Information Technology Reference
In-Depth Information
Stage
Augend1 Augend2 +
Stage
Augend1 Augend2 +
Addend1 Addend2
Addend1 Addend2
=
=
module
module
BA
24
32
BA
24
32
goal
goal
visual
visual
37
7
39
40
45
46
37
7
39
40
45
46
imaginal
imaginal
declarative
declarative
procedural
procedural
-
-
Fig. 6.
Results of ACT-R Simulation for on-task Level 2 (with carry)
Figure 6 shows the ACT-R simulation result of two digits addition problem for
on-task Level 2 (with carry). The vertical axis denotes the modules related to the
computation activities in ACT-R, and the horizontal axis denotes the time-series
of the computation activities.
The modules are fired when computing are declarative, imaginal, and procedu-
ral ones. It is suggested that the region BA (Brodmann Area) 45, BA 46, BA 7,
BA 39, BA 40 and basal ganglia corresponding to these modules deeply contribute
to such a computation related problem solving process. When augend, operator,
and addend are presented, the displayed position tracking to the number and the
attention shift are reflected and the visual module is activated. Moreover, to main-
tain the recognized number as a mental representation, the imaginal module is
fired. It is suggested that the addition knowledge is referred to the declarative
module when the equal sign is presented, and the declarative memory (semantic
memory) contributes to the problem solving. When a sum is presented, both the
manual module and the motor area are remarkable. The goal module acts through
the whole, and the process that arrives at the state of the goal while managing a
small step is suggested. Moreover, the procedural module that plays the role to
control the activity of all modules also acts through the whole.
4.4
A Comparative Study of Results
We focus on deeply investigating the delay problem that is caused by the difference
between the addition operation without carry (Level 1) and with carry (Level 2).
In the ERP analysis, it can be confirmed that, compared with Level 1, there exists
a delay about 25 msec in Level 2. However, it is dicult for a more detailed analysis
since the inconsistency of time and individual difference of subjects. For this reason,
we individually investigated the subjects who can be classified into 2 types: good
and not good at computation. The results shows that the delay is hardly seen in the
subjects who are good at computation, but the delay about 50 msec to 100 msec
appears in the subjects who are not good at computation.
The output of an ACT-R simulation is a time course for when and how long
the activations of each module involved in the task. Based on it, one can predict
the performance of the subjects (such as reaction time and accuracy) as a general
model. Therefore, we can use such a general model as a basis and combine with
ERP/topography to analyse individual difference of subjects.
