2.2 The Basics of Behaviorism

Behaviorism in the United States may be traced to the work of E.B. Twitmeyer (1902), a graduate student at the University of Pennsylvania, and E.L. Thorndike (1898). Twitmeyer's doctoral dissertation research on the knee-jerk (patellar) reflex involved alerting his subjects with a bell that a hammer was about to strike their patellar tendon. As has been the case so many times in the history of the development of behavioral theory (see, for example, Skinner, 1956), something went wrong. Twitmeyer sounded the bell but the hammer did not trip. The subject, however, made a knee-jerk response in anticipation of the hammer drop. Twitmeyer redesigned his experiment to study this phenomenon and presented his findings at the annual meeting of the American Psychological Association in 1904. His paper, however, was greeted with runaway apathy and it fell to Ivan Pavlov (1849-1936) to become the "Father of Classical Conditioning." Interestingly enough, Pavlov also began his line of research based on a casual or accidental observation. A Nobel Prize winner for his work in digestion, Pavlov noted that his subjects (dogs) seemed to begin salivating to the sights and sounds of feeding. He too altered the thrust of his research to investigate his serendipitous observations more thoroughly.

Operant or instrumental conditioning is usually associated (particularly by educators) with B.F. Skinner. Yet, in 1898, E.L. Thorndike published a monograph on animal intelligence which made use of a "puzzle box" (a forerunner of what is often called a "Skinner Box") to investigate the effect of reward (e.g., food, escape) on the behavior of cats. Thorndike placed the cats in a box that could be opened by pressing a latch or pulling a string. Outside the box was a bowl of milk or fish. Not surprisingly, the cats tried anything and everything until they stumbled onto the correct response. Also, not surprisingly, the cats learned to get out of the box more and more rapidly. From these beginnings, the most thoroughly researched phenomenon in psychology evolves.

Behavioral theory is now entering its ninth decade. The pioneering work of such investigators as Mateer (1918), Watson and Rayner (1920), Cason (1922a, 1922b), and Liddell (1926) in classical conditioning, and Blodgett (1929), Hull (1943), Skinner (1938), and Hebb (1949) in operant conditioning has led to the development of the most powerful technology known to behavioral science. Behaviorism, however, is in a paradoxical place in American education today. In a very real sense, behavioral theory is the basis for innovations such as teaching machines, computer-assisted instruction, competency-based education (mastery learning), instructional design, minimal competency testing, "educational accountability," situated cognition, and even social constructivism(see Chapter 7), yet behaviorism is no longer a "popular" orientation in education or instructional design. An exploration of behaviorism, its contributions to research and current practice in educational technology, (despite its recent unpopularity), and its usefulness in the future are the concerns of this chapter.

2.2.1 Basic Assumptions and Concepts

Behavioral psychology has provided instructional technology with several basic assumptions, concepts, and principles. These components of behavioral theory are outlined in this section (albeit briefly) in order to ensure that the discussion of its applications can be clearly linked back to the relevant behavioral theoretical underpinnings. While some or much of the following discussion may be elementary for many, we believed it was crucial to lay the groundwork that illustrates the major role behavioral psychology has played and continues to play in the research and development of instructional technology applications.

Three major assumptions are directly relevant to instructional technology. These assumptions focus on the following: the role of the learner, the nature of learning, and the generality of the learning processes and instructional procedures.

2.2.1.1.The Role of the Learner. As mentioned earlier in this chapter, one of the most misinterpreted (misrepresented) assumptions of behavioral learning theory concerns the role of the learner. Quite often, the learner is characterized as a passive entity that merely reacts to environmental stimuli (cf., Anderson's receptive-accrual model, 1986). However, according to B.F. Skinner, knowledge is action (Schnaitter, 1987). Skinner (1968) stated that a learner "does not passively absorb knowledge from the world around him but must play an active role" (p. 5). He goes on to explain how learners learn by doing, by experiencing, and by engaging in trial and error. All three of these components work together and must be studied together to formulate any given instance of learning. It is only when these three components are describable that we can identify what has been learned, under what conditions the learning has taken place, and the consequences that support and maintain the learned behavior. The emphasis is on the active responding of the learner -- the learner must be engaged in the behavior in order to learn and to validate that learning has occurred.

