Response to yeo et al.'s critique of behavioral imaging

Ruben C. Gur, Andrew J. Saykin, Larry R. Muenz, Sushma S. Trivedi, Raquel E. Gur

Research output: Contribution to journalComment/debate

1 Scopus citations

Abstract

Yeo et al.'s criticisms of the behavioral imaging (BI) method (Neuropsychological methods of localizing brain dysfunction: Clinical versus empirical approaches. Neuropsychiatry, Neuropsychol. Behav. Neurol. J.-290-303, 1990) are rebutted: (a) Localization of a lesion from neurobehavioral data is unimportant given neuroimaging techniques: We think it remains a major function of neuropsychology. (b) Expert judgments are inappropriate for testing scientific hypotheses and the Bl approach is unempirical: Experts reflect the current theoretical understanding of brain-behavior relationships. Thus, the weights used in BI are hypotheses to be tested empirically, (c) The algorithm may miss small lesions, provides only "rough" measures, and may yield misleading conclusions, as illustrated in a simulation: The BI method is still being developed and improved. We reiterate that it may supplement, but is not a substitute for, professional practice. Yeo et al.'s simulations are not examples since BI produces results consistent with clinical interpretation, (d) The graphical display of BI shows a picture that looks like a CT slice but actually is a two-dimensional projection and does not show subcortical tissue: The display is clearly described in our publications and is similar to many other standard topographic displays such as computed EEG/EP topography. We are developing three-dimensional displays that will include subcortical tissue. Yeo et al.'s proposed alternative approach, which essentially reverses the process, had been discussed and formalized in our (cited) publication. We have explained our reasons for beginning with expert judgments because of the enormous amount of clinical data needed for implementing this alternative. Their specific version ignores remote effects of lesions and does not yield estimates of variability. Once such shortcomings are addressed, such approaches will help refine BI weights.

Original languageEnglish (US)
Pages (from-to)304-312
Number of pages9
JournalNeuropsychiatry, Neuropsychology and Behavioral Neurology
Volume3
Issue number4
StatePublished - Jan 1 1990

    Fingerprint

Keywords

  • Behavior relationships
  • Behavioral imaging
  • Brain
  • Lesions
  • Neuroimaging
  • Neuropsychology

ASJC Scopus subject areas

  • Psychology(all)
  • Neurology
  • Clinical Neurology
  • Psychiatry and Mental health

Cite this