School Didactics And Learning: a school Didactic Model Framing An Analysis of Pedagogical Implication of Learning Theory

recognize
his/her relevant ideas and beliefs, then
evaluate
these ideas and beliefs in terms of what is to be learned and how this learning is intended to
occur, and then 
decide
whether or not to 
reconstruct
their ideas and beliefs.
Metacognition.
The view that learning is highly cognitive and occurs within the individual is also evident
from the interest that has developed in meta-cognitive activities, i.e. in higher-level processes in learning.
This interest is primarily focused on individual regulation, planning, predicting and monitoring one’s own
learning process.
3
The question is here how an individual organizes their activities in connection with the
goal that an individual has in the learning process (cf. e.g. Brown, 1978; Flavell, 1979). Another aspect of
this interest is concerned with what an individual does know about the field that they are trying to enter into
by learning. We will return to this second question in relation to the question of the role of prior knowledge.
Sternberg’s (1987) theory exemplifies this metacognitive interest very well. He differentiates between
nine executive processes or metacomponents that regulate learning activity. The metacomponents are:
“higher order or executive processes used to plan what one is going to do, to monitor it while being done
and evaluate it after it is done”. These metacomponents then regulate three so-called lower-level
performance processes. The lower-level knowledge-acquisition components are: selective encoding of
information (i.e. paying attention only to relevant information and neglecting what is irrelevant), selective
combination (i.e. integrating disparate pieces of new information in a meaningful way) and finally selective
comparison (i.e. relating this previously encoded information to old, stored information in the memory).
Gagné and Briggs’ (1979) position clearly exemplifies the emphasis laid on cognitive processes
explaining learning. They distinguish between the following kinds of processes “presumed to occur during
any kind of learning” (p. 154):
1.
Attention
—determines the extent and nature of reception of incoming stimulation; 
2.
Selective perception
—transforms this stimulation into the form of object features, for storage in short-
term memory;
3.
Rehearsal
—maintains and renews the items stored in short-term memory;
108
SCHOOL DIDACTICS AND LEARNING

---

4.
Semantic encoding
—the process which prepares information for long-term storage;
5.
Retrieval,
including search—returns stored information to the working memory, or to response
generator mechanism;
6.
Response organization
—selects and organizes performance;
7.
Feedback
—an external event which sets in motion the process of 
reinforcement;
8.
Executive control processes
—select and activate cognitive strategies; these modify any or all of the
previously listed processes.
Brief Overview of Cognitivist Approaches to Learning
We should now look at the theories that have been current for the last thirty years. This discussion gives the
general characteristics of some theories while others are dealt with in more detail.
We may begin by acknowledging the work done by Jerome Bruner (1960, 1966). In the present context
he is recognized as one of the important background figures of cognitive learning theory by drawing
attention to, and developing the view of, the learner as an active, selective and organizing individual. Like
most other western psychologists in the 50s, he was inspired by information theory and the development of
computers, which clearly affected his theories of learning at that time. Learning was discussed in terms of
discovery, invention and “going beyond the information given” (Bruner, 1960). Later, in the 70s and 80s, he
emphasized language much more as a cultural phenomenon and stressed that learning in some respects was
to be conceived as social by nature. Negotiating and sharing in relation to cultural contexts were frequently
occurring concepts. In this view learning is not seen solely as a receptive individual function through which
the individual receives information or knowledge about the surrounding world. To learn, according to
Bruner and Haste (1987), is also to participate in the construction of the social world and to take part in the
creation of a common culture.
Another early cognitive theory of learning was Ausubel’s (1963) model of meaningful verbal learning. It
relies heavily on the schema-concept. The cognitive structure in Ausubel’s theory consisted as a model
organizing perceived information; in learning, information was subsumed in an existing cognitive structure.
Another aspect stressed in Ausubel’s theory was that of conceiving learning as a result of the individual
facing information that differed, or was distinguishable from, existing individual cognitive structures.
Ausubel stresses the relation between the logical structure of information and the individual’s psychological
organization of the world. For Ausubel, as for Bruner, the common concepts in language form the
fundamental basis required for successful learning. Language and concepts are therefore required in
building the cognitive bridge between psychological and logical reality (Ausubel, Novak, & Hanesian, 1978).
A third earlier theory is Wittrock’s (1974) theory of generative learning. The notions of rule and
inference were central to the generative model. Learning was conceived of as generating (constructing)
connections between pieces of information in the long-term memory by trying to discover underlying
relationships and rules pertaining to the information. Thus learning occurs through suggesting connections
between potential or possible relationships and then testing these assumptions.
Most approaches to learning view the concept of de-contextualization or generalization as central to their
theory; knowledge that is originally acquired in a specific context must become more abstract so that it may
be used in other situations. Today, when learning is largely understood as contextual in nature, it is
important to note that the demand concerning the transfer of competence from one context to another has not
disappeared (Singley & Anderson, 1989). The question of how competence, skills and knowledge are de-
contextualized in order to be more generally applied is still an urgent one for all theories of learning, but is
not well understood. Even though the process of de-contextualization or generalization of knowledge is
5. OBJECT OF ANALYSIS
109

