Cell's Vision. Decoding memory: Reveals the brain's conceptual cells with a single neuronrecord of the human brain.

How brain cells encode external information, especially information with highly abstract concepts, has always been the most advanced scientific problem in neuroscience. One of the key challenges is how to study the response of neuronal discharges to behavioral events.for a long time, neuroscientists have used noninvasive electrophysiological recording techniques to answer this question, including EEG, megs and fMRI.these techniques are important for understanding the activation of brain regions under different task states. However, these records can only provide indirect detection of neuronal activity [1].therefore, for a long time, people can only rely on the extracellular recording of awake animals in order to provide direct information on the firing of single neurons in clusters.nevertheless, people have made extremely important discoveries, such as the 2013 Nobel Prize in physiology or medicine awarded to British scientist John O'Keefe and Norwegian scientists may Britt Moser and Edvard I. Moser for their discovery of the cells that constitute the brain localization system, and this discovery is based on the invasive extracellular electrophysiological recording technology.however, there is a big gap between the electrophysiological records of single neurons in vivo of experimental animals and the scientific questions that clinicians want to answer.on November 14, 2019, Professor Rodrigo Quiroga, director of the center for systematic neurobiology, University of Leicester, UK, and winner of the Wolfson Award for excellence in research of the Royal Society, published an opinion article in cell journal [2], plugging in to human memory: advantages, challenges, and Insights from Human Single-NeuronThis paper mainly describes the single neuron electrophysiological recording technology in the hippocampus of epileptic patients, and compares it with human non-invasive electrophysiological recording technology and experimental animal invasive electrophysiological technology, discusses the advantages, challenges and limitations of human single neuron recording, and further expounds the forefront scientific problem of Quiroga Laboratory for a long time Brain "concept cell". The advantages and limitations of human single neuron electrophysiological recording technology can be traced back to the 1950s, when neuroscientists operated on glass microelectrodes in epileptic patients under surgical conditions.in the 1970s, people began to use deep valley intracranial catheters to wrap multiple clusters of microfilament electrodes into the brain for recording.this design is still used by neurologists (Fig. 1). It can not only record the intracranial EEG, but also record the discharge and local field potential signals of multiple single neurons at the tip of the electrode.the temporal middle lobe (MTL) was often recorded in that era.this brain region plays an important role in different forms of epilepsy, and the recording time is about 1 week, so it can fully record the spontaneous generation signal and the final changes after surgical resection.Fig. 1. In this paper, Professor Quiroga compared the human single neuron recording technology with the human non-invasive electrophysiological recording, and thought that the former could provide direct signals of single neuron activity, and greatly overcome the sparse discharge events that EEG, Meg or fMRI could not capture at the general level.compared with the invasive electrophysiological recording technology on experimental animals, the obvious advantages of human single neuron recording technology are: (1) we can avoid functional differences between human brain and similar brain regions of animal brain; (2) we can avoid the functional differences between human brain and animal brain Subjects can give researchers direct speech feedback, which is particularly important for the study of memory activation and the conscious control of neuronal discharge in the state of thinking and imagination; (3) subjects can be directly trained by researchers without long-term reward training, which can avoid the influence of over reward training on neuron response after several months; (4) This human single neuron recording technology also enables the research on the information processing related to the subject's background. Professor Quiroga takes an example, as shown in Fig. 2B, that the neurons in the entorhinal cortex of the subject obviously respond to mathematical information, while the amygdala neurons of the other subject in Fig. 2C obviously only respond to the role Mr. t in lodge 3, and respond to the movie The relevant characters can also respond because the patient is a fan of the movie.Fig. 2. An example of single neuron electrophysiological recording in human temporal and middle lobe region was given. Professor Quiroga also pointed out that the main limitation of human single neuron recording was the insufficient number of patients.most hospitals that can perform this procedure often have to wait for years to collect enough neurons from patients for statistical analysis.other limitations include the experimental cycle and the choice of control subjects, and there are many uncontrollable factors in this experiment. Usually, the experiment should be conducted in a noisy clinical environment. in addition, the neuronal electrical activity data recorded in epileptic patients can only reflect the pathological state of the brain, not the normal brain function, and the medication situation of patients will have a great impact on the experimental results. in addition, the limitations of this technique include that the placement area of intracranial recording electrodes can only be determined by clinical