You are on page 1of 4

Educational Strategies

Problem Solving Concepts and Theories


Laura E. Hardin

ABSTRACT

Many educators, especially those involved in professional curricula, are interested in problem solving and in how to support students development into successful problem solvers. The following article serves as an overview of educational research on problem solving. Several concepts are defined and the transition from one theory to another is discussed. Educational theories describing problem solving in the context of behavioral, cognitive, and information-processing pedagogy are discussed. The final section of the article describes prior findings regarding expertnovice differences in problem solving of various kinds.
Any problem has at least three components: givens, goal, and operations. Givens are the facts or pieces of information presented to describe the problem. Goal is the desired end state of the problem. Operations are the actions to be performed in reaching the desired goal.1 Problems are categorized as ill defined or well defined, based on how problem and goal are represented. Problems with complex representations and/or more than one solution are termed ill defined. Problems with discrete representations and finite goals are termed well defined. The distinction between ill defined and well defined is a continuum, based on the complexity of the problem and what is required cognitively to solve it. Problem-solving knowledge is, conceptually, of two kinds. Declarative knowledge is knowing that something is the case. It is knowledge of facts, theories, events, and objects. Procedural knowledge is knowing how to do something. It includes motor skills, cognitive skills, and cognitive strategies. Both declarative and procedural knowledge are activated in working memory as problem solving occurs. Psychologists who distinguish between declarative and procedural knowledge believe that the two forms of knowledge are both distinct and interdependent. Declarative and procedural knowledge interact in a variety of ways during problem solving. A basic unit of declarative knowledge in the human information-processing system is the proposition. This expresses or proposes the relationships among concepts. For instance, the phrase, the man fixed the tire, depicts a complete idea. A proposition always contains two elements: a relation and one or more arguments.2 There are three attributes that are commonly used to differentiate expert from novice problem-solving characteristics. These attributes are conceptual understanding; basic, automated skills; and domain-specific strategies. The following paragraphs explain these attributes and how they are used to distinguish expert from novice problem-solving characteristics. Conceptual understanding refers to both the actual information in memory and the organization of that information in memory. Conceptual understanding is closely related to schema theory, in which information is considered to be stored in memory as frameworks or structures that, once instantiated, provide a lens through which to view new information. Having a conceptual understanding of a domain means that an individual can make meaning of domain-specific situations or problems, based on prior knowledge of that domain. Basic, automated skills in any domain are those that allow an individual to perform necessary and routine operations without much thought. These skills are overlearned to the point that they become habitual and even unconscious, enabling individuals to operate quickly and accurately without taxing their short-term memories. Automaticity allows individuals to focus their attention on the more complex tasks associated with a specific domain and is a general attribute associated with experts in a domain. Automaticity supports the experts speed and skill of execution. Unlike basic, automated skills, which occur unconsciously and thus do not tax short-term memory, domain-specific strategies remain under conscious control. They are the processes and procedures in a domain that an individual, even an expert, must consciously think about in order to solve a problem. They are, in other words, the procedural knowledge associated with a domain. Expertnovice differences have been studied and described within the context of these three attributes: Experts (1) exhibit better conceptual understanding of their domain; (2) use more automated skills and domain-specific strategies; and (3) have a conceptual understanding that is declarative, while basic skills and strategies are procedural.3

HISTORICAL PROGRESSION OF PROBLEM SOLVING THEORIES


As theories of how learning occurred evolved, the understanding of the problem-solving process also evolved. The following section will present the model of problem solving associated with each concurrent learning theory. The dominant learning theories discussed are within the conceptual domains of behaviorism, cognitive psychology, and information processing. Behaviorists view problem solving as a process that develops through positive and negative reinforcement mechanisms. Cognitive psychologists view problem solving as a process that includes introspection, observation, and the development of heuristics. The information-processing view of problem solving is based on general problem solving skills and artificial intelligence.

