Michael Levin Bioelectricity 101 Crash Course Lesson 35: Gene Regulatory Networks: Bioelectric Learning and Memory Summary

• Gene Regulatory Networks (GRNs) are interconnected sets of genes and their regulatory elements (like transcription factors) that control gene expression.
• GRNs are traditionally viewed as “wiring diagrams” that determine which genes turn on or off in response to other genes.
• This lesson introduces a dynamic perspective on GRNs, treating them as computational systems capable of learning and memory.
• “Learning” in a GRN means that its response to future inputs (gene activations) changes based on its past experiences (previous patterns of gene activation). This is not genetic mutation or epigenetic modification; it’s a change in the network’s dynamics.
• Different types of memory are defined for GRNs, inspired by concepts from neuroscience (like Pavlovian conditioning), including:
  • UCS-based memory (long-lasting response to a single stimulus)
  • Pairing memory (response to a combination of stimuli that wouldn’t trigger it individually)
  • Transfer memory (increased sensitivity to a stimulus after repeated exposure)
  • Associative memory (learning to associate a previously neutral stimulus with a response-triggering stimulus)
  • Consolidation Memory
• A computational algorithm can systematically test any given GRN model for these different types of memory.
• Real biological GRNs exhibit these memories more often than randomly generated networks of similar size and connectivity, suggesting that evolution may have favored memory-capable GRN architectures.
• Bioelectricity plays a crucial role in linking GRNs to physical changes in cells. Ion channels, membrane potential, and voltage gradients can both influence gene expression and be influenced by gene expression.
• The combination of GRN dynamics and bioelectric signaling creates a powerful system for information processing, pattern formation, and adaptive behavior. This has crucial implications for medicine as this memory phenomena can be a good candidate explainer for diverse responses of even a small number of key drugs/proteins on different genetic backgrounds or other experiences in the body.
• The “learning” doesn’t involve a structural or genetic change. No genes “change their minds” about regulatory relationship; instead, training just helps push the organism along in an attractor, making certain behaviors or states more or less like. This does NOT modify genetic relationships, this *modifies existing behaviors* into another form, as we can train organisms.

Michael Levin Bioelectricity 101 Crash Course Lesson 35: Gene Regulatory Networks: Bioelectric Learning and Memory

We’ve journeyed through the fascinating world of bioelectricity, learning how cells use electrical signals to communicate, coordinate their actions, and control fundamental processes like development and regeneration. We’ve also explored how the cytoskeleton, particularly microtubules, acts as the physical interface between bioelectric signals and cellular behavior. Now, we’re going to connect these concepts to another crucial aspect of biology: gene regulatory networks (GRNs).

Traditionally, GRNs are viewed as the “control circuits” of the cell. They determine which genes are turned on (expressed) or off (repressed) in a given cell at a given time. A GRN is like a complex wiring diagram, where genes regulate each other’s expression through proteins called transcription factors. These transcription factors bind to specific regions of DNA, either promoting or inhibiting the transcription of other genes. The textbook example helps illustrate this:

• Example: The lac Operon In bacteria, there are genetic circuits called lac operons which, as normally set, only when encountering lactose in the enviorment does genes “activate” for metabolizing the food.

This picture of GRNs is often static. It focuses on the connections between genes – which gene activates which, which gene represses which. But it doesn’t fully capture the dynamic, time-dependent nature of gene expression. Gene expression isn’t just a series of on/off switches; it’s a complex, constantly changing dance of molecular interactions. The output does not require gene connections, as an important finding from last lesson was; there is no changing or adding any new gene links here. Instead, there is learning and memory.

Think of it like this: a wiring diagram of a computer tells you which components are connected, but it doesn’t tell you what the computer will do when you run a specific program. The program’s behavior depends on the sequence of instructions and the data it processes. Similarly, the behavior of a GRN depends not just on its wiring but also on the sequence of inputs it receives – the patterns of gene activation and repression that it experiences over time.

This is where the concept of learning and memory in GRNs comes in. It’s a revolutionary idea, suggesting that GRNs aren’t just fixed circuits but adaptive systems that can change their behavior based on their past experiences. What do we mean by “learning” and “memory” in this context? We’re not talking about changes in the DNA sequence (mutations) or even epigenetic modifications (changes in how DNA is packaged and accessed). We’re talking about changes in the dynamics of the GRN – how it responds to future inputs based on its past history of inputs.