2.2.1.2. The Nature of Learning. Learning is frequently defined as a change in behavior due to experience. It is a function of building associations between the occasion upon which the behavior occurs (stimulus events) and the behavior itself (response events). These associations are centered in the experiences that produce learning, and differ to the extent to which they are contiguous and contingent (Chance, 1994). Contiguity refers to the close pairing of stimulus and response in time and/or space. Contingency refers to the dependency between the antecedent or behavioral event and either the response or consequence. Essential to the strengthening responses with these associations is the repeated continuous pairing of the stimulus with response and the pairing consequences (Skinner, 1968). It is the construction of functional relationships based on the contingencies of reinforcement under which the learning takes place. It is this functionality that is the essence of selection. Stimulus control develops as a result of continuous pairing with consequences (functions). In order to truly understand what has been learned, the entire relationship must be identified (Vargas, 1977). All components of this three-part contingency (i.e., functional relationship) must be observable and measurable to ensure the scientific verification that learning (i.e., a change of behavior) has occurred (Cooper, Heron, & Heward, 1987).

Of particular importance to instructional technology is the need to focus on the individual in this learning process. Contingencies vary from person to person based on each individual's genetic and reinforcement histories and events present at the time of learning (Gagné, 1985). This requires designers and developers to ensure that instruction is aimed at aiding the learning of the individual (e.g., Gagné, Briggs & Wager, 1992). To accomplish this, a needs assessment (Burton & Merrill, 1991) or front-end analysis (Mager, 1984; Smith & Ragan, 1993) is conducted at the very beginning of the instructional design process. The focus of this activity is to articulate, amongst other things, learner characteristics; that is, the needs and capabilities of individual learners are assessed to ensure that the instruction being developed is appropriate and meaningful. The goals are then written in terms of what the learner will accomplished via this instructional event.

The material to be learned must be identified in order to clearly understand the requisite nature of learning. There is a natural order inherent in many content areas. Much of the information within these content areas is characterized in sequences, however, many better form a network or a tree of related information (Skinner, 1968). Complex learning involves becoming competent in a given field by learning incremental behaviors which are ordered in these sequences, traditionally with very small steps, ranging from the most simple to more complex to the final goal. Two major considerations occur in complex learning. The first, as just mentioned, is the gradual elaboration of extremely complex patterns of behavior. The second involves the maintenance of the behavior in strength through the use of reinforcement contingent upon successful achievement at each stage. Implicit in this entire endeavor to the observable nature of actual learning public performance is crucial for the acknowledgment, verification (by self and/or others), and continued development of the present in similar behaviors.

2.2.1.3. The Generality of Learning Principles. According to behavioral theory, all animals -- including humans -- obey universal laws of behavior (Davey, 1981). All habits are formed from conditioned reflexes (Watson, 1924), or as a result of the experienced consequences of the organisms' behavior (Skinner, 1971). While Skinner (1969) does acknowledge species-specific behavior (e.g., adaptive mechanisms, differences in sensory equipment, effector systems, reactions to different reinforcers), he stands by the fact that the basic processes that promote or inhibit learning are universal of all organisms. Specifically he states that the research does show an

extraordinary uniformity over a wide range of reinforcement, the processes of extinction, discrimination and generalization return remarkably similar and consistent results across species. For example, fixed-interval reinforcement schedules yield a predictable scalloped performance effect (low rates of responding at the beginning of the interval following reinforcement, high rates of responding at the end of the interval) whether the subjects are animals or humans (Ferster & Skinner, 1957, p. 78).