Behaviorist
Behaviorists understanding of learning was based on cause and effect. In this conceptualization, a behavior was followed by reinforcement. If the behavior was followed by positive reinforcement, then the behavior was more likely to
227

JVME 30(3) 2002 AAVMC

be repeated; if there was negative reinforcement, the behavior was less likely to be repeated. Two problem-solving methodologies explain the problemsolving process within the framework of behaviorist learning theory. One such method is trial and error. This involves attacking the problem by various methods until a solution is found. Young children solving a jigsaw puzzle exhibit this type of problem-solving behavior. The children try fitting different pieces into the same spot until eventually they find the piece that fits.4 Another method consistent with behaviorist learning is Hulls response hierarchy.5 This method involves learned responses that are applied to a situation in a hierarchical manner. The hierarchy is based on the response for which habit strength is strongest. Stimuli in a problem situation may evoke several different responses, and responses will be produced, one at a time, in order of strength, until either the problem is solved or the organism exhausts its repertoire of responses. In their emphasis on trial-and-error learning and habit strength, behaviorists focused on the role that stimulus response interactions might play on problem solving.2 These early conceptions of learning and problem solving described the observable characteristics of the process and did not seek to elaborate on the cognitive mechanisms of the subject.

the problem. These heuristic methods can be applied to a problem in any content domain; thus they are considered to be general problem-solving skills. In addition to the problem-solving processes already discussed, other heuristics have been identified. People often have to make decisions in the face of uncertainty, with sketchy information about the situation, on the basis of suggestive but inconclusive evidence. The reasoning processes used to resolve the uncertainty are often called judgment heuristics. One form of judgment heuristic is similarity judgment, where an instance is evaluated based on prior knowledge of a similar instance. A similar type of judgment is representativeness, where an assumption is made based on the belief that the characteristics of the individual are representative of the group. Another heuristic is the availability heuristic. In this case, judgments are made based on which elements can most easily be retrieved from memory. Analogical reasoning is another heuristic method, where judgment is made by drawing similarities to events that have occurred previously. Still another judgment heuristic is the development of a mental model (simulation) to predict the outcome of an event.7 These heuristics are examples of general-purpose thinking skills, which seem applicable to many domains. The heuristics approach emphasizes finding a good representation of the problem. While content-specific knowledge is required to solve the problem, the belief that general problem-solving skills were also valuable was supported by studies in the domains of math and computer science.8

Cognitive
As cognitive psychology progressed as a discipline, more interest and effort was directed toward the mental processes of learning and problem solving. An early cognitive approach to problem solving was to identify the mental stages through which problem solving proceeded. Two noted cognitive psychologists, Wallas and Polya, developed a four-stage model of problem solving. The four stages of problem solving identified by Wallas were (1) preparationdefining the problem and gathering information relevant to it; (2) incubationthinking about the problem at a subconscious level; (3) inspirationhaving a sudden insight into the solution of the problem; and (4) verificationchecking to be certain that the solution was correct.5 Similarly, Polya6 described the following four steps in the problem-solving process: (1) understand the problem, (2) devise a plan, (3) carry out the plan, and (4) look backward.

Information Processing
As elaboration on problem solving and expertnovice differences continued, the information-processing theory of learning emerged. This theory emphasizes the role of factors such as working memory capacity, organization of long-term memory, and cognitive retrieval of relevant information. The bulk of current research in problem solving reflects inquiry into the nature of these cognitive processes.5 Newells early work in artificial intelligence1 (AI) sustained the general problem solving theory. AI is the study and development of computer programs to solve problems. An AI program uses a finite set of functions to work from the problem state to the solution state. Many simple puzzles and problems in logic were successfully completed with AI programs, supporting the idea that problem-solving success was directly related to general problem-solving skill. Newell and colleagues proposed a theory of human problem solving that emphasized the similarities between AI and human problem solving. There are four underlying principles of this theory: (1) A few gross characteristics of the problem-solving process are invariant over the task and the problem solver, (2) the characteristics of the problem are sufficient to determine the problem space, (3) the structure of the task environment determines the possible structure of the problem space, and (4) the structure of the problem space determines the possible programs (methods) that can be used for problem solving.

Cognitive Heuristics
Polya6 promoted the idea that the application of general problem-solving strategies was key to problem-solving expertise and intellectual performance. General problemsolving strategies have also been called heuristics. The word heuristics comes from the Greek, heuriskin, meaning serving to discover. A commonly used synonym for heuristics is rule of thumb. In problem-solving literature, the term implies the general methods used in problem solving. The heuristics Polya6 identifies in mathematical problem solving are discussed within the framework of a four-stage problem-solving model as discussed earlier. Some of the heuristics applied within this plan include understanding the unknown, understanding the nature of the goal state, drawing a graph or diagram, thinking of structurally analogous problems, simplifying the problem, and generalizing