Let’s revisit the types of memory defined in the Biswas et al. (2021) paper we summarized:

• UCS-Based Memory (UM): Imagine a gene that’s normally off. You activate it with a specific stimulus (the Unconditioned Stimulus or UCS). Now, even after you remove the stimulus, the gene stays on. The GRN has “remembered” the stimulus and maintained the response.
• Pairing Memory (PM): You have two stimuli, a UCS that triggers a response and a Neutral Stimulus (NS) that doesn’t. If you activate both stimuli together repeatedly, the network “learns” the association. Now, even if you only present both the UCS, together, they create an effect!
• Transfer Memory (TM): Repeated exposure to a stimulus makes the GRN more sensitive to that stimulus in the future. The response becomes stronger or lasts longer. This is like becoming better at recognizing a particular signal.
• Associative Memory (AM): This is the most sophisticated form. You pair a UCS (which triggers a response) with an NS (which initially doesn’t). After repeated pairing, the NS alone can trigger the response. The NS has become a Conditioned Stimulus (CS). Think of Pavlov’s dogs, associating the bell (CS) with food (UCS).
  • Short Recall Associative Memory (SRAM): The association is temporary.
  • Long Recall Associative Memory (LRAM): The association is long-lasting.
• Consolidation Memory: With additional repeated exposure the changes stay much longer.

How do we actually test for these types of memory in a GRN? We use computational models. We represent the GRN as a set of Boolean equations. Each gene is a “node” in the network, and its state (ON or OFF) is determined by the states of other genes according to the Boolean rules (AND, OR, NOT). We can then simulate the network’s dynamics over time, applying different sequences of inputs (activating specific genes) and observing the resulting patterns of gene expression. The specific node activations are applied according to definition.

The key findings from the Biswas et al. (2021) paper are:

1. Real GRNs Exhibit Memory: Many of the biological GRNs they tested showed evidence of various types of memory, including associative memory.
2. Memory is Not Random: Biological GRNs are more likely to have memory than randomly generated networks of similar size and connectivity. This suggests that evolution may have favored GRN architectures that support learning and memory. This answers a long-running debate, or assumption that the genetic structures are basically just complicated biochemical interactions: that, if you simply link biological material that can effect one another in a similar manner, similar kinds of functionality (including “smart” ones) should arise too. Not the case, here, with GRNs; there appears to be something unique.
3. Memory Types are Correlated: The presence of one type of memory in a GRN often predicts the presence of other types. This suggests that certain network structures are conducive to multiple forms of memory.
4. The biological, actual world GRNs have higher overall degree and prevalence of memories.

Now, let’s bring bioelectricity back into the picture. How does all of this relate to the electrical signals in cells? Here’s the crucial link:

• Bioelectricity as an Input to GRNs: Changes in membrane potential, voltage gradients, and electric fields can directly influence gene expression. Ion channels, which control the flow of ions and thus the cell’s electrical state, can be regulated by transcription factors. This means that the bioelectric state of a cell can act as an input to the GRN, influencing which genes are turned on or off.
• GRNs Control Bioelectricity: Conversely, the genes within a GRN can control the expression of ion channels, ion pumps, and gap junctions. This means that the GRN can regulate the cell’s bioelectric state.

This creates a bidirectional feedback loop: Bioelectricity influences gene expression, and gene expression influences bioelectricity. This loop is a powerful mechanism for information processing and control.

Think of it like this:

1. External Stimulus: A cell receives an external signal, like a change in the concentration of a chemical, a mechanical force, or an electrical field.
2. Bioelectric Response: This stimulus alters the activity of ion channels, changing the cell’s membrane potential.
3. GRN Activation: This change in membrane potential (or the associated ion flow) activates specific transcription factors within the GRN.
4. Altered Gene Expression: The activated transcription factors alter the expression of target genes.
5. Cellular Response: The changes in gene expression lead to changes in cell behavior (e.g., migration, differentiation, secretion of signaling molecules).
6. Bioelectric Feedback: The altered gene expression also changes the expression of ion channels, pumps, and gap junctions, further modifying the cell’s bioelectric state.

This creates a dynamic, interconnected system where electrical signals and gene expression are constantly interacting. It’s this interplay that allows cells to learn, remember, and adapt to their environment.

