zontal line) in the function relating decision speed to the number of items in the
memory set and/or in the visual display when automatic processes are used. In
fact, as can be seen in figure 15.8, decision speed was slower when the memory
set and the visual display both contained several items.
The greatest weakness of Shiffrin and Schneider’s approach is that it is de-
scriptive rather than explanatory. The claim that some processes become auto-
matic with practice is uninformative about what is actually happening. Practice
may simply lead to a speeding up of the processes involved in performing a
task, or it may lead to a dramatic change in the nature of the processes them-
selves. Cheng (1985) used the term ‘‘restructuring’’ to refer to the latter state of
affairs. For example, if you are asked to add ten twos, you could do this in a
rather laborious way by adding two and two, and then two to four, and so on.
Alternatively, you could short-circuit the whole process by simply multiplying
ten by two. The crucial point is that simply discovering that practice leads to
automaticity does not make it clear whether the same processes are being per-
formed more efficiently or whether entirely new processes are being used.
Cheng (1985) argued that most of Shiffrin and Schneider’s findings on auto-
maticity were actually based on restructuring. More specifically, she claimed
that subjects in the consistent mapping conditions did not really search sys-
tematically through the memory set and the visual display looking for a match.
If, for example, they knew that any consonant in the visual display had to be
an item from the memory set, then they could simply scan the visual display
looking for a consonant without any regard to which consonants were actually
in the memory set.
Schneider and Shiffrin (1985) admitted that some of their earlier findings
could be accounted for by assuming that subjects in consistent mapping con-
ditions made use of knowledge about the categories being used. However, they
pointed that other findings could not be explained in terms of restructuring.
For example, the finding that subjects could not ignore part of the visual dis-
play after automatic processes had been acquired does not lend itself to a re-
structuring explanation.
Norman and Shallice’s Theory
Norman and Shallice (1986) discussed a theory taking account of the distinction
between fully automatic and partially automatic processes. Instead of the usual
distinction between automatic and attentional or controlled processes, they
identified three different levels of functioning:
.Fully automatic processing controlled by schemas (organised plans).
.Partially automatic processing involving contention scheduling without
deliberate direction or conscious control; contention scheduling is used to
resolve conflicts among schemas..Deliberate control by a supervisory attentional system.
According to Norman and Shallice (1986), fully automatic processes occur
with very little conscious awareness of the processes involved. Such auto-
matic processes would frequently disrupt behaviour if left entirely to their own
devices. As a consequence, there is an automatic conflict resolution process
Attention and Performance Limitations 387