We know the world through at least two "lenses": eyes+model, and human perception of the environment and cognition of the world through at least two "lenses":
1. Eyes of the body
2. Eye of the mind
The former is the eyes of the body (pupil+lens+retina, etc.). ); The latter is a model of the mind, and the world we see has been filtered by the model.
In daily work and life, various models can be seen everywhere:
1. Business models: Five Forces Model, McKinsey 7S Model, Marketing 4P Model, AARRR Pirate User Growth Model, etc.
2. Economic models: supply and demand curve model, "prisoner's dilemma" game model, GDP growth model, etc.
There are Maslow's hierarchy of needs model in psychology, von Neumann model in computer science and neural network model in artificial intelligence.
In the history of scientific development, there are countless models, such as the well-known geocentric VS Heliocentrism model, the atomic structure model and the double helix structure model of DNA molecules.
When we face the "model", our psychology is generally contradictory and embarrassing. On the one hand, we know that abstract models can't solve practical problems; On the other hand, we have to rely on models when analyzing problems.
What is a model? A model is like a map, which is a simplification and abstraction of the real world. With a map, you may not be able to reach your destination; Without a map, it's almost certain that you won't get to your destination.
"The map is not equal to the territory it refers to. However, if the map is correct, it has the same structure as the territory, which is why it is useful. " -Alfred Colz Butch (Polish-American scientist and philosopher)
This model is both useful and useless. How did the contradiction arise?
First look at the role and value of the model:
The pond stinks for several kilometers, so the workers scoop out the fish and drain the water. But the bottom of the pool still stinks, so I dug up the mud at the bottom of the pool and carried it away. I spread fresh gravel on the bottom of the pool, replanted aquatic plants, filled it with water and put the fish back. Finally, the pool no longer stinks. But two months later, the pond stinks again.
Why can't the problem be cured? Because the construction of pond microecological model has defects. The reason why sludge is dug out is based on the hypothesis that the sludge accumulated at the bottom of the pond is too thick, which leads to the rapid proliferation of anaerobic bacteria that can produce odor. But why does mud stink when it is removed? It shows that silt is not the root cause.
Because of the error of system modeling, the root of the problem was misjudged, and then measures were taken to treat the symptoms rather than the root cause.
When considering the problem of pond odor, you may not have that random circuit diagram in your mind, but you will basically simplify the pond system into several key elements. A complete pond ecosystem includes: 1. Bacteria (aerobic+anaerobic): countless; 2. Animals such as fish and shrimp: tens of thousands; 3. Aquatic plants: thousands; 4. Oxygen and odorous gas molecules: unmeasurable.
If it is not simplified, then the brain will face an unsolvable problem that is difficult to calculate. If there are only two objects in a system and there is a relationship between them, then three equations are needed to describe the system: two equations are used for the two objects, and 1 equation is used for their interaction. With the increase of the number of objects in the system, each object needs 1 equations to describe its behavior, but the equations describing the interaction will increase exponentially, and n objects need 2+0 equations. That is to say, the system composed of 10 objects has two 10 = 1024 equations (assuming n= 10000? )。 This is the so-called "calculation square law". When the number of interacting elements in the system increases, the calculation difficulty will increase exponentially.
Model is the basic way and powerful ability for human beings to understand the world.
If a system contains a large number of elements, and there are complicated relationships between them, in order to simplify, we create a model to describe the system and take corresponding actions to influence the system. Simplification naturally saves cognitive resources and reduces cognitive burden. However, from a more essential level, modeling thinking is one of the most basic and important abilities for human beings to understand the world.
Hawking wrote in his last book "Grand Design" before his death: "We build models in scientific exploration, but in fact, we also create models in daily life. "Model-dependent realism" is not only applicable to scientific models, but also to psychological models that we consciously or unconsciously establish to explain and understand the world. "
We make models in science, but we also make models in our daily life. Model-dependent realism applies not only to scientific models, but also to conscious and subconscious mental models created by us to explain and understand the daily world.
Suppose there is a fish tank, in which goldfish observe the outside world through the curved glass of the fish tank. Now physicists among them have begun to develop "goldfish physics". They sum up the observed phenomena and establish the laws of physics. These laws can describe and explain the external world observed by goldfish through fish tanks, and even correctly predict possible new phenomena in the external world. Obviously, the physical laws of these goldfish must be very different from those of us. For example, the linear motion we see may be curved in goldfish physics.
In this regard, Hawking raised a question: Is this "goldfish physics" correct? According to the conventional concept, such "goldfish physics" is of course incorrect. Because "goldfish physics" conflicts with our physical laws, and because we think that our physical laws are more in line with objective laws and can better reflect the real world, all descriptions that are inconsistent with today's physics, whether from goldfish physicists or previous human physicists, are judged to be incorrect.
