A Literature Review

Deborah Alpert Sleight
Educational Psychology
Michigan State University

Spring 1994





Schemata and Prior Learning
Incidental Learning
Implicit Learning
Types of Learning

Job Aids

Job Aids as Performance Tools, not Learning Tools
Prior Knowledge and Job Aids
Computerized Job Aids
Adult Learning through Computers
Incidental Learning from Computerized Job Aids




Job requirements in business and industry are changing, which means employees have to learn more information, learn information that is more technical, analyze and use this information, and perform their jobs more quickly. According to Gloria Gery, "...the development of competence is taking place in an increasingly uncontrolled environment at the very time competency curves must be accelerated. The reasons by now are familiar to nearly everyone: many jobs today have become more complex, involve more variables, and require more analysis, synthesis, and interpretation in increasingly diverse contexts. Even entry-level jobs." (Gery, 1991, p. 10)

In order to compete in the new global information age, workers have to know how to decide which information they need, how to find the information they need, and how to analyze the information once it is found. They also have to know how to use the tools that enable them to find and process the information. The problem is that, although the job requirements have changed, the training and job support tools have not.

One solution to these new training needs is called just-in-time training. Training is delivered by a computer only when the worker needs and wants it. Interactive tutorials allow workers to receive training when they have time and when they need it, instead of having to attend classes when it is convenient for the company or the trainer. They can receive training on the specific topics they need, instead of receiving training generalized to a larger group within the workforce. And they can learn at their own pace, in the privacy of their own workstations.

Electronic performance support systems can provide this just-in-time training. An electronic performance support system is a computer-based system that can provide the worker access on the job to integrated information, advice, and learning experiences such as tutorials, videos, and simulations. Such a system provides "immediate, individualized on-line access to the full range of information, software, guidance, advice and assistance, data, images, tools, and assessment and monitoring systems to permit job performance with minimal support and intervention by others" (Gery, 1989).

Much information used on the job is changing too rapidly, or is too complex, or is done too infrequently, to make learning it an efficient use of time and effort. Workers do not need to memorize information such as long, step-by-step instructions on how to run a machine they rarely use. Sometimes they will simply need to be reminded of information they have already learned, whose details they have forgotten. In some jobs they may have learned the steps, but accuracy in remembering to do all the steps in the correct order may be critical, such as in the pilot's pre-flight routine. In other jobs, the reference manuals and documentation are too bulky and cumbersome to carry around or to allow information in them to be found quickly. In all these circumstances job support, rather than job training, is needed.

Performance support systems can provide job support in different forms: easy database access; job aids that provide step-by-step instructions on how to do something; and expert system advisors that can provide advice and guidance on performing infrequent or complex procedures. Job support aids are designed to guide workers through a procedure as they are doing the procedure. Such aids are designed to facilitate performance--not learning--in that they do not overtly show relationships among concepts or actions, or explain the rationale for taking an action.

Even though job aids are not designed to foster learning, it is likely that people will learn from these aids. Since they are not intended to foster learning, any learning that takes place does so incidentally. How much of the procedure will be learned incidentally? Will there be any learning about the procedure that is incorrect? Can job aids actually hinder learning? What impact will prior knowledge have on what is learned incidentally from job aids? Many studies on prior knowledge have shown that it plays a large part in understanding and remembering new information. Will this be the case with incidental learning?

Sometimes job aids are used as a convenience, so workers do not have to learn information that can be provided when it is needed. But sometimes it is more efficient if workers have learned the information. If workers can learn incidentally from job aids, then the time, effort, and cost of intentional training could be saved. The implications of the answers to these questions could have an impact on the design of performance support systems and computerized job aids that foster incidental learning. This literature review looks for studies that address the questions of incidental learning from computerized job aids.


Shuell defines learning as a change in the ability to do something that results from practice or experience and is enduring (Shuell, 1986). One popular theory of learning is the Schemata Theory. According to Moffitt, schemata are "organized, structured, clustered and abstract bodies of information that are generally conceptualized as networks of information in which the relationships among facts and actions are specified" (Moffitt, 1989, pp. 16-17). This theory hypothesizes that the schemata a person uses during learning will determine how the learner interprets the task to be learned, how the learner understands the information, and what knowledge the learner acquires.

