Computational Motor Control Club

[About]   [Members]   [Joining]   [Meetings]  [Spring 2002]  [Winter 2002/3]  

Spring 2003

About the club

The CMCC is a small group of active researchers that meet periodically at the Department of Electrical Engineering in the Technion and discuss current research in this field.  The wide range of topics include: Biological Motor Control, Brain Theory (Neuroscience), Learning Theory, Control Theory, Artificial Neural Networks, Man-Machine Interfaces, Robotics.  (*)  

Members  (In alphabetical order)

Naama Brenner, Ronen Chen, Yaki Engel, Shraga Hocherman, Gideon Inbar, Amir Karniel, Vladimir Litvak, Ronny Meir, Hillel Pratt, Sara Rosenblun, Nahum Shimkin, Rafi Sivan, Yoav Tock, Tomer Valency, Victor Yosef, Miriam Zacksenhouse, Michael Zibulevsky, Omer Ziv

 

Meetings Spring 2003 (Future dates are tentative)

(*) Some of the meetings consist of a formal talk and discussion in the club.  The formal talk is published in the Technion bulletin and is open to any interested person.

 

Joining the Club

The purpose of the club is to conduct deep discussions that would generate new research ideas and collaborations.  Therefore the members are asked to come regularly, be prepared for the meetings and occasionally to present a topic from their field of interest.  If you are an active researcher in this field (e.g., faculty member, graduate student) and wish to commit yourself to participate regularly and to contribute to our discussions, please contact Amir Karniel or any other member for details about the next meeting.  You would be asked to introduce yourself shortly at the first meeting you attend. (*) 

 


If you have comments or suggestions email karniel@ee.technion.ac.il  

 

Meetings Winter 2002/3

 

Members Spring 2002 (In alphabetical order)

Gideon Inbar, Amir Karniel, Ronny Meir, Nahum Shimkin, Rafi Sivan, Yoav Tock, Tomer Valency, Elad Yom-Tov, Victor Yosef, Miriam Zacksenhouse, Michael Zibulevsky, Omer Ziv

 

Meetings Spring 2002

 

 

Abstracts

 

Does the Brain Use Clocks, Counters, or Switches During Motor Adaptation?

Amir Karniel

From the dawn of cybernetics, the search for artificial intelligence was a salient driving force.  In the forties, numerical calculations were a daunting task; a decade ago, playing the game of chess was considered the ultimate test.  Modern computer easily beat any person in these tasks.  Nevertheless modern robots are far behind the motor abilities of a five years old child.  Beside the scientific interest, understanding the biological motor control system has great potential in promoting future applications to help the physically disabled and to build intelligent robots and computers.

In this talk I will describe a series of psychophysical experiments along with computational models that challenge some common theories and analogies between the computer and the brain. 

Studies of arm movements have shown that subjects learn to compensate for predictable mechanical perturbations by developing a representation of the relation between the state of motion of the arm and the perturbing forces.  We tested the hypothesis that subjects can employ clocks, counters or switches in order to construct this internal representation.

Our findings suggest that in contrast to standard computers and robots, the biological motor control system may not use the equivalent of clocks, counters or switches in order to adapt to time-varying environment.  Instead, the biological system tends to build an internal dynamic model that is usually based on a state mapping. 

This research was performed in collaboration with Ferdinando A. Mussa-Ivaldi.  The experiments were performed at the robotics lab of the Rehabilitation Institute of Chicago

 

 

 

The Importance of Assigning Importance

Prof. Eytan Ruppin, Tel-Aviv University

How does one aim to "understand" the processing of a system?
A classical, good point to start is to find the ``contributions'', i.e., the relative importance of the different elements that compose that system.
In this talk I shall present some novel methods to address this question, employing concepts ranging from function minimization to information theory and game theory. Their workings will be demonstrated in the analysis of neurocontrollers of evolved autonomous agents and on biological data of reversible inactivation experiments.
Developed initially for the study of neurocontrollers, this kind of localization analysis may have a broad array of potential applications in the field of functional genomics and drug design, and, more generally, to solving cost allocation problems.

