LIN: Forschungsabteilungen > Akkustik, Lernen, Sprache > Unterpunkt Ebene 3 > Unterpunkt Ebene 4
Titel: LIN Layout
Distributed ECoG Engrams of Auditory Learning in the Mongolian Gerbil (SPP 1665)
The formation of categories is a fundamental element of cognition, and has been studied extensively to probe the functional basis of cognition. However, the circuit mechanisms of category formation, especially at the mesoscopic scale bridging single neuron activity to organismal behavior, remain largely unknown. While most previous work on category discrimination has focused on unit activity reflecting category selectivity in higher cortical areas, recent work has started to focus on such mesoscopic circuit mechanisms, especially the emergence of selectivity much earlier in the sensory processing stream, particularly within the primary auditory cortex. Such distributed patterns have been documented in the form of patterned population activity, via multiphoton calcium imaging, electrode recordings, and cortical surface array recordings.
In contrast to simple discrimination, category discrimination is based on the abstraction from the learned particulars and has therefore a higher complexity requiring computation beyond simple sensory processes, and therefore serves as a good model for higher cognitive processes. We have established a robust model of auditory category discrimination learning in the Mongolian gerbil, using frequency modulated (FM)-sweeps and a go/no-go shuttlebox paradigm. We have used this model to investigate (1) behavioral characteristics of auditory category learning, (2) to localize such function to the auditory cortex, and (3) to investigate neurochemical and proteomic consequences of learning. Particularly relevant for this project, we have shown that (4) mesoscopic spatial patterns of neural population activity as measured by surface ECoG arrays can accurately predict the animals’ behavioral/cognitive decision.
In this project, we explore the causative mechanisms leading to such mesoscopic neural activity patterns and their behavioral outcome. In particular, we aim to first demonstrate formal neurophysiological causality by testing for both the necessity and sufficiency of the mesoscopic activity for behavioral output, and second, to investigate the single-neuronal circuit mechanisms underlying these mesoscopic patterns, using a combination of behavioral, electrophysiological and optogenetic techniques. We thereby hope to offer an important mesoscopic link between (A) the firing patterns of single neurons and resultant local oscillations, and (B) the total behavioral output of the brain as an organ. This level of analysis has been less represented in systems neurophysiology to date, but with the advent of massively parallel recordings and fast imaging techniques, the importance of this level of analysis has become increasingly clear as a bridging step to understanding of higher-level perception and learning.
Dopaminergic Modulation of Rodent Behavior (LIN Special Project)
Memory formation and retrieval rely on the coordinated interplay of many brain areas. While memory processes have been studied extensively in individual structures, our understanding at the network level connecting these structures is much less complete. To address these questions, we have recently conducted fMRI imaging studies of learning-related modifications in network state (Angenstein F. et al., 2013). In this study, we imaged the brain while the animal underwent negative reinforcement learning towards electrical stimulation of the hippocampus. Our results showed a qualitative change in network state, coinciding in timing with the behavioral acquisition of meaning—we hypothesize that such coincident changes reflect the formation of associative memory. The most prominent network changes in response to hippocampal stimulation were seen in the medial prefrontal cortex (mPFC). It is unclear, however, what neural mechanisms may lead to the formation of this hypothetical associative memory trace.
In particular, our recent work has highlighted the importance of the connection between hippocampus and prefrontal cortex. Namely, we demonstrated the capture of learning-related modifications in stimulus-induced brain activation patterns by fMRI. Our results show a qualitative change in functional network state, as soon as an electric stimulus of the hippocampal formation gains behavioral meaning during negative reinforcement learning. The observed functional network of brain areas is, on one hand, comprised of structures which respond to the electrical stimulation irrespective of its behavioral significance, such as the Nucleus accumbens (NAcc). On the other hand, the network contains structures, which become active only after the stimulus gains behavioral significance, i.e. reflect the formation of the associative memory in their activation. The most prominent of these is the medial prefrontal cortex (mPFC). The neural processes that cause this network state change are still unclear.
In this project, we are testing the hypothesis that these memory-associated network changes are causally related to dopaminergic activity.
Neuro-Optic Investigation of Deep Brain Stimulation for Parkinson Disease (CBBS Neuronetworks)
Parkinson disease (PD) was initially described as a disease with dominant motor impairments. More recently, it has been demonstrated that it can cause emotional and cognitive disturbances as well. Since Dopamine plays a crucial role in higher functions of the brain, its absence in basal ganglia, which is highly innervated with cortical and limbic neurons, can cause some imbalances in cognitive and emotional states. Deep Brain Stimulation (DBS) has emerged as one of the most effective therapies especially for the patients who do not respond to medications. On the other hand, currently available DBS in some cases is not effective or has side effects. To address this situation, new hypotheses have been formed regarding better DBS paradigms. However, these improvements are hindered by the lack of knowledge regarding the exact mechanism whereby DBS exerts its effects.
In this project, we ask two ambitious questions. First, using an animal model, we study the effects of DBS therapy with stochastic pulses (instead of regular pulses) on behavior. In addition, using voltage-sensitive dye imaging and optic fiber recording, we aim to investigate neural activity in the motor cortex and basal ganglia, in order to gain further insight into the systems physiology of DBS. This voltage-sensitive dye method provides a unique possibility to investigate population activity patterns without the contamination by electrical stimulation artifacts.
