更改

[[Image:Cajal actx inter.jpg|thumb|300px|right|From "Texture of the [[Nervous System]] of Man and the [[Vertebrates]]" by [[Santiago Ramón y Cajal]]. The figure illustrates the diversity of neuronal morphologies in the [[auditory cortex]].|链接=Special:FilePath/Cajal_actx_inter.jpg]]

[[Image:Cajal actx inter.jpg|thumb|300px|right|From "Texture of the [[Nervous System]] of Man and the [[Vertebrates]]" by [[Santiago Ramón y Cajal]]. The figure illustrates the diversity of neuronal morphologies in the [[auditory cortex]].|链接=Special:FilePath/Cajal_actx_inter.jpg]]

Early treatments of neural [[Biological network|networks]] can be found in [[Herbert Spencer]]'s ''Principles of Psychology'', 3rd edition (1872), [[Theodor Meynert]]'s ''[[Psychiatry]]'' (1884), [[William James]]' ''Principles of [[Psychology]]'' (1890), and [[Sigmund Freud]]'s Project for a Scientific Psychology (composed 1895).<ref name=":0">{{cite web |url=http://psych.stanford.edu/~jlm/papers/ThomasMcCIPCambEncy.pdf |title=Connectionist models of cognition |author1=Michael S. C. Thomas |author2=James L. McClelland |publisher=[[Stanford University]] |access-date=2015-08-31 |archive-url=https://web.archive.org/web/20150906120214/http://psych.stanford.edu/~jlm/papers/ThomasMcCIPCambEncy.pdf |archive-date=2015-09-06 |url-status=dead }}</ref> The first rule of neuronal learning was described by [[Donald Olding Hebb|Hebb]] in 1949, in the [[Hebbian theory]]. Thus, Hebbian pairing of pre-synaptic and post-synaptic activity can substantially alter the dynamic characteristics of the synaptic connection and therefore either facilitate or inhibit [[neurotransmission|signal transmission]]. In 1959, the [[neuroscientist]]s, [[Warren Sturgis McCulloch]] and [[Walter Pitts]] published the first works on the processing of neural networks.<ref name=":1">{{citation | title = What the frog's eye tells the frog's brain. |author1=J. Y. Lettvin |author2=H. R. Maturana |author3=W. S. McCulloch |author4=W. H. Pitts | year = 1959 | work = Proc. Inst. Radio Engr. | issue = 47 | pages = 1940–1951 }}</ref> They showed theoretically that networks of artificial neurons could [[implementation|implement]] [[logic]]al, [[arithmetic]], and [[symbol]]ic functions. Simplified [[Biological neuron model|models of biological neurons]] were set up, now usually called [[perceptrons]] or [[artificial neurons]]. These simple models accounted for [[Summation (Neurophysiology)|neural summation]] (i.e., potentials at the post-synaptic membrane will summate in the [[cell body]]). Later models also provided for excitatory and inhibitory synaptic transmission.

Early treatments of neural [[Biological network|networks]] can be found in [[Herbert Spencer]]'s ''Principles of Psychology'', 3rd edition (1872), [[Theodor Meynert]]'s ''[[Psychiatry]]'' (1884), [[William James]]' ''Principles of [[Psychology]]'' (1890), and [[Sigmund Freud]]'s Project for a Scientific Psychology (composed 1895).<ref name=":0">{{cite web |url=http://psych.stanford.edu/~jlm/papers/ThomasMcCIPCambEncy.pdf |title=Connectionist models of cognition |author1=Michael S. C. Thomas |author2=James L. McClelland |publisher=[[Stanford University]] |access-date=2015-08-31 |archive-url=https://web.archive.org/web/20150906120214/http://psych.stanford.edu/~jlm/papers/ThomasMcCIPCambEncy.pdf |archive-date=2015-09-06 |url-status=dead }}</ref> The first rule of neuronal learning was described by [[Donald Olding Hebb|Hebb]] in 1949, in the [[Hebbian theory]]. Thus, Hebbian pairing of pre-synaptic and post-synaptic activity can substantially alter the dynamic characteristics of the synaptic connection and therefore either facilitate or inhibit [[neurotransmission|signal transmission]]. In 1959, the [[neuroscientist]]s, [[Warren Sturgis McCulloch]] and [[Walter Pitts]] published the first works on the processing of neural networks.<ref name=":1">{{citation | title = What the frog's eye tells the frog's brain. |author1=J. Y. Lettvin |author2=H. R. Maturana |author3=W. S. McCulloch |author4=W. H. Pitts | year = 1959 | work = Proc. Inst. Radio Engr. | issue = 47 | pages = 1940–1951 }}</ref> They showed theoretically that networks of artificial neurons could [[implementation|implement]] [[logic]]al, [[arithmetic]], and [[symbol]]ic functions. Simplified [[Biological neuron model|models of biological neurons]] were set up, now usually called [[perceptrons]] or [[artificial neurons]]. These simple models accounted for [[Summation (Neurophysiology)|neural summation]] (i.e., potentials at the post-synaptic membrane will summate in the [[cell body]]). Later models also provided for excitatory and inhibitory synaptic transmission.