---

taken to refer to extending its range of application to other situations, it is not always specified what this
generalization process contains or how it is realized. On the other hand, one might expect contextual theories
on cognition to be better suited than others for dealing with the relation between context and competence.
How are the learning processes then defined? Previously we looked at Sternberg’s (1987) theory, in
which metacognitive activities are stressed. In Norman’s (1982) and Rumelhart and Norman’s (1987)
schema-based theory these metacognitive activities are combined with the idea of longterm memory. They
suggest three kinds of learning:
• Accretion—i.e. encoding new information in existing schemata;
• Restructuring—i.e. creation of new schemata;
• Tuning—i.e. the modification of a schema as a result of using it in different situations.
Accretion means that new information is interpreted in terms of an already existing schema. After this the
actual piece of information is added to the long-term memory without changes in its structure. Shuell (1986,
p. 421) has compared this process to memorization. Restructuring again refers to a change in knowledge in
terms of a changed structure. New structures are constructed in interpreting both old and new information.
However, for restructuring to occur, there is no need for new information; this process may just change the
relations between existing pieces of information to form new patterns. Tuning has been characterized as a
series of accretions (or accommodations), (Hergenhahn & Olson, 1993, p. 366). Norman (1982, p. 81)
writes:
Tuning is the fine adjustment of knowledge to a task. The proper schemas exist and appropriate
knowledge is within them. But they are inefficient for the purpose, either because they are too general
or because they are mismatched to the particular use that is required of them, so the knowledge must
be tuned, continually adjusted to the task.
Brown and Van Lehn (1980), who have developed their theory primarily within arithmetical learning, claim
that learning occurs at impasses. They claim that when an individual reaches a certain situation which
cannot be handled by existing procedures, then they must invoke new procedures in order to deal with the
problem, must invent repairs or patches. The heart of this theory is that a student is thought to use general-
purpose problem-solving heuristics to guide repairs. However, what is counted as general heuristics and
domain-specific strategies depends on the problem at hand.
Another theory that is clearly an information-processing theory is John Anderson’s “Adaptive Control of
Thought” (ACT), (Anderson, 1983). This theory to a large extent makes use of semantic networks to
represent declarative knowledge and production rules to represent procedural knowledge. His three-stage
model of procedural learning involves the following steps: first a subject interprets and solves a problem by
using existing declarative knowledge, secondly, in a knowledge compilation stage, declarative knowledge is
converted into procedural knowledge. The third step or phase involves tuning. In the tuning stage,
procedural knowledge is refined by generalization, discrimination and strengthening. These are processes
where production rules become broader or narrower in applicability or whereby some rules are strengthened
or weakened.
Anderson’s theory makes use of both the idea of semantic networks and productions (If… Then rules).
According to the semantic (declarative, associative) theory of network, concepts are represented by nodes
(pieces of information). These are linked to each other on associative principles, i.e. because the concepts
are similar, because they occur together or because they are in contrast or opposition to each other. On the
110
SCHOOL DIDACTICS AND LEARNING

---

basis of these associative principles, combinations of concepts form networks. It should be noticed that this
view has very long traditions especially in British empiricist philosophy. If knowledge is seen in terms of
associations between concepts, what then is learning? In traditional semantic network theory it is thought
that concepts (or nodes) are connected with each other with various strength in relation to their similarities,
i.e. similar concepts are connected with each other more firmly than more distant concepts, e.g. “a dog and a
cat node may be connected by a link with an activation of 0.5 whereas a dog and a pencil may be connected
by a link with a strength of 0.1” (Eysenck & Keane, 1991, p. 17). Thus, “in learning that two concepts are
similar, the activation of a link between them may be increased” (ibid.). In Anderson’s ACT this way of
reasoning about learning is combined with productions. In principle the point is that a production system
operates by matching the information stored in the working memory with If-parts of the If-Then rules in the
long-term memory. Suitable consequences (i.e. then-parts) are executed. If there are many If-parts that
match the information, then new rules (conflict-resolution rules) select the rule that best fits the information
as discussed in Anderson (1982, pp. 249–250):
One of the fundamental assumptions of cognitive learning theory is that new knowledge is in large
part “constructed” by the learner. Learners do not simply add new information to their store of
knowledge. Instead, they must connect the new information to already established knowledge
structures and construct new relationships among those structures. This process of building new
relationships is essential to learning.
An important difference from many other cognitivist theories lies in the assumption that only this process is
involved in all types of learning. It is thus a content-neutral theory of learning. In Anderson’s theory,
learning is seen as a constructive process.
Laird, Newell, and Rosenbloom (1987) have, in one sense, developed Anderson’s ACT in their so-called
SOAR, a problem-solving system making use of production rules. Like Brown and Van Lehn’s (1980)
approach, it works on impasses in problem solving. Encountering a too difficult problem, subgoals are set
up; if a solution is successful, it is stored as an operator that will possibly be used in the future. In this sense
SOAR falls between ACT and impasse-learning theory. From a theoretical perspective, chunking is a crucial
concept in SOAR architecture. By chunking, those representations and procedures which occur together can
be accessed together. Chunking of information makes it possible to overcome resource limitations of the
information processing system.