|
Alexa Riehle, PhD Research Director at the CNRS
Team DyVA
Post-docs and PhD Students: Bjřrg Kilavik Adrián Ponce Alvarez |
|
|
Staircase,
photo by Alexa Riehle, |
|
Research Interests:
Introduction: By integrating concepts and methods from cognitive psychology with those of neuroscience, our goal is to arrive at a unified approach for understanding higher brain functions involved in motor control. The primary objective is to decipher cortical mechanisms by which multiple sources of information are integrated during the process of movement preparation, and to discover the relationship between movement representations, based on specific changes in neuronal population activity, as well as on the modulations of neuronal cooperativity within such populations, and behavioral aspects such as reaction time. Preparation for action is considered to be based on central processes, which are responsible for the maximally efficient organization of motor performance. A strong argument in favor of such an efficiency hypothesis of preparatory processes is the fact that providing prior information about movement parameters and/or removing time uncertainty about when to move significantly shortens reaction time. In order for motor performance to be efficiently organized, both contextual and sensory information have to be assembled and integrated to shape the motor output. The notion of uncertainty, which is related to the manipulation of contextual information, is at the core of preparatory processes. The best-suited paradigm for studying such processes is the so-called 'preparation paradigm'. In this paradigm, two signals are presented successively to the subject in each trial: the first, the preparatory signal, provides prior information about what to do after occurrence of the second, the response signal, and/or about when to do it. By means of such prior information, the context in which the subject is placed can be experimentally manipulated. The subject knows with more or less precision both what to do and when to initiate the requested movement, and has to adjust movement preparation accordingly. In numerous studies we have demonstrated that the activity of monkey motor and premotor cortical neurons is affected by prior information about the upcoming movement (for a review see Riehle 2005). We manipulated prior information about movement parameters such as direction and/or extent (Riehle & Requin 1989, 1993; Riehle et al. 1984), and the level of the force to be exerted (Riehle et al. 1994; Riehle & Requin 1995), but also temporal aspects of movement planning (Riehle et al. 1997, 2000; Roux et al. 2003). Finally, we have shown that preparatory neuronal activity is highly predictive for performance speed, i.e. reaction time (Riehle & Requin 1993).
Dynamics and cooperativity in motor
cortical networks:
Complementarity of spike synchrony and spike rate
The temporal coding hypothesis suggests that not only changes in firing rate but also precise spike timing, especially synchrony, constitute an important part of the representational substrate for perception and action. In this framework, the concept of cell assemblies uses synchrony as an additional dimension to firing rate, as a candidate for information processing. Consequently, the observation of spike synchrony between neurons might be interpreted as an activation of a functional cell assembly. An essential ingredient of the notion of coordinated ensemble activity is its flexibility and dynamic nature, in other words neurons may participate in different cell assemblies at different times, depending on stimulus context and behavioral demands (see Figure below). When in an instructed delay task prior information is provided about movement parameters, such as movement direction (spatial parameters), or the moment when to move (temporal parameters), movement initiation is faster. Cortical neurons selectively modulate their activity in relation to this information. As to indicate the end of an instructed delay, motor cortical neurons synchronize significantly their activity at the moment of signal expectancy (for the method see Figure 1), often without any detectable modulation in firing rate (Riehle et al. 1997). The observed increase of the temporal precision of synchrony towards the end of an instructed delay is interpreted to facilitate the efficiency of the motor output leading to an increase of performance speed (Riehle et al. 2000). Furthermore, there is a systematic relationship between directionally selective synchrony, firing rate and behavioral directional preferences (Grammont & Riehle 2003). Finally, we showed that the timing of the task is dynamically represented in the temporal structure of significant spike synchrony at the population level which is shaped by learning and practice (Kilavik et al. 2009). The emergence of significant synchrony becomes more structured, that is it becomes stronger and more localized in time with practice in parallel with a decrease in firing rate and an improvement of the behavioral performance. Performance optimization through practice might therefore be achieved by boosting the computational contribution of spike synchrony, allowing an overall reduction in population activity.
|
Cell
Assemblies are dynamic entities defined by the everchanging level
of correlation among the activities of their member neurons. Neurons may
participate in different assemblies at different times, depending on context
and behavioral demands
|
|
|
Spiking activity, local field potentials and timing processes
The
accurate estimation of time intervals is an essential aspect of motor
performance. Most often temporal judgments are tightly linked to the dynamics
of spatial features in the environment, e.g. the time for a ball to arrive
within reach ("time to contact"). Thus, the prediction of
forthcoming events is crucial for organizing most efficiently motor
performance. Removing temporal uncertainty by providing prior information about
when to move significantly shortens reaction time. In a condition, in which the
delay between two signals is manipulated such that a finite number of durations
is randomly presented, reaction time decreases due to the increase of
probability for the second (GO) signal to occur as time elapses from the first
(instruction) signal. In order to assess this probability, the elapsed time has
to be correctly estimated. In other words, time estimation is an integral part
of movement preparation and correct time estimation improves movement
performance – expressed by reaction time. We have shown that timing processes
are indeed represented in motor cortical single neuron activity, albeit in a
manner that is strongly dependent on context (Roux et al. 2003). Furthermore, motor
cortical neurons significantly synchronize their activity precisely at moments
when a GO signal is expected at the end of a correctly estimated delay (Riehle et al. 1997). And finally, local field potentials (LFPs)
recorded simultaneously with spiking single-neuron activity modulated during the delay period as
a function of time (Roux et al. 2006).