But Hawking asked, "How do we know if we have a distorted real image? ..... The real image of goldfish is different from ours, but can we be sure that it is more unreal than ours? "
The model is the projection of the environment in the brain.
We think that the world seen by goldfish in the fish tank is distorted by the distortion of light. So, isn't the world we perceive distorted? The three people in the picture below are actually the same height, but because the room is not a standard cuboid, the angle between the two walls and the wall where the window is located and the shape of the window have changed, which leads us to think that the three people are not the same height.
The world we perceive (see, hear, smell, taste, touch, etc. ) will deviate or even distort like the world seen by goldfish, because what we perceive is not the world, but the signal from the "world". The brain and related nervous system form a communication system. The information of the surrounding environment enters the brain through the media of light and air, and through the channels of vision, hearing, taste, smell and touch. These signal sources are like raw materials when baking bread, and the brain processes them into shaped "bread". This "bread" is a model of the world. The psychological representation of the model to the world is the projection of the environment in the brain.
The so-called "psychological representation" means that the state presented by the physical world reappears in the brain, but it is not 100% reduction, but processing and even deformation. Therefore, we are not so much perceiving the world as perceiving a model about the world and using it to adjust our behavior in the environment. We know the world through "models" and take action in it.
This is what Hawking means by "model-dependent realism". We don't know what the "reality" of the world is, but we need to rely on and rely on models to approach that "reality" indirectly. Why is the simplified model effective in complex reality? "According to the realism of dependent models, it is meaningless to ask whether a model is true, but it is meaningful only if it is consistent with observation. If there are two models consistent with observation, just like the image of goldfish and our image, then people can't say that this is more real than that. Under the circumstances considered, which one will be more convenient. " -"Great design"
The model does not reflect the reality in detail, so why is the simplified model effective? To answer this question, it needs to be divided into two parts:
1. Why does the brain build a simplified model?
2. Why is the model effective for cognition and behavior?
Why does the brain build a simplified model?
The brain is bombarded by the flood of information from the internal and external environment at any time. Externally, it is the light, color, sound and other signals received by sensory organs, especially vision and hearing. Internally, it is the body's temperature, respiration, heartbeat, hormone secretion and other signals (generally controlled by the autonomic nervous system). Brain neurons need glucose, oxygen and time to process information. In order to survive and reproduce more efficiently, the human body has evolved a set of attention filtering mechanism, which can automatically block out a lot of information that the brain thinks is irrelevant.
For example, you drove across a highway yesterday. Now let's try to recall the whole process. How many ramps can you remember? Where do they lead? What's your top speed? Is the toll booth staff a woman or a man? I think you are as difficult to answer as I am, because the information is neither important nor novel.
According to statistics, the bandwidth of information processing in the conscious state of the brain is about 120 bits/second, but the information transmitted by the retina alone reaches 10 megabit/second. In other words, only a very small amount of information enters the conscious space.
Working memory is like an octopus, combining information into meaning.
Working memory system can hold 4 to 7 pieces of information (or information strings, also called chunks) at the same time. The size of an information string or block depends on the person's previous knowledge structure. For example, letters representing the four directions of southeast, northwest, north, n, east, e, west, w, south and s can be counted as four pieces of information, and can also be regarded as a large piece of news, provided that you have the word news in your long-term memory.
Working memory works like an octopus, quickly combining input information into meaningful structures. However, this octopus has less than eight legs, maybe only four or seven. If the information processed by working memory is important, it will be stored in the long-term memory system. It is conceivable that only a few "big fish" were salvaged by the aforementioned "fishing nets" and put into the freezer.
Only a very small amount of information enters people's attention, only a few of which are captured by working memory system, and finally the remaining few are successfully saved in long-term memory. In order to effectively reflect the external world, the brain must have enough ability to represent the infinite environment with limited information.
The intelligence and strength of the brain lies in discovering and recognizing patterns.
The picture below is made up of black stripes of different shapes and sizes. When I ask you what it is or what you see, you may look blank and wonder what it is.
"Conditionally; If there are no conditions, we must create conditions. " And the brain is, if there is a pattern, use it; If there is no pattern, create one. Patterns represent meaning, phenomena change from uncertainty to certainty, and the brain hates uncertainty extremely. Therefore, although a small amount of information enters the brain, especially at the level of consciousness, the brain is good at using this information, processing it into a model, and using it to simulate and map the outside world.
Why is the simplified model effective for cognition and behavior?