The implications of this are profound:

• Understanding Development: Bioelectric signaling, mediated by GRNs, plays a crucial role in guiding the development of complex body structures. The “bioelectric code” we talked about earlier is, in part, implemented by the dynamic interactions between GRNs and electrical signals.
• Regenerative Medicine: By understanding how bioelectricity and GRNs control regeneration, we might be able to stimulate the regrowth of lost or damaged tissues.
• Cancer Therapy: Disruptions in bioelectric signaling and GRN dynamics are hallmarks of cancer. By targeting these disruptions, we might be able to develop new cancer treatments. The goal of returning rogue, unconnected cells back to being proper participants of a working physiological tissue.
• Explaining Variability: Individual differences in GRN “memories” (due to past experiences) could explain why people respond differently to drugs or why diseases manifest differently in different individuals. It also points to key ways of overcoming “drug resistance” when pathogens stop responding.
• Beyond gene editing: Current research has an emphasis, the paper and the researchers argue, for editing gene, modifying the code/data. But with GRN network properties – in addition to epigenetic controls like microRNAs – it can allow doctors or researches to potentially train existing circuits to perform some functions – without resorting to transgenic additions or changes.

This lesson provides a powerful new perspective on how cells and tissues function. It integrates the concepts of bioelectricity, the cytoskeleton, and gene regulatory networks into a unified framework for understanding how biological systems learn, adapt, and build complex structures. It’s a shift from a purely molecular, reductionist view to a more holistic, computational understanding of life.

Michael Levin Bioelectricity 101 Crash Course Lesson 35: Gene Regulatory Networks: Bioelectric Learning and Memory Quiz

1. What are Gene Regulatory Networks (GRNs)?

A) Networks of neurons in the brain.
B) Interconnected sets of genes and their regulatory elements.
C) The physical connections between cells in a tissue.
D) The electrical wiring of the body.

2. Traditionally, GRNs are viewed as:

A) Dynamic systems capable of learning.
B) “Wiring diagrams” showing which genes regulate each other.
C) Unimportant for understanding cell behavior.
D) Systems that operate independently of bioelectricity.

3. What does “learning” mean in the context of a GRN, as defined in this lesson?

A) A change in the DNA sequence of the genes.
B) A change in the way the GRN responds to future inputs based on its past experiences.
C) A change in the physical connections between genes.
D) The accumulation of knowledge in the cell’s memory.

4. Which of the following is NOT a type of memory defined for GRNs in this lesson?

A) UCS-based memory
B) Pairing memory
C) Associative memory
D) Genetic memory

5. What is the key difference between Short Recall Associative Memory (SRAM) and Long Recall Associative Memory (LRAM)?

A) SRAM involves chemical signals, while LRAM involves electrical signals.
B) SRAM is temporary, while LRAM is long-lasting.
C) SRAM is found only in plants, while LRAM is found only in animals.
D) SRAM involves changes in gene expression, while LRAM involves changes in protein structure.

6. How can we test for different types of memory in a GRN?

A) By sequencing the DNA of the genes.
B) By using computational models and simulating the network’s dynamics.
C) By observing the physical structure of the network under a microscope.
D) By measuring the concentration of proteins in the cell.

7. How is “training” defined in this study and in GRN memory overall?

A) Stimuli.
B) Node Activations.
C) Transient inputs that affect GRN function long term.
D) All of the Above.

8. True or False: Biological GRNs are less likely to exhibit memory than randomly generated networks of similar size.

A) True
B) False

9. What is the role of bioelectricity in relation to GRNs?

A) Bioelectricity has no relationship to GRNs.
B) Bioelectricity can influence gene expression, and gene expression can influence bioelectricity.
C) Bioelectricity only affects the structure of the cytoskeleton, not GRNs.
D) GRNs control bioelectricity, but bioelectricity has no effect on GRNs.

10. What is a key implication for Associative Memory found in GRNs, regarding use of drugs, for doctors?

A) Pairing a normal or mostly harmless stimuli, may act like a drug itself and produce a physiological repsonse.
B) All medicine becomes useless and irrelevant.
C) Cancer will immediately disppear if one shocks cancer tissue enough
D) We do not have bodies.

11. What is a configuration model of a real GRN in this study’s experiments?

A) Random.
B) Copies all existing structural attributes such as edge, and genes.
C) Preserves size, node/edge numbers, keeps direction of relationships intact, but Boolean operator is allowed to change
D) C and B.

