Notes on: Pask, G. and Scott, B.  (1972) 'Learning Strategies and Individual Competence'.  In International Journal of Man - Machine Studies, 4: 217-53.

Dave Harris

Teaching styles vary and so do teaching materials.  Students need cognitive competence, if necessary to alter their styles or to structure the materials.  It is more effective when these are matched than if they are mismatched.  Pask and Scott used a particular technique called TEACHBACK rather than relying on the usual diagnostic tests.  The idea was to test the structuring principles available for text and to identify needs for matching.  There are only two or three types of student competence, so matching is feasible.

Serialist learners operate with strings [of items] where items are related by 'simple data links', and 'low - order relations', and are intolerant of irrelevance.  Holists attempt to remember the sequences as a whole and work with 'high order relations'.  There are two subtypes of holists—redundant and irredundant.  The latter prefers material which contains only the most relevant and essential constituents, while the former can work with material that is logically irrelevant or overspecific, as in the idea of data being used to enrich the curriculum, as in the inclusion of case studies.  These styles allow learners to 'access, retain and manipulate 'the material (218-9).  The styles have emerged from empirical studies but also from Pask's general theory, and they have been modified through rigorous testing.

Experiments turn on a laboratory exercise in mastering an abstract zoological taxonomy [see below] . Students were first allowed to free learn or browse the material, and then interrogated more systematically, giving the reasons for what they were doing.  Approaches were then coded, as, for example 'search a coordinate of the message space', or 'test an hypothesis about a simple predicate' (220).  The TEACHBACK procedure was based on a content analysis of what was said [using categories apparently provided by Schneidmann 1966—unknown to me, but see below], and the results used to assign students to a particular learning style.  This can be automated.  Matching was then tested by assigning identified students at random. All students' factual knowledge and ability to generalize were then tested using a '30 item questionnaire'.  In a later addition, students were invited to demonstrate concept reproduction, to measure the effects on retention [developed in another article].  Concept reproduction was not tapped by conventional tests, but this procedure examined it: concept reproduction help students to acquire concepts any way, and is a good technique because 'the object of education is to teach somehow'(222).

The problems of the pilot study included over-reliance on one sort of subject matter—taxonomies.  To examine whether or not the learning styles fitted other material, additional tests of learning a biological concept were administered, and results so far indicate that matched students do better with that task as well (224).  The pilot study also indicated that some students changed strategy.  This was quite common, but not investigated on this occasion [often the case -- so what exactly is the status of the original fixed types?].

We can see free learning followed by TEACHBACK as providing a skeleton for a 'metalanguage' where a student externalizes and discusses the way he [sic] learns.  Natural language will need to be formalized first, but this seems possible (226) [ in the laboratory].  The result could be a human 'questioning language' or a 'conversational' computer language: the semantic structures for both would be 'virtually identical' (226).  The former human language has already been formalized in the initial experiment to learn taxonomies, and it should also be possible to extend the work to 'large parts of chemistry, biology, and the other sciences; most of the abstract disciplines, and some parts of history and the social sciences' (226) [only after 'normalizing' their specific languages, of course. Classic technological optimism and gee-whizzery. The great day never quite dawns of course].  The approach can be used as a tutorial aid and as a 'vehicle for reproducing the subject matter structure' (226). A further possibility is  being able to measure the information content 'over expressions in this [special] language' [presumably, compared to the conventional language], and eventually, to 'compare two or more learning situations, for instance, in terms of their relative difficulty' (227).

Details of further experiments follow 227f [and I was reminded of Adorno's point about how laboratory experiments are used to simplify and distort material in order to make the object fit the concept].  As an example, we can choose a task which is learnable according to either serialist or holist strategies.  The examples based on particular analytic schemes [already rationalized in other words] include learning an abstract taxonomy of 'Martian' fauna, systematic biological concepts, the human menstrual cycle as a system.  The subjects for the experiment were students from the Kingston College of Technology.  Early attempts to predict learning style were made, relying on the subjects''personality and occupation' (232).  Further operationalisms were evident, defining holists for example as those who asked to test hypotheses 'based on complex predicates rather than simple ones' (234), where a simple predicate was seen as a single one [there is already a dangerous circularity here, where students are tested for particular learning styles, but the tests themselves predefine them].

There is no simple preference for particular styles here [there certainly is if you compare holists to those embracing the classic bourgeois taste for abstraction and formalism in Bourdieu].  Holists are prone to overgeneralize, while serialists get too much unrelated information and cannot easily reproduce the structure.  The irredundant holist can have these problems corrected by a suitable design of material, but redundant holists and serialists are most commonly catered for by existing educational materials.

Details of the categories of content analysis used in TEACHBACK are then listed, and they include using 'information from the teaching programme'; the use of irrelevant information or redundant information; whether statements referring to information are to be presented or have been previously presented; whether information is inferred from the teaching programme; whether there are feedback statements; whether there are cross references of false statements 'with a correction, repetition, other concepts and interjections' (244).  Overall, the data from TEACHBACK does show distinct differences between serialists and holists in terms of:  the complexity of the predicates they use, the number of inferential statements they make, and whether or not the order of preservation of concepts in the programme is preserved.  Students who had been matched learned with fewer repetitions of the programmes.

back to Education Studies