What tests the value of a model is not whether it is true, but whether it is consistent with observation (through practice or experiment) and whether it is useful. Since the reality is complicated and the model is simplified, why is the model effective? To answer this question, we need to turn it into: Why can the real world be simplified into a model? In other words, why is the real world equivalent to a model to some extent? It's like several children saying, let's build a spaceship with Lego. The premise of playing this game is that a complex spaceship (more than 65,438+million parts) can be "simulated" by dozens or hundreds of Lego components. Why can the real world be modeled? There are two main reasons:
1. The system usually has a hierarchical structure.
There are many redundancies in the real world.
The surrounding system usually has a hierarchy.
Hierarchical structure, vividly understood as pyramid structure. Things around us and our own behaviors can be decomposed into hierarchical structures. (A) the performance of the spatial structure:
(Hierarchical structure of human body: body-system-organ-tissue-cell-molecule-atom)
(B) the performance of the time series:
Human behavior also presents a hierarchical structure in steps and processes. 1, cooking:
2. Project:
The hierarchical structure of stages and steps of a project can be called a recursive process in mathematical language: from top to bottom, from left to right.
The upper structure is composed of the lower structure, which is the foundation of the upper structure, and the upper structure forms constraints and controls on the lower structure. Generally speaking, the more stable the upper layer, the less the number of elements; On the contrary, the lower the level, the more elements and the greater the change.
The number of cells in the human body must be much greater than the number of organs and tissues; At the same time, cells are constantly updated, but the morphology and function of organs and tissues remain basically unchanged. Take eating in a restaurant as an example. Choosing dishes (Sichuan, Guangdong or Hunan) is a higher-level structure than ordering food. It is impossible to choose Cantonese cuisine with Mao Wangxue as the main ingredient. The choice of dishes is the lower structure of dinner theme and scene. If it is a business banquet, it is impossible to eat in a street shop.
Compared with the real world, the model is simplified because the upper structure of the modeled object (system) is selected in the modeling process. When we model the pond system as a whole, we will not choose a few fish or a piece of sludge. Anaerobic bacteria in the model refer not to some bacteria, but to all anaerobic bacteria. The upper structure is the collection and abstraction of the lower components, which is naturally streamlined in number. Furthermore, if the number of model elements is not equal to or less than that which can be controlled by working memory, then the analysis and processing of the brain will be extremely difficult.
There is a lot of redundancy in the real world.
The above picture lists the writing methods of the word "book" in different periods and different fonts. Whether it's traditional Chinese characters or simplified Chinese characters, whether it's official script or cursive script, we can recognize this word. When typing on the computer, we can also choose different fonts, such as bold, regular, round, square and elegant black. Although there are different forms, which contain a stable and consistent pattern, although it is difficult for us to describe this pattern in words, we can recognize the word "book".
The so-called "redundancy" is something outside this stable and consistent model. Even if it is deleted or removed, it will not affect the information contained in it. Information is the pattern left after removing message redundancy. For example, a series of Fibonacci series: 1, 1, 2, 3, 5, 8, 13, 2 1 ... After removing redundancy, the formula is: F(0)=0, f (/kloc-0.
A new typhoon is generated in the South China Sea and will soon reach the Pearl River Delta. Even if you read this sentence quickly, you can understand its meaning, despite grammatical errors and word gaps. The reason is that language is generally redundant, even if some words are lost or the local order is changed, people can understand the meaning.
. It is relatively easy to understand that there are redundancies in the information we come into daily contact with (precisely, the signals and symbols carrying information). So, how can we say "there is a lot of redundancy in the real world"? A child over 2 years old, even if he has only seen a few cats in the community, can basically recognize a cat when he goes outside, whether he sees a big cat or a kitten, a white cat or a black cat, a brown-eyed cat or a cross-eyed cat. This shows that there is a stable and consistent cat model in children's memory, that is, the concept of cat. Outside the pattern, no matter the difference in size, size, color and gender characteristics, it can be called redundancy.
This pattern (consistency)+redundancy (diversity) exists in common things: animals, plants, languages, products, buildings and so on. In other words, this is the basic attribute of the real world.
. The ultimate goal of biological evolution is to adapt to the environment. To some extent, the structure of the human brain is a "mirror image" of the environmental structure. The human brain is different and "advanced" from our close relatives-apes (gorillas, chimpanzees, gibbons, etc.). ) and the cerebral cortex of other mammals. The proportion of neocortex in human brain is the largest (80%) and the number of neurons is the largest (about 65.438+0.6 billion). The human brain is not the largest in absolute volume and weight, or in proportion to the size and weight of the body; The number of neurons is not the largest. The biggest and heaviest brain is the blue whale; Elephants have more than 250 billion neurons, while humans have about 86 billion. However, there are far more neurons in human cerebral cortex than elephants.
