33.9 Emotional-Motivational Aspects Of Choice And Learning

"Motivation" is a very slippery concept. J. M. Keller (1983) defines motivation as the "magnitude and direction of behavior. In other words, it refers to the choices people make as to what experiences they will approach or avoid, and the degree of effort they will exert in that respect" (p. 389). Both intuition and research (Tobias, 1987b) inform us that poorly motivated students are also very often poor performers in educational settings, too. However, the derivation of instructional prescriptions to help students improve their motivation to learn requires a much more detailed exploration of both the determinants of motivated behavior and the effects of motivation on choice and learning. That is, we need to uncover the reasons (motives) behind particular choices a student may make, to clarify which variables determine, or at least predict, both general patterns and levels of choice and situation-specific choices students will make. Additionally, we need to investigate the relationship between motivation and learning. In path-analytic terms, both direct and indirect (via the actual instructional choices made) relationships of motivation and learning require clarification.

A terminology issue needs to be raised at this point. Many researchers would argue that a "motivated" behavior might be based on rational, logical decision-making processes, and thus is not best described in terms of "emotional-motivational" processes. This is true to a large extent (although some could argue it is moot). However, for clarity's sake in this chapter, learner "motivation" refers primarily to the emotional states and reactions (and their consequent overt behaviors) experienced before, during, and after instruction which have an impact on learning and choice. So-called "rationally" motivated behaviors were discussed earlier in this paper.

A large body of research and several psychological theories exist that attempt to describe and explain the relationships among emotional-motivational variables, choice, and learning, and will only be touched on here. Instead, the implications of these findings about motivation and learning for the design of learner-controlled instruction will be explored. (For further discussion of motivation issues in educational technology, see also 32.5.5.)

33.9.1 Achievement Motivation and Learner-Controlled Instruction

The history of motivation research contains a sizable body of literature concerning what is called achievement motivation-in simple terms, a person's desire to perform and achieve. J. W. Atkinson (1974b) presents a theory, sprung from a behaviorist tradition, which connects so-called, 'motivated behavior" with performance based on what he calls "resultant achievement motivation." This construct has been defined and operationalized in many ways, but according to Heckhausen, Schmalt, and Schneider (1985), the measures of achievement motivation which tend to yield the most fruitful research are those indicators that are most overt, such as an individual's tendencies to persist Q. W. Atkinson, 1974a) or otherwise exert effort (Revelle & Michaels, 1976; Thomas, 1983) at some task or tasks. Because "achievement motivation" is largely defined by overt behaviors such as persistence and perseverance, there would seem to be at least on the surface clear reasons to attempt to extend experimental findings from the theory into the domain of learner-controlled instruction, so to try to provide more grounded explanations of student behaviors in such situations.

In J. W. Atkinson's theory (1974b), a person's level of achievement motivation at any given time is described as being the function of several variables. The first set of these variables includes motive to succeed and motive to avoid failure. The second type of variable influencing achievement motivation is the perceived probability of task success (also called expectancy). Last, extrinsic motivational factors (such as rewards or social approval) also play a role in the level of resultant achievement motivation. Various combinations of these variables produce both approach tendencies (i.e., a person's inclination to engage in some type of performance situation) and avoidance tendencies (i.e., a person's likelihood of shunning a particular performance situation). It is these two tendencies that taken together indicate a person's level of achievement motivation.

Expanding on the achievement-motivation tradition, Lepper (1985) suggests a link between motivation and achievement which is related to covert states in the learner. It is possible that a person's level of motivation during the performance of a learning task affects key components of information processing related to learning, a position also taken by Salomon (1983). Emotional-motivational variables may influence the direction and intensity of attention processes, arousal, depth of processing, and problem representation. Even though Lepper (1985) points out that many of these information-processing ideas are at present hypothetical, there does seem to be an emerging unification of the underlying mechanisms linking motivation and achievement (Humphreys & Revelle, 1984).

Some researchers suggest that the relationship between persistence or effort and achievement is generally linear (Revelle & Michaels, 1976; Salomon, 1983). That is, they say that highly achievement-motivated individuals will usually outperform those with lower motivation levels. However, theorists following J. W. Atkinson's original model (1974b) treat motivation as having a curvilinear (inverted-U shape) relationship with learning performance (Brophy, 1983; Humphreys & Revelle, 1984; J. M Keller, 1983). That is, both excessively low and high motivational levels can have dysfunctional effects on learning. This effect is moderated depending on either task difficulty or task complexity Q. W. Atkinson, 1974b; Humphreys & Revelle, 1984).

