Application of a novel balance test to study motor learning abilities

The research carried out during the Ph.D. course regarded the application of pos-tural balance tests to study motor learning abilities. Two protocols are used in this work: a novel divided attention (DA) test and the modified clinical test for sensory interaction in balance (mCTSIB). A total of 42 subjects participated in the experi-mental campaigns. To provide experimental contributions to imitation learning, which is thought to be supported by the mirror neuron (MN) system, the DA test is administered to two groups of subjects, to which the information of the test is conveyed either through verbal instructions or imitation by observation. The results show that the imitation based learning approach only allows to catch the gross gestures, while the fine motor gestures are often not understood. Moreover, imitation requires a higher cognitive ef-fort, thus resulting in slower gestures. This work points out that action understanding is not supported by the MN system, but instead, other brain circuitries are involved. Postural tests are used in clinical analysis to diagnose neurological dysfunctions such as Parkinson’s disease and fall risk in elderly. These tests are sometimes per-formed over multiple sessions to follow up the patient’s rehabilitation. However, since the subject performance is analyzed by performing the test with only one at-tempt, the results can be biased by learning or adaptation effects which might lead to erroneous conclusions or diagnosis. In this work, the results of both DA and mCTSIB show that subjects need some practice trials to exclude any learning effect from the analyzed data. The number of practice trials depends on the test conditions and the considered parameters. The learning effect should be taken into account also when validating models with data measured from subjects. Moreover, averaged results flatten the transient phase, which might mask important features. The results of this work show that the target pursuit movements of the DA test, which are achieved through 2D postural move-ments, follow the Fitts’ model of speed-accuracy trade-off, whereas the choice reac-tion time follows the Hick’s model. Both models show that the performance im-proves at each repetition with a power function. A modified model to account for learning is also proposed. In this work, postural models are analyzed by means of biofidelity and learning abilities with reference to behavioral motor models. Vehicle collisions are mostly caused by driver’s errors and are among the most causes of death in the world. Nowadays law enforcement faces the challenge to as-sess driving impairment that arises from multiple sources such as drowsiness and psychoactive substances. However, the current tools are not adequate to objectively assess driver’s abilities. The features of this novel DA test seem suitable for a road test to assess driver’s psychophysical conditions. Moreover, preliminary data on im-paired subjects show that the learning trend might be used for assessment purposes.