INTRODUCTION
This article discusses the principles of two qualitatively different and somewhat competing instructional designs from the 1950s and 1960s, linear programmed instruction and programmed branching. Our hope is that an understanding of these ideas could have a positive influence on current and future instructional designers who might adapt these techniques to new technologies and want to use these techniques effectively. Although these older ideas do still see occasional mention and study (e.g., Brosvic, Epstein, Cook, & Dihoff, 2005; Dihoff, Brosvic, & Epstein, & Cook, 2004), many contemporary instructional designers are probably unaware of the learning principles associated with these (cf., Fernald & Jordan, 1991; Kritch & Bostow, 1998; McDonald, Yanchar, & Osguthorpe, 2005).
BACKGROUND
An important difference between these instructional designs is associated with the use of feedback to the learner. Although we could provide a student with a score after completing an online multiple-choice quiz, applications that provide more immediate feedback about correctness upon completion of each individual question might be better. Alternatively, we could provide adaptive feedback in which the application provides elaboration based upon qualities of a particular answer choice.
Following is a discussion of two qualitatively different instructional designs, one providing immediate feedback regarding the correctness of a student's answer, the other providing adaptive feedback based on the qualities of the student's answer. Suitability of one design or the other is a function of the type of learner and of the learning outcomes that are desired.
SOME CLASSIC CONCEPTS OF INSTRUCTIONAL DESIGN AND OUTCOMES
Although the idea of non-human feedback would seem to imply a mechanical or electronic device, other methods could be used. Epstein and his colleagues, for example, have used a multiple-choice form with an opaque, waxy coating that covers the answer spaces in a series of studies (e.g., Epstein, Brosvic, Costner, Dihoff, & Lazarus, 2003); when the learner scratches the opaque coating to select an answer choice, the presence of a star (or not) immediately reveals the correctness of an answer. Examples of the designs discussed next are based on paper topics, but they are easily adaptable to technologies that use hyperlinks, drop-down menus, form buttons, and such.
Linear Programmed Instruction
The programmed psychology text topic of Holland and Skinner (1961) asked the student a question on one page (the following quote starts on page 2) and then asked the student to turn the page to find the answer and a new question:
A doctor taps your knee (patellar tendon) with a rubber hammer to test your_.
The student thinks (or writes) the answer and turns the page to find the correct answer ("reflexes") and is then asked another question.
Questions or statements are arranged in sequentially ordered frames such as the previous single frame. A frame is completed when the student provides a response to a stimulus and receives feedback. Skinner contended that this method caused learning through operant conditioning, provided through positive reinforcement for stimuli that are designed to elicit a correct answer (c.f., Cook, 1961; Skinner, 1954, 1958).
Skinner (and others who use his methods) referred to his method as programmed instruction, which incorporates at least the following principles (cf., Fernald & Jordan, 1991; Hedlund, 1967; Holland & Skinner, 1961; Skinner, 1958; Whitlock, 1967):
• Clear learning objectives.
• Small steps; frames of information repeat the cycle of stimulus-response-reinforcement.
• Logical ordered sequence of frames.
• Active responding by a student who works at his/her own pace.
• Immediate feedback to the response in each frame with positive reinforcement for correct answers.
A technique in programmed instruction is to help the student a great deal at first, and then gradually reduce the cues in latter frames; this is called fading (Fernald & Jordan, 1991; Reiff, 1980). If correct responding suggests that a student is learning at a quick rate, gating can be used to skip over frames that repeat prior information (Vargus & Vargus, 1991). The programmer is expected to use information about student performance to make revisions; if the student is not succeeding, then it is due to a fault of the program, not to an inability of the student (Holland & Skinner, 1961; Vargus & Vargus, 1991).
Programmed Branching
Crowder (e.g., 1959, 1963) and others (e.g., Pressey, 1963) were critical of Skinner's approach, arguing that students not only learn from knowing a correct answer, but also learn by making mistakes. Crowder distinguished between his automatic tutoring device and the Skinner-type teaching machine, proposing that the automatic tutoring device is more flexible in allowing the student to receive an explanation when an error is made.
In this programmed branching method of Crowder, the student is taken to one of several possible discussions depending on the qualities of the answer.
While Skinner's design would be expected to work only when stimuli elicit correct answers, Crowder's design allows for mistakes and must be designed to anticipate particular mistakes. Crowder believed that this method caused learning through cognitive reasoning. Whatever answer is chosen by the student, the programmed text topic (or machine) makes a branch to a discussion associated with issues relevant to the answer that was chosen. This is followed by a return to the same question if the student had made an incorrect choice, or a jump to new a frame containing the next question if the student had made a correct choice.
Learning Outcomes
Many issues have been raised over the years about programmed instruction methods. Reiff (1980) discussed several criticisms:
• It does not take into consideration the sequence of development and readiness to learn (e.g., children of different ages or children vs. adults).
• It develops rote learning skills rather than critical thinking skills.
• Students can in some implementations cheat.
• The encouragement to respond quickly could develop bad reading habits.