Rumelhart and Norman (1978) account for learning based on a schema-based theory of long-term memory.They posit three distinct types of learning:

1) Restructuring or schema creation, where the learner creates a new schema because none existed into which new information could fit. This is the most difficult type of learning;

2) Accretion, where the learner encodes new information in terms of existing schemata. It is hypothesized that this is how prior knowledge of a topic makes it easier to learn new information on that topic.

3) Tuning or schema evolution, where the learner makes slow refinements or alterations to an existing schema as it is used in different situations. Tuning takes place over the lifetime of an individual and is important in the movement from novice to expert.

Schemata and Prior Learning

Prior knowledge means that facts can be used independently of the context in which they were learned (Johnson & Kieras, 1983). How much prior knowledge is activated during learning affects what is learned and how meaningful the learned material is. Lack of relevant prior knowledge may cause failure to link new information with existing information that could make the new information easier to understand and remember (Bransford et al., 1982).

Schemata theory is important in explaining how prior knowledge aids in the acquisition of new knowledge. According to the theory, prior knowledge is stored in schemata. New knowledge is either stored in existing schemata (that is, prior knowledge), or in new schemata. According to Norman, it is easier to store new knowledge in existing schemata than to create new schemata. "When you already have the proper conceptual framework, accretion is easy, painless, efficient...[but] when there isn't a good conceptual background, then accretion is slow and arduous" (Norman,1993, p. 28).

Gagné describes three ways new learning is influenced by prior knowledge (Gagné, 1980):

1) Prior knowledge helps the learner remember new information by creating more memory cues;

2) Prior knowledge fills gaps in the new knowledge, supplies conventional details, and changes some ideas; and

3) Prior knowledge of simpler component skills is necessary in order to learn new procedural or rule-governed intellectual skills.

Even though numerous studies have shown that prior knowledge is very important in learning new information, "reminders of previous relevant learning tend to be included casually and incidentally in instruction, rather than deliberately" (Gagné, 1980, p. 10). How can people who lack prior knowledge be helped to learn new information? Moffitt suggests reducing the perceived arbitrariness of the new material by providing people with organizational schemes, or teaching them temporary models (Moffitt, 1989). Rieber suggests incidental learning as a way of orienting students to subsequent instruction. Incidental learning could provide the students with prior knowledge they could then use in later intentional instruction (Rieber, 1991). This is born out by studies by Jenkins et al., who provided incidental prior exposure to new words, in order to see if students would learn those words more easily when they read them in context (Jenkins et al., 1984). This is discussed in more detail below under Incidental Learning.

Incidental Learning

The kind of learning that has been discussed so far is what is called intentional learning, or learning what the teacher or instructional designer intended to be learned. The learner is aware that there is something to be learned (Pressley, et al., 1987). Humans learn from many types of experiences, however, whether they have been designed to foster learning or not. Learning from experiences that were not designed to foster learning is called incidental learning. Incidental learning is often a side effect of the experience; the person may not have been aware that there was something to learn, but learned something nonetheless. When you move to a new city, you incidentally learn, perhaps through mistakes, what route is the fastest or easiest to take to work. Or when you get a new job, you learn what the politics of that office are.

Incidental learning can occur anywhere, even in educational institutions, alongside intentional learning. Students may learn how to avoid having the teacher call on them to answer questions, or may learn from a textbook that history is boring. Students may learn incidentally things the teacher or instructional designer never intended.

Most studies of incidental learning have looked at what is learned incidentally in the same domain as the material being intentionally taught. But Turner, in her study on the hidden impact of computers in schools, did include tangential incidental learning. She named three kinds of incidental learning that accompanies intentional learning: concomitant, incidental, and antithetical. Turner places these kinds of incidental learning on a continuum of how well they match the instructional intentions and purposes of the teachers or materials.





