What is the scientific approach to neuropsychology? - Chapter 3

This chapter describes the types of questions and the importance of a dissociation and double dissociation. The recovery, course and treatment are also discussed. Often it is difficult to assess whether recovery is the result of treatment or spontaneous recovery processes. Another problem in determining this is the test-retest effect: Someone can show a better performance in the second test because he / she is already familiar with the test.

What is the difference between clinical neuropsychological examination and fundamental research?

A distinction is made between clinical neuropsychological research and fundamental research. In clinical neuropsychological examinations the focus is on further characterization of the clinical picture, examining the usability of the test instruments and test procedures and analyzing the course of a disease. Fundamental research focuses on increasing the understanding of the disorder and related brain structures.

Clinically oriented neuropsychological issues

The questions in clinical neuropsychological range from differential diagnostic questions to evaluating treatments and providing answers to advice questions. Diagnostic research has to be done according to the diagnostic (empirical) cycle. A major disadvantage of research on clinical questions is that the value of the conclusion is very dependent on the quality of the test, the available standard data and other psychometric properties of both tests and questionnaires. An even bigger problem is the matter of balancing between what is practically feasible on the one hand with the optimum test combinations on the other. It is impossible to offer all the tests that would measure part of the memory; moreover, the memory remains so complex that this cannot be included in all tests. A third limitation is the missing values that are the result of a test battery that is not fully completed by the patient (due to, for example, fatigue or aphasia).

Fundamental issues

The causes of disorders and underlying cognitive processes are investigated through experimental paradigms. It is not necessary to use standardised procedures, because the comparison is made within the experiment, as the different conditions are compared.

What does a research design look like?

Subtraction

Frans Donders introduced the reaction time paradigm by applying a subtraction method: in this procedure the score (for example reaction time) is subtracted from the score or a more complex task in a simpler condition. This subtraction method is often used for imaging research. In this case, not the reaction time, but the activation level represents a score. A comment on this procedure is the unreliability of the difference score: one condition has a certain unreliability just like the other condition and these are added together. Factorial designs (analysis of variance) can be used to solve this problem.

Single and double dissociations

In neuropsychology, a dissociation implies that there is a specific dropout but general cognitive functioning is intact. With a single dissociation there is a selective dropout for task A (complex) but not for task B (simpler), while in practice it has been shown that someone who fails in task B also fails in task A. This can be explained by the fact that task A is more complex than task B: sometimes capacities are saved for simple tasks. With a double dissociation, two independent tasks are clearly distinguished that were initially thought to be strongly interdependent. For example, one patient is not able to carry out task A but is able to carry out task B while another patient is not able to carry task B, but is able to carry out task A. An example of a double dissociation is the perception of objects and the perception of spatial relationships between objects. Only a double dissociation is not sufficient evidence.

Single-case study

The patient Tan (Leborgne) and HM are well-known examples of single-case studies. Carmazza (1986) states that a single-case study is the only correct way to investigate cognitive impairment, due to the fact that no lesion (in form and impact) is the same for all individuals within a group study. In addition, it is often assumed that premorbid functioning is 'normal' in the heterogeneous group, whose focus is then on adaptability and functioning after a lesion. However, group studies are not useless: they offer a general framework and the possibility of generalization, and ultimately the single-case studies must also fall within this framework.

The design of a single-case study

There are various designs for conducting a single-case study. First, the test score of the patient can be compared with the score of a normative group. Secondly, intra-individual research can be conducted: all kinds of specific tasks are performed and the conditions are compared with each other. In this case the patient's scores will have to be compared with the scores of a control group in which participants are matched to important characteristics.

How is the course of a disease studied?

The difference between a longitudinal and a cross-sectional study

There are two main course studies that both try to make statements about the course of a disease. A longitudinal study means that one or more patients are monitored and tested for a number of years, after which their data is analyzed. An important confound - disturbance factor - here is the test-retest effect which endangers the validity of the test. The solution to this problem is to add a control group that is also tested twice at the same interval. Cross-sectional research means that different groups of patients with the same syndrome are tested at different times. In this case, the 'average course' can only be discussed. Both designs have their advantages and limitations. That is why Salthouse advocates a combined design.

Treatment studies

Treatment studies look at the specific effect of treatment through pre-measurements and post-measurements. Here it is important that generalization takes place: the improvement must not only be noticeable on objective neuropsychological tasks, but also in daily life. The difference between the follow-up measurement and the pre-measurement does not necessarily have to be caused by treatment but can be the result of a non-specific effect such as improved motivation or reduced feelings of depression. If the improvement is noticeable in a wide range of cognitive functions, the improvement appears to be caused by a non-specific effect. A solution for this is the multiple baseline design in which multiple pre-measurements are made to find out whether there is already a spontaneous recovery.

Control task and cross-over design

To assess if a specific effect occurs after treatment, a control task can be used: if a patient is treated for function A, then there should be no improvement for function B after treatment. The placebo effect should also be considered, or the Hawthorne effect, where the extra attention for the patient can lead to improved cognitive functioning. The cross-over design is a variant of the control test: here a training is given for function A, after which the performance on function A and function B is measured. Subsequently, a training is given for function B and again the performance on function A and function B is measured. The training should only lead to an improvement in the trained function.

Item-specific training

There is a specific effect if only the specific items that are trained actually improve while the other items do not improve. With specific items you can think of naming certain pictures.

Randomization test

A randomization test analyzes the probability of a certain pattern of scores when the test samples have been taken at random moments in time. For one patient, by chance, a different starting point is chosen than for the other patient. Subsequently, the course of the scores and the chance of a specific pattern are examined. It is also possible to look at two treatments within one person in this way.

Test-retest problem

One of the problems with treatment studies is the so-called learning effect. The patient may have learned item-specific. This can be solved by working with parallel versions. A test-retest effect is not so much about item-specific learning, but about task-specific learning: the patient is familiar with the instructions, the situation and knows what is to come. This must be taken into account when drawing up the design: a control group must also be offered the repeated measurements but not the treatment.

Generalization

Within neuropsychology it is important that a certain result can be generalized. This means that a learned skill in a treatment is also useful in daily practice. For this reason, it can be useful to practice tasks that (almost) exactly match reality, such as shopping. If this is not possible, it is nevertheless important to pay attention to transfer. This means that similar exercises are done, in which the patient learns how to transfer the skill to daily life.

Which quality criteria are treatments subject to?

The SCED scale contains ten quality criteria that single-case studies must meet. This includes the adequate description of clinical data, a suitable design for establishing causal relationships and the quality of the baseline measurement. A relatively new criterion has been included under the term 'randomized controlled trial' (RCT). In an RCT, the patients are randomly placed in one of the conditions (experimental or control condition).