Robert J. Seidel

An Historical Perspective and a Model for Evaluation of Intelligent Tutoring Systems.

The purposes of this article are to 1) examine the changes which have occurred in the development and evaluation of ITS systems for the last twenty years, 2) speculate on future directions, and 3) propose a conceptual model to evaluate and institutionalize this technological innovation into training and educational settings. First, we review theoretical and technical dimensions that could be considered relevant to internal evaluation of ITS technology. Secondly, we discuss evaluation dimensions which, while external to ITS per se, are relevant for situational evaluation of any technological innovation. After theoretical and methodological discussions of intelligent features and effectiveness of ITS, its research and development activities are examined from a historical perspective. The development of ITS is divided into three stages and the development focus for each stage is discussed to show the shifts. It is observed that theoretical and methodological criticisms of ITS, needs unique to practical applications, and development of computer technology have influenced the shifts of ITS development focuses: purpose, staffing, hardware and software technology used, and contributions. In stage III, most recent development focuses and future prospects are discussed, including investigation of specific instructional strategies, creation of flexible environments for instructional strategy research, modeling of the human tutor, simulation of human learning and cognition, and creation of multimedia environments. After discussing problems associated with moving ITS from the laboratory to an institutionalized part of the practical world, a conceptual model for evaluating technological innovation in training and educational environments is proposed. This model considers the technology purpose, processes, outcome measures, and the maturity of the technological innovation (the latter especially relevant for ITS). It has three major stages: adoption, implementation, and institutionalization and two major processes: assimilation and accommodation of the innovation. These stages and processes occur within a context of multiple levels of users/stakeholders or decision-makers. Taken together they make up the dimensions of the evaluation model for any technological innovation, including ITS.

A multidisciplinary model for development of intelligent computer-assisted instruction

Development of an intelligent computer-assisted instructional (ICAI) system requires a multidisciplinary team effort. A schematic model is proposed to help ICAI developers identify the types of required expertise and integrate them into a system. Three types of expertise are discussed in the model: (a) domain expertise, (b) domain engineering expertise, (c) and instructional expertise. Domain expertise is classified into two types: conceptual knowledge and performance (procedural) expertise. Specific knowledge representation methods are proposed to represent the two types of domain—semantic network for conceptual knowledge and production rules for procedural knowledge—and frames-scripts for the combined domain of the two types. For domain engineering expertise, three technical methods are discussed: knowledge acquisition, task analysis, and knowledge representation. For instructional expertise, conceptual issues and technical methods to model student learning and to select instructional strategies are discussed. A multidisciplinary cooperative effort is reemphasized in the conclusion.

COMPUTER-ADMINISTERED INSTRUCTION VERSUS TRADITIONALLY ADMINISTERED INSTRUCTION: ECONOMICS,

Abstract : An attempt is made to assay the economics of computer-administered instruction (CAI) versus traditionally administered instruction (TAI) in controlling the structure of the learners stimulus environment in teaching and training situations. There is a discussion of the need for a sound, objective economic appraisal of the value to society as a whole of increments in the breadth and depth of education in the population, and of the influence of varying rates with which these increments are brought about. The necessity for reliable, objective information concerning cost data is emphasized. Projected cost/effectiveness comparisons based on the assumption of equal effectiveness for CAI and TAI are discussed for both civilian and military instruction. (Author)

PROGRAMMED LEARNING: PROLOGUE TO INSTRUCTION

The paper indicates some pertinent issues in the field of programmed instruction (PI) and suggests promising directions for future growth of PI, both as a medium for the application of principles of learning and as a means of furthering our understanding of learning processes. Practical and theoretical implications are touched upon and combined to give a position statement on PI as a pedagogical and psychological research tool. In this vein the utility and inevitability of computer-aided instruction are discussed.

Evaluation of CBT Authoring Systems: Need a Data Base

A computer-based training (CBT) authoring system is a special kind of software tool-kit for training developers with no programming skills and limited instructional design experience to use in developing CBT materials (i.e.,courseware). Many authoring systems have taken into account the needs of computer-naive users so that the learning requirements for using the systems is supposed to be minimal. However, the selection of the best authoring system for a given purpose requires a systematic evaluation because many systems have been developed to serve basically the same purpose with different procedural approaches, functional features and power. The purposes of this paper are: (a) to discuss evaluation methods for selecting an authoring system; (b) to examine important evaluation criteria; (c) to propose a set of evaluation procedures; (d) to discuss the need for developing a CBT authoring tool data base and provide initial inputs for the development of the data base.

Informational context as a determinant of what can be learned

Abstract The thesis put forth is that the question of ‘what is learned’ does not fall into the category of scientifically specious problems as Kendlep asserted earlier (1952). Rather, it can be properly addressed if the question is given a unique operational meaning, i.e., by being placed in a proper paradigm to elicit explicit answers. Two experiments in a novel, free-field paradigm are presented to illustrate the argument. Discussion centers on a need for a change in focus in the area of learning from what the organism learns (i.e., response learning or specific connectionism) to how the organism manages to achieve a performance capability. An informational view is put forth to analyze the organisms environment in terms of information available, information assimilated and information used to solve the problem of the moment. The free-field paradigm is proposed as but one of many yet untapped paradigms for studying the problem further.

ROLE OF REPETITION AND REINFORCEMENT IN LEARNING

Ten paired-associates were used with CVC syllables as stimuli and single-digits as responses in two experiments. Exp. I consisted of five reinforcement (R) treatments prior to TT. Exposure conditions were 4 sec. for S-R (or S) and 4 sec. blank between exposures. Exp. II consisred of R1TT and R2TT with the blank inrra-trial interval omitted. Exp. I yielded a linear increasing function of N1:C2 switching with increasing R. Exp. II, however, showed no increase in N1:C2 switching although learning and retention were comparable to Exp. I. Caution is advised in prematurely using the data from the RTT paradigm to support a given theory of associative learning.