12. What is meant by a system that exhibits the Markov property?

A) A new kind of neural system in monkeys
B) No memory exists.
C) Memory only can exist
D) There exists memory abilities in computers but not GRNs

13. In general, which is “smarter” (as measured by Memory Types available, degree, and quantity) from GRNs observed in living animals?

A) A round worm (invertebrate).
B) Frog (vertebrate).

14. Which of the following is NOT a potential application of understanding GRN memory?

A) Developing new cancer therapies.
B) Designing strategies for regenerative medicine.
C) Predicting weather patterns.
D) Understanding variability in drug responses.

15. GRN learning represents a form of:

A) Changing or creating any connections, rewiring.
B) Applying patterns of external sitmuli that result in changes, from those initial pattern inputs, long term.
C) Simply modifying “gene code”.
D) Only C and B

16. Does Levin/et. al. find Memory effects can emerge in *random networks*?

A) yes
B) no

17. True/False; an output node representing key processes cannot itself be a “feedback node”.

A) True
B) False

18. In this GRN context, a node repesents?

A) a single atom
B) a protein.
C) a gene
D) a bioelectric signal.

19. Pavlovian classical conditioning, often referenced in Levin’s work, and with regards to GRN and Basal Congition/learning, it represents:

A) a complex neural response only.
B) memory, the organism can change behavior upon receiving the initial “neutral signal.”
C) can emerge in very simple gene networks
D) B and C

20. The memory effect changes on a biological tissue occur via Boolean regulation within what time period:

A) The reproductive and lineage growth of whole, multicellular animals
B) Occurs between generations
C) A time period shorter than A or B: over seconds/minutes.
D) It is unknown.

Michael Levin Bioelectricity 101 Crash Course Lesson 35: Gene Regulatory Networks: Bioelectric Learning and Memory Answer Sheet

1. B

2. B

3. B

4. D

5. B

6. B

7. D

8. B

9. B

10. A

11. D

12. B

13. B

14. C

15. B

16. A

17. B

18. C

19. D

20. C

迈克尔·莱文 生物电 101 速成课程 第35课：基因调控网络：生物电学习和记忆 摘要

• 基因调控网络 (GRN) 是相互连接的基因及其调控元件（如转录因子）的集合，控制基因表达。
• 传统上，GRN 被视为“线路图”，决定哪些基因响应其他基因而开启或关闭。
• 本课引入了 GRN 的动态视角，将其视为能够学习和记忆的计算系统。
• GRN 中的“学习”意味着它对未来输入（基因激活）的反应会根据其过去的经验（先前的基因激活模式）而改变。 这不是基因突变或表观遗传修饰； 而是网络动态的变化。
• 为 GRN 定义了不同类型的记忆，灵感来自神经科学的概念（如巴甫洛夫条件反射），包括：
  • 基于 UCS 的记忆（对单一刺激的持久反应）
  • 配对记忆（对原本不会触发反应的组合刺激的反应）
  • 转移记忆（重复暴露后对刺激的敏感性增加）
  • 联想记忆（学习将先前中性的刺激与触发反应的刺激联系起来）
  • 巩固记忆。
• 计算算法可以系统地测试任何给定的 GRN 模型是否存在这些不同类型的记忆。
• 与类似大小和连接性的随机生成的网络相比，真实的生物 GRN *更频繁地*表现出这些记忆，这表明进化可能有利于具有记忆能力的 GRN 架构。
• 生物电在将 GRN 与细胞的物理变化联系起来方面起着至关重要的作用。 离子通道、膜电位和电压梯度既可以影响基因表达，也可以受基因表达的影响。
• GRN 动力学和生物电信号传导的结合创建了一个强大的信息处理、模式形成和适应性行为系统。 这对医学具有重要意义，因为这种记忆现象可以很好地解释即使是少数关键药物/蛋白质在不同基因背景或其他体内经历下的不同反应。
• “学习”不涉及结构或遗传变化。 没有基因“改变主意”关于调控关系； 相反，训练只是帮助生物体沿着吸引子前进，使某些行为或状态或多或少变得像。 这不会改变遗传关系，这会将现有的行为修改为另一种形式，因为我们可以训练生物体。

迈克尔·莱文 生物电 101 速成课程 第35课：基因调控网络：生物电学习和记忆

我们已经踏上了生物电的神奇世界之旅，了解了细胞如何利用电信号进行交流、协调其行为并控制发育和再生等基本过程。 我们还探索了细胞骨架，特别是微管，如何充当生物电信号和细胞行为之间的物理界面。 现在，我们将把这些概念与生物学的另一个关键方面联系起来：基因调控网络 (GRN)。