Continuum of Incidental Learning

Concomitant learning is unintended by the teachers, but is consistent with their intentions. Incidental learning is neutral to the teachers' intentions, while antithetical learning is opposite to the intentions of the teachers. It is important that teachers be aware of what unintended learning may be accompanying their teaching, because it can persist beyond school and influence how students deal with future experiences (Turner, 1984, pp. 169-171).

Even though incidental learning is not designed or planned for, studies have been made on what people learn incidentally, how this learning compares to the same information learned intentionally, and how incidental learning may be encouraged. To facilitate incidental learning, Moffitt suggests reducing the perceived arbitrariness of unfamiliar verbal materials by precisely elaborating sentences. This type of sentence is illustrated in Moffitt example below.

Base Sentence: The hungry man got into the car.

Imprecisely Elaborated Sentence: The hungry man got into the car to go for a ride.

Precisely Elaborated Sentence: The hungry man got into the car to go buy groceries.

In the base sentence above there is no explicit relationship between the fact that the man is hungry and the fact that he got into the car. The sentence following it is elaborated because it provides more detail, but imprecise because it still does not make clear the relationship between the man's hunger and his getting into the car. The last sentence does clarify the relationship, and thus is the kind of precisely elaborated sentence that aids incidental learning (Moffitt, 1989, p. 20).

Kurtz found that incidental learning from passages of information was facilitated when text was presented in passages longer than 20 lines before being interrupted by a question, whereas intentional learning was facilitated by shorter passages. She also found that students not given practice in using instructional objectives retained more incidental factual information than those given practice. It was hypothesized that the position of instructional objectives given before students read a passage may have stimulated activities which inhibited incidental learning (Kurtz, 1974).

Incidental learning seems to occur frequently through context. Jenkins et al. found that word meanings can be learned incidentally from context, although it seems to require a sizable number of word repetitions, even when the surrounding contexts are highly informative. The researchers found that more frequent presentation of words in context increased the incidental learning of those words. Better readers profited more from context than did less skilled readers, perhaps because they paid more attention to words they did not know. Students were first exposed to the new words by having them read aloud the new words on the blackboard, after hearing the teacher pronounce them, and by giving students word lists with example sentences and synonyms. This prior exposure resulted in even greater incidental learning of the new words from context (Jenkins et al., 1984).

According to Neuman and Koskinen (1992), it has been said that the acquisition process in language and reading is identical to what has been termed incidental learning. They hypothesized that reading materials with contextual supports would lead to greater incidental learning of words. In a study of children learning a second language, the researchers showed television programs that displayed printed captions to language minority students in the 7th and 8th grades. These students were divided into four groups: a control group that read from a textbook on science; a group that read along and listened to the textbook being read aloud, a group that watched a non-captioned television program on the same subject, and a group that watched captioned television.

The group that watched captioned television incidentally learned more words than either of the two other treatment groups or the control group. They also remembered more science information than the others. "Providing different modes of information appeared to enhance incidental learning from context rather than overwhelming the student's attentional capacity. Visual and printed context that provided explicit, and thus redundant, information supported incidental word learning" (Neuman & Koskinen, 1992, p. 104).

There may be negative consequences to incidental learning. What is learned may be antithetical to what the teacher intended. In addition, incidental learning may interfere with intentional learning. There are usually trade-offs between intentional and incidental learning, where trying to facilitate one can impede the other. Rieber, however, found that animated displays did not interfere with intentional learning (Rieber, 1991).

Implicit Learning

This proposal is not concerned with implicit learning, but because it is sometimes used interchangeably with incidental learning, I will take the time to differentiate the two. Usually when we think about learning, whether it is in the classroom or at work or in everyday situations, we think about explicit, conscious learning, where what is learned is verbalizable. People can state what they know. Incidental learning is also conscious, verbalizable learning that was not intended by the teacher or designer of the materials. Implicit learning may or may not be intended, but it is not verbalizable. People do not know they have learned anything. The matrix below shows the different combinations of intent and consciousness, resulting in the different types of learning.

