Andrew is a theoretical neuroscientist who lives in Vancouver, Canada and is associated with the Australian National University in Canberra. In his first career he worked in industry on many different aspects of the design of extremely complex electronic real time control systems with many billions of components. He then shifted to working on applying the techniques used to manage the design information about such systems to organizing the vast amount of neuroscience knowledge in such a way that higher cognition can be understood.
Andrew is a theoretical neuroscientist who lives in Vancouver, Canada and is associated with the Australian National University in Canberra. In his first career he worked in industry on many different aspects of the design of extremely complex electronic real time control systems with many billions of components. He then shifted to working on applying the techniques used to manage the design information about such systems to organizing the vast amount of neuroscience knowledge in such a way that higher cognition can be understood.
A brain has no resemblance to an electronic system. In a computer, information processes are performed by transistors. In a brain, information processes are performed by neurons. The information processes performed by neurons are completely different from the information processes performed by transistors. The ways neurons are organized in the brain are completely different from the ways transistors are organized in a computer. Nevertheless, how we go about understanding complex electronic systems has some important lessons for how to go about understanding the brain. A computing system can have over 100 billion components like transistors. An engineer does not design a computer or a new feature by simultaneously imagining the activity of all the billions of transistors that could be involved. Rather, the design information is organized in such a way that all design tasks are within the mental bandwidth of a human designer. The techniques to achieve this involve hierarchies of description, carefully managed use of approximation, and the ubiquitous use of just two types of information process: instructions and data read/writes.
A brain can have of the order of 100 billion neurons. We will not be able to understand how the brain performs some cognitive process by simultaneously imagining the activity of all the billions of neurons that might be participating. However, natural selection pressures have resulted in brain architectures to which techniques analogous with those used in computer systems can be applied. These pressures have resulted in organization of brain resources into modular hierarchies and the ubiquitous presence of two types of information processes in the brain, condition definition/detections and behavioural recommendation definition/integrations. As a result, hierarchies of description can be created to support intuitively satisfying understanding of cognitive phenomena in terms of the activity of neurons.
A brain can have of the order of 100 billion neurons. We will not be able to understand how the brain performs some cognitive process by simultaneously imagining the activity of all the billions of neurons that might be participating. However, natural selection pressures have resulted in brain architectures to which techniques analogous with those used in computer systems can be applied. These pressures have resulted in organization of brain resources into modular hierarchies and the ubiquitous presence of two types of information processes in the brain, condition definition/detections and behavioural recommendation definition/integrations. As a result, hierarchies of description can be created to support intuitively satisfying understanding of cognitive phenomena in terms of the activity of neurons.