Introduction
These pages describe an interactive music/media project that I am in the process of producing in collaboration with Eric Schultz of Tempe, Arizona.
The timeline for this project is no less than five years, so these pages will develop slowly as materials are gathered. I should also say that as time passes and other distractions rear their heads the estimated time of completion of this project gets a little further away. Someday I'll have the strength and time to continue this work. The project began in 2001 within conversations between myself and my close friend and colleague, Eric Schultz. I should explain that each of us have our own reasons for being involved in this project, and that we work out our ideas slowly through long-distance conversations.
The words "Character and Nature" serve as the working title for the piece in my mind and I write those words as the title of this page to remind myself why I'm doing this. Eric may have other words and other reasons in his mind. Furthermore, those words may or may not serve a purpose in the finished project.
Sketches: Terms and Concepts
- Behaviouristic Psychology
- Behaviourism
A school of psychology that confines itself to the study of observable and quantifiable aspects of behavior and excludes subjective phenomena, such as emotions or motives.
An approach to psychology that emphasizes observable measurable behavior.
Behaviorism is a theory of animal and human learning that only focuses on objectively observable behaviors and discounts mental activities. Behavior theorists define learning as nothing more than the acquisition of new behavior.
- Lens
A device that causes radiation other than light to converge or diverge by an action analogous to that of an optical lens.
- Perceive
To achieve understanding of; apprehend.
To become aware of directly through any of the senses.
- Perception
The process, act, or faculty of perceiving.
Recognition and interpretation of sensory stimuli based chiefly on memory.
The act of perceiving; cognizance by the senses or intellect; apprehension by the bodily organs, or by the mind, of what is presented to them.
The faculty whereby, through the senses, we obtain a knowledge of the external world.
- Action
The state or process of acting or doing.
Organized activity to accomplish an objective.
A process or condition of acting or moving, as opposed to rest; the doing of something.
- Reaction
Any action in resisting other action or force; counter tendency; reverse action.
Reaction is always equal and opposite to action, that is to say, the actions of two bodies upon each other are always equal and in opposite directions. --Sir Isaac Newton (3rd Law of Motion).
Response to action.
- Interaction
Mutual or reciprocal action or influence.
- Interact
Act together or towards others or with others.
- Sentient
Having a faculty, or faculties, of sensation and perception.
Endowed with feeling and unstructured consciousness.
- Feedback
The return of information about the result of a process or activity; an evaluative response.
The process by which a system is modulated, controlled, or changed by the product, output, or response it produces.
- Determinism
The philosophical doctrine that every state of affairs, including every human event, act, and decision is the inevitable consequence of antecedent states of affairs.
A theory which states that all events without exception are effects - events necessitated by earlier events. Hence any event of any kind is an effect of a prior series of effects, a causal chain with every link solid. Profound "action" is impossible in this theory.
- Indeterminacy
The state or quality of being indeterminate.
- The Uncertainty Principle
A principle in quantum mechanics holding that increasing the accuracy of measurement of one observable quantity increases the uncertainty with which another conjugate quantity may be known. (i.e. Determinism must negotiate the element of indeterminacy - for not all properties can be determined with absolute precision.)
Neural Networks
From an article by Peter Lafferty in QPB Science Encyclopedia, Quality Paperback Book Club: New York, 1999. p. 518.
Neural networks- strictly artificial neural networks- represent a radically different approach to computing. They are called neural networks because they are loosely modeled on the networks of neurons-neurons-that make up brains. Neural networks are characterized by their ability to learn, and can be described as trainable pattern recognizers. The study and use of neural networks is sometimes called neurocomputing.
Brains perform remarkable computational feats: recognizing music from just a few seconds of a recording, or faces seen only once before-accomplishments that defeat even the most modern computers. Yet brains stumble with arithmetic and make errors with simple logic. The reason for these anomalies might be found in the differences between brain and computer architecture- their internal structure and operating mechanisms.
Conventional computers possess distinct processing and memory units, controlled by programs, but animal nervous systems and neural networks are instead made up of hightly interconnected webs of simple processing units. They have no specific memory locations, information instead being stored as patterns of interconnections between the processing units. Neural networks are not programmed, but are trained by example. They can, therefore, learn things that cannot easily be stated in programs, making them invaluable in a wide range of applications.
Although neurocomputing might seem a recent development, research started at around the same time as the early work on computers. In the 1940s scientists devised simple electrical networks, crudely modeled neural circuits, which could perform simple logical computations.
More sophisticated networks, called perceptrons, followed in the 1950's- the ancestors of modern neural networks. Perceptrons were just simple networks of amplifiers, but they could learn to recognize patterns. This generated tremendous excitement; however, significant limits to their abilities were soon discovered. Marvin Minsky and Seymour Papert of the Massachusetts Institute of Technology, USA, proved that certain problems could never be solved by perceptrons. When they published their results, research into neural networks effectively ceased for over a decade. At the end of the mid 1970s, however, theoretical breakthroughs made it possible for more complex neural networks to be developed, and by the mid 1980s they had become sufficiently sophisticated for general applications.
Neural networks are of interest to computer technologists because they have the potential to offer solutions to a range of problems that have proved difficult to solve using conventional computing approaches. These problems include pattern recognition, machine learning, time-series forecasting, machine vision and robot control. Underpinning all this is their ability to learn.
In a famous example, a neural network was trained to recognize speech- with eerily realistic results. The network, called NET- talk, was developed by T.J. Sejnowski and C.R. Rosenberg at Johns Hopkins University in the USA. It was linked to a computer that could produce synthetic speech, so its progress could be heard. After producing formless noise for a few hours, it started babbling like a baby. Overnight training improved its performance so that it could read text with a 95% accuracy. No conventionally programmed computer could do this.