Perception and Attention - Computational Explorations in Cognitive Neuroscience - page 257

a) b) c)

on the third event where the objects overlap in the same
region of space. To see this pattern more precisely, we
can open a graph log of the settling times.

T L T T L

Figure 8.26: The three different MULTI OBJS conditions
illustrated with letters. a) has two different objects in different
locations. b) has the same object in different locations. c) has
two different objects in the same location. This last condition
is clearly the most difficult for object identification.

Do View , GRAPH_LOG to open the log. Then do Run
on the control panel to re-run the 3 events.
This graph log shows the settling times (number of
cycles to reach threshold) for each event. You should
see that the network did indeed have more difficulty
with the objects appearing in the same spatial location
(although sometimes it doesn't take as long — there is
some noise added to the processing and this can activate
the target unit relatively quickly in a minority of cases).
To get a better sense of the overall reaction times in this
simulation, we can run multiple runs and record the re-
sults in text logs.

, !

nition when multiple objects are presented simultane-
ously. Therefore, we will start with a quick exploration
of the network's object recognition capacities as a func-
tion of the spatial distribution of two objects. This will
provide an introduction to the kinds of interactions be-
tween spatial and object processing that can happen us-
ing this relatively simple model. Let's begin by viewing
the events that we will present to the network.

Do View , TEXT_LOG and View , BATCH_TEXT_LOG
to open the logs. Then run a batch of 10 runs by press-
ing Batch . Note that the log column labels (Neutral,
Valid, Invalid) relate to the Posner spatial cuing task, but
they correspond to the three events in this case.

Press View , EVENTS in the control panel.
The environment window shows 3 events (fig-
ure 8.26). The first event has two different objects (fea-
tures) present in different spatial locations. Note that
the target object has slightly higher activation (i.e., it is
more salient), which will result in the reliable selection
of this object over the other. The next event has two
of the same objects (targets) presented in different loca-
tions. Finally, the last event has the two different objects
in the same spatial location. As the figure makes clear,
recognizing objects when they overlap in the same lo-
cation is considerably more difficult than when they ap-
pear in different locations. Although it is clearly easier
to recognize objects if only copies of the same object
are present as opposed to different objects, this differ-
ence is not likely to be significant for small numbers of
presented objects.
Now, let's test these predictions in the model.

Question 8.8 (a) Describe the network's reaction to
each event and explain it in terms of the interactions
between the excitation and inhibition as it flows through
the spatial and object pathways in the network. (b) Re-
port the resulting average settling times in the batch text
log ( Batch_0_TextLog ). Were your original results
representative?

You should have observed that spatial representations
can facilitate the processing of objects by allocating at-
tention to one object over another. The key contrast con-
dition is when both objects lie in the same location, so
that spatial attention can no longer separate them out,
leaving the object pathway to try to process both objects
simultaneously.

Switch back to viewing act in the network window.
Do a Step in the control panel.
This will present the first event to the network, which
will stop settling (i.e., updating the network's activa-
tions a cycle at a time) when the target unit's activation
exceeds the threshold of .6 in the Output layer.

The Posner Spatial Cuing Task

Now, let's see how this model does on the Posner spatial
cuing task.

Set env_type on the overall control panel to
STD_POSNER ( Apply ). Do View , EVENTS to see how
the task is represented.
There are three groups of events shown here, which
correspond to a Neutral cue (no cue), a Valid cue,

Then Step through the remaining events.
You should have seen that the network settled rela-
tively quickly for the first two events, but was slowed