   Intended  Unintended


  Intentional-Explicit   Incidental-Explicit


  Intentional-Implicit   Incidental-Implicit

Types of Learning

Learning can be either conscious or unconscious, intended or unintended. By definition, all conscious learning is explicit, and all unconscious learning is implicit. Also by definition, all intended learning is intentional , and all unintended learning is incidental. Intentional-explicit learning is what teachers usually aim for in schools. They intend that the students consciously learn specific topics. Students may learn what the teacher intended, but may also learn some things the teacher did not intend, such as the fear of trying to solve math problems; this is incidental-explicit learning. Implicit learning will be described in this section.

Research has shown that sometimes there is a discrepancy between people's explicit declarative knowledge of a task, and their performance of that task. Their performance is better than their knowledge about the task would predict. For example, people know when a sentence in their native language "sounds wrong," but frequently they cannot explain why it sounds wrong. Their knowledge of the grammatical structure of the language is, to a large degree, implicit. This discrepancy between verbalizable knowledge and performance has launched many studies of implicit, unconscious, non-verbalizable learning.

When is learning explicit and when is it implicit? Do people have a choice as to which type of learning in which to engage? Since by definition people are unaware when implicit learning is taking place, they cannot choose to learn implicitly. If they think they have made a choice, then it is always explicit learning they have chosen.

According to numerous studies by Reber and Berry & Broadbent, implicit learning only occurs in complex rule-governed systems because such systems are "too complex to be learned [explicitly] in an afternoon in the laboratory...a rich and complex stimulus domain is a prerequisite for the occurrence of implicit learning. If the system in use is too simple, or if the code can be broken by unconscious effort, then one will not see implicit processes" (Reber, 1989, p. 220). Reber defines implicit learning as"the process by which knowledge about the rule-governed complexities of the stimulus environment is acquired independently of conscious attempts to do so" (Reber, 1989, p. 219). He implies, then, that if implicit learning occurs when a system is complex, then explicit learning occurs when the material is simple or is organized in an understandable manner. This type of organization Reber refers to as structural salience (Reber et al., 1980). Salient information is organized and presented so as to make its key variables, assumptions, and rules obvious. It is the probability that, "if a person learns by the selective [explicit] rather than the unselective [implicit] mode, the key variables in the task will be chosen" (Berry & Broadbent, 1988, p. 254). Methods of increasing salience include telling subjects what the key variables and rules are; reducing the number of irrelevant variables; and taking advantage of prior knowledge by making key events act in accordance with it, "such as happening at the same time rather than at widely separated times" (Berry & Broadbent, 1988, p. 254).

Sometimes complex information can be learned explicitly if it is salient. But is performance of complex tasks enhanced more by explicit or implicit learning? According to several surprising studies (Berry & Broadbent, 1988, Lewicki et al., 1987, Reber et al., 1980), the performance of complex tasks is much poorer when the task is learned explicitly than when it is learned implicitly. Why does performance of complex tasks improve with implicit learning? How does implicit learning take place?

According to Berry & Broadbent (1988), when we attempt to learn explicitly, we are selective about what variables we pay attention to. In complex systems there are too many variables to allow us to explicitly learn them all in the short time allowed during the experiment, so we select what we think are the key variables and ignore the others. When we learn implicitly, however, we are not in conscious control, hence we are unselective and pay attention to all the variables. The knowledge becomes non-verbalizable because "a person may observe the variables unselectively and attempt to store all of the contingencies between them. Therefore, it may be hard to report so many links or to have confidence in any of them" (Berry & Broadbent, 1988, p. 253). Many of the associations implicitly learned will be perceptual characteristics which may not be important, but happen to have been present (Berry & Broadbent, 1988).

Implicit learning is enhanced when the learner is active in the situation. Watching someone else manipulate variable or make decisions does not result in implicit learning, but when the learner is the one to make the decisions and manipulate the variables, then the associations are made and implicit learning takes place (Berry, 1991).

A type of implicit knowledge is expert knowledge that has become automatic and therefore unconscious. "When a person learns to automaticity, the knowledge that was previously declarative (verbally accessible) seems to become proceduralized-compiled (Neves & Anderson, 1981), and intermediate steps are no longer verbally accessible" (McGeorge & Burton, 1990). Automatic skills may be non-verbalizable, but might be retrieved through probing questions.