= = 早期研究 = = 对神经网络的早期研究见于赫伯特 · 斯宾塞Herbert Spencer的《心理学原理》第三版(1872)、西奥多 · 梅纳特Theodor Meynert的《精神病学》(1884)、威廉 · 詹姆斯William James的《心理学原理》(1890)和西格蒙德 · 弗洛伊德Sigmund Freud的《科学心理学计划》(1895)。<ref name=":0" /> 1949年，赫布Hebb在其理论（即赫布理论Hebbian theory）中提出了神经元学习的第一定律。赫布理论认为，配对中突触前神经元和突触后神经元的活动可以充分改变突触连接的动态特性，即要么促进，要么抑制信号传递signal transmission。1959年，神经科学家，沃伦·麦卡洛克Warren Sturgis和 沃尔特·皮茨Walter Pitts 发表了关于神经网络处理的第一部著作。<ref name=":1" /> 他们从理论上证明了人工神经元网络可以实现逻辑、算术和符号功能。生物神经元的简化模型models of biological neurons由此建立起来，现在通常被称为感知器perceptrons或人工神经元artificial neurons。这些简单的模型解释了神经加成作用neural summation(即突触后膜上的电位将在细胞体中加成)。后来的模型也提供了兴奋性和抑制性突触传递。

= = 早期研究 = = 对神经网络的早期研究见于赫伯特 · 斯宾塞Herbert Spencer的《心理学原理》第三版(1872)、西奥多 · 梅纳特Theodor Meynert的《精神病学》(1884)、威廉 · 詹姆斯William James的《心理学原理》(1890)和西格蒙德 · 弗洛伊德Sigmund Freud的《科学心理学计划》(1895)。<ref name=":0" /> 1949年，赫布Hebb在其理论（即赫布理论Hebbian theory）中提出了神经元学习的第一定律。赫布理论认为，配对中突触前神经元和突触后神经元的活动可以充分改变突触连接的动态特性，即要么促进，要么抑制信号传递signal transmission。1959年，神经科学家，沃伦·麦卡洛克Warren Sturgis和 沃尔特·皮茨Walter Pitts 发表了关于神经网络处理的第一部著作。<ref name=":1" /> 他们从理论上证明了人工神经元网络可以实现逻辑、算术和符号功能。生物神经元的简化模型models of biological neurons由此建立起来，现在通常被称为感知器perceptrons或人工神经元artificial neurons。这些简单的模型解释了神经加成作用neural summation(即突触后膜上的电位将在细胞体中加成)。后来的模型也提供了兴奋性和抑制性突触传递。

