Perception and Attention - Computational Explorations in Cognitive Neuroscience - page 259

Valid Invalid Diff

paired performance on the invalid trials of the Posner
spatial cuing task. Specifically, they are slower when
the cue is presented on the side of space processed by
the intact hemisphere (i.e., ipsilateral to the lesion), and
the target is then processed by the lesioned hemisphere.
The patients showed a 120 ms difference between in-
valid and valid cases, with a validly cued RT of roughly
640 ms, and an invalidly cued RT of 760 ms. These data
can be compared to matched (elderly) control subjects
who showed a roughly 60 ms invalid-valid difference,
with a validly cued RT of 540 ms and an invalidly cued
RT 600 ms (table 8.3).
You should notice that, as one might expect, older
people are slower than the young adult normals, and
older people with some kind of brain damage are still
slower yet, due to generalized effects of the damage.
In this case, we are interested in the specific involve-
ment of the parietal lobes in these attentional phenom-
ena, and so we have to be careful to dissociate the spe-
cific from the generalized effects. To determine if the
patients have a specific attentional problem, we must
first find a way of normalizing the data so that we can
make useful comparisons among these different groups
(including the model). We normalize by dividing the el-
derly control's data by a constant factor to get the same
basic numbers reported for the adult normals (or the
model). If there is a specific effect of the brain dam-
age, we should find that the pattern of reaction times is
different from the adult normals even when it is appro-
priately normalized.
To find the appropriate scaling factors, we use the
ratio between the valid RT's for the different groups.
Ideally, we would want to use the neutral case, which
should be a good measure of the overall slowing, but
only the valid and invalid trial data are available for the
patients and elderly controls. So, to compare the el-
derly controls with the adult normals, we take the adult
valid RT's of 350 ms, and divide that by the elderly con-
trol subjects valid RT's of 540 ms, giving a ratio of .65.
Now, we multiply the elderly controls invalid RT's (600
ms) by this factor, and we should get something close to
the adult normals invalid RT's (390 ms). Indeed, the fit
is perfect - 600 ￿ : 65 = 390 . The elderly controls thus
appear to behave just like the adult normals, but with a
constant slowing factor.

Adult Normal 350 390 40

Elderly Normal 540 600 60

Patients 640 760 120

Elderly normalized (*.65) 350 390 40

Patients normalized (*.55) 350 418 68

Tab le 8 . 3 : Reaction times for the Posner spatial cuing task
for various populations as shown. The normalized scores en-
able the elderly and patient data to be compared with the adult
normals, revealing that the patients have a larger invalid-valid
difference than would be expected just by their overall slow-
ing.

of the cue) to overcome the residual spatial activation in
the cued location. This shows that the spatial pathway
needs to have a balance between sensitivity to bottom-
up inputs and ability to retain a focus of spatial atten-
tion over time. This network allows this balance to
be set separately from that of the influence of the spa-
tial pathway on the object pathway (controlled by the
spat_obj parameter), which is not the case with the
Cohen et al. (1994) model.

Set v1_spat back to 2 before continuing.
One additional manipulation we can make is to the
eccentricity (visual distance) of the cue and target. If
they are presented closer together, then one might ex-
pect to get less of an attentional effect, or even a facil-
itation if the nearby location was partially activated by
the cue.

This can be tested by setting env_type in the con-
trol panel to CLOSE_POSNER ( Apply ). Run this case.
You will see that an overlapping set of spatial rep-
resentations is activated, and a Batch run reveals that
there is no longer a significant slowing for the invalid
case relative to the neutral case (but the validity effect
remains).

Switch the env_type back to STD_POSNER before
continuing.

, !

Effects of Spatial Pathway Lesions

As we mentioned earlier, Posner et al. (1984) showed
that patients who had suffered lesions in one hemi-
sphere of the parietal cortex exhibit differentially im-