I was unable to find any literature on implicit learning of job aids, and few studies of incidental learning of technical prose, such as job aids and manuals. How much of a technical procedure is learned incidentally? Is there be any incidental learning about technical procedures that is incorrect? Can job aids hinder intentional learning? What impact will prior knowledge have on what is learned incidentally from job aids? These questions are discussed in the next section.


Job aids are performance support tools used on the job that are succinct, step-by-step descriptions of how to do a task. They specify what to do, how to do it, and to what standards it should be done. The most successful example of a job aid, according to Lawson (1986), is the instruction plate on self-serve gas pumps, because it enabled fifty million drivers to become self-sufficient within five years. Other types of job aids include recipes, checklists, and bicycle assembly directions.

Job aids are helpful for doing certain kinds of tasks such as tasks with too many steps to remember; tasks that require inaccessible reference material; jobs that require extreme accuracy, such as jet takeoffs and landings; and infrequent procedures, where people are apt to forget how to do them. But jobs aids are not useful in jobs where speed is essential, or for tasks that have many options. Emergency room procedures are not good candidates for job aids because the medical worker does not have time to read the job aid because speed is critical for the patient's survival. In this case the worker must memorize the steps needed to do the procedure. Tasks with many options would make a job aid difficult to follow, although tasks with some options can be aided by a branching flowchart (Lawson, 1986).

Job Aids as Performance Tools, not Learning Tools

The purpose of a job aid is to guide and facilitate performance. To this end the job aid breaks the task down into discrete steps and minimizes extraneous information, such as explanations and related information. A person putting together a bicycle doesn't care at that moment about the history of the bicycle or examples of different kinds of bicycles, but only what has to be done to accomplish the task. The person wants to do the job, not learn how to do the job. Therefore job aids are designed to foster performance, not learning.

[Job aids] are not, in the truest sense, instructional. That is, users are not supposed to learn the procedures by heart as a result of reading the job aid. Users are supposed to refer to the job aid every time they perform the procedure. They use the job aid as a reminder and as a checklist, not as a "teacher." Because job aids are not instructional, they do not take the form of reference material, tutorials, explanation, or detail. They do take the form of diagrams, checklists, algorithms, flip charts, on-line helps, etc." (Lawson, 1986, p. 15).

Because they are performance tools, job aids do not contain explanatory detail or tutorial information. Nonetheless, people can learn from job aids. Since job aids are not instructional in purpose, any learning from them would be incidental, not intentional. As was discussed earlier, people do learn incidentally from experiences and materials that were not designed to foster learning. But people do not always learn from job aids. Campbell found that incorporating job aids into instructional material to teach verbal reasoning about directions did not improve learning. She found that many of the participants indicated either they did not use the job aids, or made their own that were similar to the ones provided (Campbell, 1992).

Aside from not fostering learning, can job aids actually hinder learning? According to studies by Weill-Fassina, initially work aids can facilitate performance on specific sub-tasks, but in the long run they may interfere with learning the whole job. Because they do not present the big picture and make clear the relationships among task variables, they may prevent the worker from developing a concept of the whole job, and from organizing and limiting the field of action. She suggests that job aids may be effective in routine, secondary sub-tasks, but that job aids for major tasks "should be limited to proposing a broad overall structure that leaves an operator free to organize the work" (Weill-Fassina, 1980, p. 343).

Prior Knowledge and Job Aids

Job aids are used frequently for technical tasks. What effect does prior knowledge have on the incidental learning of technical tasks? Johnson and Kieras ran a study on the effects of prior knowledge on recall of new and known technical propositions. They divided the participants into three groups: a self-paced group, who knew they would be tested after reading a technical prose passage and could read at their own pace; a forced-pace group, who also knew they would be tested after reading the technical prose passage, but were given a set period of time to read the passage; and an incidental group, who could read at their own pace, but did not know they would later be tested. The results of the study indicated that "the effects of prior knowledge are strongest in the incidental task, in which recall was 70% higher for previously known propositions than for previously unknown ones, although the average level of recall was lower than that of the other conditions (self-paced and forced-pace)" (Johnson & Kieras, 1983, p. 462). It seems that prior knowledge may be helpful in incidental learning, although not as helpful as in intentional learning.

