In lieu of an abstract, here is a brief excerpt of the content:

A DEVELOPMENTAL MODEL OF INFORMATION PROCESSING IN THE CHILD* MICHAEL A. ARBIB\ and ROY M. KAHN\ I. Introduction In this paper we shall study mental development and disorder in a cybernetic framework, wherein we make continual use ofanalogies between information processing in the human brain and the processes ofcommunication , computation, and control in machines such as computers. The key use ofcomputer analogies is to develop a language for thinking about complex information-processing systems so that the psychologist or biologist will have a better chance of precisely systematizing his results. We are not trying to reduce brains to machines; rather we try to make explicit the extent to which our present notion ofmachine function falls short ofbrain function, so as to improve our models and help machines become more "brain-like." We do not think this will reduce men to automata —rather, as we understand the underlying hardware (i.e., circuitry) better, we will gain a higher appreciation ofthe individual's "software" or "program," that is, the unique blend of experience and learning which further shapes his own personality.1 We learn to answer questions about the world without always experimenting to find the answer. It is as ifour brains contained a "submachine" * Research was supported in part by the U.S. Air Force Office of Scientific Research (AFAFOSR -1198-67). We are also grateful to Dn. Bill Kilmer and Bernard Rimland for valuable discussions ofinfantile autism. t Department of Electrical Engineering, Stanford University, Stanford, California 94305. % Department of Mental Health, Boston, Massachusetts, and Gaebler Children's Unit, Metropolitan State Hospital, Waltham, Massachusetts. Present address: Counseling Center, University of California, Berkeley. 1 For an excellent discussion of the philosophical questions involved here and many important ideas about cybernetics, see Craik [1, 2]. Craik was perhaps the first to construct a cybernetic psychology . His papers were written in the early 1940's and contain a wealth ofideas, many ofwhich have even now not been thoroughly studied. 397 whose output (representing a correct answer) as well as input (representing the question) must be coded descriptions of the corresponding external events. In explaining our actions, then, a crucial role will be played by the notion that our braincontains a model ofthe world [i, 2, 3, 4]. Similarly , the crucialproblem ofrobot designers is that offinding a description of the environment in a form which may be stored in the robot, updated as the robot explores its environment, and which the robot can readily manipulate to guide its actions [5]. In computer programming there is a hierarchy which runs from logic elements and microinstructions to machine language, mnemonic languages , and assembly languages. A single instruction in the assembly language may control a complicated sequence of actions of switching elements. The complexity of a program can increase, without its being more complex to write, by increasing the sophistication ofthe assembly language. As we learn more ofassociative and distributed memory systems , new program and input-output systems, timesharing, multiprocessor systems, cellular computers, and of the design techniques of synchronous versus asynchronous machines, so we shall build better brain models. However, we must not expect our models to conform too closely to the structure ofpresent-day programs. For instance, human language has two aspects—language with intent, in which I try to make you understand what I mean, and gesture and emphasis, which modify the message and convey information unwittingly (cf. the distinction between digital and analogical communication in [6]). This distinction is not readily programmable in computer languages available today. One ofthe most intriguing problems in neurocybernetics, and that on which we focus here, is that ofunderstanding the development ofintelligence in children. A computer does not, as commonly considered, have an inherent developmental sequence ofits own which will proceed essentially independently ofthe environment—whereas a reasonably nourished child has a nervous system maturing physically whether or not his environment provides the stimulation required for proper mental development , which development is limited by the stage ofphysical development of the substrate. This stresses that certain new factors, not usually found in the cybernetic model, do exist in people. How does one deal with a machine which is evolving itselfand not solely dependent on the outside world for programming? Studies...

pdf

Share