"Schank employs the concepts of case based reasoning to argue for the benefits of learning 'by doing' over learning 'by being told'. Review the main claims of his argument to consider how successfully he is able to explain the learning of concepts and principles through formal instruction. You should make reference to examples of instructional learning, and you may wish to situate Schank's account within the background of learning theories."
Schank's investigation of the argument for learning 'by doing' in favour of learning 'by being told' attempts to provide a more
rigorous explanation as to why learning 'by doing' is (and should be) considered more effective. Such an argument could easily be seen as rejecting learning 'by being told' (formal instruction) out of hand, but Schank does not go this far. Rather, he suggests that learning 'by doing' should be utilised in more situations where previously formal instruction was seen as the only
practical solution.
The
crux of his argument is that where both options are available, learning by doing will always be more effective as it teaches
implicitly. For this effectiveness to be fully realised, the educational system needs to be designed correctly (basing the environment, structure and content around
implicit knowledge) and likewise evaluated in terms of whether the participant can now achieve their desired
goal (rather than make
explicit statements about what they have learnt). He recounts that a lack of
understanding about how learning by doing works has long been a
stumbling block in the process of designing and implementing new educational systems. Even though the limitations of teaching solely by instruction have long been recognised (and such systems frowned upon) by
psychologists, it still persists as the method of choice in a wide range of educational systems. Schank examines the two explanations often given for this
inconsistency between theory and practice: that not all things that we want to teach can be 'done' (or at least not practically); and that it is unclear as to what learning by doing actually teaches.
From an
educator's perspective, the first of these arguments seems plausible enough. If the
teaching environment does not offer the tools needed to learn something by doing, and this cannot be rectified safely or inexpensively (or no alternative environment is known), then it seems reasonable for the existing system to be retained. The second argument provides the additional
disincentive that we would not know if the new system taught the right things. We knew what the previous system taught, as we specified a
curriculum and checked that it had been learned by examination. The upshot of this of course is that the content of the
curriculum is dictated by whether it can be easily taught and examined - whether it is relevant (i.e. whether it helps the learner achieve their desired goal) becomes a secondary concern.
We can see this in practice in many areas of
education. At school level, concepts and facts (and sometimes entire subjects) are taught seemingly for their own sake. The course of study will usually be heavily weighted toward
formal instruction. Crucially, the division of content into discrete subjects leads to knowledge that needs to be acquired to achieve a specific goal being separated from that goal's frame of reference. For instance
mathematics taught early on in school is not presented in the context of helping the student understand
biology or
chemistry, subjects which are introduced at a later stage.
In
higher education, the emphasis on easily
quantifiable achievement leads to the focus of courses shifting from teaching a discipline ('skill') to ensuring the students are familiar with previous work in the field ('reading/recall'). Schank concedes that in some courses, the aim may not be to train the student to be a practitioner of the discipline in question, but instead to give them an overview of the subject within a short time. In academic study it could be argued that a familiarity with important previous work, even though learning it involves no 'doing' component and it may not in itself help the student solve problems, is still an important part of the
curriculum.
As well as examining situations where learning by being told has remained
entrenched, Schank considers some where this has never been the case, and learning by doing is considered the
norm. One such situation is learning in the
workplace, and another, with similar goals but some fundamental differences, is
training.
Learning in the workplace occurs spontaneously because the learner has an incentive - good performance equates to better rewards, while bad performance could lead to
unemployment.
Proficiency also improves with
repetition of the task. Furthermore, if the employee cares about the task, they will dislike failing at it and try to avoid doing so - provided that they are made
aware of their
mistakes.
Training is less effective because the task is
artificial - elements of the environment do not sufficiently
emulate the real task, and there is less emphasis on the
consequences of the trainee's actions. The employer has dictated that the training must be undertaken, regardless of whether the employee has any
incentive. This might not be such a problem if training was initiated in response to an employee failing at a task - but although this would probably be more effective as a learning tool, the employer's main aim is to prevent errors and losses before they happen.
