Wednesday, October 7, 2009

Davis (2009) Atkins decision: IQ part scores and modular nature of intelligence: Guest post by Dr. Dale Watson

A number of recent Atkins-related IQ/MR death penalty court decisions have raised important issues re: the interpretation of variability in part scores when compared to the total (full scale-general intelligence; g) composite index from an intelligence test.  In particular, Davis (2009) and Vidal (2007) are excellent examples of the complex issues...and how they relate to a diagnosis (or not) of mental retardation.

Dr. Dale Watson has studied the Davis (2009) decision and has provided the following thoughtful guest blog post (click here for other guest blog posts].  Thank you Dr. Watson for the excellent analysis and commentary.  I (the blog dictator) have reproduced his post "as is" with a few exceptions.  I've added a few URL links to other sources.  I've also added emphasis to certain statements via underlining, followed by a note that I added the emphasis.  Finally, I've yet to figure out if it is possible to add real footnote superscripts to blog text via the blog editor I use.  So, I've adopted the format of putting footnote numbers in brackets [ ] and placing the footnotes at the end of the blog post.

I encourage other psychologists and measurement specialists to review some of the Atkins court decisions posted at this blog and send me their analysis and thoughts for potential blog posts. 

By Dr. Dale Watson

In a recent Maryland case, U.S. v. Davis, (2009 WL 1117401 (D.Md.)) the trial court found the defendant to be mentally retarded and thus ineligible for the death penalty in line with Atkins v. Virginia, 536 U.S. 304, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002).  An analysis of the IQ results in this case highlights issues regarding the impact of part-score variability in the IQ profile and the diagnosis of mental retardation as well as the modular nature of intelligence. [emphasis added by blogmaster.  Blogmaster note--Also, see prior post regarding the recommended use of total vs part scores from a national panel report ].

Earl Davis, the defendant, had been charged with federal crimes, including murder, which made him potentially eligible for the death penalty unless he were found to be intellectually and developmentally disabled.  In a pre-trial proceeding the court heard evidence regarding whether Mr. Davis was in fact mentally retarded.  The defense presented the testimony of five imminently qualified experts supporting their position.  The prosecution, relying on the opinions of two board certified neuropsychologists, alternatively argued that Mr. Davis did not meet the criteria for being mentally retarded and instead demonstrated a learning disability.  This argument was founded, in part, on the finding of a significant discrepancy between the Verbal Comprehension (VCI) and Perceptual Reasoning Indices (PRI) of the WAIS-IV.[1]   The prosecution contended that the 15-point discrepancy between these indices made the Full Scale IQ (FSIQ) unreliable as an overall measure of his intellectual functioning.[2]   A discrepancy of this magnitude (VCI = 66; PRI = 81) could be expected to occur with a base rate of 12.7 percent within his ability range and could thus be considered “abnormal.” [3]   This argument was advanced despite the fact that Mr. Davis’ WAIS-IV FSIQ of 70 nominally met the first prong of the definition of an intellectual disability (mental retardation).
The prosecution argument was not without precedent.  For example, it was noted in testimony that the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) reported:
When there is a significant scatter in the subtest scores, the profile of strengths and weaknesses, rather than the mathematically derived full-scale IQ, will more accurately reflect the person’s learning abilities.  When there is a marked discrepancy across verbal and performance scores, averaging to obtain a full-scale IQ can be misleading (U.S. v. Davis citing DSM-IV-TR, p. 42).
Fiorello et al. (2002) summarized the literature addressing this viewpoint as applied to the Wechsler scales:
Sattler (2001) cautions that the FSIQ may misrepresent a child’s cognitive functioning level if the Verbal IQ (VIQ) and Performance IQ (PIQ) are significantly different; however he indicates that no empirical evidence exists to indicate when the FSIQ should not be reported or used in eligibility decisions.  Prifitera, Weiss, and Saklofske (1998) recommend that FSIQ should not be interpreted when differences between VIQ/PIQ or Verbal Comprehension Index (VCI) and Perceptual Organization Index (POI) scores are extreme, which they define as differences found in less than 10% of the population (p. 117). [4]
The reluctance to make the diagnosis of mental retardation in the face of significant discrepancies between measures of verbal and non-verbal abilities is thus perhaps not surprising or new.  There has been a long-held view that mental retardation is marked by a “flat” cognitive profile.  When confronted with a profile that instead shows a pattern of ipsative strengths and weaknesses perhaps many psychologists would be hesitant to make the diagnosis of mental retardation.  However, recent empirical evidence does not support that position. [emphasis added by blogmaster]