Given this relationship, it would be interesting to look for interactions of level of motivation and learner- or program-controlled instructional treatments. Such an ATI has been found by Carrier and Williams (1988). Using task persistence as the overt motivational index, they found that under two program-controlled treatments (with low and high amounts of instruction) students performing best were those in the middle levels of persistence; under learner control, however, the best performers had the highest levels of persistence. In other words, the curvilinear relation between motivation and learning was found under program control, but a mostly linear relationship was found under learner control. (Similar data were collected in a study by Morrison et a]., 1992. However, they only reported on a posited linear relationship-none found-between task persistence and achievement. It would be interesting to reanalyze their data to see if such a curvilinear relationship emerges.)

A possible explanation for these differential treatment effects can be inferred from a paper by Humphreys and Revelle (1984). Following their theory describing the underlying relationships between effort and performance, it is speculated that the students in the Carrier and Williams (1988) study behaved as though learner control were an easier or less complex condition; i.e., it placed fewer demands on their learning resources. In addition, it's possible the learner-controlled treatment produced less overall anxiety that could have interfered with learning. This interpretation is also consistent with Salomon's (1983) general notion of "perceived demand characteristics" of instructional treatments.

Although still hypothetical, three instructional factors are proposed here which might be expected to interact with a person's average general level of achievement motivation: learner or program control, task complexity, and extrinsic motivation variables.

Learner- and program-controlled treatments might be perceived by different students to be easier or more difficult to manage. It is possible that general motivational level could have an influence on performance by interacting with these treatments in a linear or curvilinear fashion, depending on the perceived "ease" of learning under the treatment.

Second, fairly simple tasks given under both learner-and program-controlled treatments might find no differences for highly motivated students. However, for difficult tasks, or those tasks requiring careful and deliberate thinking, one might expect learner control to surpass program control, at least for highly motivated (persistent) students. It is not clear yet what to expect for students of low or middle levels of motivation under tasks of varying difficulty or complexity.

Last, the object of using extrinsic motivators would be to try to increase the learner's persistence or effort expenditure through instructional manipulations, particularly for those students with low motivation levels. J. W. Atkinson (1974a) lists as examples of extrinsic motivators authority, competition, social approval, and external rewards. Several studies from the learner control literature support the use of these extrinsic motivators. Tennyson and Buttrey (1980) and Tennyson (1981) found that providing students under learner control with computer-delivered advisement's-that is, instructional recommendations about whether they should select more material (based on a mastery diagnosis)--did result in higher amounts of material chosen and in learning equivalent to the program-controlled version. In this case, the computer can be viewed as an extrinsic motivator because of the presumed authority its recommendations carry to the learner. Peters (1988), too, found that students receiving advisements requested more practice and answered more practice questions correctly on the first attempt than did students with no advisements, although there were no differences on posttest performance. Similarly, Carrier et al. (1986) found that simple encouragements within a learner-controlled treatment did increase the amount of material chosen by the students over a learner--controlled treatment without encouragements. Hicken, Sullivan, and Klein (1992) employed another external incentive approach by varying the type of task orientation in a lesson (i.e., students were told either that simply completing the lesson was sufficient to receive credit, or that a performance criterion level of 70% was required). The result was that students in the performance criterion condition, even without selecting any additional options or by spending more time-on-task, outperformed the group with the less-stringent conditions. The authors suggest that this extrinsic type of instructional manipulation functions to improve students' effort or concentration levels. There is evidence, then, that the type of task orientation given or simple instructional guidance in learner-controlled settings can alter performance, or at least the overall level of task persistence and other on-task behaviors,

33.9.2 Emotional-Motivational Patterns and Learner-Controlled CBI

The remainder of this section attempts to peer beneath the overt motivational variables (e.g., persistence) to see how learner emotional states might have direct or indirect impacts on learner-control effectiveness. Dweck (1986) and Dweck and Leggett (1988) offer a useful integrative approach to understanding student behaviors in terms of the student's own internal beliefs about the nature of their performances and their striving to confirm those beliefs. In their model, students are continually forming implicit theories about themselves which orient them to seek particular goals related to confirming these theories. Dweck (1986) describes so-called adaptive (or "mastery-oriented") and maladaptive ("helpless") motivational patterns. The maladaptive pattern is characterized by an avoidance of challenge and a deterioration of performance in the face of obstacles. Students who exhibit an adaptive pattern, in contrast, tend to seek challenging tasks and the maintenance of effective perseverance under failure circumstances.

What follows is an example of one avenue of promising theory, to date fairly unresearched within CBI contexts, which holds promise for explaining the heretofore mixed effects of learner-controlled CBI and suggesting means of improving instructional designs that adopt learner control. The investigation of other related theoretical frameworks is encouraged, however, as well. Kinzie (1990), for example, folds into her discussion of self-regulation and learner control another promising avenue of motivation theory, namely, self-efficacy (Bandura, 1986), and suggests ways of pursuing the topic in future research. But regardless of the stance on motivation research an investigator might adopt, the idea is to try to understand the nature of emotional states the learner experiences which produce healthy (adaptive) or dysfunctional (maladaptive) expression in terms of choices, persistence, and perseverance during learner-controlled instruction.