Connections display temporal and spatial characteristics. Temporal characteristics refers to the continuously modified activity-dependent efficacy of synaptic transmission, called [[spike-timing-dependent plasticity]]. It has been observed in several studies that the synaptic efficacy of this transmission can undergo short-term increase (called [[neural facilitation|facilitation]]) or decrease ([[Neural facilitation#Short-term depression|depression]]) according to the activity of the presynaptic neuron. The induction of long-term changes in synaptic efficacy, by [[long-term potentiation]] (LTP) or [[long-term depression|depression]] (LTD), depends strongly on the relative timing of the onset of the [[excitatory postsynaptic potential]] and the postsynaptic action potential. LTP is induced by a series of action potentials which cause a variety of biochemical responses. Eventually, the reactions cause the expression of new receptors on the cellular membranes of the postsynaptic neurons or increase the efficacy of the existing receptors through [[phosphorylation]].连接表现出时间和空间特征。时间特征是指突触传递的持续修饰的活动依赖的效能，称为峰时依赖的可塑性。多项研究发现，根据突触前神经元的活动，这种传递的突触效能可以经历短期的增加(称为易化)或减少(抑制)。通过长期增强(LTP)或抑制(LTD)诱导突触效能的长期变化，在很大程度上取决于兴奋性突触后电位和突触后动作电位的相对起病时间。LTP是由一系列动作电位引起的各种生化反应引起的。最终这些反应导致突触后神经元细胞膜上表达新的受体或通过磷酸化增加现有受体的效能。

Connections display temporal and spatial characteristics. Temporal characteristics refers to the continuously modified activity-dependent efficacy of synaptic transmission, called [[spike-timing-dependent plasticity]]. It has been observed in several studies that the synaptic efficacy of this transmission can undergo short-term increase (called [[neural facilitation|facilitation]]) or decrease ([[Neural facilitation#Short-term depression|depression]]) according to the activity of the presynaptic neuron. The induction of long-term changes in synaptic efficacy, by [[long-term potentiation]] (LTP) or [[long-term depression|depression]] (LTD), depends strongly on the relative timing of the onset of the [[excitatory postsynaptic potential]] and the postsynaptic action potential. LTP is induced by a series of action potentials which cause a variety of biochemical responses. Eventually, the reactions cause the expression of new receptors on the cellular membranes of the postsynaptic neurons or increase the efficacy of the existing receptors through [[phosphorylation]].连接表现出时间和空间特征。时间特征是指突触传递的持续修饰的活动依赖的效能，称为峰时依赖的可塑性。多项研究发现，根据突触前神经元的活动，这种传递的突触效能可以经历短期的增加(称为易化)或减少(抑制)。通过长期增强(LTP)或抑制(LTD)诱导突触效能的长期变化，在很大程度上取决于兴奋性突触后电位和突触后动作电位的相对起病时间。LTP是由一系列动作电位引起的各种生化反应引起的。最终这些反应导致突触后神经元细胞膜上表达新的受体或通过磷酸化增加现有受体的效能。

Backpropagating action potentials cannot occur because after an action potential travels down a given segment of the axon, the [[Depolarizing pre-pulse#Hodgkin–Huxley model|m gate]]s on [[voltage-gated sodium channel]]s close, thus blocking any transient opening of the [[Depolarizing pre-pulse#Hodgkin–Huxley model|h gate]] from causing a change in the intracellular sodium ion (Na⁺) concentration, and preventing the generation of an action potential back towards the cell body. In some cells, however, [[neural backpropagation]] does occur through the [[dendrite|dendritic branching]] and may have important effects on synaptic plasticity and computation.反向传播的动作电位是不可能发生的，因为当动作电位沿着轴突的某一特定节段传递之后，电压门控钠通道上的m门关闭，从而阻止h门的任何瞬态打开，以免引起细胞内钠离子浓度的变化。并阻止动作电位的产生回到细胞体内。然而，在某些细胞中，神经反向传播确实通过树突分支发生，并可能对突触可塑性和计算产生重要影响。

Backpropagating action potentials cannot occur because after an action potential travels down a given segment of the axon, the [[Depolarizing pre-pulse#Hodgkin–Huxley model|m gate]]s on [[voltage-gated sodium channel]]s close, thus blocking any transient opening of the [[Depolarizing pre-pulse#Hodgkin–Huxley model|h gate]] from causing a change in the intracellular sodium ion (Na⁺) concentration, and preventing the generation of an action potential back towards the cell body. In some cells, however, [[neural backpropagation]] does occur through the [[dendrite|dendritic branching]] and may have important effects on synaptic plasticity and computation.

