The Influence of Feature Type, Feature Structure and Psycholinguistic Parameters on the Naming Performance of Semantic Dementia and Alzheimer

1
The Influence of Feature Type, Feature Structure
and Psycholinguistic Parameters on the Naming
Performance of Semantic Dementia and Alzheimers
Patients.
Krist Noonan Peter Garrard
Department of Psychology, University College
London, UK
Analysis 2 Predicting Group Individual
Patient Naming
Introduction
METHOD
How is conceptual knowledge organized?

• Individual and group regression analyses were
  conducted to predict individual item naming.
• Including feature level predictor variables
  (relevance, dominance, distinctiveness feature
  intercorrelation ), psycholinguistic variables
  (familiarity, word frequency
• age of acquisition).

• The ability to understand and interpret certain
  categories of conceptual knowledge can be lost
  following focal or degenerative brain damage.
• Living kinds are most frequently affected with
  non-living concepts often remaining relatively
  intact. This phenomena is know as category
  specific semantic impairment.
• The underlying principles of semantic memory are
  often disputed different accounts exist based on
  feature type, feature structure and
  psycholinguistic parameters. No current studies
  have investigated the predictive power of these
  theories on a single set of patients

Three Explanations of Category Specific Semantic
Impairments

• 1 Knowledge for living concepts is
  differentially reliant on perceptual information.
  Nonliving on functional information e.g. what a
  concept is used for (Fig 1, Analyses 1).
• 2 Living concepts have more features which are
  shared and intercorrelated across items.
  Nonliving have more distinctive features.
  Patterns of knowledge impairment change as
  feature structure interacts with disease
  progression (Fig 2, Analyses 3).
• 3 Semantic Relevance is the organizing
  principle of semantic memory. A feature is high
  in relevance when its consistently used across
  individuals to identify a concept (dominance) and
  distinguishes that concept from other exemplars
  (distinctiveness) (Fig 3, Analyses 2).

RESULTS Which variables influenced naming?
CONCLUSIONS
No category advantage remained when
psycholinguistic variables were accounted for.
Semantic Relevance Theory is a poor predictor
of successful naming. Better performance based
on functional Knowledge may be related to areas
of temporal lobe atrophy seen in SD AD.

• Descriptions emphasizing functional knowledge.
• Psycholinguistic variables, especially
  familiarity.
• Semantic Relevance for SD group, although
  largely a result of the influence of patient VH.

Analysis 3 Feature Structure and Disease
Severity
PREDICTIONS METHOD

• Non-Living concepts should be consistently
  predicted by feature distinctiveness (Devlin et
  al, 1998).
• Living concepts should be predicted by the
  interaction between the proportion of shared and
  intercorrelated features in the initial stages of
  dementia (Devlin et al, 1998).
• Tested using individual regressions for each
  patient on separate living and non-living item
  subsets.

• Variables included (dominance, distinctiveness,
  feature intercorrelation, intercorrelation
  distinctiveness relevance).

RESULTS
KH RB only patients to have living concepts
predicted by shared / intercorrelated
features. No patient showed non-living
performance to be predicted by distinctiveness
Analysis 1 Performance Across Domain Modality
CONCLUSION
METHOD

• The conceptual structure approach could not
  account for the majority of the patients naming
  performance.

Patients regression predictors plotted on a
stylised representation of Devlin et als (1998)
concept loss curves.

• 5 Semantic Dementia (SD) 5 Alzheimers (AD)
  patients were tested on a naming to description
  task.
• 58 living 64 non-living concepts. Each concept
  had two descriptions one emphasizing perceptual,
  the other emphasizing functional information

Summary and Discussion
RESULTS (accuracy across domain modality of
description)

1. All three of the feature level theories
   considered proved insufficient to explain the
   patterns of patient performance.
2. In contrast psycholinguistic variables
   consistently predicted naming performance across
   individual patients and both dementia groups.
   These findings indicate that concepts which are
   frequently encountered and acquired at an early
   age are more resistant to loss.
3. Descriptions composed of functional attribute
   knowledge provided an advantage for naming in
   many of the patients studied. It is proposed
   that this may result from the progressive damage
   to the inferior lateral and medial temporal lobes
   seen in SD and AD respectively. These brain
   regions are often associated with high-level
   visual knowledge and this may account for the
   patients impairments on naming concepts from
   perceptual (often visual) attribute knowledge.

Living vs Non-Living Perceptual vs
Functional

• Two patients show a category advantage for
  nonliving concepts.
• Only one of the two shows an advantage for
  functional descriptions.
• Five patients show better performance for
  functional descriptions but no accompanying
  advantage for non-living concepts.

() () ()
() () () () () () () () ()
References
1 Warrington, E. K., Shallice, T. (1984).
Category-specific semantic impairments. Brain,
107 829-853. 2 Devlin, J.T., Gonnerman, L.M.,
Andersen, E.S. Seidenberg, M.S. (1998)
Category-specific semantic deficits in focal and
widespread brain damage A computational account.
Journal of Cognitive Neuroscience, 10,
77-94. 3 Sartori, G. Lombardi, L. (2004).
Semantic relevance and semantic disorders.
Journal of Cognitive Neuroscience, 16, 439-452.
CONCLUSION

• The Sensory functional theory cannot accommodate
  the findings of this analysis which indicates
  that deficits in identifying concepts from
  perceptual knowledge can arise without the
  accompanying deficit for identifying living
  concepts.

() Significant at .05, () Significant at .01.


correspondence
ltnoonank_at_lsbu.ac.ukgt