Lecture 004

Topic 1: Explanation, Prediction, and the Aims or Goals of Science

Explanation

Explanation as Deducing From Law: Apply Function

Carl Hempel: a deduction from the laws (of nature) that apply and the specific initial condition of the world surrounding the phenomenon to be explained.

• deductive-nomological model: explanation is (law, causality: a mapping between initial condition and consequence) and the initial condition

Explanation as Description: More fine grained

Explanation: both compression of former data and prediction

• something that describes in terms of what we understand

• in terms of more "fundamental"

• in terms of cause of the phenomenon

Explanation as Specifying Causal Mechanisms

Explanation has causal relationship

Goal

Goal for science

• explanation

• description: describe and compress patterns

• prediction

Prediction

Prediction: you don't need an explanation for prediction Prediction vs. Explanation

• logical positivist:
  • explanation: causal relationship
  • prediction: induction, blindly fitting data

Causation

Qualifying potential causation variable

• stands out as a difference among the variety of other potential causes

• covaries with the effect (covariant)

• precedes the effect

• contiguous with the effect in time and space

• is similar to the effect in terms of duration, magnitude, and physical characteristics

• can be linked to the effect by a causal chain or mechanical model

• is more plausible than rival causal candidates

Topic 2: Scientific Representations

representation: partial, abstract

• inaccuracy

• have limitations of applicability

extra-mental representations:

• better come to consensus

• involves: natural language, drawings, math, physical models, computer simulation

Topic 3: The Nature of Scientific Theories - Realism and Anti-Realism (Reality and Generalization)

Theoretical

• very substance of theoretical claims

• claims are not directly observable

Reality

realists: theories are real because

• there are measures of their effects

• explanation and prediction is good

anti-realist (instrumentalists): talking about measurements (theory) is not real

Generalization

realists: bigger generalization, we can use representation variable to infer represented variable

anti-realist: all you have is data

Questions

1. description vs. explanation A scientific description is a compressed representation of the natural phenomenon. Such a description is intended for better communication of ideas between scientists. A scientific explanation is both a compressed representation of the natural phenomenon and a causal relationship between all variables in the explanation. For example, in the case of "water changing from a liquid to ice", a scientific description may be: "At -12 Celsius under 1 atm, we observe that water molecules start to line up in a relatively stable pattern". A scientific explanation may be: "water molecules start to line up in a relatively stable pattern because they are at -12 Celsius under 1 atm." Notice how a description cannot be directly used as a prediction since it does not imply any relationship (other than that two phenomenons, temperature and the change in molecular structure, are happening at the same time) between temperature and patterns of water molecules. An explanation, however, can imply a causation relationship between variables. Because a scientific explanation is "a) something that describes that phenomenon in terms that we already understand or b) in terms that are more “fundamental” or c) in terms of the causes of the phenomenon" ("Topic 1: Explanation, Prediction, and the Aims or Goals of Science"), therefore an explanation contains a description (by a) and differs from a description because it has extra causation relationship (by b and c).

2. correlation vs. causation in accuracy of prediction (using big data in stock market vs. provide an explanation, however, in my opinion there isn't a sound explanation to be provided) Because causation

3. covaries with the effect (covariant)

4. is similar to the effect in terms of duration, magnitude, and physical characteristics

therefore the relationship is more stable than a prediction

1. science's purpose: prediction or explanation Since explanation has causality, it includes prediction. Then science is about explanation since it is boarder.

2. Right or wrong: we cannot accurately know what's in other people's mind objective: come to same conclusion given the same premise subjective: come to different conclusion given the same premise

color blink:

DISAGREE: For this question, lets define "same" as an equivalence relationship on "inter-mental representation" as the average of all relevant "extra-mental representation". (learned from NLP)

• therefore if two internal representation, although different, have the same impact on the behavior of individual, then they are the same.

• we can iteratively approximate our idea of one's internal representation

• are social science exceptionless? First of all, I don't think that physical science are exceptionless. Secondly, I don't agrees that "increasing the supply of a good causes a reduction in its price" should be considered a law

My answer: There exists universal pattern, we can only approximate them. (This question is highly opinionated, therefore I don't have a valid justification for it.)

Class Notes

causal process: a process that has the ability to cause a persisting change in the structure of another process

• leave permenent effect

causal interaction: is an intersection between two causal processes that changes the structure of both