Crowder's programmed branching design, which has received far less attention and study than Skinner's ideas, would seem to answer at least some of these criticisms. Crowder's design provides an explanation to both correct and incorrect answers, so the learner is not rewarded for cheating or working too quickly. Since the explanation is tied to the learner's thinking at the time a choice was made, Crowder's design would appear to be better to develop critical thinking skills, but might not be so good at developing rote learning skills. Crowder's design would appear to be better suited to students who have a greater readiness to learn, while perhaps not so well suited to a student who is at an earlier stage of learning a subject.
The previous discussion suggests that each of these designs is useful, but that each is useful in different kinds of situations and that the learning outcomes of each approach might be different. Skinner's teaching machine, for example, might be more useful in situations where students are learning lists and definitions. The automatic tutoring device, on the other hand, might be more useful when the student is already at a higher level of understanding whereby s/he can now use reasoning to derive an answer, or in situations where the student understands that there are degrees of right and wrong without concrete answers. The Skinner-type teaching machine might be better suited to "lower-order" levels of learning, while the Crowder-type automatic tutoring device might be better suited to "higher-order" levels of learning.
Although many ideas have been proposed with regard to a hierarchical perspective on "lowef' and "higher" levels of learning, the most well-known, "Bloom's Taxonomy" (A Committee of College and University Examiners, 1956), originated in about the same timeframe as the ideas of Skinner and Crowder. "Bloom's Taxonomy" proposes that the objectives of learning lie on a hierarchical continuum: (1) knowledge of terminology and facts, (2) comprehension of translation and paraphrasing, (3) application, (4) analysis, (5) synthesis, and (6) evaluation.
"Bloom's Taxonomy" is actually only Part I of a two-part work. The previously mentioned first part is known as the cognitive domain. Part II (Krathwohl, Bloom, & Masia, 1964) focuses on the affective domain: (1) willingness to receive ideas, (2) commitment to a subject or idea, (3) feeling that an idea has worth, (4) seeing interrelationships among multiple ideas, and (5) the integration of ideas as one's own.
future trends
Fernald and Jordan (1991) discussed several reasons as to why programmed instruction might have fallen out of use since the decades of the 1950s and 1960s:
• It was seen to dehumanize the teaching process.
• Educators feared that it might be too effective and threaten their jobs.
• The importance of the learning principles was not understood.
• Applications were often not effectively designed.
Technology, economics, and attitudes have since changed. As economics and student demand push us to use distance education methods, the first two arguments would seem to become more diminished in the future.
It is hoped that this article assists in diminishing the latter two arguments by introducing instructional designers to the principles discussed in this article and by encouraging instructional designers to create more effective designs with regard to appropriateness for a particular student audience and with regard to the type and level of learning outcomes that are desired. By better understanding the past, we can better affect the future.
Curiously, there has been less attention devoted to Crowder's ideas of adaptive feedback than to Skinner's ideas of immediate feedback and reinforcement. We continue see occasional research devoted to related issues, such as issues of immediate vs. delayed feedback (e.g., Brosvic et al., 2005; Dihoff et al., 2004; Kelly & Crosbie, 1997) or of allowing students to keep selecting answers from a multiple-choice set until the correct answer is finally discovered (Epstein et al., 2003). However, we still can only speculate with regard to conditions under which a Skinner-style of instructional design would be better and when a Crowder-style of design would be better. It is hoped that this article generates greater awareness of and use of these designs in new technologies, but also that greater interest in these ideas will stimulate more research into the learning mechanisms associated with them.
conclusions
New technologies such as Web browsers now make it relatively easy for educators with the most modest of skills to present instructional frames in a linear sequential ordering or as branches that are dependent on the student's selection of answers from a list. In adapting some of these older ideas to newer technologies, we hope that instructional designers will be better equipped to select appropriate methods by considering:
• the student's level of readiness for learning
• the basis for learning when different instructional designs are used
• the qualitatively different kinds of learning outcomes that are possible with different instructional designs
KEY TERMS
Adaptive Feedback: Immediate feedback in the form of an explanation or discussion that is tailored to the qualities of the student's answer.
Automatic Tutoring Device: A device that uses programmed branching and adaptive feedback. Learning results from cognitive reasoning.
Cognitive Reasoning: Learning through the process of thinking about an issue; the student learns new ideas and relationships by relating an issue to previously learned material.
Frame: A small piece of information or a statement to which the student is exposed, such as a page with a single question. In linear programmed instruction, a frame includes a stimulus, a response, and reinforcement (positive feedback).
Hierarchy of Learning: The concept that learning can be sequentially ordered along a continuum from lower-order to higher-order. "Bloom's Taxonomy" is one of many that have been proposed.
Linear Programmed Instruction: A design whereby a series of frames are presented to the student in a specific sequential order. The student actively responds to stimuli in each frame and receives immediate feedback to that response. Learning results through operant conditioning.
Operant Conditioning: Learning through immediate positive feedback (reinforcement) regarding the correctness of an answer; the student learns to respond in a particular way to a particular question or issue (stimulus). Fading can be used by gradually reducing stimulus cues in subsequent frames when material is repeated.
Programmed Branching: A method whereby the student is taken to one of several possible explanations or discussions depending on the qualities of an answer that is given to a question. Gating is a simple skip of frames that repeat prior information when a student's answers suggest that the material has been adequately learned.
Teaching Machine: A device that uses linear programmed instruction whereby frames present a question followed by feedback of the correct answer. Learning results from reinforcement of the student's correct answer.