Information processing and problem solving

Psy 416: Reasoning and Problem Solving

Information processing and computer simulation Sources on some of these issues are accessible from the references
E. Segal

    People function in the world. In order to do so, we must gain knowledge about things and events and be able to act in the world. Much of the time we act appropriately to the people, objects, and events we encounter. A major question for Cognitive Psychologists is: How do we function so well?
    One general answer, one that has incredible power, is that humans and many other organisms, are Information Processing Systems. Many, but not all, of our interactions with the world is via information.
   Information Processing (IP) is the dominant perspective held by cognitive psychologists (and many others in the Information Sciences) for the last 30 years or so. This perspective holds that cognition means the input, storage, transduction and transmission of information, and that the problems that cognitive psychologists face are all connected to Information Processing. Questions asked by Information Processing Psychologists almost all relate to Information: What information do we respond to? How is it stored? What are its properties? How is it accessed? How is it transduced? What conditions enhance IP? How is it used? How is information represented in the organism? etc.

1. What is Information?

    Very few Information Processing sources actually define their core concept. Some researchers use the term informally and its meaning varies from one use to another. I find that many who use the term are not clear about what it means. I think that Information is probably best defined as a pattern that "rides" on matter or energy. Information has the property that the same pattern can ride on different kinds of matter or energy. In information sciences, patterns and structures are the primary focus of study. Norbert Wiener, an early champion of cybernetics , argued that the concept of information changed the view of causality in science. One entity can cause a change in another with only an infinitesimal transference of energy. The causer or controller does it with a signal rather than a push.
    Information is a key to understanding much of the modern world including communication and computation. Modern technological devices from telephones, radios, computers, and the internet have developed in great part because of informational analysis.
    Although any pattern that rides on matter or energy may be considered information, for many modern analyses, the patterns are set up to be, or analyzed as, concatenations of symbols. For example, this page consists of information represented by the concatenation of letters to form English words, sentences, and paragraphs. When the symbol system being contains a small finite number of symbols which are concatenated to make larger structures, the system is a discrete or digital one.
    The cognitive and informational sciences owe their very existence to the study of information storage, transmission, and transduction. You have heard of flow charts. In psychology, communication and computer science, flow charts chart the flow of information. More on information

2. Information processing systems: Simon and Newell
An analysis follows what happens from the beginning of a task, such as being given a problem to solve to the end with the problem solved. The basic theory is that much of the sequence of events can be thought of as the movement, storage and transformation of information.

Major components

receptors --senses
processors--transform, interpret, integrate, select--attention, set, automatic and controlled processes.
memories--long term, short term, working, STSS.
effectors--muscles, glands