The example of learning '
on the job', and the use of simulation and '
real-world' tasks in learning (e.g.
driving lessons) correspond closely with both case-based reasoning and the 'understanding cycle' - a theory of how learning occurs naturally.
Case-based reasoning is a model for explaining how people use previous experiences to help understand and interact with new situations, and to form generalisations (scripts and other
cognitive schema) that can be applied in many situations with minimal
adaptation (and therefore little cognitive effort). An important aspect of this model is way in which new cases are stored and indexed in relation to each other, which is subject to ongoing refinement as more cases are processed (reasoned over). The learning that takes place in such a model is entirely the
by-product of this organisation.
What all the presented examples of successful learning by doing have in common is that they all present a goal at the right time and in the right context, and present the learner with experiences that are applicable in helping them deal with future situations. These experiences can be considered as 'cases', and the natural learning process described above is exploited in the acquisition of cases in this way to facilitate their organisation and
correlation in memory. We now want to identify what is best learnt through exposure to new cases so that appropriate cases can be presented and other methods (such as formal instruction) can be used to cover any shortfall.
Schank uses the example of a '
dining school' to illustrate some of the points made about learning by doing raised so far, and to show that applying the concepts of case-based reasoning gives us a clearer view of what is learnt. By identifying and classifying what is learnt, we can present cases in the right sequence and with the right context to significantly enhance the learning of a task. If successful, such an approach can be much more effective than formal instruction alone.
Schank calls the system for presenting a collection of cases to allow the user to perform a specific task or set of related tasks
Goal Based Scenarios (GBSs). A
GBS will result in the accomplishment by the learner of a goal, which must seem to be as real as possible. Any format can be used to present the scenario - including paper-based work, software, simulation or the real task. Note that the goal does not have to be physical - it can in fact be purely cognitive. In such cases the 'doing' involved may also be purely
cognitive - but this may still require cognizing over and above that involved in learning by being told. The teacher's role is to intervene when a problem is encountered and present a new case that can be applied to the problem. They should also provide useful tools and answer questions throughout the
scenario.
Such a scenario adheres to a number of principles that are required for
problem solving by case-based reasoning. Tasks have to be repeated with slight variations to allow comparisons to be made, and to strengthen relevant correlation's between cases while discarding less important ones. Successful and unsuccessful attempts at the task have to be experienced. Formal instruction (such as in the dining school example) can only impart pieces of the instructor's indexing information - the sensory data and context of each case cannot be transferred in this way.
The design of a GBS is centred around the selection of cases that will require the learner to develop the right skills. 'Skill' is too vague a term, as it can be used to describe the solution to any problem regardless of scope. Schank decomposes the content of cases into '
microscripts' - discrete abilities (either cognitive, physical, perceptual, or a combination thereof) with limited use on their own that cannot be taught explicitly. Instead they are acquired naturally in the pursuit of a goal. Individually, the microscript's result may be rewarding, it may be part of a '
package' of actions the result of which is rewarding, or it may be required to allow the learning of other microscripts. It will be reinforced by non-artificial repetition and will
degrade over time with disuse.
If we look at the GBS in terms of its microscript content, cases can be presented that contain microscripts relevant to attaining the goal (as opposed to ones that are easy to identify - a pitfall similar to the one discussed in curriculum design). It can also be argued that much of learning by being told is just a reinforcement of 'memorization' microscripts. Certainly, from Schank's description, formal instruction methods would have little to offer directly in the way of microscripts.
In addition to microscripts, cases delivered in a GBS present the learner with some insights into non-context specific processes. These insights develop into
Participation Strategies - universally applicable concepts (and a limited amount of executable procedures) that may result in a more successful interaction with that process. Unlike microscripts, participation strategies are not guaranteed to work every time - they can only suggest a course of action that may or may not prove successful in a given situation. Schank gives
Communication,
Human Relations and
Reasoning as examples of processes which will always arise in a GBS provided certain conditions are met (e.g. any GBS involving a group of participants will always involve Human Relations).