State-Dependent Propagation of Population Activity in the Neocortex
Propagating waves of excitation have been observed extensively in the neocortex, during both spontaneous and sensory-evoked activity, and they play a critical role in spatially organizing information processing. However, the mechanisms governing wave propagation remain an issue of ongoing debate. In particular, little is known about how propagation dynamics are influenced by global network state and other network dynamics. We have developed analytical tools to quantified local propagation patterns of spontaneous activity in single trials using voltage-sensitive dye (VSD) imaging. Our results so far demonstrate that, despite a prevailing similarity in laminar networks, propagation of spontaneous excitable waves is very susceptible to cortical state.
Figure Properties of wave propagation in rat visual cortex. (A) Spontaneous state alternations of the electrocorticogram (ECoG) between ECoG-synchronized and ECoG-desynchronized states under urethane anesthesia. (B) VSD imaging field with a 464-channel photodiode array covering primary (V1) and secondary (V2M) visual areas. (C) Representative single-trial examples of spontaneous cortical waves for synchronized and desynchronized states as obtained by VSD imaging. Vertical lines and lower-case labels indicate time periods for which frames are drawn. Inset frames: frames show propagation of activity within the imaging field (normalized scale, variable scaling). Note that in the desynchronized state, propagation patterns are more spatially fragmented as compared to the synchronized state. Black contour frames: Rose plots indicating flow trajectories for each example wave a - f as obtained by a temporospatial correlation algorithm. (D) Spontaneous waves in the desynchronized state tend to propagate faster than spontaneous waves in the synchronized state. (E) Rose histograms showing propagation preferences of spontaneous cortical waves. Note that the average flow histogram (n = 9, transparent blue and red) indicates highly anisotropic propagation in both states, the axis of which is approximately equivalent in both states.
Spiral Dynamics in Epilepsy
Detection of spiral activity in cortical models of epilepsy. A. Stable spiral dynamics elicited in a tangential slice from rat visual cortex. VSD imaging frames were taken very 10 ms. Pattern analysis revealed a counter-clockwise rotation. B. Spiral dynamics from an epileptic Mongolian gerbil. Epileptic activity was triggered by transdural application of bicuculline methiodide and showed clockwise rotation. C.and D. Spiral patterns can be quantified in the flow maps, where arrows reveal the overall translational flow pattern, and color indicates spiral nature, with red signifying positive match to the rotation template, i.e., clockwise rotation, and blue signifying negative match to the rotation template, i.e., counterclockwise rotation (Takagaki et al. 2011).
Every day our environment challenges us with an overabundance of sensory stimuli. It is critical for us to integrate matching information across sensory streams and to segregate unrelated information. To foster the study of multisensory integration, reliable experimental model systems are necessary. In one project our work therefore focuses on a region in rat parietal cortex, which exhibits complex multisensory responses. We employ functional hemodynamic imaging to locate the area and find interactions between sensory modalities affecting its neural activity. Our results demonstrate that these interactions depend on the relative timing of sensory stimuli from different sensory modalities (stimulus onset asynchrony, SOA). For example, we found that somatosensory stimulation increased subsequent visual activity and that visual stimulation diminished subsequent somatosensory activity (figure below, a). Furthermore, this asymmetry is limited to current source densities, while population firing rates show sublinear effects under all stimulation conditions. This could hint at the existence of cortical multisensory normalization mechanisms, as suggested by recent studies. Temporal properties of the described phenomena and further pharmacological experiments indicated a strong cortical influence on the observed interactions, in contrast to propagated downstream effects. By establishing behavioral models we are currently working to transfer our findings to further levels of brain function.
In a second project we have studied the cross-frequency coupling in gerbil auditory and visual cortex and its learning-induced plasticity. In the prestimulus activity, alternating groups of coupled frequencies were identified both in the auditory and the visual cortex. Moreover, there was evidence for both second- and third-order cross-frequency interactions. All groups of coupled frequencies exhibited characteristic differences between the auditory and the visual cortex. The presentation of the auditory and the visual stimuli induced modulations of cross-frequency interactions which could not be explained by linear superposition of a fixed multi-frequency response on the ongoing activity. Strong activity was induced around frequencies correlating with all groups of interacting frequencies identified in the prestimulus activity which might increase the likelihood of interactions between groups of coupled frequencies. At the same frequencies we also observed strong evidence for cross-modal interactions between the auditory and the visual cortex (figure below, b). Cross-cortical interactions were shown to be plastic during the course of the stimulation sessions as they increased gradually with increasing number of stimulus presentations.
a) Multisensory modulation of current source densities in a parietal multisensory zone. Depending on relative stimulus onset asynchrony (SOA), either supra- or sublinear effects emerge. Responses to visual stimuli preceded by a somatosensory response (orange) show enhancement, while the reverse leads to depression. PS indicates the point of simultaneity (mean and s.d.) in individual experiments, at which the different latencies between visual and somatosensory activity are equal. Note that the curve is symmetrical around PS and not around synchronous stimuli (SOA of 0 ms). b) Cross-Frequency-coupling from 1 to 100 Hz 100 ms after presentation of an auditory stimulus. Left panel: cross-frequency coupling in the auditory cortex. Middle panel: cross-frequency coupling in the visual cortex. Right panel: cross-frequency coupling between the auditory and the visual cortex.
Development of Population Activity Patterns in the Human and Rodent Brain
We are interested in developmental patterns of brain dynamics, such as development of the human posterior basic rhythm. A publication is pending.