228

JVME 30(3) 2002 AAVMC

HISTORICAL PROGRESSION OF EXPERTNOVICE PROBLEM-SOLVING THEORIES


General problem solving knowledge was determined to be an incomplete explanation of how problem solving occurred. A close look at the expertnovice research reveals that the expert does need domain knowledge, basic automated skills, and domain-specific expertise to exhibit the characteristics of expert problem solving. Several studies reveal that expertise relies on both domain-specific knowledge and problem-solving skill. The following examples, from various domains, illustrate the body of research that has been conducted to reach this conclusion.

for student behavior. A study comparing expert and novice teachers also revealed that expert teachers thought about learning from the perspective of the student and performed a cognitive analysis of each learning task, which they adapted to students needs during teaching. Novice teachers used specific objectives to form lesson plans and did not adapt to student needs during teaching. 12 Barba and Rubbas13 research with 30 expert (in-service) and 30 novice (preservice) earth- and space-science teachers was conducted to study the cognitive differences between experts and novices. Data from audio-taped interviews revealed that expert teachers were more accurate, verbalized more declarative knowledge (facts and concepts), used fewer steps, and generated more subroutines than their novice counterparts. Experts generated more alternative solution paths and moved less between procedural (rules and strategies) and declarative knowledge than did the novices. As the authors noted, these findings support Norman and Gagnes theories of cognitive learningspecifically, that knowledge is hierarchically arranged in schemata. According to this theory, facts unite to form concepts, concepts join to form rules, and rules join to form problem-solving structures. These findings also support Normans (1982) theory that expert performance in procedural knowledge is characterized by smoothness, automaticity, and decreased mental effort, as compared to that of the novice.14

Chunking
A key area of expertnovice research has been the study of chess players. The game of chess is thought to require substantial skill and cognitive ability for successful play. Cognitive researchers began studying expert and novice chess players to develop understanding of the underlying cognitive processes. Research by De Groot, as cited by Chase and Simon,9 indicated that master chess players are better at reconstructing chess positions due to chunking information within a relational structure. Chase and Simon9 conducted further research to discover and evaluate the structures, or chunks, that are perceived by the chess player. Three chess players (a master, a Class A player, and a beginner) conducted memory tasks on positions derived from 20 published chess games. Participants were asked to view a game board layout, then reproduce, from memory, the layout on another board. Results of this study indicated that the information extracted from a position, briefly exposed to the viewer, varies with playing strength (the more expert player derives more information). The data also suggested that experienced players are able to encode the positions into larger chunks (than novices). The number of chunks in short-term memory appeared consistent across levels of playing expertise, thereby strengthening the belief that content-specific relationships are chunked in a cognitively accessible framework. Contrary to original speculation, studies describing the cognitive processes of experts and novices have indicated that experts (chess players) do not consider plays farther ahead than the novice; rather, experts choose among vastly superior (complex) moves. The expert is able to chunk relevant information, while novices envision single pieces of information.10 Researchers began to look beyond chess into other knowledge-rich domains. Studies have described the problemsolving process in physics, mathematics, computer programming, and medical diagnosis. Language skills, such as reading and writing, and the question how students used these skills to acquire more knowledge were also studied intensively.10, 11

Experience and Training


A study by Schoenfeld and Herrmann15 showed that domain-specific content is relevant to successful problem solving. Nineteen college students participated in this study, which evaluated the differences between experts and novices in mathematical problem perception. Experts determined the deep- and surface-structure properties of each problem prior to initiation of the study. All students were asked to sort the 32 math problems, based on which, if any, were mathematically similar (each of the similar problems would be solved in the same way). After completion of the first sort, subjects either participated in a course in mathematical problem-solving strategies that stressed a systematic, organized approach to solving problems or took a structured computer programming course that taught a structural, hierarchical, and orderly way to solve non-mathematical problems using the computer. Upon completion of the course, subjects sorted an equivalent set of mathematical problems. Those completing the mathematical problem-solving course were more accurate in identifying the appropriate deep and surface structures of the problems than they had been prior to the course, whereas those completing the programming course were not. This study showed that training within the problem domain affects problem solving more than training outside that domain.