Behavioral theory has contributed several important concepts and principles to the research and development of instructional technology. Three major types of behavior: respondent learning, operant learning and observational learning, serve as the organizer for this section. Each of these models relies on the building associations -- the simplest unit that is learned -- under the conditions of contiguity and repetition (Gagné, 1985). Each model also utilizes the processes of discrimination and generalization to describe the mechanisms humans use to adapt to situational and environmental stimuli (Chance, 1994). Discrimination is the act of responding differently to different stimuli, such as stopping at a red traffic light while driving through a green traffic light. Generalization is the act of responding in the same way to similar stimuli, specifically, to those stimuli not present at time of training. For example, students generate classroom behavior rules based on previous experiences and expectations in classroom settings. Or, when one is using a new word processing program, the individual attempts to apply what is already known about a word processing environment to the new program. In essence, discrimination and generalization are inversely related, crucial processes that facilitate adaptation, enable transfer to new environments.

2.2.1.3.1. Respondent learning. Involuntary actions, called respondents, are entrained using the classical conditioning techniques of Ivan Pavlov. In classical conditioning, an organism learns to respond to a stimulus that once prompted no response. The process begins with identification and articulation of an unconditional stimulus (US) that automatically elicits an emotional or physiological unconditional response (UR). No prior learning or conditioning is required to establish this natural connection (e.g., US = food; UR = salivation). In classical conditioning neutral stimulus is introduced, which initially prompts no response from the organism (e.g., a tone). The intent is to eventually have the tone (i.e., the conditioned stimulus or CS) elicit a response that very closely approximates the original UR (i.e., will become the conditional response or CR). The behavior is entrained using the principles of contiguity and repetition (i.e., practice). In repeated trials, the US and CS are introduced at the same time or in close temporal proximity. Gradually the US is presented less frequently with the CS, being sure to retain the performance of the UR/CR. Ultimately, the CS elicits the CR without the aid of the US.

Classical conditioning is a very powerful tool for entraining basic physiological responses (e.g., increases in blood pressure, taste aversions, psychosomatic illness), and emotive responses (e.g., arousal, fear, anxiety, pleasure) since the learning is paired with reflexive, inborn associations. Classical conditioning is a major theoretical notion underlying advertising, propaganda, and related learning(see 7.4). Its importance in the formations of biases, stereotypes, etc. is of particular importance in the design of instructional materials and should always be considered in the design process.

The incidental learning of these responses is clearly a concern in instructional settings. Behaviors such as test anxiety and "school phobia" are maladaptive behaviors that are entrained without intent. From a proactive stance in instructional design, a context or environmental analysis is a key component of a needs assessment (Tessmer, 1990). Every feature of the physical (e.g., lighting, classroom arrangement) and support (e.g., administration) environment are examined to ascertain positive or problematic factors that might influence the learner's attitude and level of participation in the instructional events. Similarly, in designing software, video, audio and so forth, careful attention is paid to the aesthetic features of the medium to ensure motivation and engagement. Respondent learning is a form of methodological behaviorism to be discussed later.

2.2.1.3.2. Operant Conditioning. Operant conditioning is based on a single, simple principle: that there is a functional and interconnected relationship between the stimuli that preceded a response (antecedents), the stimuli that follow a response (consequences), and the response (operant) itself. Acquisition of behavior is viewed as resulting from these 3-term or 3-component contingent relationships. While there are always contingencies in effect which are beyond the teacher's (or designer's) control, it is the role of the educator to control the environments so that the predominant contingent relationships are in line with the educational goal at hand.

2.2.1.3.3. Antecedents are those objects or events in the environment that serve as cues. Cues set the stage or serve as signals for specific behaviors to take place. Antecedent cues determine and may include temporal cues (time), interpersonal cues (people), and covert or internal cues (inside the skin). Verbal and written directions, non-verbal hand signals and facial gestures, highlighting with colors and bold-faced print are all examples of cues used by learners to discriminate the conditions for behaving in a way that returns a desired consequence. The behavior ultimately comes under stimulus control (i.e., controlled by the discriminative stimulus or cue) though the contiguous pairing in repeated trials, hence serving in a key functional role in this contingent relationship. Often the behavioral technologist seeks to increase or decrease antecedent (stimulus) control and must be cognizant of those cues to which generalized responding is desired or present. Antecedent control will increase with consequence pairing.

