My understanding of scientific research 我对科研工作的认识

Written by Yiyun Chen

个人认为科学是对现象和实践做归纳法，其命题通常有全称的属性，尝试用尽可能少的因去解释更多的结果，并且其命题是可证伪的。

I personally think that science is an induction method of phenomena and practice. Its propositions usually have the attributes of universal propostion. Try to explain more results with as few causes as possible, and its propositions are falsifiable.

求知 Seek knowledge 对真理的求知工作，我想应该分为两部分：

1 科学上的归纳法和大统一理论，用尽可能少的因和尽可能简化的模型来解释最多的现象。

2 神经科学的角度上[i]，或者说哲学的主观上，去研究人类认知模式和主观感受是如何通过神经系统运作的。[ii]因为这是对世界做客观描述所无法解决的问题。

I think the search for the truth should be divided into two parts:

1 Scientific induction and grand unified theories use as few factors as possible and simplified models to explain the most phenomena.

2 From the perspective of neuroscience, or the subjective philosophy of philosophy, to study how human cognitive models and subjective feelings operate through the nervous system. Because this is a problem that cannot be solved by objective descriptions of the world.

我对科研工作的分类 My classification of research work 1在于发现，从自然界中发现一些现象，归纳出一些特性，比如发现一种新元素，新粒子，另一部分。通常来说这种进一步的发现在科学史上都包含一些偶然性的小故事。这些新的发现常常预示着新的解释和新的理论。大部分的科研发现工作都是局部的，特称的。

2 阐述局部的因果关系，对局部现象进行进一步的建模和预测。

对于纯数学和计算机系统这种纯粹人为规定的体系，我们常常在这其中做逻辑分析，定理证明等工作，这些工作建立在已有的大的框架基础上。在这里我们所做的工作往往是局部性的，分支性的，我们在做判断和陈述时，常常预设地认为给定的公理，定律都是正确的。

对于物理现实世界，我们也发展出了一定的体系。与数学体系不同的是，物理里的定律是归纳出来的，同时这些定律的成立随着物理学的进一步发展是限于一定条件和精度的。但是我们再做科研工作时，常常把这些未被证伪的且普遍接受的理论作为已知的条件，即已知的大前提，在这些大前提下，我们再结合实际给定的特例条件做一些因果关系的解释，这样的解释是建立在已有的物理学框架的基础上的，即我们默认框架中所有的定理定律都是正确的。再用这些框架去解释现象。

对于绝大多数科研工作者而言，是无法做到推翻已有的体系的，尤其是数学体系[iii]，物理体系的定律由于是被归纳出来的可以证伪，部分程度上还可以被颠覆。

3 对科学体系框架做出修正和改进添加，比如数学上一些新概念[iv]的提出，比如物理学上相对论的提出，比如化学上的元素周期律和原子模型。这一类工作基本都能在科学史上留名，绝大部分人都不会有机会进行此类工作，个人认为此类工作除了天赋和努力外，还极大与时代背景相关。

4 提出一种新的局部方法，可以改进实验精度，或者提高效果，或者减低成本，提高生产力水平，具有某些具体的使用价值。这通常属于工科的范畴多一些。工科的大部分论文都是提出了什么方法，算法然后分析的时候比较各种方法和算法的结果好坏和成本问题。个人感觉，其实大部分工科这种提出方法类的论文，往往包含大量的拼凑，整合。具体的方法通常视不同的使用场景有不同的利与弊。

1 lies in discovery, discovering some phenomena in nature, summarizing some characteristics, such as discovering a new element, a new particle, and another part. Generally speaking, this further discovery contains some accidental stories in the history of science. These new discoveries often herald new explanations and new theories. Most scientific discovery work is partial and special.

2 Explain the local causality, and further model and predict the local phenomenon.

For purely artificially prescribed systems such as pure mathematics and computer systems, we often do logical analysis and proof of theorems in them, which are based on the existing large framework. The work we do here is often partial and branched. When we make judgments and statements, we often presuppose that the given axioms and laws are correct.

For the physical real world, we have also developed a certain system. Different from the mathematical system, the laws in physics are generalized, and the establishment of these laws is limited to certain conditions and accuracy with the further development of physics. But when we do scientific research, we often regard these unfalsified and generally accepted theories as known conditions, that is, known major premises. Under these major premises, we will combine the actual conditions given by the special cases. Some causal explanations are based on the existing physics framework, that is, all theorems and laws in the framework are correct by default. Then use these frameworks to explain the phenomenon.

