Neural Correlates of visual perception and behaviour: Analysis of multiple single-neuron recordings in humans
Lead Research Organisation:
University of Leicester
Department Name: Engineering
Abstract
Patients suffering from severe epilepsy that do not improve with medication may be considered for a surgical intervention to remove the seizure originating area. Based on clinical criteria, these patients may be implanted with intracranial electrodes to accurately localize the focus of the seizures and evaluate the potential outcome of the surgery.
Since recently, a setup unique in Europe is in place at King?s College London that from the intracranial electrodes allows not only the recording of EEG but also of multiple single-neurons. From the activity of these neurons, we plan to study whether it is possible to predict the onset of epileptic seizures. This has large clinical potential to develop seizure-warning devices or closed loop systems that can preclude the onset of the impending seizures (e.g. using a fast-acting drug or microstimulation). Previous studies at this respect have been inconclusive because they used EEG recordings, which have inherent major limitations due to their poor spatial resolution and high complexity. This will be the first time that such study is done using the activity of single cells.
Besides its clinical importance, the recording setup at King?s College London allows the unique opportunity to record directly the activity of single neurons in awake and behaving human subjects, who can report and give details of their perception and behaviour in different tasks and brain states.
Using simple experimental paradigms, we plan to address one of the most fascinating and puzzling questions in neuroscience: how do neurons in the brain represent conscious visual perception? For this, we will study the activity of neurons while the subjects see different pictures. Then we will evaluate whether the neurons increase their firing in response to particular pictures and whether these responses change if the pictures are recognized or not.
Since recently, a setup unique in Europe is in place at King?s College London that from the intracranial electrodes allows not only the recording of EEG but also of multiple single-neurons. From the activity of these neurons, we plan to study whether it is possible to predict the onset of epileptic seizures. This has large clinical potential to develop seizure-warning devices or closed loop systems that can preclude the onset of the impending seizures (e.g. using a fast-acting drug or microstimulation). Previous studies at this respect have been inconclusive because they used EEG recordings, which have inherent major limitations due to their poor spatial resolution and high complexity. This will be the first time that such study is done using the activity of single cells.
Besides its clinical importance, the recording setup at King?s College London allows the unique opportunity to record directly the activity of single neurons in awake and behaving human subjects, who can report and give details of their perception and behaviour in different tasks and brain states.
Using simple experimental paradigms, we plan to address one of the most fascinating and puzzling questions in neuroscience: how do neurons in the brain represent conscious visual perception? For this, we will study the activity of neurons while the subjects see different pictures. Then we will evaluate whether the neurons increase their firing in response to particular pictures and whether these responses change if the pictures are recognized or not.
Technical Summary
For clinical reasons we can record directly the activity of single neurons in humans. These recordings are performed in epileptic patients refractory to medication, who are implanted with intracranial electrodes to evaluate the feasibility of curative surgery. Besides its clinical importance, this procedure allows the unique opportunity to record directly the activity of single neurons in awake and behaving human subjects, who can report and give details of their perception and behaviour in different tasks and brain states.
Our overall aims are: i) to study how visual information and behaviour are represented by populations of neurons in the human brain, and ii) to study the mechanisms of epileptic seizures and the possibility of seizure forecasts for clinical applications.
For the first aim, a large set of pictures will be presented several times on a laptop computer while recording the neuronal activity. Pictures will be shown in conditions of difficult recognition (using short presentation times and noise) so that they will be recognized in some of the trials and not recognized in some other. Then, we will assess whether the firing of the responsive neurons is stronger when the pictures are recognized, thus reflecting conscious perception, or whether they keep the same firing for recognized and not-recognized trials, thus reflecting unconscious visual processing. The information obtained from the firing of single-cells will be compared to the one obtained from local field potentials using a decoding approach.
For the second aim, we will first characterize changes in the firing of the neurons and their connectivity patterns during seizures and pre-seizures states, in comparison with the activity during inter-ictal states. Based on this information we will then evaluate the possibility of predicting the seizure onsets. We will focus on forecasts of at least minutes in advance, which should give an adequate time window for realistic clinical applications. Predictions obtained with the multiple single-neurons will be compared to those obtained with the intracranial EEGs.
Finally, we plan to develop and optimize further advanced methods to analyze the single-cell data. The general goal is to record as many neurons as possible from each electrode in an unsupervised way and eventually on-line. These developments will be tested using extensive computer simulations mimicking the activity of large neural populations.
Our overall aims are: i) to study how visual information and behaviour are represented by populations of neurons in the human brain, and ii) to study the mechanisms of epileptic seizures and the possibility of seizure forecasts for clinical applications.
For the first aim, a large set of pictures will be presented several times on a laptop computer while recording the neuronal activity. Pictures will be shown in conditions of difficult recognition (using short presentation times and noise) so that they will be recognized in some of the trials and not recognized in some other. Then, we will assess whether the firing of the responsive neurons is stronger when the pictures are recognized, thus reflecting conscious perception, or whether they keep the same firing for recognized and not-recognized trials, thus reflecting unconscious visual processing. The information obtained from the firing of single-cells will be compared to the one obtained from local field potentials using a decoding approach.
For the second aim, we will first characterize changes in the firing of the neurons and their connectivity patterns during seizures and pre-seizures states, in comparison with the activity during inter-ictal states. Based on this information we will then evaluate the possibility of predicting the seizure onsets. We will focus on forecasts of at least minutes in advance, which should give an adequate time window for realistic clinical applications. Predictions obtained with the multiple single-neurons will be compared to those obtained with the intracranial EEGs.
Finally, we plan to develop and optimize further advanced methods to analyze the single-cell data. The general goal is to record as many neurons as possible from each electrode in an unsupervised way and eventually on-line. These developments will be tested using extensive computer simulations mimicking the activity of large neural populations.