As well as being the '
raw material' in a GBS for providing domain-specific and domain-independent solutions, new cases can be retained in memory for ongoing reference and integration, even if the relevance of the new information is not immediately apparent.
As has been mentioned, a learning goal can be entirely cognitive. Whereas there are cognitive microscripts to help achieve such goals, they may also depend on the acquisition of knowledge. Schank refers to this knowledge, learnt in the employ of reaching a goal and which the learner is explicitly aware of as
Explicit Functional Knowledge (
EFK). This falls into four categories.
Domain knowledge helps form expectations of aspects of a domain, and is required for the learner to be able to observe and interact with that domain with an adequate level of understanding. There would not appear to be any reason why this type of EFK would have to be acquired through cases. Many goals require knowledge that has to be available before cases could be successfully attempted (e.g.
formulae,
syntax of
programming languages,
symbols and
notation, etc.) which would fall into the category of domain knowledge, and would have to be learnt explicitly before they could be put into practice enough to be known implicitly.
Knowledge for
day-to-day decision making (the second type of EFK) is acquired by making a large number of previous decisions in similar situations. A GBS could be designed to present a number of cases requiring decisions to be made, which would provide decision-making
knowledge completely separate from any microscript. The basis of this knowledge is the ability to closely compare alternatives and weigh up all the external factors involved in the decision, skills which could only be very roughly covered by formal instruction.
The third type of EFK is knowledge that is useful for
understanding life. This might be better described as generalisations and opinions gained from observing example cases that can then applied universally. The final type of EFK is '
trivia' - which Schank does not elaborate on, but can be defined as items of information with no apparent application in terms of a
goal (except in highly specialised or artificial situations). This category could be employed to flag any content within a learning system that could be safely omitted.
Having examined what is learnt from participating in a
goal-based scenario, Schank proposes some rules that can be used to help decide on the content of the cases presented and at what stage they should be introduced. No case should be introduced until the knowledge it provides is needed to for the learner to make progress toward their goal ('No case before its time'). Every new case should in some way
violate the expectations the learner has inferred from the cases previously encountered. Otherwise the case does not prompt any changes in
memory organisation, and hence nothing is learnt. Cases should relate to actions, meaning not only that learning by doing should be the preferred teaching method, but that presenting a new case in terms of actual behaviours will make subsequent cases that refer to it more memorable. Finally, cases should have the potential to change behaviour, as any instance of learning can only be deemed effective if this is the case.
Schank's view of formal instruction is
unfavourable. In effect, he argues that concepts and principles can be taught through formal instruction, but the lack of an
incentive (or a distorted incentive such as not having any other choice) coupled with the emphasis on explicit knowledge will often limit its effectiveness. Environments for instructional learning such as
schools have developed with the purpose of transferring a rigidly specified programme of knowledge and then testing the students' recall. This uninspiring approach stems back to the origins of most school and training systems - to
expediently prepare (through conditioning of behaviour as much as through academic instruction) a large number of participants for work in low-skill employment.
This
environment has changed very little over the decades, in terms of
methodology (although in schools the curriculum has gradually broadened and the lengthened). Change has not been rapid, even with the acceptance that learning 'by doing' is more efficient, because of the problems in its perception described above. In addition to this, the alteration of any educational system in such an extensive way as to allow learning 'by doing' would require
time,
resources and
expertise in the short term to set up. The benefits of providing
implicit knowledge to the learner should ideally be seen as outweighing these costs. Not only is implicit knowledge more readily learnt, it is remembered longer, easy to test and provides domain-independent skills that can be applied in future learning situations. Possibly the most serious
accusation against formal instruction is that it fosters an
educator-centric attitude that does not serve the learner's needs as well as could be hoped.