Neural-Network Activity
Multi-tetrode and Multi-site Recordings
Local cortical processing has been extensively studied with respect to coding stimulus information. Few attempts have been made to study the role of more global brain signals like feedback from higher areas or brain states in cortical information processing. The latter are particularly important for explaining the generation of behavior at a neuronal level, because expectations are known to influence perception, indicating that internal models are constantly tested against evidence from sensory input [1]. Our aim is to understand the mechanisms by which global neural activity patterns influence local processing of information and, in turn, how much local processes can contribute to explaining cognitive performance. With local processing we mean spiking of neurons in the context of mesoscopic signals like local field potentials that reflect synaptic integration of both local and longrange inputs rather than how a neuron “fires” when an animal performs a task.
In order to differentiate between the contributions of local and those of remote cortical circuits, we routinely perform: 1. simultaneous recordings from multiple sites, both within and across areas, 2. reversible manipulations of local neuronal activity by pharmacological intervention (we will soon use cooling and microstimulation), and 3. well-controlled changes in the system’s state through changes in motivation, attention or the activity of diffuse neuromodulatory systems. We believe that such an approach, coupled with appropriate analysis methods, may help us discern local stimulus-evoked processes from those induced by feedback and neuromodulatory activity that underlie a subject’s cognitive state. Our ultimate goal is to draw causal inferences from different neuronal signals on cognitive performance. In order to differentiate between the contributions of local and those of remote cortical circuits, we routinely perform: 1. simultaneous recordings from multiple sites, both within and across areas, 2. reversible manipulations of local neuronal activity by pharmacological intervention (we will soon use cooling and microstimulation), and 3. well-controlled changes in the system’s state through changes in motivation, attention or the activity of diffuse neuromodulatory systems. We believe that such an approach, coupled with appropriate analysis methods, may help us discern local stimulus-evoked processes from those induced by feedback and neuromodulatory activity that underlie a subject’s cognitive state. Our ultimate goal is to draw causal inferences from different neuronal signals on cognitive performance.
In order to study the signal integration of cortical circuits, we train monkeys to perform demanding behavioral tasks that require cortical processing rather than highly over-trained stereotypes. Therefore, monkeys perform memory and sensorimotor tasks in which the animals can be challenged by dynamically increasing memory load [2] or the similarity of stimuli that need to be compared or the precision of required sensorimotor integration. We routinely record from arrays of tetrodes (Figure1), so far within one area like ventral prefrontal cortex, and analyze the sorted spiking activity at the level of single units, at the level of local ensembles (multi-unit activity) as well as of distributed ensembles within a cortical area. As these experiments challenge both experimental technology and analysis methods, our group is collaborating with both industrial partners and theoreticians in work funded by the German Ministry for Education and Research (BMBF). Ensemble codes appear to be particularly important when cortical activity is not stimulus-driven, suggesting that maintenance of information requires a different coding mechanism, probably involving more feedback from other areas. What kind of signals can mediate feedback which needs to reach widespread representations at least in sensory systems? In order to study the signal integration of cortical circuits, we train monkeys to perform demanding behavioral tasks that require cortical processing rather than highly over-trained stereotypes. Therefore, monkeys perform memory and sensorimotor tasks in which the animals can be challenged by dynamically increasing memory load [2] or the similarity of stimuli that need to be compared or the precision of required sensorimotor integration. We routinely record from arrays of tetrodes (Figure1), so far within one area like ventral prefrontal cortex, and analyze the sorted spiking activity at the level of single units, at the level of local ensembles (multi-unit activity) as well as of distributed ensembles within a cortical area. As these experiments challenge both experimental technology and analysis methods, our group is collaborating with both industrial partners and theoreticians in work funded by the German Ministry for Education and Research (BMBF). Ensemble codes appear to be particularly important when cortical activity is not stimulus-driven, suggesting that maintenance of information requires a different coding mechanism, probably involving more feedback from other areas. What kind of signals can mediate feedback which needs to reach widespread representations at least in sensory systems?
Oscillations are an interesting candidate for structuring distributed representations in time and space [see 3 for review], as it has been shown that by synchronizing they can coordinate processes across different cortical areas. Their spatiotemporal pattern can modulate synaptic integration such that signals may be routed through cortical pathways as required for solving a behavioral task. The timing of oscillatory local field potentials recorded from pairs of micro wires implanted in multiple cortical areas is so precise that latencies can be determined and compared across areas.