General Instructional Consequences

General Instructional Consequences
Instruction is only necessary toward the unlearned end of the learning
continuum of the cognitive matrix of continua (Fig. 1), and one of its primary
functions is to provide a partial substitute for the missing central executive at
this end of the continuum. Consider again someone wishing to learn the road
route frompointAto point B. They can have someone explain the route, use a
road map, or use a combination of prior knowledge with a problem-solving
search to fill in the gaps. These activities function as a central executive in
244 John SwellerdiVerent ways and have diVerent instructional consequences. Both an
explanation and a map are two diVerent forms of direct instruction, whereas
problem solving provides an example of exploratory learning.
An explanation provides a strong substitute for a cognitive central
executive. As one would expect from a central executive, it provides an
overarching set of instructions for the critical processes that must be taken.
Furthermore, the instructions can be followed with a minimum of
additional learning, such as learning to use a map. If the explanations are
adequate, all random processes are eliminated because the explanation, as a
central executive, tells the learner precisely what needs to be done. Once a
road route is learned, the learner moves to the right side of the matrix of
continua, and the schemas acquired take over from the explanation and act
as the central executive, rendering an explanation redundant.
A map, while it also acts as a substitute for a cognitive central executive,
requires more intermediate learning than an explanation before it can be
used. People need to learn to use a map before they can use it to learn a
particular route. Thus, learning to use a map has its own set of learning
continua, and until a person has acquired the map-reading schemas that
allow movement to the right side of the matrix of continua for map reading,
learning a route by using a map will be diYcult or even impossible.
Nevertheless, if map-reading skills have been acquired, a map, like
explanations, can provide a powerful central executive substitute. Used
properly the need to consider the consequences of random actions can be
totally obviated and can continue to be avoided until the schema-based
central executive on the right of the matrix of continua takes over the
executive functions.
Problem solving provides the least eVective substitute for a cognitive
central executive. There is no choice but to propose actions randomly and
then use the environment or prior knowledge to test the eVectiveness of
those actions as far as they can be tested. The learner is likely to move to the
right of the matrix of continua very slowly, and so for much of the learning
process, there is no eVective central executive function. Only toward the end
of the learning process, when schemas have been acquired, is an eVective
central executive available. Using this reasoning, problem solving may be
considered as a last resort instructional technique when other more direct
forms of instruction are unavailable.
The inadequate central executive function provided by problem solving
has other ramifications. Combining elements randomly and testing the
eVectiveness of combinations against reality require substantial working
memory resources (Sweller, 1988). The activity imposes a heavy working
memory load just at the point where working memory resources are at their
weakest because problem-solving search occurs at the new, yet-to-be-learned
Evolution of human cognitive architecture 245end of the learning continuum where working memory limitations are
relevant. The heavy working memory load associated with problem solving
can interfere with learning. Direct, fully guided instruction alternatives to
problem solving circumvent both the lack of a central executive and the
heavy cognitive load associated with search. On this analysis, direct guided
instruction, rather than problem solving, should be used as a means of
acquiring schemas. Substantial empirical evidence exists for this suggestion
(see Sweller, 1999; Sweller, van Merrienboer, & Paas, 1998; Tuovinen &
Sweller, 1999).
The contrast between direct guided instruction and exploration applies to
all material that needs to be learned, including material covered in
educational institutions. Learning to solve classes of mathematical
problems, write essays in history, run scientific experiments, or learning to
read and write must all be aVected without an adequate cognitive central
executive provided by schemas. Showing students how to solve mathemat-
ical problems, write particular types of essays, run experiments, or providing
direct instruction in how to read and write can all provide an eVective
central executive substitute and reduce the cognitive load associated with
problem solving, although care must be taken to ensure that the instruction
itself does not impose a heavy working memory load (e.g., Sweller,
Chandler, Tierney, & Cooper, 1990; Sweller, Mawer, & Ward, 1983). In all
cases, direct guided instruction can provide a temporary replacement for
schemas until they are acquired.
Indirect instruction provided by various discovery/exploratory techniques
oVers a less eVective central executive substitute with an inevitably high
random component. Direct guided instruction is eVective because it
reduces the number of random element combinations that must be tested.
