I started getting contract work doing hardware designs for companies and universities designing ‘neurocomputers ‘ or ‘neuromorphic‘ chips. I was most interested in how ‘spike timing‘ can be represented in these systems. Schmoozing eventually led to some contract work with Oxford University and prominent neuroscientists there carrying the neuromorphic torch that was lit by ‘Carver Mead at CIT.’ We had independently developed similar approaches to represent spike events.
I was what you could call a ‘one man gang’ because I did everything turn-key. I did the circuit design, schematic entry, CPLD/PAL/GAL/FPGA HDL design, PCB placements (eventually also PCB layout), subcontractor management, PCB fab management, board soldering and all hand assembly, contract writing and NDA negotiation, licensing agreements, DARPA and SBIR proposals, purchasing, kitting, shipping & receiving, accounting, taxes, and not least marketing myself and what I could do. This was all quite challenging, but I made an adequate living at it for about 15 years. Don’t ever follow my example unless you have no choice or fat VC’s helping.
All along I wanted to get closer to the science since I was exposed to Hebb’s work, physiological and cognitive psychology (Neisser) at Michigan, and human information processing at Purdue. I knew there was much more to the story than back-propagation algorithms or measures of connections per second. In particular, timing was obviously important among spiking neurons. Even many neuroscientists were still discounting the importance of timing and how to model it accurately. I often argued with them. But a few of their own finally got the message across: Gray, Singer, Bi & Poo, and others finally demonstrated pretty much what Hebb was telling them in 1949 (according to my reading of Hebb in 1969). There are many skeptics to this day, but spike timing has finally gained some respect.
I often had a lot of time on my hands where I could take advantage of local seminars in neuroscience and even take some classes. I transferred credit for some graduate work at UC San Diego back to U of Michigan to finalize my undergraduate degree in the Psychology Honors program to complete the double major. Eventually I got a job at UCSD in this field and later had two more positions in the university, including this one (knock on wood). Applied Neuro has been on hold for about 10 year now, but it might need to be resurrected again to pay bills.