It is no coincidence that the neocortex of the brain forms such a six-layer structure. The information it receives will flow from the bottom up. As information is transmitted to the upper level step by step, different sensory information (vision/hearing/smell/touch, etc. ) will gradually integrate and finally form an unchanging representation at the top. The upward flow of information in this hierarchical structure is a process of constantly eliminating redundancy. When children see different cats, the information about the specific characteristics of cats (body shape, coat color, voice, etc.). ) into the lower neurons, upward transmission, gradually extract the cat's identity characteristics, stored in the upper structure. According to the hierarchical characteristics of the system, the higher the structure, the more stable it is. This is the so-called "constant representation", and it is also the place where the model is generated and saved.
"Humans ... have a strong core ability of pattern recognition. In order to carry out logical thinking, we need the help of the new cerebral cortex, which is itself the largest pattern recognizer. " ("How to Create Thinking" ray kurzweil)
The real world is a complex system, which consists of small complex systems. Institution and institution are interrelated and influence each other. Systems usually have hierarchical and redundant properties. Through AHP, we can capture the upper structure of a system; By eliminating redundancy, we can extract the core mode of the system. Therefore, the process of modeling is also a process of hierarchical analysis and redundancy elimination (of course, these two processes may overlap or cross).
The practical value of the model lies in helping to solve problems.
"If we must seriously think about the world and take effective actions, some simplified realistic maps, theories, concepts, models and styles are necessary." -samuel huntington (author of "Clash of Civilizations and Reconstruction of World Order")
Maps, theories, concepts or patterns can basically be collectively referred to as models. The value of the model for real life lies in "thinking about the world" and "taking action", specifically, helping to solve problems. The problems are numerous and complex, but they can be roughly divided into four categories:
1. What is this? (situation identification)
2. What is the reason? (problem analysis)
3. What are the countermeasures? What can I do? (decision analysis)
4. What about the future? (predictive analysis)
Solving every problem is inseparable from the acquisition of information (or intelligence). The core of intelligence: all intelligent activities involve creating a target model and then extracting knowledge from it (as in all problem-solving processes). "-Robert Clarke (American senior intelligence analyst, author of" Intelligence Analysis: A Goal-centered Approach ") In the aforementioned Ponzi case, the root cause and solution of analyzing the problem depend on the creation and optimization of the model.
The so-called "goal model" means that the problems we face and the objects we will take action are often complex systems. In order to obtain key information and design effective actions, we must model the system, analyze the valuable information in the model and conceive an action plan.
The 2020 American election is just around the corner. Will Trump be re-elected, or will Biden finally win? The whole world is paying attention to and predicting this event that affects the world situation. The American election is obviously a complicated system. Which candidate can win is influenced by several factors: popular support rate, swing states, candidates' health, the influence of military industrial groups, the support tendency of powerful consortia and so on. Only by modeling can we sort out the key influencing factors.
These 13 indicators include:
1. political party authorization: after the mid-term election, the number of ruling party seats in the us house of representatives increased.
2. Competition: There is no fierce competition between the presidential candidates of the ruling party.
3. Seeking re-election: the candidate of the ruling party is the current president.
4. Third Party: There are no important third parties or independent candidates.
5. Short-term economy: There was no recession during the election campaign.
6. Long-term economy: The actual per capita economic growth during the term of office is equal to or more than the average growth of the previous two presidential terms.
7. Policy change: The current president has a great influence on national policy.
8. Social unrest: there was no continuous social unrest during his tenure.
9. Scandal: There are no major scandals in the current government.
10. diplomatic and military failures: the current government has not made major diplomatic and military mistakes.
1 1. diplomatic and military achievements: this government has achieved great success in diplomacy and military affairs.
12. Personal charm of the incumbent: the candidate of the ruling party is extremely charming or a national hero.
13. Challenger's personal charm: The candidate of the opposition party has no personal charm and is not a national hero.
Leachman believes that in this year's US presidential election, Trump's 13 key indicators have seven negative indicators, which indicates that his re-election may fail.
What matters is not how many models you have mastered, but the ability to model problems effectively.
There are two ways to know the black box: opening the black box and not opening the black box. The way to open the black box is to take the system apart and see it clearly. For example, if you don't know how the clock works, take it apart with a screwdriver and observe the composition and operation of the parts. If the black box is not opened, the structure and operation of the black box can be inferred according to the relationship between input and output, and the hypothesis formed on this basis is the model. Corresponding to the black box is the white box (or white box), which can observe and analyze all the elements in the system. Most systems are between the black box and the white box, that is, the gray box. Some systems can be observed and understood, while others cannot be observed and understood.
(Charles Munger's core thinking model) However, even if you master these models, it may not be very helpful to solve specific problems, because all the problems you face are specific black boxes or gray boxes, and no universal model can completely correspond to these black boxes or gray boxes. In other words, there is no universal model "lens" to obtain information of different black boxes and gray boxes. Therefore, what matters is not how many models you have mastered, but the ability to effectively model specific problems!