It is likely to be essential for very high element interactivity material
for which the number of random combinations that must be tested will
be unacceptably high. The knowledge that lies behind such material could
only be derived by scholars engaged in the very lengthy, working memory-
taxing activities inevitably required when dealing with a multitude of
interacting elements that are not appropriately organized by a central
executive. Such problem-solving activity is unavoidable when neither
schemas nor direct instruction are available to act as a central executive
that indicates appropriate relations between elements. Humans learn
through problem solving not because it is eVective (empirical evidence
indicates unambiguously that it is not eVective as a learning device, see
Sweller, 1999; Sweller et al., 1998) but because they are forced to by the
environment and the lack of a central executive. Direct guided instruction
acts as a substitute for a central executive and should always be used if
available.
246 John SwellerB. Creativity
Creativity has always been a diYcult concept to deal with or even to define.
Nevertheless, most definitions of creativity incorporate the generation of
new ideas and, under such definitions, it is easy to assume that the general
instructional consequences discussed in the previous section leave no room
for human creativity or may even stifle creativity. In fact, the common
information processing structures of human cognitive architecture and
evolution by natural selection can provide a solution to the problem of
human creativity.
Evolution by natural selection has created innumerable functions,
procedures, and outcomes that vastly exceed the capability of human
cognition. We are not only unable to create what evolution by natural
selection has created, to this point we are unable to even understand many
of the products of evolution, with massive scientific enterprises being
devoted to precisely this cause. Given the much shorter time frame in which
human cognitive activity operates, it is not surprising that our creative
endeavors are unable to match those of evolution by natural selection.
Nevertheless, humans are and have been creative and that creativity can be
explained by the current theoretical framework. Based on the perspective of
this chapter, human creativity and the creativity exhibited by evolution by
natural selection are generated by the same mechanisms. Those mechanisms
are reflected on the left side of the matrices of continua. A knowledge base in
long-term memory or as part of a genetic code may become inadequate and
is altered by random processes; the knowledge base requires procedures for
testing the eVectiveness of alterations and only incorporating those that are
eVective; and the knowledge base must have mechanisms to protect it from
large random alterations that may destroy it. Using these mechanisms, both
evolution by natural selection and human cognition have been able to create
new and eVective structures.
It needs to be noted that on this analysis, random processes provide the
initial impetus for human creativity just as random mutation is critical for
the creativity of evolution by natural selection. There is no central executive
determining what is creative (left-hand side of the second continuum of
Figs. 1 and 3). Nevertheless, despite the initiating random processes,
creativity is critically determined by the current knowledge base, as it is
from that base that new creative actions are taken, just as it is the
information encapsulated in a genome from which random mutations can
determine new biological procedures and functions (fourth continuum of
Figs. 1 and 3).
Langley, Simon, Bradshaw, and Zytkow (1987) also suggested that
creativity depends on an appropriate knowledge base associated with
Evolution of human cognitive architecture 247conventional problem-solving search mechanisms. Some evidence for the
validity of their proposal comes from a production system that they
constructed that rediscovered some of the early laws of physics. That system
only had the knowledge base required to generate particular laws and so has
not been able to discover new scientific laws. If the theoretical suggestions
made in the current chapter are valid, no computational system is likely to
discover, as opposed to rediscover, new scientific laws unless it incorporates a
massive knowledge base with the mechanisms for small random alterations
of that base over long periods of time along with procedures for testing the
eVectiveness of those alterations. Such a system is currently not available.
Suggested procedures for ‘‘teaching’’ creativity arise periodically in both
psychology and education. None of these attempts has been able to obtain
widespread, empirical support. The current proposals imply that teaching
creativity is likely to be diYcult or impossible but that humans may no more
need to be taught how to ‘‘explore,’’ ‘‘investigate,’’ ‘‘discover,’’ or ‘‘create’’
than does evolution by natural selection. Only a knowledge base can be
taught and learned and that knowledge base will determine what can and
cannot be created.
It is, of course, possible that life on earth includes multiple mechanisms
that have creativity as one of their end results and that the creativity
exhibited by evolution by natural selection and by humans uses diVerent
mechanisms. Nevertheless, the thesis outlined in this chapter suggests a
single rather than multiple mechanism