Unlike the involuntary actions entrained via classical conditioning, most human behaviors are emitted or voluntarily enacted. People deliberately "operate" on their environment to produce desired consequences. Skinner termed these purposeful responses operants. Operants include both private (thoughts) and public (behavior) activities, but the basic measure in behavioral theory remains the observable measurable response. Operants range from simple to complex, verbal to non-verbal, fine to gross motor actions - the whole realm of what we as humans choose to do based on the consequences the behavior elicits.

While the first two components of operant conditioning (antecedents and operants) are relatively straightforward, the nature of consequences and interactions between consequences and behaviors is fairly complex. First, consequences may be classified as contingent and non-contingent. Contingent consequences are reliable and relatively consistent. A clear association between the operant and the consequences can be established. Non-contingent consequences, however, often produce accidental or superstitious conditioning. If perchance, a computer program has scant or no documentation and the desired program features cannot be accessed via a predictable set of moves the user tends to press many keys, not really knowing what finally cause a successful screen change. This reduces the rate of learning, if any learning occurs at all. Another dimension focuses on whether or not the consequence is actually delivered. Consequences may be positive (something is presented following a response) or negative (something is taken away following a response). Note that positive and negative do not imply value (i.e., "good" or "bad"). Consequences can also be reinforcing, that is, tend to maintain or increase a behavior, or they may be punishing, that is, tend to decrease or suppress a behavior. Taken together, the possibilities then are positive reinforcement (presenting something to maintain or increase a behavior); positive punishment (presenting something to decrease a behavior); negative reinforcement (taking away something to increase a behavior); or negative punishment (taking away something to decrease a behavior). Another possibility obviously is that of no consequence following a behavior, which results in the disappearance or extinction of a previously reinforced behavior.

Examples of these types of consequences are readily found in the implementation of behavior modifications. Behavior modification or applied behavior analysis is a widely used instructional technology that manipulates the use of these consequences to produce the desired behavior (Cooper, Heron & Heward, 1987). Positive reinforcers ranging from praise to desired activities, to tangible rewards are delivered upon performance of a desired behavior. Positive punishments such as extra work, physical exertion, demerits are imposed upon performance of an undesirable behavior. Negative reinforcement is used when aversive conditions such as a teacher's hard gaze or yelling are taken away when the appropriate behavior (e.g., assignment completion). Negative punishment or response cost is used when a desirable stimulus such as free time privileges are taken away when an inappropriate behavior is performed. When no consequence follows the behavior, such as ignoring an undesirable behavior; ensuring that no attention is given to the misdeed, the undesirable behavior often abates. But this typically is preceded by an upsurge in the frequency of responding until the learner realizes that the behavior will no longer receive the desired consequence. All in all, the use of each consequence requires consideration of whether one wants to increase or decrease a behavior, if it is to be done by taking away or giving some stimulus, and whether or not that stimulus is desirable or undesirable.

In addition to the type of consequence, the schedule for the delivery or timing of those consequences is a key dimension to operant learning. Often a distinction is made between simple and complex schedules of reinforcement. Simple schedules include continuous consequation and partial or intermittent consequation. When using a continuous schedule, reinforcement is delivered after each correct response. This procedure is important for the learning of new behaviors because the functional relationship between antecedent - response - consequence is clearly communicated to the learner through predictability of consequation.

When using intermittent schedules, the reinforcement is delivered after some but not all responses. There are two basic types of intermittent schedules: ratio and interval. A ratio schedule is based on the numbers of responses required for consequation (e.g., piece work, number of completed math problems). An interval schedule is based on the amount of time that passes between consequation (e.g., payday, weekly quizzes). Ratio and interval schedules may be either fixed (predictable) or variable (unpredictable). These procedures are used once the functional relationship is established and with the intent is to encourage persistence of responses. The schedule is gradually changed from continuous, to fixed, to variable (i.e., until it becomes very "lean"), in order for the learner to perform the behavior for an extended period of time without any reinforcement. A variation often imposed on these schedules is called limited hold, which refers to the consequence only being available for a certain period of time.

