Write a comment

In the paper Does early-stage Alzheimer’s disease affect the dynamics of motor adaptation? by Sutter et al., an error-based motor adaptation process (AP) is analysed with 20 early-stage Alzheimer's disease (AD) patients.

The AD patients and the control group were asked to perform a defined movement of a handle of a planar robotic manipulator, while various force fields were influencing the movement. The strength and speed of adaptation to these force fields is the central object of the study, whereby a distinction is made between a fast, declarative and a slow, non-declarative AP. The hypothesis is that only the former is negatively influenced in AD patients. The AP of the test persons is assigned to an adaptation index (AI). It was observed that AD patients were less able to adapt. A Bayesian two-state model was used to determine a retention rate and a learning rate from the AI for both APs. Both test groups differed significantly from each other only in the retention rate of the fast process, which supports the authors' hypothesis. These parameters were then used to simulate the learning process of the fast and slow processes.

The structure of the paper is excellent. The abstract consists of a good explanation of the motivation, a first overview of the methods and an outlook on the scientific and medical relevance of the results. Because of this, the paper is accessible for a general scientific audience. The experimental procedure is described very well in the methods section. In a very detailed discussion, the authors compared their results with the findings of other studies and thus defined the context of their work. In terms of content, the control group was matched to the patients in terms of age, educational level and IQ. The two groups did not differ in these aspects. But both groups differ significantly in their gender distribution, but unfortunately is not addressed by the authors. This means that it cannot be said with absolute certainty, that the results can be attributed to AD, as they can also be attributed to different genders.

The paper is of interest to a general audience, but in its current form suffers from several important deficiencies that need to be corrected before the paper can be accepted for publication.

Major issues

• The control group was matched to the group of patients in terms of age and educational level, but not in terms of gender distribution. The distribution of the genders is significantly different (p $<$ 0.05, t-test). While there are three times as many men as women among AD patients, the control group has slightly more women than men. This potential issue is not addressed by the authors. It is important that the results can be attributed entirely to the disease and not to the difference in gender. One way of doing this would be to cite studies that prove that the characteristics being analysed are independent of gender. Another would be to analyse the available experimental data according to the gender of the participants and thus show that the results are independent of gender. However, the latter is difficult to do without further experiments, as the number of participants in the individual groups is then very small. For example, only five women with AD were included in the experiment.

• It is explicitly pointed out that in the simulation of the APs, the fast process is weaker in the AD patients than in the control group. According to the authors, this leads to a stronger slow process in the simulation of the AD patient, compared to the control group, which can be seen in Figure 6. This contradicts the results from the fit of the Bayesian hierarchical version of the dual-rate model to the experimental data. If the two plots from Figure 4 are superimposed, it can be seen that the slow process is still stronger in the control group than in the AD patients. This contradiction has to be addressed and possible causes have to be discussed.

• In the abstract, the sentence "but recent findings suggest that it involves both fast, declarative and slow, non-declarative adaptive processes" suggests that the declarative process is fast and the non-declarative process is slow. However, the following hypothesis "If the declarative memory system shares resources with the fast process in motor adaptation" describes that it is not certain that the declarative process can also be assigned to the fast process. This contradiction needs careful attention to be resolved.

Minor issues

• The resolution of Fig. 4 is quite low. Even if this does not limit the basic message of the paper, a higher resolution could make the data easier to read.

• The control group was also matched to the patient group with regard to their level of education. The authors need to therefore explain why they did this or in other words, why they expect the level of education to have an influence on motor adaptation. In addition, the number of years of education is more meaningful and easier to quantify than a points system in which a certain qualification results in a certain number of points.

Competing interests

The authors declare that they have no competing interests.