33.9.2.1. Attribution Theory and Learner Control. A major portion of Dweck and Leggett's (1988) model is based on research in the area of student attributions of their success and failures. Here the conception of motivation becomes that of a somewhat unstable factor affected on a moment-by-moment basis by the person's perception of events happening during instruction and their own inferred role in those events. Generally, an "attribution" refers to an individual's perceived causes of his or her own success or failures. Early conceptualizations by Kukla (1978) and Weiner (1974) explain that the degree to which students ascribe the causes of their own success or failures to ability, effort, task difficulty, or luck will differentially predict whether or what kinds of subsequent performance opportunities the student is likely to select voluntarily. These four variables can be grouped along two primary dimensions: internal versus external (analogous to, but not the same as, the familiar "locus of control" dimension of Rotter, 1966); and stable versus unstable.

Other researchers have recently extended, refined, and reconceptualized attribution theory. For example, Covington and Omelich (1984a, 1984b, 1985) attempt to frame student attributions in terms of emotional states they imply, such as pride, shame, guilt, and humiliation. Additionally, Dweck and Leggett (1988) present a model that seeks to explain the precursors of an individual's attributions along the "controllability" dimension. That is, they attempt to explain why some individuals feel more in control of their performances outcomes and others feel more "helpless." These developments in attribution theory have potentially important consequences for the design of motivational interventions during instruction.

Very few studies have explicitly examined attribution-like variables in connection with learner-controlled CBI. Treating perception of internality/extemality of reinforcement (or "locus of control"; Rotter, 1966) as a predictor variable has yielded generally unimpressive results in differentially predicting learning under several instructional conditions (Tobias, 1987b) and in predicting overall choice levels or learning in learner-controlled instruction (Carrier et al., 1985, 1986; Gray, 1989; Klein & Keller, 1990; Lopez & Harper, 1989; Santiago & Okey, 1992). In fact, Lopez and Harper (1989) conclude that there is little to be gained by further research investigating Rotter's locus-of-control construct in connection with learner-controlled instruction. Nevertheless, these negative findings could be masking potentially valid discriminations within groups broadly labeled externals or internals. For example, the differences between the two internal attribution styles, ability and effort, might be expected to affect options selection in either adaptive or maladaptive ways.

One early study (Fisher et al., 1975) treated various attributional. variables as dependent variables under conditions of learner- and program-controlled problem selection. The authors found that subjects in the choice group made significantly more internal and stable attributions during or following instruction than did students in the program-controlled group. They also found no treatment differences for an attribution variable they called control-no control, but they do not provide an operational definition of this variable to aid interpretation. Additionally, even though these researchers did not take baseline measures of attribution, nor plot the changes in attributions occurring over time, their study still supports the short-term modifiability of attributions as a possible result of treatment variables.

Within J. M. Keller's ARCS model of motivational design (1983, 1987a, 1987b), both attribution theory and learner control would potentially play useful roles when attempting to improve student confidence. In some strategies, Keller suggests, students might receive attributional feedback to enhance the feeling of that "they can do it." Additionally, they could be given some degree of control over their learning situation to enhance feelings of their own self-efficacy. The attributional feedback would seem to apply mostly to situations under learner control where students are asked to choose performance-related options. These options could include the selection of such specific instructional events as optional practice items, feedback, test situations, and, possibly, remediation or review following test conditions.

However, J. M. Keller's model is fairly nonspecific about the types of attributional feedback that should be offered to students and under which circumstances it would function optimally. Milheim and Martin (1991), too, suggest the utility of attribution theory for explaining the mixed effects found in the learner-control literature, but they, too, offer few specific suggestions for possible instructional design strategies that incorporate the theory.

Manipulations directed toward attaining these treatment goals would seem to fall into three classes of instructional strategies: (1) those affecting an entire lesson condition; (2) those preceding specific choice situations, taking the form of guidance, advice, or recommendations; and (3) those immediately following performance situations, taking the form of interpretations and attributions of success or failure generated by instruction.

The first strategy class includes attempts to adapt instruction to whatever overall attributional style a person seems to possess, Here, diagnosis of attribution levels would take place once, prior to the start of the lesson. All instruction might be subsequently modified accordingly in the manner of an ATI.

Additionally in this class, instruction could at the outset inform learners that they have control over what they see, and that their performance will be determined by how much they try. Given this, it would be necessary that the instruction monitor performance throughout the lesson and adjust task difficulty so as to minimize the discouraging effects of frequent failure.