This study was done using technical prose because, according to the researchers, the relevancy of schematic knowledge to technical prose has not yet been established. Schemata may play only a weak role in technical prose since the authors say that schemata are usually described in terms of sequential events, such as might appear in simple stories, goal-directed activities, or the plot structure of a story, independent of passage content. Prose containing a sequence of events is quite different from technical prose, which is often simply an exposition of facts about a subject and which often does not include sequential events or a goal. Thus is appears that schematic knowledge is not necessarily applicable to technical prose. This conclusion is supported by Graesser, Hoffman, and Clark (cited in Johnson & Kieras, 1983), who found that the narrativity of a passage strongly predicted its recall, and that technical prose was not narrative. Job aids, although technical, are highly sequential. Thus, while prior knowledge may not facilitate recall of technical passages in general, it may help people remember the steps in a job aid.

Job aids are helpful to people because they relieve people of "remembering" details; they become an extension of memory. By this definition they are cognitive artifacts, tools that aid the mind, or external devices that "expand human abilities beyond that which biological heritage alone makes possible" (Norman, 1993, p. 123). According to Norman, cognitive artifacts are the devices that let us increase our memory, thought, and reason. They are the "things that make us smart" (Norman, 1993, p. 43). They do this by representing actions or information we want to remember. By representation they help the mind keep track of complex events.

Computerized Job Aids

In the past job aids have been printed on paper, metal, plastic, or whatever was appropriate to the task. Today many job aids are computerized. Examples of computerized job aids are online help systems that come with some computer software, or directions that appear on small screens on machines such as photocopiers.

Advantages of computerized job aids are their interactivity; they sense actions they user has taken and can present directions appropriate for the action the user needs to take next. They can also present feedback in terms of correct and incorrect actions the user has taken. Disadvantages occur because the user considers the interactivity to be similar to a conversation with another human. Problems and misunderstandings between two people can be cleared up by mutual question asking and explanations. A computer, however, cannot participate fully in a conversation because it can sense only actions it is designed to sense, that is, actions that change the state of the machine. Many actions a user takes when using a machine are not able to be sensed by the machine, such as numbering the pages of a document to be copied two-sided. If the user organizes the pages wrong, the copier will copy the wrong pages together. The machine cannot know how the user has organized the pages. All it can sense is whether or not paper is in the document feeder.

Computerized job aids can be integrated into electronic performance support systems to provide online support at the user's workstation when that support is needed. The components of an electronic performance support system (EPSS) might include databases, productivity software, training modules, expert systems, online help, intercomputer network access, hypertext data links, artificial intelligence mentoring, and anything else that would help the user accomplish job tasks. All these components would be integrated into an interface that would make it easier for the user to find the information, tool, or help when it is needed.

One component of an EPSS is the advisory component, or expert system. An expert system can be considered a computerized job aid because it is used on the job and it is designed to foster performance, not learning. An expert system helps users diagnose specific problems and guides them through specific procedures. For example, an expert system could be written that could guide an agent through filling out a rarely used form in an insurance company. The employee would indicate to the expert system which form was in question. The expert system would then ask the employee a series of questions and tell the employee what kind of information should go in each section.

Expert systems, and sometimes online help systems, contain an artificial intelligence component that enable the system to infer beyond the rules they have been given. Thus they can provide information that makes them seem intelligent, if not conscious. But the information they provide to the user is often incomplete because the computer lacks what Collins calls "cultural competence" (Collins, 1990, p. 94), that is, knowledge of how to function in a culturally and historically situated world. This knowledge often cannot be reduced to a set of rules or heuristics, and if it can be, then the set of rules would be much too huge to be manageable, even by a computer. Thus, the user must meet the expert system more than halfway and provide all the interpretation and meaning of the information the system provides.

Adult Learning through Computers

Can computers provide the kind of learning situation that workers need? Adult workers have different learning needs than do children. Children's learning needs reflect their growth patterns; different types of concepts are learned at different ages. For this reason schools typically group children into grades by age or, less typically, by levels of expertise. Learning is done for the sake of learning, not to accomplish specific tasks. Children are given time to devote to learning.

