The brief survey of the literature above found that dysfunctions in attention have been consistently noted throughout the history of the diagnosis of schizophrenia (Kraepelin, 1919), and a common complaint of those so diagnosed (McGhie & Chapman, 1961). Moreover, between-group deficits on tests of attention have been commonly identified (Heinrichs & Zakzanis, 1998), and appear to be promising predictors of future development of the illness amongst those at high genetic risk (Freedman et al., 1998). However, it had also been identified that ‘attention’ is a complex construct, which is often defined differently between researchers (Cohen, 1993), and that specific aspects of
experiencing schizophrenia, rather than being necessary, characteristic traits of the diagnosis (Egan et al., 2000; Palmer et al., 1997). The identified relationships between symptoms and measures of attention to date, however, had been only moderate or weak (Tables 36-48). On the basis of these facts, with the knowledge that much of the symptom modelling to date provided a great mismeasure of the structural nature of symptoms within those diagnosed with schizophrenia, it was proposed that if there were relationships between particular symptoms and measures of aspects of attention, then these may be being obscured by the use of models providing poor descriptions of symptoms. As such, if this was the case, then it was considered that the use of the more focused and homogeneous symptom groupings developed in the current study would identify stronger relationships with aspects of attention, which would, in turn, provide a measure of validation for the symptom model itself.
These predicted relationships between symptom groups and neuropsychological measures of components of attention were not identified, with the findings instead suggesting that the relationships arose from issues independent of, or peripheral to, those
symptoms considered to be central to the concept of schizophrenia (such as hallucinations, delusions, negative signs, or thought disorder). As such, two key questions arise from these findings: firstly, were there any methodological reasons for the lack of identification of the proposed relationships between symptomatology and the measures of attention (i.e. that relationships were in fact there, but methodological design deficiencies conspired to obscure them); and secondly, if this is not the case, is there evidence to suggest the independence of neuropsychological measures of attention from symptomatology, and if so, what does this mean for our approach toward ‘schizophrenia’?
Methodological Issues
Possible methodological reasons for the lack of identification of relationships with symptom groupings will be briefly discussed below, under several main headings: the complexities of the neuropsychological measures adopted; the nature of the symptom groupings used as predictors; characteristics of the participant group; analysis logic; and item set artefacts.
Complexities of the Neuropsychological Tasks
Neuropsychological tests are often defined according to the functions or brain regions that they are proposed to measure, such as ‘frontal lobe’ tasks or tests of ‘working memory’. However, as had been noted in the preceding review of findings in regard to each domain of attention, the vast majority of neuropsychological measures that are applied clinically are complex tasks, which generally require the operation of a number of different cognitive processes (Fossati, Amar, Raoux, Ergis & Allilaire, 1999; Shallice, Burgess & Frith, 1991). Table 75 summarises many of the cognitive functions required for the successful performance of the neuropsychological tasks used in the current study. The implication of this is that poor performance on any given task may arise from multiple causes. This has been clearly shown in the case of the Wisconsin Card Sorting Test, where poor utilisation of feedback may arise from a failure to shift cognitive set or due to a failure to retain that new information, both of which are functions relating to very different regional areas; and studies have demonstrated that the pattern of perseverative errors on the task is indistinguishable between patients with frontal lobe and non-frontal damage (Anderson, Damasio, Jones & Tranel, 1991). Other factors that may contribute to variability are tasks were there performance can be enhanced by the
symbol-coding), and that many measures do not include any internal control measures (MacDonald & Carter, 2002), or where they do, they may be inadequate (such as the Trail-Making Test: Gaudino, Geisler & Squires, 1995).
One possible alternative approach to this situation would be the use of the case-study methods, where large numbers of tests are given to a small number of patients, with tests selected on the basis of continual refinement of hypotheses derived from performance on each of the tasks delivered (Lezak, 1995; Shallice, Burgess & Frith, 1991). This may address one of the particular limitations of the methodological approach more commonly applied in schizophrenia research, where a small number of tests are uniformly applied to a large number of subjects, as in this scenario, the applied measures are required to be broad enough so that they can accommodate the full range of performance in both clinical and non-affected samples (i.e. avoid both floor and ceiling effects) as well as retaining sensitivity of assessment (Zakzanis, 1998). However, in studies seeking to examine reasons behind the marked heterogeneity of clinical presentation and cognitive ability among those diagnosed with schizophrenia, single-case or small-n studies are unlikely to produce robust or generalisable findings.
Another possible alternative approach is that taken by cognitive neuropsychology, where investigation of much more specific cognitive processes can be performed, through the ability to simplify tasks (reducing the number of possible alternative approaches), to develop detailed and specific control measures (hence isolating variance due to specific functions), and to collect more precise information in regard to performance (MacDonald & Carter, 2002; Knight & Silverstein, 1998). This sort of approach has proved fruitful in describing specific systems that may be related to the expression of particular symptoms (Cahill & Frith, 1996; Green, 1998; McGrath, 1996; Nayani &
David, 1996; Neufeld & Williamson, 1996). As such, in the search for the processes underlying the expression of particular symptoms, a clinical neuropsychological approach such as that applied in the current study, may prove too coarse to identify these subtle deficits.