 

反向传播的动作电位是不可能发生的，因为当动作电位沿着轴突的某一特定节段传递之后，电压门控钠通道上的m门关闭，从而阻止h门的任何瞬态打开，以免引起细胞内钠离子浓度的变化。并阻止动作电位的产生回到细胞体内。然而，在某些细胞中，神经反向传播确实通过树突分支发生，并可能对突触可塑性和计算产生重要影响。

 

A neuron in the brain requires a single signal to a [[neuromuscular junction]] to stimulate contraction of the postsynaptic muscle cell. In the spinal cord, however, at least 75 [[afferent nerve|afferent]] neurons are required to produce firing. This picture is further complicated by variation in time constant between neurons, as some cells can experience their [[Excitatory postsynaptic potential|EPSPs]] over a wider period of time than others.大脑中的某一神经元需要一个单一的信号到神经肌肉连接，刺激突触后肌肉细胞的收缩。然而，在脊髓中，产生放电需要至少75个传入神经元。由于神经元之间的时间常数变化，情形变得更加复杂，因为一些细胞会比其他细胞在更长的一段时间内感受到兴奋性突触后电位（EPSPs）。

大脑中的某一神经元需要一个单一的信号到神经肌肉连接，刺激突触后肌肉细胞的收缩。然而，在脊髓中，产生放电需要至少75个传入神经元。由于神经元之间的时间常数变化，情形变得更加复杂，因为一些细胞会比其他细胞在更长的一段时间内感受到兴奋性突触后电位（EPSPs）。

While in synapses in the [[development of the human brain|developing brain]] synaptic depression has been particularly widely observed it has been speculated that it changes to facilitation in adult brains.

While in synapses in the [[development of the human brain|developing brain]] synaptic depression has been particularly widely observed it has been speculated that it changes to facilitation in adult brains.

电路

电路

An example of a neural circuit is the [[trisynaptic circuit]] in the [[hippocampus]]. Another is the [[Papez circuit]] linking the [[hypothalamus]] to the [[limbic lobe]]. There are several neural circuits in the [[cortico-basal ganglia-thalamo-cortical loop]]. These circuits carry information between the cortex, [[basal ganglia]], thalamus, and back to the cortex. The largest structure within the basal ganglia, the [[striatum]], is seen as having its own internal microcircuitry.<ref name="Stocco">{{cite journal |last1=Stocco |first1=Andrea |last2=Lebiere |first2=Christian |last3=Anderson |first3=John R. |title=Conditional Routing of Information to the Cortex: A Model of the Basal Ganglia's Role in Cognitive Coordination |journal=Psychological Review |volume=117 |issue=2 |pages=541–74 |year=2010 |pmid=20438237 |doi=10.1037/a0019077 |pmc=3064519}}</ref>神经回路的典型例子是海马体中的三突触回路。另一个是连接下丘脑和边缘叶的帕佩兹回路。在皮层-基底神经节-丘脑-皮层环路中有几个神经回路。这些回路在皮层、基底神经节、丘脑之间传递信息，并将信息传回皮层。基底神经节内最大的结构，纹状体，被认为有自己的内部微电路。<ref name="Stocco" />

An example of a neural circuit is the [[trisynaptic circuit]] in the [[hippocampus]]. Another is the [[Papez circuit]] linking the [[hypothalamus]] to the [[limbic lobe]]. There are several neural circuits in the [[cortico-basal ganglia-thalamo-cortical loop]]. These circuits carry information between the cortex, [[basal ganglia]], thalamus, and back to the cortex. The largest structure within the basal ganglia, the [[striatum]], is seen as having its own internal microcircuitry.<ref name="Stocco">{{cite journal |last1=Stocco |first1=Andrea |last2=Lebiere |first2=Christian |last3=Anderson |first3=John R. |title=Conditional Routing of Information to the Cortex: A Model of the Basal Ganglia's Role in Cognitive Coordination |journal=Psychological Review |volume=117 |issue=2 |pages=541–74 |year=2010 |pmid=20438237 |doi=10.1037/a0019077 |pmc=3064519}}</ref>

 

神经回路的典型例子是海马体中的三突触回路。另一个是连接下丘脑和边缘叶的帕佩兹回路。在皮层-基底神经节-丘脑-皮层环路中有几个神经回路。这些回路在皮层、基底神经节、丘脑之间传递信息，并将信息传回皮层。基底神经节内最大的结构，纹状体，被认为有自己的内部微电路。<ref name="Stocco" />