Also in this class are manipulations related to task or ego involvement. Norm-referencing (ego-involving) suggestions to the student by the instructional system could be presented to students with high success rates. Examples of such presentations might be general statements that a student's performance will be compared to others, or perhaps comments to the students that they did better than most people on a particular task. Low-performing students might be best placed under task-involved conditions that encourage value placed on task improvement.

The second class also subscribes to a typical adaptive instruction paradigm, although here we are dealing with microinstructional adaptations of task-specific attributions. In particular, strategies in this class are forward looking, and include encouragement and advisement techniques such as those mentioned in the earlier section on achievement motivation. Some specific techniques might include recommending to the student that they choose a task of hard-difficulty (medium, easy) level depending on what the student's current performance level and attributional tendencies are at the moment. They might also include such motivating statements as "try harder on this one ...," or "the next task is an easy one... ." Another possibility might be to describe a subsequent task in terms compatible with the student's attributional style, but again on a very local level.

In the last class of instructional manipulations, the instruction could make evaluative and interpretive comments on a student's performance immediately following the success or failure of the task. The goal of these reflective or backward-looking instructional strategies is to after attributions intentionally. Comments to the student might attribute failure to his not trying hard enough or, when appropriate, to a task being difficult. Successful performance would always be attributed by the instruction to an internal factor. A study by Carrier et al. (1986) gave students a variety of backward-looking encouraging feedback (though not attributionally related) and found positive effects for task persistence. It is expected that feedback engineered more specifically to counteract maladaptive attributional patterns in the students would be even more fruitful.

A doctoral dissertation conducted by this author (Williams, 1992) examined the impact of attributionally related feed-back on learners of differing attributional tendencies (or styles) within learner- and program-controlled conditions. The type of feedback employed in the study was specifically intended to affect students' temporary perceptions of the causes of their learning successes and failures, that is, their attributions of their performance outcomes, so to minimize the dysfunctional behaviors of learners with maladaptive attributional styles. Providing specific attributionally related feedback to learners in an attempt to alter attributions temporarily has a well-established research base (Andrews & Debus, 1978; Barker & Graham, 1987; Borkowski, Weyhing & Carr, 1988; Dweck, 1975; Fowler & Peterson, 1981; Graham & Barker, 1990; Medway & Venino, 1982; Meyer & Dyck, 1986; Schunk, 1982, 1983, 1990a; Schunk & Cox, 1986) which hitherto had not been investigated in a computer-based context.

Findings from the study generally support the notion that, overall, certain types of attributional styles are maladaptive. (Examples of types of these maladaptive attributional styles include tendencies to attribute personal success to external causes, or to attribute personal failure to lack of ability.) That is, students who exhibit such motivational patterns tend not to exert as much effort or mental investment in their learning activities, and thus are prone to perform poorly.

The study by Williams (1992) also showed that the granting of a relatively small degree of learner control within the CBI lesson succeeded in improving the performance of students who otherwise showed certain types of maladaptive motivational patterns, namely, those who attribute their successes to either effort or to external causes. Also, students who attributed their successes to external causes, a generally maladaptive attributional style, showed markedly improved performance when given appropriate attributionally related feedback following their on-task performances. Such feedback was designed in accordance with recommendations from researchers on attribution theory, and consisted of reflective interpretations given by the computer to the student that a particular successful performance on the first try was due to ability (e.g., "You seem to know this material well!"), and on a second try was due to effort (e.g., "Terrific! It pays to try a little harder the second time."). Similarly, a failed performance on the first try was ascribed by the computer to lack of effort (e.g., "Perhaps you weren't concentrating enough on the question."), and on a second try to external causes (e.g., "You made a good try. That question was particularly hard."). In other words, giving these types of attributional feedback moderated the maladaptive tendencies of these students.

The Williams (1992) study supports the utility of the adaptive instruction paradigm of Gehlbach (1979). In this framework, unlike the classic ATI approach of Cronbach and Snow (1977), students who are deficient in some relevant aptitude are administered an instructional treatment intended to "correct' 'the difficulty, not operate around it or on top of it. In the current case, students who exhibited suboptimal motivational patterns were provided with appropriate feedback in an attempt to encourage more healthy emotional self-perceptions and hence more functional behaviors.

To summarize, the previous section posits that the general ineffectiveness of learner-controlled CBI can be explained, at least in part, by the fact that some learners have acquired maladaptive motivational tendencies and as a result exhibit dysfunctional or suboptimal choices (e.g., showing low persistence or perseverance, or terminating a lesson early). There is some evidence, although scant, that one particular motivational theory, namely, attribution theory, can be exploited to improve the on-task motivational behaviors for learners within learner-controlled situations. Other related theoretical approaches, e.g., learned helplessness and self-efficacy, also need explicating as to their potential relationships with learner-control effectiveness. The goal is to increase both motivation to achieve, where such motivation is low and motivational patterns are maladaptive, and to help students optimize their selection of instructional support.