    Information enters the system via the receptors and then is transformed and operated on by the processors, some intervening outputs are temporarily stored and others are more permanently stored in memory, outputs are generated which lead to behavior and interaction with the environment. Historically, information processing psychologists have used flow charts to identify the path of the information through the cognitive processing system.
    Let us attempt to concretize the idea of information and information processing. An IPS has several components including receptors, memories, processors, and effectors (Newell & Simon, 1972). Such systems receive information from the environment through their receptors. They then go through processes of transforming, storing, comparing and evaluating this information. For example, assume that you see a duck. What happens informationally? In order for you to know that you see a duck, or to be aware that it is a duck you see, you have to compare part of the visual input with some representation of a duck in memory. Processors must parse the visual stimulus into meaningful components in order to isolate the duck from its visual context, and to compare the resultant duck information to a memorial representation. The representation must include not only information concerning the visual appearance of a duck, but also information identifying the visual information to be that of a duck. In order to behave
appropriately, therefore intelligently, such as to say "Oh, there's a duck." there has to be a link between your representation of the visual appearance and a representation of the verbal form "duck." In addition, this information has to tie to devices which control the effectors in your vocal apparatus. 3. Computation theory The concept of 'effective procedure' is one of the more important concepts in the symbolic sciences, and one which is needed at least on an informal basis in order to work within any cognitive science. "An effective procedure is a finite, unambiguous description of a finite set of operations. The operations must be effective in the sense that there is a strictly mechanical procedure for completing them" Algorithm: An effective procedure which is guaranteed to solve a problem.
Universal Turing Machine : An particular abstract information processing system consisting of a linear tape, a read head, and a finite set of states. It can read the tape for either of two symbols, it can write either of the two symbols, it can move one unit to either the left or right, and it can switch from one state to another. That is all. Correctly programmed a Turing machine can solve any problem for which one can specify an algorithm. There are, however, unsolvable problems (Godel proved that)
    There are other systems of effective procedures, Church's Lambda calculus, and Emil Post's production systems to name two, which using a different architecture, can solve any solvable problem.
    Two critical ideas that are necessary to have computational systems serve as the basis of complex problem solving are rules of choice and recursive rules.
Rules of choice are rules in which the data are evaluated to decide what to do next.
Recursion rules are rules in which the output of the rule may be used as an input to the same rule either immediately or after other operations have occurred.
Church-Turing Thesis : Any problem solving device which uses an effective set of operations to solve certain arithmetic problems is logically equivalent to any other system which can solve such problems. Any such device contains all the procedures necessary to solve any solvable well-specified problem.
Strong AI extension of Church-Turing Thesis: Humans can solve these arithmetic problems. Therefore their mechanisms are computationally equivalent to those of machines that solve them. Thus any problems humans can solve, are solvable by computational machines. 4. Physical Symbol Systems
Newell and Simon's thesis: All intelligent systems are Information Processing Systems whose implementations are Physical Symbol Systems.
    Humans and computers solve problems in basically the same way. These are identified by Newell and Simon as real systems that can do intelligent things such as reasoning, problem solving, text comprehension, planning, etc. Newell (1981) identifies a hierarchy of at least five descriptive levels within all PSSs: (a) the device level, (b) the circuit level, (c) the logic level (d) the program level, and (e) the PMS (Processor, Memory, Switch) level. Each of the levels has its own principles and characteristics which are only partially constrained at the other levels. (a) The device level identifies the set of physical units which must be duplicated and interconnected for a PSS. In a computer this used to be tubes and wires, now it tends to consist of semiconductive impurities on silicon chips. In organisms it consists primarily of neurons and synapses.
(b) The circuit level consists of the flow of matter or energy with particular voltages and resistances, or potentials and neurotransmitters. In a PSS something has to move through the system.
(c) The logic level refers to structural and functional patterns. Registers being on or off, the passing of bits according to patterns of their combination, for example some units may turn on only if all connecting units are on (AND gate), or a unit may turn on only if only one of several connecting units is on (XOR (exclusive or) gate).
(d) The program level contains data structures, symbols, addresses and programs. Symbols (structured patterns) are stored in accessible locations, and there are programs to retrieve information (identify and possibly duplicate subpatterns) and operate on it according to some principle. The result of that operation may be the addition of new data to the data structure or some external output, or both.
(e) The PMS level is the functional level at which intentions, plans, and purposes are realized. "Here there is simply a medium, called data or information, which flows along channels called links and switches and is held and processed by units called memories, processors, controls, and transducers." 5. Analysis of problem solving from an Information Processing perspective: Newell and Simon's analysis 1) Identifying the problem space. The first stage of an analysis of a problem is to identify the initial and goal states (Newell & Simon, 1972). These two states define the boundary of the problem space. The larger the "distance" between the two states the larger the problem space.
2) Identifying some of the intermediate states between the initial and goal state. Only for trivial problems can the solver go directly from the initial state to the goal state. There are usually going to be relatively stable describable intermediate states which need to be reached. Both the problem solver and the analyst may need to know of these.
3) Identifying what needs to be done; the "moves," which enable the problem solver to get from one state to another. In order for a problem to be solved there has to be some procedure by which the situation is transformed from one state to another.
4) Identifying the resources, e.g., knowledge, skills, materiel, personnel and time, needed to execute each of the moves. What is needed in order to reach each of the states from the immediately previous state? David Marr's approach: There are three stages of analysis. 1) Computation . The problem solver must analyze the task that needs to be done rather carefully. This requires an analysis of the specific parts. What are the inputs to the problem? What are the relations between the parts?
2) Algorithm (and representation). The second task for the problem solver is to specify an effective procedure that one can carry out in order to achieve the goal of the task. This requires a specific characterization of the sequence of operations that operate on a given data base; if the sequence is followed, it will lead to a solution of the problem.
3) Implementation . This requires identifying a set of physical objects which can carry out the algorithm automatically. Strategies : Trial and error, Hill climbing, Means-ends analysis, subgoals, goal stack, forward chaining, structural analysis. Content: Some people can use logical forms; some evaluate with the content, probably trying to understand the picture from a realistic model; the beliefs of a reader and her emotions often interfere with logical analyses. Models often have dynamic and causal relations as well as structural or logical ones. The pragmatics of the situation often interferes with the analysis. The topics of the premises and the feelings about the terms used, e.g. large and small, better and worse, are treated differently. Attempts to transform the presented information to one that is easier for the person to deal with may not be done correctly. We act rational to the extent that we do things that seem to have worked before; we use our knowledge base and expectancies. References Copeland, B. J. (1996). The Church-Turing Thesis. Stanford Encyclopedia of Philosophy.
         http://plato.stanford.edu/entries/church-turing/
Graham, Neill (1979). Introduction to Computer Science . St. Paul, MN: West.
         Chapter 1 on reserve
Marr, D. (1982). Vision. San Francisco: W. H. Freeman.
Newell, A. and Simon, H. A. (1972). Human Problem Solving. Englewood Cliffs, NJ: Prentice Hall.
Penrose, R. (1989). The Emperor's New Mind. New York: Oxford University Press.
Segal, E. M. (1994) Archaeology and Cognitive Science. In C. Renfrew and E. Zubrow (Eds.) The Ancient Mind: Elements of Cognitive Archaeology, Cambridge: Cambridge University Press. Click here for version of this paper.
Shannon, C. E. and Weaver, W. (1949). The Mathematical Theory of Communication. Urbana, IL: University of Illinois Press.
Simon, H. A. (1969) The Sciences of the Artificial. Cambridge MA: MIT Press.
Stanford Encyclopedia. Turing machine
Wiener, N. (1948). Cybernetics. New York: Wiley Psy 416 Syllabus