Neural circuits in the [[spinal cord]] called [[central pattern generator]]s are responsible for controlling motor instructions involved in rhythmic behaviours. Rhythmic behaviours include walking, [[urination]], and [[ejaculation]]. The central pattern generators are made up of different groups of [[spinal interneuron]]s.<ref name="Guertin">{{cite journal |last1=Guertin |first1=PA |title=Central pattern generator for locomotion: anatomical, physiological, and pathophysiological considerations. |journal=Frontiers in Neurology |date=2012 |volume=3 |pages=183 |doi=10.3389/fneur.2012.00183 |pmid=23403923|pmc=3567435 }}</ref>

Neural circuits in the [[spinal cord]] called [[central pattern generator]]s are responsible for controlling motor instructions involved in rhythmic behaviours. Rhythmic behaviours include walking, [[urination]], and [[ejaculation]]. The central pattern generators are made up of different groups of [[spinal interneuron]]s.<ref name="Guertin">{{cite journal |last1=Guertin |first1=PA |title=Central pattern generator for locomotion: anatomical, physiological, and pathophysiological considerations. |journal=Frontiers in Neurology |date=2012 |volume=3 |pages=183 |doi=10.3389/fneur.2012.00183 |pmid=23403923|pmc=3567435 }}</ref>

在发散回路中，一个神经元与许多突触后细胞形成突触。这些神经元中的每一个都可能与更多的神经元形成突触，从而使一个神经元刺激多达数千个细胞成为可能。例如，从单个运动神经元的初始输入可以刺激到成千上万的肌纤维。<ref name="Saladin" />

在发散回路中，一个神经元与许多突触后细胞形成突触。这些神经元中的每一个都可能与更多的神经元形成突触，从而使一个神经元刺激多达数千个细胞成为可能。例如，从单个运动神经元的初始输入可以刺激到成千上万的肌纤维。<ref name="Saladin" />

In a converging circuit, inputs from many sources are converged into one output, affecting just one neuron or a neuron pool. This type of circuit is exemplified in the [[respiratory center]] of the [[brainstem]], which responds to a number of inputs from different sources by giving out an appropriate breathing pattern.<ref name="Saladin" />在收敛电路中，来自多个源的输入收敛成一个输出，只影响一个神经元或神经元池。脑干的呼吸中枢就是这种回路的典型例子，它通过发出适当的呼吸模式来响应来自不同来源的大量输入信号。<ref name="Saladin" />

In a converging circuit, inputs from many sources are converged into one output, affecting just one neuron or a neuron pool. This type of circuit is exemplified in the [[respiratory center]] of the [[brainstem]], which responds to a number of inputs from different sources by giving out an appropriate breathing pattern.<ref name="Saladin" />

 

在收敛电路中，来自多个源的输入收敛成一个输出，只影响一个神经元或神经元池。脑干的呼吸中枢就是这种回路的典型例子，它通过发出适当的呼吸模式来响应来自不同来源的大量输入信号。<ref name="Saladin" />

A reverberating circuit produces a repetitive output. In a signalling procedure from one neuron to another in a linear sequence, one of the neurons may send a signal back to initiating neuron.

A reverberating circuit produces a repetitive output. In a signalling procedure from one neuron to another in a linear sequence, one of the neurons may send a signal back to initiating neuron.

Each time that the first neuron fires, the other neuron further down the sequence fire again sending it back to the source. This restimulates the first neuron and also allows the path of transmission to continue to its output. A resulting repetitive pattern is the outcome that only stops if one or more of the synapses fail, or if an inhibitory feed from another source causes it to stop. This type of reverberating circuit is found in the respiratory center that sends signals to the [[Muscles of respiration|respiratory muscles]], causing inhalation. When the circuit is interrupted by an inhibitory signal the muscles relax causing exhalation. This type of circuit may play a part in [[epileptic seizure]]s.<ref name="Saladin" />混响电路产生重复的输出。在以线性顺序从一个神经元到另一个神经元的信号传递过程中，其中一个神经元可能会将信号发回初始神经元。每当第一个神经元发出信号时，另一个神经元就会再次发出信号，把信号送回信号源。这将重新刺激第一个神经元，并允许传输路径继续到它的输出。由此产生的重复模式只有在一个或多个突触失效，或来自另一个来源的抑制性馈电导致其停止时才会停止。这种类型的混响电路在呼吸中枢被发现，它向呼吸肌肉发送信号引起吸入。当回路被抑制信号打断时，肌肉就会放松导致呼气。这种类型的回路可能在癫痫发作中起作用。<ref name="Saladin" />