Schema Theory
Differences between expert and novice teachers have also been studied. Research has shown that teachers, like experts in other fields, possess well-elaborated schemas that provided a framework for meaningful interpretation of information. Expert teachers also had an understanding of what to expect in the classroom and set up procedures and rules

Metacognition
Bruer summarizes the research on domain-specific problem solving as follows: Expertise, these studies suggest, relies on highly organized, domain-specific knowledge that can arise only after extensive experience and practice in the domain10

JVME 30(3) 2002 AAVMC

229

Studies in domain-specific problem-solving expertise also introduce the underlying principle of metacognition. Metacognition is the ability to think about thinking, the selfawareness of problem solving, and the ability to monitor and control ones mental processing. According to Bruer, Bransfords self-study of physics revealed several aspects of metacognition, which ultimately constitute an awareness and control of learning that can cross content domains.10 The ability to solve problems successfully depends on a number of factors related to the human information-processing (IP) system. This higher order learning theory elaborates the cognitive processes of problem solving. There are six attributes that define expertnovice differences in problem-solving skill within the IP framework, regardless of content domain. Through years of study of the cognitive processes applied during problem solving, six characteristics of expert performance have become widely accepted: Experts perceive large, meaningful patterns in their domain. The ability to see meaningful patterns reflects organization of the knowledge base. An expert has the knowledge required to solve a problem, and this knowledge is assessable in a way that does not tax working memory. Experts are faster and more accurate than novices at solving problems within their domain. There are two likely reasons for this phenomenon: Experts have developed the basic, automated skills applicable to the problem, and they have an organized database from which to retrieve the solution. Experts have superior short- and long-term memory, again based on superior memory organization, rather than volume. Experts see and represent data at a more conceptual (principled) level than novices. Experts spend more time analyzing and evaluating a problem quantitatively before beginning to solve the problem. Experts have strong self-monitoring skills. They are more aware when they make errors, why they fail to comprehend, and when they need to check their solutions.4 The preceding review of research in general problem-solving methods and domain-specific problem-solving characteristics has concluded with a summary of the characteristics of expert problem-solving. The discussion of both general and content-specific problem-solving attributes leads to the conclusion that both content knowledge and general problem-solving skill are necessary for expert problem solving to occur. Content-specific knowledge allows the expert to perceive information in a way that maximizes memory and information is conceptualized more on the level of principles. Superior problem analysis

and evaluation and strong self-monitoring skills can be recognized as general problem-solving expertise.

REFERENCES
1. Newell A, Simon H. Human Problem Solving. Englewood Cliffs, NJ: Prentice Hall, 1972. 2. Gagn ED, Yekovich CW, Yekovich FR. The Cognitive Psychology of School Learning. New York: Harper Collins, 1993. 3. Miller K. Differences between expert and novice writers. Unpublished, 1996. 4. Chi MH, Glaser R, Farr MJ. The Nature of Expertise. Hillsdale, NJ: Lawrence Erlbaum, 1988. 5. Ormrod JE. Human Learning. Upper Saddle River, NJ: Prentice Hall, 1987. 6. Polya G. Mathematics and Plausible Reasoning. Princeton, NJ: Princeton University Press, 1954. 7. Glass AL, Holyoak KJ. Reasoning and decision making. In Cognition. New York, NY: Newberry Awards Records, Inc, 1996:333363. 8. Nickerson RS, Perkins DN, Smith EE. The Teaching of Thinking. Hillsdale, NJ: Lawrence Erlbaum, 1985. 9. Chase WG, Simon HA. Perception in chess. Cognit Psychol 4:5581, 1973. 10. Bruer JT. The minds journey from novice to expert. Am Educator Summer:646, 1993 p15. 11. Perkins DN, Salomon G. Are cognitive skills context. Educ Res JanFeb:1625, 1989. 12. Westerman DA. Expert and novice teacher decision making. J Teach Educ 42:292305, 1991. 13. Barba RH, Rubba PA. A comparison of preservice and in-service earth and space science teachers general mental abilities, content knowledge, and problem-solving skills. J Res Sci Teach, 29:10211035, 1992. 14. Norman GR, Learning and Memory. San Francisco, CA: W.H. Freemen, 1992. 15. Schoenfeld AH, Herrmann DJ. Problem perception and knowledge structure in expert and novice mathematical problem solvers. J Exp Psychol Learn Mem Cogn 8:484494, 1982.

AUTHOR INFORMATION

Laura E. Hardin, DVM, MS, PhD, is an Assistant Professor of

Veterinary Basic Sciences, Mississippi State University , College of Veterinary Medicine. E-mail: SMTP:lhardin@cvm.msstate.edu

230

JVME 30(3) 2002 AAVMC

You might also like