For most scientific researchers, it is impossible to overthrow the existing system, especially the mathematical system. The laws of the physical system can be falsified because they are generalized, and can be subverted to some extent.

3 Make amendments and improvements to the framework of the scientific system, such as the proposal of some new concepts in mathematics, such as the theory of relativity in physics, such as the periodic law of elements and atomic models in chemistry. This type of work can basically leave a name in the history of science, and most people will not have the opportunity to do this type of work. I personally think that this type of work is greatly related to the background of the times in addition to talent and hard work.

4 Propose a new local method, which can improve the accuracy of the experiment, or increase the effect, or reduce the cost, increase the level of productivity, and have some specific use value. This is usually more of an engineering category. Most of the engineering papers put forward what methods, algorithms and then compare the results of various methods and algorithms and cost issues during analysis. Personally, in fact, most engineering papers that propose methods often contain a lot of patchwork and integration. The specific method usually has different advantages and disadvantages depending on different usage scenarios.

个人对科研的几点感想 Personal thoughts on scientific research 1 科研可以说，并没有多么崇高，绝大部分人所从事的科研工作，并不能提升我们对世界的认识水平，或者提高多少生产力的发展，但是每一个科学的小分支里的一些现象逻辑因果分析工作和实验发现等此类带有重复性的工作总要有人去做，或者说就是“搬砖”，由于这种搬砖工作需要熟悉大前提，也就是现代科学的数学物理框架，所以门槛显得比较高。

2 人的一生非常短暂，个体尽其一生能归纳的现象是相当有限的，所有科学的发展都建立在前人的基础上。个体想要去实证一些物理定律的正确性，是相当困难的，这导致我们不得不直接地不加批判地接受——这些已有的物理定律是正确的，因为质疑它们的成本对个体来说实在是太高了。这就导致我们的科研工作必将局限于一定的范围和框架内，我们必须要接受一些给定的框架为真。同时另外还有一点是，人的大脑资源是非常有限，我们在一个时间段内能注意到的对象，能进行的逻辑推演的体量都十分有限。因此，我们的推理工作必须依赖于既定的真命题，既定的框架，我们无法直观地掌握所有真相，只好接受一部分既定的真实，在小范围内，验证小前提下的一些特例的结论，对于一些大前提和定律定理，我们只需要记住以提高效率，不必质疑或者说也无法支付质疑其正确性的成本。

3 数理类的科研工作，很难揭示事物本质（物自体不可知）。现代的物理学已经高度抽象化，数学化，参数化，且进展缓慢，由于基础框架逻辑认识论上一些模糊地带，进一步地归纳出定律和规律变得尤为困难（牛顿力学体系相比量子力学相对论要直观得多）。神经科学上，个人认为，无法通过建立数理模型的方式揭示感官范畴为何如此呈现的本质，只能研究一些因果联系。（这就好比用计算机语言构建痛觉一样，无法想象）。

4 在工程类的科研工作中，通常是提出方法，关注于应用。个人觉得，拼凑灌水的现象比较严重，大部分人的成果都是拼凑裁剪式的。这些方法的提出通常也是基于一定的数学物理框架体系。这些方法之间做优劣比较时会发现，并没有本质性的发展。工程问题主要在于发展生产力，但这归根结底还是数理类研究打下的基础上才有方法的提出和比较。

1 It can be said that scientific research is not so lofty. The scientific research work done by most people cannot improve our understanding of the world or increase the development of productivity. However, there are some phenomena in every small branch of science, logical cause and effect. Repetitive tasks such as analysis work and experimental discovery must be done by someone, or “brick-moving”, because this kind of brick-moving work requires familiarity with the major premises, that is, the mathematical and physical framework of modern science, so the threshold It seems taller.

2 The life of a person is very short, and the phenomena that an individual can sum up during his entire life are quite limited. All scientific developments are based on the foundations of predecessors. It is quite difficult for individuals to prove the correctness of some physical laws, which leads us to accept directly and uncritically-these existing physical laws are correct, because the cost of questioning them is for the individual It is too high. This leads to our scientific research work must be limited to a certain scope and framework, we must accept some given framework is true. At the same time, there is another point that human brain resources are very limited. The objects we can notice in a period of time and the volume of logical deductions we can perform are very limited. Therefore, our reasoning work must rely on established truth propositions and established frameworks. We cannot intuitively grasp all the truths. We have to accept a part of the established truths. In a small range, we must verify the conclusions of some special cases under small premises. For some We only need to remember the major premises and laws and theorems to improve efficiency, without questioning or paying the cost of questioning their correctness.