Each time that the first neuron fires, the other neuron further down the sequence fire again sending it back to the source. This restimulates the first neuron and also allows the path of transmission to continue to its output. A resulting repetitive pattern is the outcome that only stops if one or more of the synapses fail, or if an inhibitory feed from another source causes it to stop. This type of reverberating circuit is found in the respiratory center that sends signals to the [[Muscles of respiration|respiratory muscles]], causing inhalation. When the circuit is interrupted by an inhibitory signal the muscles relax causing exhalation. This type of circuit may play a part in [[epileptic seizure]]s.<ref name="Saladin" />

 

混响电路产生重复的输出。在以线性顺序从一个神经元到另一个神经元的信号传递过程中，其中一个神经元可能会将信号发回初始神经元。每当第一个神经元发出信号时，另一个神经元就会再次发出信号，把信号送回信号源。这将重新刺激第一个神经元，并允许传输路径继续到它的输出。由此产生的重复模式只有在一个或多个突触失效，或来自另一个来源的抑制性馈电导致其停止时才会停止。这种类型的混响电路在呼吸中枢被发现，它向呼吸肌肉发送信号引起吸入。当回路被抑制信号打断时，肌肉就会放松导致呼气。这种类型的回路可能在癫痫发作中起作用。<ref name="Saladin" />

In a parallel after-discharge circuit, a neuron inputs to several chains of neurons. Each chain is made up of a different number of neurons but their signals converge onto one output neuron. Each synapse in the circuit acts to delay the signal by about 0.5 msec so that the more synapses there are will produce a longer delay to the output neuron. After the input has stopped, the output will go on firing for some time. This type of circuit does not have a feedback loop as does the reverberating circuit. Continued firing after the stimulus has stopped is called ''after-discharge''. This circuit type is found in the [[reflex arc]]s of certain [[reflex]]es.<ref name="Saladin" />

In a parallel after-discharge circuit, a neuron inputs to several chains of neurons. Each chain is made up of a different number of neurons but their signals converge onto one output neuron. Each synapse in the circuit acts to delay the signal by about 0.5 msec so that the more synapses there are will produce a longer delay to the output neuron. After the input has stopped, the output will go on firing for some time. This type of circuit does not have a feedback loop as does the reverberating circuit. Continued firing after the stimulus has stopped is called ''after-discharge''. This circuit type is found in the [[reflex arc]]s of certain [[reflex]]es.<ref name="Saladin" />

The modern balance between the connectionist approach and the single-cell approach in [[neurobiology]] has been achieved through a lengthy discussion.

The modern balance between the connectionist approach and the single-cell approach in [[neurobiology]] has been achieved through a lengthy discussion.