criteria, so scientists cannot determine the details of electrode implantation. different from the chronically implanted electrodes in mice, the human single neuron recording electrode cannot move back and forth to find the response neurons. Therefore, an obvious limitation is that people can not accurately locate the precise location of the microfilament electrode in the human brain. another limitation is that it is difficult for human single neuron recording to be carried out over a time span of up to several days, which makes it more challenging to study the stability and plasticity of MTL in memory consolidation. similarly, since the electrodes in the patient's brain cannot be implanted for more than a week, it is impossible to track the activity of the same neuron for several days. therefore, it is necessary to record the discharge characteristics of the same neuron 24 hours a day for 7 consecutive days. 2. Using human single neuron electrophysiological recording technology to reveal the role of temporal middle lobe in memory coding, neuroscientists observed an epoch-making discovery on patient H.M. half a century ago, suggesting that MTL plays an important role in declarative memory [3]. the important role of MTL in encoding and consolidating episodic memory has been verified in animal experiments and brain imaging, but it is still unable to answer how human MTL neurons participate in memory, because the most direct evidence still needs to come from the study of human single neuron electrophysiological records. episodic memory is mainly based on a single experience. By recording single neurons in human brain, researchers have found that MTL neurons respond to novel stimuli, and these neurons respond selectively to familiar stimuli. major advances in this field include the use of higher-order peak potential classification algorithms to analyze MTL neurons that are almost inactive in firing, and recognize that there are a few sparse and indefinite neurons in MTL, which can respond to a "concept" rather than specific visual or auditory information. therefore, these neurons are defined as concept cells, which are characterized by: (1) they only respond to specific and well-known concepts (such as familiar people or familiar places); (2) they are highly invariant (that is, they respond to different pictures of the same person); (3) they are not regulated by the situation. the existence of concept cells can be supported by the following evidences: (1) concept cells have a relatively delayed response (about 300 ms); (2) they discharge personal related concepts; (3) they are highly invariant and highly selective; (4) their functions are not affected by sensory information processing. Human MTL neurons encode associative episodic memory, which depends on the formation of fast association. For example, people tend to forget irrelevant details and remember concepts. therefore, the researchers recorded single neurons in the human brain and studied memory association through different behavioral paradigms. after the concept cells had the maximum discharge to a certain stimulus, the response decreased after multiple learning, while the response to the stimulation of non optimal discharge increased significantly after multiple learning. this kind of fast learning is very related to situational memory, for example, if you need to remember someone you see in a certain place, it usually takes only one experience to form. the researchers also observed that these conceptual cells that responded to the association rarely continued to respond to the association after the association was established or even enhanced. after training the same neuron to respond to a certain concept, it was found that these neurons could encode long-term associations, which were not created for learning tasks, but related to the subjects themselves. Therefore, it is believed that MTL neurons encode episodic memory for a long time. Professor Quiroga summarized two important functions of concept cells through electrophysiological recording of single neurons in human brain and task training for patients: (1) forming meaningful association and tracing back; (2) directing and jointly activating sensory representation of neocortex. further explains that conceptual cells do not function alone, but form a cluster of cells to represent familiar concepts [4]. further clarify that MTL and neocortex jointly encode episodic memory, which can be explained by Figure 3 [4]. V1 cortical neurons respond to directional stimuli, as shown in the right side of the figure. The neurons only respond to vertical stimuli. This visual information is further processed by visual pathway and enters the temporal lobe cortex (it). Neurons in this region process pure visual information, such as maximum response to face (bottom left corner of Figure 3). the temporal cortex is widely associated with MTL including hippocampus, so neurons in this region only respond selectively to soccer player Diego Maradona. therefore, it can help readers understand the role of MTL concept cells in the whole visual information processing and memory coding. Figure 3. Visual perception and memory coding. So, do other species also have conceptual cells? This is still a controversial issue. this is because the training methods and behavioral tasks of experimental animals are essentially different from those of single neuron electrophysiological recording. the authors believe that it is necessary to clarify this issue in future studies, but it is speculated that even single neuron recording in primates may not be able to observe concept cells with the same abstract degree as human beings. another major reason is that human beings have a fine language system, which can help us share knowledge and culture, and language can also