This thesis is about the kind of signals used in our central nervous system for guidingskilled motor behavior. In the first two projects a currently very influential theory on the flow of visualinformation inside our brain was tested. According to A. D. Milner and Goodale(1995) there exist two largely independent visual streams. The dorsal stream issupposed to transmit visual information for the guidance of action. The ventralstream is thought generate a conscious percept of the environment. The streamsare said to use different parts of the visual information and to differ in temporalcharacteristics. Namely, the dorsal stream is proposed to have a lower sensitivityfor color and a more rapid decay of information than the ventral stream. In the first project the role of chromatic information in action guidance wasprobed. We let participants grasp colored stimuli which varied in luminance. Criti-cally, some of these stimuli were completely isoluminant with the background. Thesestimuli thus could only be discriminated from their surrounding by means of chro-matic contrast, a poor input signal for the dorsal stream. Nevertheless, our partici-pants were perfectly able to guide their grip to these targets as well. In the second project the temporal characteristics of the two streams wereprobed. For a certain group of neurological patients it has been argued that theyare able to switch from dorsal to ventral control when visual information is re-moved. These optic ataxic patients are normally quite bad at executing visuallyguided movements like e.g. pointing or grasping. Different researchers, however,demonstrated that their accuracy does improve when there is a delay between tar-get presentation and movement execution. Using different delay times and pointingmovements Himmelbach and Karnath (2005) had shown that this improvement in-creases linearly with longer delay. We aimed at a replication of this result and ageneralization to precision grip movements. Our results from two patients, however,did not show any improvement in grasping due to longer delay time. In pointing aneffect was found only in one of the patients and only in one of several measures ofpointing accuracy. Taken together the results of the first two projects don´t support the idea oftwo independent visual streams and are more in line with the idea of a single visualrepresentation of target objects. The third project aimed at closing a gap in existing model approaches on pre-cision grip kinematics. The available models need the target points of a movementas an input on which they can operate. From the literature on human and roboticgrasping we extracted the most plausible set of rules for grasp point selection. Wecreated objects suitable to put these rules into conflict with each other. Therebywe estimated the individual contribution of each rule. We validated the model bypredicting grasp points on a completely novel set of objects. Our straightforwardapproach showed a very good performance in predicting the preferred contact pointsof human actors.
Verknüpfung zu Publikationen oder weiteren Datensätzen