传统上，GRN 被视为细胞的“控制电路”。 它们决定了给定细胞在给定时间开启（表达）或关闭（抑制）哪些基因。 GRN 就像一个复杂的线路图，其中基因通过称为转录因子的蛋白质来调节彼此的表达。 这些转录因子与 DNA 的特定区域结合，促进或抑制其他基因的转录。 教科书上的例子有助于说明这一点：

• 例子：lac 操纵子 在细菌中，存在称为 lac 操纵子的基因回路，按照正常设置，只有在环境中遇到乳糖时，才会“激活”用于代谢食物的基因。

GRN 的这张图通常是静态的。 它侧重于基因之间的连接——哪个基因激活哪个基因，哪个基因抑制哪个基因。 但它并没有完全捕捉到基因表达的动态的、时间依赖性的性质。 基因表达不仅仅是一系列开关； 它是一个复杂的、不断变化的分子相互作用之舞。 输出不需要基因连接，正如上一课的一个重要发现； 这里没有改变或添加任何新的基因连接。 相反，存在学习和记忆。

可以这样想：计算机的线路图告诉你哪些组件是连接的，但它并没有告诉你当你运行特定程序时计算机将做什么。 程序的行为取决于指令序列和它处理的数据。 同样，GRN 的行为不仅取决于它的线路，还取决于它接收到的输入序列——它随时间经历的基因激活和抑制模式。

这就是 GRN 中学习和记忆概念的来源。这是一个革命性的想法，表明 GRN 不仅仅是固定的电路，而是适应性系统，可以根据其过去的经验改变其行为。 我们所说的“学习”和“记忆”在这个背景下是什么意思？ 我们不是在谈论 DNA 序列的变化（突变），甚至不是表观遗传修饰（DNA 的包装和访问方式的变化）。 我们谈论的是 GRN 动力学的变化——它如何根据其过去的输入历史来响应未来的输入。

让我们回顾一下我们在 Biswas 等人 (2021) 论文中定义的记忆类型：

• 基于 UCS 的记忆 (UM): 想象一个通常关闭的基因。 你用特定的刺激（非条件刺激或 UCS）激活它。 现在，即使在你移除刺激之后，基因仍然保持开启状态。 GRN 已经“记住”了刺激并维持了反应。
• 配对记忆 (PM): 你有两种刺激，一种会触发反应的 UCS 和一种不会触发反应的中性刺激 (NS)。 如果你反复一起激活两种刺激，网络就会“学习”这种关联。 现在，即使你只呈现 *两种* UCS，*一起*，它们也会产生效果！
• 转移记忆 (TM): 重复暴露于刺激会使 GRN 在未来对该刺激更敏感。 反应变得更强或持续时间更长。 这就像变得更擅长识别特定信号一样。
• 联想记忆 (AM): 这是最复杂的形式。 你将 UCS（触发反应）与 NS（最初不触发）配对。 在重复配对后，单独 NS 就可以触发反应。 NS 变成了条件刺激 (CS)。 想想巴甫洛夫的狗，把铃声 (CS) 和食物 (UCS) 联系起来。
  • 短时回忆联想记忆 (SRAM): 这种关联是暂时的。
  • 长时回忆联想记忆 (LRAM): 这种关联是持久的。
• 巩固记忆：通过额外的重复暴露，变化会持续更长时间

我们如何在 GRN 中实际测试这些类型的记忆？ 我们使用计算模型。 我们将 GRN 表示为一组布尔方程。 每个基因都是网络中的一个“节点”，其状态（开或关）由其他基因的状态根据布尔规则（与、或、非）决定。 然后，我们可以模拟网络随时间的动态，应用不同的输入序列（激活特定基因）并观察产生的基因表达模式。 根据定义，应用特定的节点激活。