-- Joint work with Ranit Aharonov, Lior Segev, Alon Keinan and Isaac Meilijson --

 

 

Planning and execution of arm movements in normal humans and in neglect patients

Dvorkin Assaf, ICNC, Hebrew Univ., Jerusalem; Dept. CS & Appl. Math., Weizmann Inst., Rehovot

Reaching toward a visual target involves the transformation of visual information into motor commands, using multiple spatial reference frames (RF). The origin of these RF may be shoulder-, hand-, or eye-centered.

In an attempt to identify which RF are used to represent target position, we measured constant and variable errors in a double-step (DS) paradigm, while subjects (Ss) reached toward visual targets, located in a horizontal plane. Moreover, in this study we investigated the spatial and temporal characteristics of the attentional deficit in unilateral neglect patients during reaching.

Results from young Ss showed that in pointing movements, the major axes of the confidence ellipses co-aligned with the average movement direction. This may suggest a hand-centered RF. In contrast, in DS movements various orientations were found, suggesting that endpoint variability may also reflect motion planning rather than mere execution.

The recorded movements in both young and elderly normal Ss included averaged/non-averaged and direct trajectories. In contrast, results from two neglect patients showed no averaged modified trajectories. These findings suggest that the patients were unable to perform on-line motor corrections. Furthermore, a higher proportion of direct trajectories was observed, for movements that were initially directed toward the left side of space. In addition, whereas in elderly subjects and patients, some of the DS movements contained a pause, the patients displayed significantly longer pauses, especially when the first target appeared ipsilesionally. These findings reflect the existence of a competitive bias in favor of ipsilesional stimuli.

Results from a perceptual control experiment demonstrated the existence of a spatial perception deficit in both patients. 

This research was done under the supervision of Flash Tamar and Bentin Shlomo, and in collaboration with Behrmann Marlene and Soroker Nachum.

 

Causal Localization of Neural Function: The Shapley Value Method

Alon Keinan, Tel-Aviv University

Identifying the functional roles of elements of a neural network is one of the first challenges in understanding neural information processing. Aiming at this goal, lesion studies have been used in neuroscience, most of which employing single lesions and hence, limited in their ability to reveal the significance of interacting elements. I will present the Multi-lesion Shapley value Analysis (MSA), an axiomatic, scalable and rigorous method, addressing the challenge of calculating the contributions of network elements from a multi-lesion data set. The successful workings of the MSA will be demonstrated on artificial and biological data. MSA is a novel method for causal function localization, with a wide range of potential applications for the analysis of reversible deactivation experiments and TMS-induced "virtual lesions".

 

Processes that lead to action: lessons from a reaction time study

Prof. Shraga Hocherman, Medical School, Technion

Parkinson's disease (PD) is known to inflict cognitive impairments, involving mainly executive functions. Of these, set formation and set shifting are studied by use of verbal/thought-dependent tools, as well as by reaction time (RT) tests that employ varying levels of cuing, choice and response complexity. Different studies report different, sometimes conflicting findings, regarding the ability of PD patients to use advanced cuing in order to shorten their RT. Thus, conclusions about the changed state of underlying processes such as motor programming and motor initiation are not certain. In the present study we have manipulate the validity of cuing in order to interfere with sensory-motor associations, motor programming and motor initiation. Otherwise, our RT tests involved simple, well differentiated, non spatial visual cues, distinct imperative sound stimuli and very simple motor responses. All cues and stimuli were brief (100ms) and the cue-stimulus interval was fixed at 800ms. Thus, responses were unlikely to depend on thinking. The studied 19 PD patients were found to have impaired set formation and shifting in WCST and were found to be slower than the 21 age matched controls on the RT tests. However, the patients were able to utilize cuing for preprogramming of their responses to acoustic stimuli. Motor initiation was found to be normal but motor programming, although accessible by cognitive set was significantly slower in the patients.