1) What function do brain rhythms serve?
Neuronal output cannot be described as a Poisson process, since neuronal responses are strongly oscillatory. These oscillatory responses are synchronized across areas and neurons.
Groups of neurons can participate in a multitude of rhythms, which may serve communication through different channels. The function of these rhythms has remained elusive so far.
We address the problem of the function of brain rhythms both from the experimental, methodological and theoretical side.
Experimentally:
Together with Jeroen Bos and Cyriel Pennartz we record spiking activity from well isolated single units (using tetrodes) from four rat brain areas at the same time, namely the visual cortex, barrel cortex, perirhinal cortex and CA1 area of the hippocampus, in the context of a sensory discrimation task in an eight maze. We question how spiking activity is phase-locked between areas, and how this phase-locking subserves memory retrieval, decision making and sensory coding.
With Pascal Fries and Thilo Womelsdorf I am involved in the analysis of visual cortex data (gamma band oscillation, Vinck et al., 2010, JNS; Womelsdorf et al., 2011, PNAS) and ACC/PFC data, where we look at the coding of spatial attention through phase-locked activity.
Methodologically:
An important problem in neuroscience is the quantification of phase-locking from datasets under limited sampling. As an example, in a typical in vivo experiments we may record 50 s of controlled activity from a given neuron, which fires at about 1 Hz per second. This means our statistical degree of freedom for phase-locking measures is rather limited. This problem is increased by the fact that phase-locking is typically quite weak, meaning that the variance of available phase-locking estimators is quite high.
In my work, I’ve solved four problems in quantifying phase-locking between spikes and their surrounding electric fields (LFPs) (Vinck et al., 2010, Neuroimage; Vinck et al., 2011, JCN).
- create a measure of phase-locking independent of the number of spikes/trials.
- create a measure of phase-locking that is not affected by non-Poissonian history effects within spike-trains, such as bursting/refractoriness.
- create a measure of phase-locking that is orthogonal to phase-shifts / phase-precession.
- link the traditional spike-field coherence measures (operating on complete spike-train snippets) to measures that operate on single spikes.
Theoretically:
I’m performing simulations on the effect of oscillatory synchronization on a post-synaptic target, using both statistical and leaky-integrate and fire models. I compare the reliability by which changes in oscillatory synchronization can be detected with the reliability of changes in the firing rate. In addition, I examine the effectivity and reliability of changes in coherence between areas.
2) What is the structure of the neural code?
A key question in neuroscience is the nature of the neural code, i.e.: How does the pattern of spikes provide information about e.g. an external stimulus?
I approach this question from the experimental and methodological side. Following a long tradition in neuroscience I try to examine to what extent precise temporal spike patterns provide information about a stimulus.
Experimentally:
We analyze spiking and local field potential data from many brain areas in the context of well controlled, sensory experiments, and consider the information contained in the phase of firing relative to oscillatory cycles.
Methodologically:
In collaboration with Han Vinck and Vladimir Balakirsky we develope estimators of entropy and mutual information for datasets under limited sampling.
We have developed an improved entropy/information estimator (Vinck et al., 2012, in submission).
3) How is memory organized?
A beautifulmodel system in the nervous system is episodic memory. The associative nature of episodic memory strongly contrasts with the way modern computer memories operate, and understanding the architecture of memory would have both strong clinical implications (Alzheimer, aging) and technological implications. The traditional notion is that episodic memory arises from an interaction between the hippocampus, a brain area whose removal can lead to a complete loss of episodic memory, and the neocortex.
Experimentally:
Together with Jeroen Bos and Cyriel Pennartz we record isolated spiking activity from the hippocampus, sensory areas, and the perirhinal cortex, an area that forms a bridge between the hippocampus and the rest of the neocortex. The idea is to study the neocortical-hippocampal loop. Together with an active processing task, we also perform sleep recordings before and after the processing task.
Theoretical:
I’m working on a model of the neural architecture of episodic memory. This model specifies the nature of the interaction between the hippocampus and the neo-cortex, and the nature of the hippocampal representations.