Skill acquisition: estimating the contribution of old skills to performance on new tasks

Craig Speelman

Research output: ThesisDoctoral Thesis

124 Downloads (Pure)


[Truncated] There were two main aims of the research described in this thesis. The first was to investigate the determinants of transfer of training. The second was to examine the effects of transfer on the shape of learning functions. In parallel with these aims was the aim to evaluate and extend certain features of
Anderson's ACT* theory of skill acquisition.

Part 1 of the thesis was concerned with the determinants of transfer. This
issue was investigated by comparing two accounts of a transfer phenomenon
called the contextual interference effect. In brief, this effect typically involves
comparing performance in two presentation conditions: (1) random
presentation, where various tasks are performed in a random order; and (2)
blocked presentation, where a block of trials of one task are performed
together, followed by a block of another task, and so on. Performance in
blocked training is usually faster than during random training. However, this
difference is usually reversed when presentation is switched to random only.
One account of this effect is provided by the ACT* theory. This theory holds
that skilled performance is based on hierarchies of condition-action pairs
called productions. These productions involve a condition that needs to be
satisfied before a particular action can proceed. The ACT* theory predicts that
developing a set of productions to perform one task will benefit performance
on another task to the extent that the two tasks require the same productions.
In other words, transfer of training is a function of the number of "old"
productions that are useful in a new situation. The ACT* account of the
contextual interference effect basically suggests that blocked training does not
encourage development of productions which are required with random
transfer items. This account of the effect was contrasted with one termed the
intratrial processing account. This account suggests that the important feature of random training which provides the advantage during transfer is that
subjects in this condition have extensive practice at loading new solution
methods into working memory. This practice is not provided with blocked
training but is important for efficient performance with random transfer items.
These two accounts of the contextual interference effect were compared in
Experiments 1, 2 and 3. Subjects were given extensive practice with
syllogistic reasoning problems that varied with respect to several presentation
features. These features were shown to affect the type of processing strategies
that subjects adopted to solve the problems. In turn, the various strategies
were shown to vary in their appropriateness for problems with different
presentation features. The occurrence of these strategy differences was
accounted for by the ACT* theory but not the intratrial processing account.
The ACT* account was not only able to predict these differences during
training, but was able to relate these to performance differences during
transfer on the basis of the extent to which the various strategies shared
identical productions. Thus it was concluded that the ACT* theory provided a
superior account of the contextual interference effect compared to the one
provided by the intratrial processing account.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • The University of Western Australia
Publication statusUnpublished - 1991

Take-down notice

  • This thesis has been made available in the UWA Profiles and Research Repository as part of a UWA Library project to digitise and make available theses completed before 2003. If you are the author of this thesis and would like it removed from the UWA Profiles and Research Repository, please contact [email protected]


Dive into the research topics of 'Skill acquisition: estimating the contribution of old skills to performance on new tasks'. Together they form a unique fingerprint.

Cite this