In 1972, Barlow announced the ''single neuron revolution'': "our perceptions are caused by the activity of a rather small number of neurons selected from a very large population of predominantly silent cells."<ref name="Barlow1972">{{cite journal |last1=Barlow|first1=HB|title=Single units and sensation: a neuron doctrine for perceptual psychology? |journal=Perception|date=December 1, 1972|volume=1|issue=4 |pages=371-394|doi=10.1068/p010371|pmid=4377168}}</ref> This approach was stimulated by the idea of [[grandmother cell]] put forward two years earlier. Barlow formulated "five dogmas" of neuron doctrine.  Recent studies of '[[grandmother cell]]' and sparse coding phenomena develop and modify these ideas.<ref name="QuianQuiroga2005">{{cite journal |last1=Quian Quiroga|first1=R|last2=Reddy|first2=L|last3=Kreiman|first3=G|last4=Koch|first4=C|last5=Fried|first5=I|title=Invariant visual representation by single neurons in the human brain|journal=Nature|date=Jun 23, 2005|volume=435|issue=7045|pages=1102-1107|doi=10.1038/nature03687|doi-access=free|pmid=15973409}}</ref> The single cell experiments  used intracranial electrodes in the medial temporal lobe (the hippocampus and surrounding cortex). Modern development of [[concentration of measure]] theory (stochastic separation theorems) with applications to [[artificial neural networks]] give mathematical background to unexpected effectiveness of small neural ensembles in high-dimensional brain.<ref name=":2">{{cite journal |last1= Gorban|first1= Alexander N.|last2= Makarov|first2= Valeri A.|last3= Tyukin |first3= Ivan Y.|date= July 2019|title= The unreasonable effectiveness of small neural ensembles in high-dimensional brain|journal= Physics of Life Reviews|volume= 29 |pages= 55–88|doi= 10.1016/j.plrev.2018.09.005|doi-access=free|pmid= 30366739|arxiv= 1809.07656}}</ref>在神经生物学中，连接主义方法和单细胞方法之间的现代平衡已经通过长时间的讨论实现。1972年，巴洛宣布了“单一神经元革命”：“我们的感知是由从大量沉默细胞中选择的少量神经元的活动引起的。‘’<ref name="Barlow1972" /> 这种方法是受到两年前提出的祖母细胞的启发。巴洛提出了神经元学说的“五大信条”。最近对“祖母细胞”和稀疏编码现象的研究进一步完善了这些观点。<ref name="QuianQuiroga2005" />单细胞实验使用位于内侧颞叶(海马和周围皮层)的颅内电极。度量集中理论(随机分离定理)的现代发展及其在人工神经网络中的应用为高维大脑中小型神经系统集成的意想不到的有效性提供了数学背景。<ref name=":2" />

In 1972, Barlow announced the ''single neuron revolution'': "our perceptions are caused by the activity of a rather small number of neurons selected from a very large population of predominantly silent cells."<ref name="Barlow1972">{{cite journal |last1=Barlow|first1=HB|title=Single units and sensation: a neuron doctrine for perceptual psychology? |journal=Perception|date=December 1, 1972|volume=1|issue=4 |pages=371-394|doi=10.1068/p010371|pmid=4377168}}</ref> This approach was stimulated by the idea of [[grandmother cell]] put forward two years earlier. Barlow formulated "five dogmas" of neuron doctrine.  Recent studies of '[[grandmother cell]]' and sparse coding phenomena develop and modify these ideas.<ref name="QuianQuiroga2005">{{cite journal |last1=Quian Quiroga|first1=R|last2=Reddy|first2=L|last3=Kreiman|first3=G|last4=Koch|first4=C|last5=Fried|first5=I|title=Invariant visual representation by single neurons in the human brain|journal=Nature|date=Jun 23, 2005|volume=435|issue=7045|pages=1102-1107|doi=10.1038/nature03687|doi-access=free|pmid=15973409}}</ref> The single cell experiments  used intracranial electrodes in the medial temporal lobe (the hippocampus and surrounding cortex). Modern development of [[concentration of measure]] theory (stochastic separation theorems) with applications to [[artificial neural networks]] give mathematical background to unexpected effectiveness of small neural ensembles in high-dimensional brain.<ref name=":2">{{cite journal |last1= Gorban|first1= Alexander N.|last2= Makarov|first2= Valeri A.|last3= Tyukin |first3= Ivan Y.|date= July 2019|title= The unreasonable effectiveness of small neural ensembles in high-dimensional brain|journal= Physics of Life Reviews|volume= 29 |pages= 55–88|doi= 10.1016/j.plrev.2018.09.005|doi-access=free|pmid= 30366739|arxiv= 1809.07656}}</ref>

 

在神经生物学中，连接主义方法和单细胞方法之间的现代平衡已经通过长时间的讨论实现。1972年，巴洛宣布了“单一神经元革命”：“我们的感知是由从大量沉默细胞中选择的少量神经元的活动引起的。‘’<ref name="Barlow1972" /> 这种方法是受到两年前提出的祖母细胞的启发。巴洛提出了神经元学说的“五大信条”。最近对“祖母细胞”和稀疏编码现象的研究进一步完善了这些观点。<ref name="QuianQuiroga2005" />单细胞实验使用位于内侧颞叶(海马和周围皮层)的颅内电极。度量集中理论(随机分离定理)的现代发展及其在人工神经网络中的应用为高维大脑中小型神经系统集成的意想不到的有效性提供了数学背景。<ref name=":2" />

==Clinical significance==

==Clinical significance==