Adult workers, on the other hand, are not helped by these traditional educational practices. They have varying levels of expertise and, among themselves, different reasons for learning. They must stay abreast of the changes in their fields, and do all this while continuing to work, for there is usually not enough time set aside for training. Therefore training must not only be related to work, but usually must be done on the job. On-the-job training is easily delivered by computer for many jobs. Training delivered by computer must allow adult professionals to located the desired information quickly and easily, and provide multiple ways to access and use the information, in order to accommodate the wide range of learning modes. Users should be in control of their own instruction (Shaw, 1992).

Shaw's study of workers found that they wanted to be able to learn what they needed when they needed it, and that the information they desired should be easy to find. This description matches the way performance support systems provide information to workers.

Incidental Learning from a Computerized Job Aid

How is incidental learning affected when information is presented on a computer? Rieber found that students were better able to learn incidentally-delivered material without sacrificing intentional learning when that material was presented in an animated display on a computer, and was divided into chunks of textual and visual sequences. He also found that recall of incidental information was more robust for pictures than for words. (Rieber, 1991).

Moffitt studied incidental learning of decision making from a computerized expert system that gives advice on decision making. The participants would ask the system questions, and it would guide them through the decision-making process by answering their questions and explaining why it had given that particular answer. Moffitt found that more incidental learning occurs when explanations are specific to the context in which they are given, and are embedded in the questions the user is asking or the responses the system is giving, than when they are not (Moffitt, 1989).


Although there have been many studies on incidental learning, most of them have involved non-technical material, such as deriving new word meaning from the context of a novel. Few studies have been made on incidental learning from technical materials, such as job aids. All tasks, even those involving non-technical topics, have technical aspects. And job aids, no matter what topic they address, are always technical in nature, because they are created to simplify the technical aspects of topics, whether those topics are technical or not. Job aids, whether they are interactive or not, involve step-by-step instructions on performing a procedure or listing items to check.

Do people learn from job aids? The few studies that have been made on incidental learning from job aids are inconclusive. One study found that the use of printed job aids incorporated into instructional materials did not improve learning. Another study found that students did learn from embedded explanations in an expert system. There is much to study in this area, particularly since there seems to be a great difference in interactive and non-interactive job aids, since interactive job aids can return feedback and provide context-sensitive instructions.

Does prior knowledge facilitate incidental learning? Prior knowledge does seem to play a role in incidental learning similar to the role it plays in intentional learning. It is easier to learn incidentally when one has prior knowledge of the topic to be learned, whether that topic is technical or not. Prior knowledge may be provided by giving the learners a preview of what they will be encountering. To foster incidental learning or recall from job aids, before, a list may be presented to learners before the job aid is used. This list would contain technical or unfamiliar words, with definitions, that learners will encounter in the job aid.

Can job aids hinder learning? Yes, it can hinder the person forming a cognitive representation of the job as a whole, and being able to organize or limit the field of action in performing the job. The researcher (Weill-Fassina) suggests using job aids for small sub-tasks only, and giving only broad directions for performing the whole job.

Since much of the learning needed for current and future jobs will be learned individually and incidentally from such computerized systems as performance support systems, it is important that we learn if and what people will learn from these systems, and how to design them for optimal performance and optimal learning of the job.



Berry, D.C. (1991). The role of action in implicit learning. The Quarterly Journal of Experimental Psychology, 43A(4), 881-906.

Berry, D., & Broadbent, D. (1988). Interactive tasks and the implicit-explicit distinction. British Journal of Psychology, 79, 251-272.

Bransford, J.D., Stein, B.S., Vye, N.J., Ranks, J.J., Auble, P.M., Mezynski, K.J., & Perfetto, G.A. (1982). Differences in approaches to learning: An overview. Journal of Experimental Psychology: General, 111, 390-398.

Campbell, A.W. (1992). Job Aids, Feedback, and the Teaching of Verbal Reasoning. Unpublished dissertation, Western Michigan University, Kalamazoo, Michigan.