It is the case, as noted in the WAIS-IV Technical and Interpretive Manual, that “The prevalence of large and unusual discrepancies between verbal and nonverbal composite scores [on the Wechsler scales] has been shown to decrease with decreasing levels of ability (internal citations omitted)” p. 102.  However, Bergeron and Floyd (2006), using the Woodcock-Johnson III Tests of Cognitive Abilities (WJ-III Cog.), demonstrated that individuals with mental retardation “will not likely display a flat cognitive profile on comprehensive assessments of CHC broad cognitive abilities—especially when measures vary widely in g loadings—regardless of the information presented in test manuals for mental retardation groups (emphasis added).” [5]  These researchers found that, in children with mental retardation, there was significantly greater intra-individual score ranges (scatter) than in average-achieving individuals.  Specifically, nearly 37% of these children obtained at least one CHC factor score within the average range.  Bergeron and Floyd concluded,
…it is likely that the increasing number of specific cognitive abilities measured by intelligence test batteries and the variation of these scores in individual profiles inadvertently muddies the waters of mental retardation diagnosis.  As a result, when faced with IQs in the range described in the diagnostic criteria for the disorder and part scores that are much higher, practitioners may believe that an individual cannot be diagnosed with mental retardation because of evidence of “intact” or “unimpaired abilities” p. 428.
…denying special education eligibility or failing to make a diagnosis of mental retardation based on significant part score variability may do these children disservice when other ecologically valid evidence of mental retardation (e.g., adaptive behavior skill deficits) indicates genuine need. P. 429.
These findings have implications for the role of g as well as domain specific disabilities in the genesis of mental retardation.  Bergeron and Floyd posited that the centrality of the impaired CHC factors, as measured by the factor’s g-loading, would determine its sensitivity to mental retardation.  In fact, the Comprehension-Knowledge and Fluid Reasoning clusters, with the highest g-loadings, were most commonly associated with the diagnosis of mental retardation, though low average or average scores on one of these clusters did not preclude the diagnosis.

In another line of research, Anderson (1998) has suggested the need to distinguish “between mental retardation as a general deficit of thinking and mental retardation that might result from the global effects of a specific deficit in a cognitive module.” [6]  In a similar vein, Frith and Happé (1998) have argued “that general impairments (e.g. low IQ) in developmental disorders need not be the result of primary damage to domain-general mechanisms.  Rather, they may be the developmental consequence of damage to very specific, even modular, mechanisms which act as gatekeepers in development” p. 270. [7]

Neuroimaging data further supports the view of the modularity of cognitive functions.  In such a view, though g may represent a relatively unitary phenomenon in the normal brain it can fractionate in the face of modular neuropathology.  Gläscher et al. (2009) used CT and MR lesion maps to localize the impairments found on the WAIS-III Index scores in focal brain-damaged patients.   These investigators “found (1) impairments in VCI (Verbal Comprehension Index) were associated with damage in left hemisphere, in particular in the left inferior frontal cortex, (2) impairments in POI (Perceptual Organization Index) were associated with damage in right parietal, occipito-parietal, and superior temporal cortex, (3) impairments in WMI (Working Memory Index) were associated with left hemispheric lesions particularly focused in superior parietal cortex, and (4) impairments in PSI (Processing Speed Index) correlated with a number of small regions distributed across both hemispheres” (p. 686). [8]  Each of these indices, with the exception of the PSI, significantly predicted a lesion in the associated region.  These results largely confirm clinical lore regarding the significance of specific impairments in the Wechsler indices.

Williams syndrome, a genetically determined developmental disorder commonly marked by mental retardation, has frequently been characterized by relatively intact language abilities and profoundly impaired visual-spatial functions.  Thompson et al. (2005), using MRI scans, identified specifically increased cortical thickness within the perisylvian and inferior temporal regions of the right hemisphere.  This pattern of regional neuropathology appears to correspond well with the differentiated pattern of cognitive deficits.

Mental retardation may thus result from both general impairments or as the “developmental consequence of damage to very specific, even modular, mechanisms.”  [emphasis added by blogmaster]. In the former case one could anticipate a relatively “flat” profile of intellectual abilities.  In the latter instance, a more extreme pattern of strengths and weaknesses might emerge and yet result in sufficient impairment of overall intellectual functioning as to meet the criteria for mental retardation.  Evaluators would do well to remember that the diagnostic criteria for mental retardation do not limit the diagnosis to a particular ipsative pattern of scores, that mental retardation can arise from diverse etiologies and that “strengths co-exist with weaknesses.” (emphasis added by blogmaster).  These cautions apply equally within a clinical context and within a capital punishment context when the diagnosis is “a matter of life or death.”
  • [1] The opinion reflected some confusion on this issue indicating that the discrepancy was between the VIQ and PIQ despite the fact that these scores are no longer included in the WAIS-IV.
  • [2] See People v. Superior Court of Tulare County (Vidal) for an example of even greater discrepancies between verbal and performance IQs in an Atkins case.
  • [3] The “abnormality” of such discrepancies is somewhat arbitrary but authors have variously set the cut-off for an unusual finding at a base-rate of either 10 or 15 percent level.
  • [4] Fiorello, C.A., Hale, J.B., McGrath, M., Ryan, K. & Quinn, S. (2002).  IQ interpretation for children with flat and variable test profiles.  Learning and Individual Differences, 13, 115-125.
  • [5] Bergeron, R. & Floyd, R.G. (2006). Broad cognitive abilities of children with mental retardation: An analysis of group and individual profiles.  American Journal on Mental Retardation, 111(6), 427.
  • [6] Anderson, M. & Miller, K.L. (1998).  Modularity, mental retardation and speed of processing.  Developmental Science, 1(2), 239-245.
  • [7] Frith, U. & Happé, F. (1998).  Why specific developmental disorders are not specific: On-line and developmental effects in autism and dyslexia.  Developmental Science, 1(2), 267-272.
  • [8] Gläscher, J., Tranel, D., Paul, L.K., Rudrauf, D., Rorden, C., Hornaday, A., Grabowski, T., Damasio, H., and Adolphs, R. (2009).  Lesion mapping of cognitive abilities linked to intelligence.  Neuron, 61, 681-691.

Technorati Tags: , , , , , , , , , , , , , , , , , , , , , , , , ,