3 It is difficult to reveal the essence of things in scientific research in mathematics and science (thing-in-itself are unknowable). Modern physics has been highly abstracted, mathematical, and parameterized, and its progress is slow. Due to some fuzzy areas in the basic framework logic epistemology, it is particularly difficult to further generalize the laws and regulations (the Newtonian mechanics system is more important than the quantum mechanics theory of relativity. Much more intuitive). In neuroscience, I personally believe that it is impossible to reveal the nature of the sensory categories through the establishment of mathematical models, but only to study some causal connections. (This is like using computer language to construct pain sensation, unimaginable).

4 In engineering research work, methods are usually proposed, focusing on application. I personally feel that the phenomenon of patchwork irrigation is more serious, and most people’s results are patchwork and tailoring. These methods are usually also based on a certain mathematical physics framework system. When comparing the advantages and disadvantages of these methods, you will find that there is no essential development. The main problem of engineering lies in the development of productivity, but in the final analysis, methods can be put forward and compared only on the basis of mathematical research.

[i] 个人觉得这可能行不通，即使我们弄清楚了大脑神经系统所有的运作机理，依然不能涵盖认知范畴，举个例子，理解痛觉感受器的发生机制更多是在视觉和逻辑上，靠此不能感知痛觉本身，不过这样一来，我们会发现，我们自身的感觉到底是为何如此独特，神经传导的模型只能给出联系，而给不出源头。

I personally feel that this may not work. Even if we figure out all the operating mechanisms of the brain and nervous system, we still cannot cover the cognitive field. For example, understanding the mechanism of pain receptors is more visually and logically, and you cannot perceive pain by this. In this way, we will find out why our own feeling is so unique. The model of nerve conduction can only give the connection, but not the source.

[ii] 比如说以神经科学的角度给出视觉，听觉等直观感觉的形成和信宿究竟在哪里，人类的神经系统如何形成空间时间这种范畴的，这种时间流逝感是如何形成的，视神经的输入如何形成的空间感。反思这些感受野的组成单元作为基本存在的原理。我个人难以想象如何解释基本存在的存在。康德将这些原理归纳为先天范畴，并认为物自体不可知。

For example, from a neuroscience point of view, where is the formation of intuitive sensations such as vision and hearing and where the information is located, how the human nervous system forms the category of space and time, how this sense of time lapse is formed, and what is the input of the optic nerve A sense of space formed. Reflect on the constituent units of these receptive fields as the basic principle of existence. Personally, I can’t imagine how to explain the existence of basic existence. Kant summarized these principles into an innate category, and believed that the thing itself is unknowable.

[iii] 为什么数学体系无法被推翻，或者说该体系是唯一的呢？我认为可以从这一角度去解释，数学逻辑所研究的是各种对象之间的关系，重点在于对这些关系的分析，比如数量关系，函数变化关系，空间几何关系。这些关系的形成，是具有唯一性的，我们的大脑神经系统的结构预示着我们对于这些关系的理解必然只有一种体系。也就是说基础的数学框架体系是唯一的，这可以视为我们先天范畴的一部分。

Why can’t the mathematical system be overthrown, or that the system is unique? I think it can be explained from this perspective. What mathematical logic studies is the relationship between various objects, and the focus is on the analysis of these relationships, such as quantitative relationships, functional change relationships, and spatial geometric relationships. The formation of these relationships is unique. The structure of our brain and nervous system indicates that we must have only one system for understanding these relationships. In other words, the basic mathematical framework system is unique, which can be regarded as part of our innate category.

[iv] 比如微积分，其实就是数量关系抽象化程度提高了一些，由加法和乘法构造而来，微积分的提出并没有创造新的对象，只是这种抽象描述更精巧实用。

For example, calculus is actually the degree of abstraction of quantitative relations increased a bit. It is constructed from addition and multiplication. The introduction of calculus does not create new objects, but this abstract description is more sophisticated and practical.