Carr, C. (1992). Smart training: The manager's guide to training for improved performance. McGraw-Hill, Inc., New York.

Collins, H.M. (1990). Artificial experts: Social knowledge and intelligent machines. Cambridge, MA: The MIT Press.

Gagné, R.M. (1980). Is educational technology in phase? Educational Technology, 2, 7-13.

Gery, G. (1989). The quest for electronic performance support. CBT Directions, July 1989, 21-21.

Gery, G. (1991). Electronic performance support systems: how and why to remake the workplace through the strategic application of technology. Weingarten Publications, Boston.

Jenkins, J.R., Stein, M.L., & Wysocki, K. (1984). Learning vocabulary through reading. American Educational Research Journal, 21(4), 767-787.

Johnson, W., & Kieras, D. (1983). Representation-saving effects of prior knowledge in memory for simple technical prose. Memory & cognition, 11(5), 456-466.

Kurtz, P.D. (1974). The effect of instructional objectives on student learning. The Alberta Journal of Educational Research, 20(4), 327-333.

Lawson, P. (1986). Job aids: Give the readers what they want. ADCIS Proceedings, November, 1986, 15-18.

Lewicki, P., Hoffman, H., & Czyzewska, M. (1987). Unconscious acquisition of complex procedural knowledge. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13 (4), 525-530.

McGeorge, P. & Burton, A.M. (1990). Semantic processing in an incidental learning task. The Quarterly Journal of Experimental Psychology, 42A(3), 597-609.

Moffitt, K.E. (1989). An empirical test of expert system explanation facility effects on incidental learning and decision making. Unpublished dissertation. Arizona State University.

Neuman, S.B. & Koskinen, P. (1992). Captioned television as comprehensible input: Effects of incidental word learning from context for language minority students. Reading Research Quarterly, 27(1), 95-106.

Norman, D.A. (1993). Things that make us smart: Defending human attributes in the age of the machine. Reading, MA: Addison-Wesley Publishing Co.

Pressley, M., McDaniel, M.A., Turnure, J.E., Wood, E., & Ahmad, M. (1987). Generation and precision of elaboration: Effects on intentional and incidental learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 291-300.

Raybould, B. (October, 1991). An EPSS case study: Prime Computer. Paper presented at the Electronic Performance Support Systems Conference, Atlanta, Georgia.

Reber, A. (1989). Implicit learning and tacit knowledge. Journal of Experimental Psychology: General, 118 (3), 219-235.

Reber, A., Kassin, S., Lewis, S., & Cantor, G. (1980). On the relationship between implicit and explicit modes in the learning of a complex rule structure. Journal of Experimental Psychology: Human Learning and Memory, 6 (5), 492-502.

Rieber, L.P. (1991). Animation, incidental learning, and continuing motivation. Journal of Educational Psychology, 83(3), 318-328.

Rumelhart, D.E., & Norman, D.A. (1978). Accretion, tuning, and restructuring: Three modes of learning. In J.R. Anderson (Ed.), Semantic Factors in Cognition (pp. 37-53). Hillsdale, NJ: Lawrence Erlbaum Associates.

Shaw, D.S. (1992). Computer-aided instruction for adult professionals: A research report. Journal of Computer-Based Instruction, 19(2), 54-57.

Shuell, T.J. (1986). Cognitive conceptions of learning. Review of Educational Research, 56, 411-436.

Suchman, L.A. (1987). Plans and situated actions: The problem of human-machine communication. New York: Cambridge University Press.

Turner, M.I. (1984). "Playing with computers": The hidden impact of the computer in a school. Unpublished dissertation. Oregon State University.

Weill-Fassina, A. (1980). Guidage et planification de l'action par les aides au travail. [Job manuals and other work aids as guides in planning action]. Bulletin de Psychologie, XXXIII, No. 344, 343-349.

(c) Deborah Alpert Sleight, 1994
Permission is given to reprint for non-profit use providing credit is given.

Deborah Alpert Sleight
Educational Psychology
Michigan State University
East Lansing, MI 48824