Biswas 等人 (2021) 论文的主要发现是：

1. 真实的 GRN 表现出记忆：他们测试的许多生物 GRN 都显示出各种类型记忆的证据，包括联想记忆。
2. 记忆不是随机的：与类似大小和连接性的随机生成的网络相比，生物 GRN 更可能具有记忆。 这表明进化可能有利于支持学习和记忆的 GRN 架构。 这回答了一个长期存在的争论，或者假设遗传结构基本上只是复杂的生化相互作用：如果你简单地连接能够以类似方式相互影响的生物材料，类似的功能（包括“智能”功能）也应该出现。 在这里，对于 GRN 来说，情况并非如此； 似乎有些东西是独特的。
3. 记忆类型是相关的：GRN 中一种类型记忆的存在通常预示着其他类型的存在。 这表明某些网络结构有利于多种形式的记忆。
4. 与随机网络相比，真实的，现实世界的 GRN具有更高的整体记忆程度和普遍性。

现在，让我们把生物电带回画面。 所有这些与细胞中的电信号有什么关系？ 这是关键的联系：

• 生物电作为 GRN 的输入： 膜电位、电压梯度和电场的变化可以直接影响基因表达。 控制离子流从而控制细胞电状态的离子通道可以受转录因子调节。 这意味着细胞的生物电状态可以充当 GRN 的输入，影响哪些基因被开启或关闭。
• GRN 控制生物电： 相反，GRN 内的基因可以控制离子通道、离子泵和间隙连接的表达。 这意味着 GRN 可以调节细胞的生物电状态。

这创建了一个双向反馈回路：生物电影响基因表达，基因表达影响生物电。 这个循环是信息处理和控制的强大机制。

可以这样想：

1. 外部刺激： 细胞接收外部信号，如化学物质浓度的变化、机械力或电场。
2. 生物电响应： 这种刺激会改变离子通道的活性，从而改变细胞的膜电位。
3. GRN 激活： 膜电位的这种变化（或相关的离子流）会激活 GRN 内的特定转录因子。
4. 基因表达改变： 激活的转录因子会改变靶基因的表达。
5. 细胞反应： 基因表达的变化导致细胞行为的变化（例如，迁移、分化、分泌信号分子）。
6. 生物电反馈： 改变的基因表达也会改变离子通道、泵和间隙连接的表达，进一步改变细胞的生物电状态。

这创建了一个动态的、相互关联的系统，其中电信号和基因表达不断相互作用。正是这种相互作用使细胞能够学习、记忆和适应它们的环境。

其含义是深远的：

• 理解发育： 由 GRN 介导的生物电信号传导在指导复杂身体结构的发育中起着至关重要的作用。 我们之前谈到的“生物电代码”在某种程度上是由 GRN 和电信号之间的动态相互作用实现的。
• 再生医学： 通过了解生物电和 GRN 如何控制再生，我们也许能够刺激失去或受损组织的再生。
• 癌症治疗： 生物电信号传导和 GRN 动力学的破坏是癌症的标志。 通过靶向这些破坏，我们也许能够开发出新的癌症疗法。 让失控的、未连接的细胞重新成为正常生理组织的一部分的目标。
• 解释变异性： GRN“记忆”的个体差异（由于过去的经验）可以解释为什么人们对药物的反应不同，或者为什么疾病在不同个体中的表现不同。 它还指出了克服病原体停止反应时“耐药性”的关键方法。
• 超越基因编辑： 目前的研究重点是编辑基因，修改代码/数据。 但是有了 GRN 网络特性——除了 microRNA 等表观遗传控制——它可以让医生或研究人员有可能训练现有电路来执行某些功能——而无需进行转基因添加或更改。

本课程提供了关于细胞和组织如何运作的强有力的新视角。 它将生物电、细胞骨架和基因调控网络的概念整合到一个统一的框架中，以了解生物系统如何学习、适应和构建复杂的结构。 这是从纯粹的分子、还原论观点向更全面、计算的生命理解的转变。

迈克尔·莱文 生物电 101 速成课程 第35课：基因调控网络：生物电学习和记忆 小测验

1. 什么是基因调控网络 (GRN)？

A) 大脑中的神经元网络。
B) 相互连接的基因及其调控元件的集合。
C) 组织中细胞之间的物理连接。
D) 身体的电线。

2. 传统上，GRN 被视为：

A) 能够学习的动态系统。
B) 显示哪些基因相互调节的“线路图”。
C) 对于理解细胞行为不重要。
D) 独立于生物电运行的系统。

3. 在本课中定义的 GRN 的背景下，“学习”是什么意思？

A) 基因 DNA 序列的变化。
B) GRN 根据其过去的经验对未来输入的反应方式发生变化。
C) 基因之间物理连接的变化。
D) 细胞记忆中知识的积累。

4. 以下哪一项不是本课中为 GRN 定义的记忆类型？

A) 基于 UCS 的记忆
B) 配对记忆
C) 联想记忆
D) 遗传记忆

5. 短时回忆联想记忆 (SRAM) 和长时回忆联想记忆 (LRAM) 之间的主要区别是什么？

A) SRAM 涉及化学信号，而 LRAM 涉及电信号。
B) SRAM 是暂时的，而 LRAM 是持久的。
C) SRAM 仅存在于植物中，而 LRAM 仅存在于动物中。
D) SRAM 涉及基因表达的变化，而 LRAM 涉及蛋白质结构的变化。

