The role of dynamic representations in understanding semantic impairments: Investigations from connectionist neuropsychology. - Page 140 |
Save page Remove page | Previous | 140 of 184 | Next |
|
small (250x250 max)
medium (500x500 max)
Large (1000x1000 max)
Extra Large
large ( > 500x500)
Full Resolution
All (PDF)
|
This page
All
|
124
superordinate question, on the other hand, is possible with any of a set of features (Caramazza, Hillis, Rapp,
& Romani, 1990; Rapp & Caramazza, 1989). For example, knowing only that an object <has wings> and
<has feathers> is sufficient to answer, “Is this a bird or a vehicle?” It is not, however, enough to determine
which bird it is nor whether it has talons or webbed feet. Thus superordinate questions are actually easier
than attribute questions because they require less precise information to answer. One consequence of this is
that any damage which affects semantic memory would be expected to lead to a greater difficulty with
attribute over category membership.
Notwithstanding the differences in task difficulty between the two types of questions, the results
from the picture naming and knowledge probe tasks in the previous chapter illustrate that damage
preferentially affected attribute information relative to superordinate information. The modeling subjects
were frequently only able to produce the category name for an item following semantic damage. Similarly,
their knowledge of individual attributes was strongly compromised. Thus the subjects demonstrated the
expected dissociation between superordinate and attribute information. Because shared features within a
category were highly intercorrelated, they provided strong mutual support which made them more robust to
damage than individual features (cf. Tippett, McAuliffe, & Farah, 1995). As a result, superordinate
information was relatively spared compared to attribute information.
“Rate of presentation” has similar difficulties. Only two subjects have been tested using varying
presentation speeds, VER (Warrington & McCarthey, 1983) and YOT (Warrington & McCarthy, 1987).
Both of these patients were considered “access” patients based on their inconsistency and the fact that their
performance improved with longer intervals between trials, even though VER showed a significant effect of
item frequency.6 According to Warrington and McCarthy, pathologically enhanced “refractoriness” led to
representations becoming inaccessible for longer than normal and consequently given more time, the
subject was able to overcome this refractoriness and eventually access the semantic representation. Of
course, if the representation itself was lost due to damage, no amount o f extra time would suffice to access
it. This assumption was essentially arbitrary, however, and different assumptions lead to different
5 Neither VER nor YOT was tested for priming/cueing effects nor depth of processing effects. The authors
did not look for an effect of frequency in YOT.
R ep rod uced with perm ission of the copyright ow ner. Further reproduction prohibited without perm ission.
Object Description
Description
| Title | The role of dynamic representations in understanding semantic impairments: Investigations from connectionist neuropsychology. - Page 140 |
| Repository email | cisadmin@lib.usc.edu |
| Full text |
124
superordinate question, on the other hand, is possible with any of a set of features (Caramazza, Hillis, Rapp,
& Romani, 1990; Rapp & Caramazza, 1989). For example, knowing only that an object |