Estimation of a fixed delay period was associated with a single cycle
modulation of the LFP. Its amplitude depended on the probability of movement
execution and its timing on the duration of the delay to be estimated.
Additionally, the LFP signal was highly oscillatory during the delay period
with a mean frequency in the ß-range, an oscillation which abruptly stopped
with movement onset. Time-resolved cross correlation studies of LFPs recorded a
few hundred microns from one another exhibited a central peak around zero and
correlated side peaks at roughly ±60ms. The correlation strength of the center peak varied for different
pairs of LFPs and strongly modulated both in time and with the behavioral
condition. Both the frequency of the correlated oscillation as well as the
correlation strength increased toward the end of the delay period, i.e. in
relation to the expectancy of the GO signal.
Time is a rubberband (Renoult et al. 2006): Studies using instructed delay tasks have shown that the production of a movement as response to a temporally predictable signal induces large variability in response times. This is usually attributed to the changes in readiness to act. However, to which extent the (variable) estimation of the delay duration influences the readiness to respond is largely unknown. Time estimation is indeed crucial for correct performance in delayed motor tasks. Here, we studied in a delayed pointing task the relation between the variability in response time and motor cortical activity. Precise spike times were determined on a trial-by-trial basis and a new time scale was defined in each trial such that the time between the precue signal and movement onset was common for all trials. Each spike was then displaced accordingly. We then estimated single trial firing rates using both the original and the new time bases for determining the moments of peak activities after the precue, at the moment when the go signal was expected (but did not occur) and around the go signal. Peak latencies during both time scales were then correlated with response time. In the original time scale, the variability of peak latencies increased as time elapsed during the trial, whereas in the modified time scale variability remained small. The activity of neurons in MI reflects a signature of continuous time between the precue and movement initiation. The discharge of many neurons seems to be linked to the temporal judgment of the animal. In the new time scale, in which response time variability is suppressed, the between trials variability in peak latencies is no longer significant, showing a direct link between spike occurrence and response time.
Neuronal population representation of movement parameters and the "dynamic neural field model"
Motor cortical areas are involved in the programming of movement parameters such as direction, extent, force etc. For example, most motor cortical neurons are broadly tuned in relation to movement direction not only during the process of movement initiation, but also in response to prior information that precues the direction of an upcoming movement. Based on these single neuron responses, we constructed a neuronal population representation of movement direction as it evolves in time (Bastian et al. 1998, 2003; method: Erlhagen et al. 1999). The shape and the peak location of the population representation reflect the content of prior information throughout the preparatory period. From this "preshaped" activity distribution evolves the final representation of the parameter direction once the signal to move is presented. These experimental data can be modelled by the "dynamic neural field" (cf Erlhagen and Schöner 2002) in which prior information is represented by localized input activity that pre-activates the corresponding sites in the field. Dynamic interactions in the field contribute to its gradual evolution towards the target state and stabilization of a single peak at the relevant location when the response signal provides specific input. Since a strongly structured preshape can be interpreted as a high degree of movement preparation, one may suggest that it is predictive for behavioral performance. In fact, we found that the preshape of the population representation correlates with reaction time and reflects the time course of preparation.
Selected
publications: Alexa Riehle
Close collaborations:
Ad Aertsen,
Dept. of Neurobiology & Biophysics and BCCN, University of Freiburg,
Germany aertsen at biologie.uni-freiburg.de
Markus
Diesmann, Computational Neurophysics, RIKEN Brain Science Institute,
Wako-Shi, Japan diesmann
at brain.riken.jp
Wolfram Erlhagen, Dept. of Mathematics, University of Minho, Guimaraes, Portugal wolfram.erlhagen at mct.uminho.pt
Sonja Grün, Statistical Neuroscience, RIKEN Brain Science Institute, Wako-Shi,
Japan gruen at
brain.riken.jp
William A. MacKay, Dept. of Physiology, University of Toronto, Toronto, Canada william.mackay at utoronto.ca
Martin P. Nawrot, AG Neuroinformatik, Frei Universität Berlin, Berlin, Germany martin.nawrot at fu-berlin.de
Stefan Rotter, Bernstein Center for Computational Neuroscience Freiburg, University
of Freiburg, Germany rotter at bccn.uni-freiburg.de
Gregor Schöner, Institute of Neuroinformatics, Ruhr-University Bochum, Bochum, Germany
Gregor.Schoener at
neuroinformatik.ruhr-uni-bochum.de
Jun Zhang,
Dept. of Psychology, University of Michigan, Ann Arbor, MI, USA junz at umich.edu
last revision: January 2010
|
Alexa Riehle and Eilon Vaadia (eds.)
|
|
|
Ophrys Massiliensis, an
endemic orchid of the Marseille area, blooming in January-February in the
famous Calanques |
|