6. 我们如何测试 GRN 中不同类型的记忆？

A) 通过对基因的 DNA 进行测序。
B) 通过使用计算模型并模拟网络的动态。
C) 通过在显微镜下观察网络的物理结构。
D) 通过测量细胞中蛋白质的浓度。

7. 在本研究和 GRN 记忆中，“训练”是如何定义的？

A) 刺激。
B) 节点激活。
C) 长期影响 GRN 功能的瞬时输入。
D) 以上都是。

8. 对或错：与类似大小的随机生成的网络相比，生物 GRN 不太可能表现出记忆。

A) 对
B) 错

9. 生物电与 GRN 的关系是什么？

A) 生物电与 GRN 没有关系。
B) 生物电可以影响基因表达，基因表达也可以影响生物电。
C) 生物电只影响细胞骨架的结构，不影响 GRN。
D) GRN 控制生物电，但生物电对 GRN 没有影响。

10. 在 GRN 中发现的联想记忆对于医生使用药物有什么重要意义？

A) 配对正常或基本无害的刺激，可能像药物一样起作用并产生生理反应。
B) 所有的药物都变得毫无用处和无关紧要。
C) 如果一个人对癌组织进行足够的电击，癌症就会立即消失
D) 我们没有身体。

11. 在本研究的实验中，真实 GRN 的配置模型是什么？

A) 随机的。
B) 复制所有现有的结构属性，如边缘和基因。
C) 保留大小、节点/边缘数，保持关系的方向完整，但允许布尔运算符更改
D) C 和 B。

12. 系统表现出马尔可夫性质是什么意思？

A) 猴子中的一种新型神经系统
B) 不存在记忆。
C) 记忆只能存在
D) 计算机中存在记忆能力，但 GRN 中不存在

13. 一般来说，从活体动物中观察到的 GRN 来看，哪一个“更聪明”（根据可用的记忆类型、程度和数量来衡量）？

A)蛔虫（无脊椎动物）。
B) 青蛙（脊椎动物）。

14. 以下哪一项不是理解 GRN 记忆的潜在应用？

A) 开发新的癌症疗法。
B) 设计再生医学策略。
C) 预测天气模式。
D) 了解药物反应的变异性。

15. GRN 学习代表一种：

A) 改变或创建任何连接，重新布线。
B) 应用外部刺激模式，这些模式会导致从最初的模式输入长期变化。
C) 简单地修改“基因代码”。
D) 仅 C 和 B

16. Levin/等人 发现记忆效应会出现在*随机网络*中吗？

A) 是的
B) 否

17. 对/错：代表关键过程的输出节点本身不能是“反馈节点”。

A) 正确
B) 错误

18. 在这个 GRN 上下文中，节点代表什么？

A) 单个原子
B) 一种蛋白质。
C) 一个基因
D) 一种生物电信号。

19. 经常被 Levin 的工作提及的巴甫洛夫经典条件反射，以及关于 GRN 和基础认知/学习，它代表：

A) 仅是复杂的神经反应。
B) 记忆，生物体在接收到最初的“中性信号”后可以改变行为。
C) 可以出现在非常简单的基因网络中
D) B 和 C

20. 生物组织上的记忆效应变化通过布尔调节发生在什么时间段内：

A) 整个多细胞动物的繁殖和谱系生长
B) 代际之间发生
C) 比 A 或 B 更短的时间段：超过几秒/几分钟。
D) 这是未知的。

迈克尔·莱文 生物电 101 速成课程 第35课：基因调控网络：生物电学习和记忆 答案表

1. B

2. B

3. B

4. D

5. B

6. B

7. D

8. B

9. B

10. A

11. D

12. B

13. B

14. C

15. B

16. A

17. B

18. C

19. D

20. C