Complex schedules are comprised of the various features of simple schedules. Shaping requires the learner to perform successive approximations of the target behavior by changing the criterion behavior for reinforcement to become m and more like the final performance. A good example of shaping is the writing process, wherein drafts are constantly revised towards the final product. Chaining requires that two or more behaviors must be performed in a specific sequence for consequation. Each behavior sets up cues for subsequent responses to be performed (e.g., long division). In multiple schedules, two or more simple schedules are in effect for the same behavior with each associated with a particular stimulus. Two or more schedules are available in a concurrent schedule procedure, however, there are no specific cues as to which schedule is in effect. Schedules may also be conjunctive (two or more behaviors that all must be performed for consequation to occur, but the behaviors may occur in any order), or tandem (two or more behaviors must be performed in a specific sequence without cues).

In all cases, the schedule or timing of the consequation is manipulated to fit the target response, using antecedents to signal the response, and appropriate consequences for the learner and the situation.

2.2.1.3.4. Observational learning. By using the basic concepts and principles of operant learning, and the basic definition that learning is a change of behavior brought about by experience, organisms can be thought of as learning new behaviors by observing the behavior of others (Chance, 1994). This premise was originally tested by Thorndike (1989) with cats, chicks, and dogs, and later Watson (1908) with monkeys, without success. In all cases, animals were situated in positions to observe and learn elementary problem-solving procedures (e.g., puzzle boxes) by watching successful same-species models perform the desired task. However, Warden and colleagues (1935, 1940), found that when animals were put in settings (e.g., cages) that were identical to the modeling animals and the observers watched the models perform the behavior and receive the reinforcement, the observers did learn the target behavior, often responding correctly on the first trial (Chance, 1994).

Attention focused seriously on observational learning research with the work of Bandura and colleagues in the 1960s. In a series of studies with children and adults (with children as the observers and children and adults as the models), these researchers demonstrated that the reinforcement of a model's behavior was positively correlated with the observer's judgments that the behavior was appropriate to imitate. These studies formed the empirical basis for Bandura's Social Learning Theory (1977) which stated that people are not driven by either inner forces or environmental stimuli in isolation. His assertion was that behavior and complex learning must be "explained in terms of a continuous reciprocal interaction of personal and environmental determinants... virtually all learning phenomenon resulting from direct experience occur on a vicarious basis by observing other people's behavior and its consequences for them" (p.11,12).

The basic observational or vicarious learning experience consists of watching a live or filmed performance or listening to a description of the performance (i.e., symbolic modeling) of a model and the positive and/or negative consequences of that model's behavior. Four component processes govern observational learning (Bandura, 1977). First, attentional processes determine what is selectively observed and extracted valence, complexity, prevalence, and functional value influence the quality of the attention. Observer characteristics such as sensory capacities, arousal level, perceptual set, and past reinforcement history mediate the stimuli. Second, the attended stimuli must be remembered or retained (i.e., retentional processes). Response patterns must be represented in memory in some organized, symbolic form. Humans primarily use imaginal and verbal codes for observed performances. These patterns must be practiced through overt or covert rehearsal to ensure retention. Third, the learner must engage in motor reproduction processes which require the organization of responses through their initiation, monitoring, and refinement on the basis of feedback. The behavior must be performed in order for cues to be learned and corrective adjustments made. The fourth component is motivation. Social learning theory recognizes that humans are more likely to adopt behavior that they value (functional) and reject behavior that they find punishing or unrewarding (not functional). Further, the evaluative judgments that humans make about the functionality of their own behavior mediate and regulate which observationally learned responses they will actually perform. Ultimately, people will enact self-satisfying behaviors and avoid distasteful or disdainful ones. Consequently, external reinforcement, vicarious reinforcement and self-reinforcement are all processes that promote the learning and performance of observed behavior.