January 21, 2019

What can Science do?

Plan for Today

1) Recap: Causal vs. Descriptive claims

2) Falsifiable/Unfalsifiable


Causal vs. Descriptive Claims

Two varieties of empirical claims:

descriptive claims: claims about what exists (or has existed/will exist) in the world:

causal claims: are claims about the how one phenomena (\(X\)) affects or causes another phenomena (\(Y\)). Causal claims state that \(X\) acts on \(Y\) in some way, not merely that they appear together in some pattern:

Descriptive Claims

claims about what exists (or has existed/will exist) in the world:

  • what phenomena exist (what kinds of things exist?)
  • what is the type of a specific phenomenon (what is this thing?)
  • amount/frequency of phenomena (how much of something is there?)
  • relative amount/frequency of phenomena across different places/times (how much of something is there here vs. there/now vs. then?)
  • what patterns are there in the shared appearance/non-appearance of different phenomena (does this thing usually appear together with that other thing?) do not conflate with causal claims

Causal Claims

  • the effect that one thing or event has on another thing (effects of causes)
  • the cause of some event or thing in the world (causes of effects)
  • the conditions under which some thing or event happens (causes of effects)
  • the process through which one thing affects another (causes of effects)

Causal claims include some kind of causal verb (or phrase). E.g.:

Causal claims always imply: if we could directly manipulate one thing \(X\), then we would see \(Y\) change

Causal Claims

"The chance of cancer increases with age"

Same as:

"The chance of cancer increases as age increases"

Is this a causal claim? Why/Why not?


Causal Claims and Prescriptive Claims

You and your friend win a sum of money in a lottery

You and your friend agree:

the most important ethical goal is minimizing human suffering. (value judgment)

You disagree on what to do:

  1. We should donate mosquito nets to stop malaria (prescriptive claim)

  2. We should give money directly to impoverished people (prescriptive claim)

Causal Claims and Prescriptive Claims

  • Assuming less malaria is good, resources are finite
  • What if mosquite nets don't prevent malaria, or cause more harm (tangling deaths)?
  • Showing "Malaria nets does not reduce malaria" is informative!

Accepting a prescriptive claim requires both a value judgment and accepting a causal claim

bonus term: instrumental rationality


Which claims can science evaluate?

#NotAllEmpiricalClaims open to science


an empirical claim is falsifiable

if that claim can be shown to be wrong by a specific empirical test. Even if we never do show the claim is wrong, it is falsifiable if, in principle it could be shown to be wrong.

  • e.g. The Earth orbits the Sun

an empirical claim is unfalsifiable

if there is no empirical test that can be done to show the claim is wrong


Because science considers alternatives/is open to being wrong…

we can only scientifically evaluate claims that can, in principle be shown to be wrong.

The best claims/theories are those that stand up to many tests to show them wrong.

No claims/theories are "proven", they have "not yet been proven wrong"

  • invitation to challenge


Contrast with verification

If we had an empirical claim, \(H_1\) (\(H\) for hypothesis)

and, if \(H_1\) were true or valid, then it implies we should make certain empirical observations \(O_1\)

Verification says that

  1. \(H_1 \rightarrow O_1\)

  2. We see \(O_1\)

  3. Therefore, \(H_1\) is valid


Many people reject verification because many different, incompatible claims are might imply the same observed evidence

  1. \((H_1, H_2, \ldots, H_k) \rightarrow O_1\)

  2. If we observe \(O_1\)

  3. \(H_1\) is not proven


If a claim \(H_1\) is falsfiable, there must be observable implication \(O_1\) that, if it were untrue, would invalidate \(H_1\). So, the logic of falsification is:

  1. \(H_1 \rightarrow O_1\)
  2. \(not\ O_1\)
  3. Therefore, \(not\ H_1\)


\(H_1:\) "Immigrants are coming across the border to kill you" ~ Donald Trump


  1. \(H_1 \rightarrow\) Murder committed by an "illegal immigrant"

  2. There has been a murder by an "illegal immigrant" see here

  3. Therefore, "Immigrants are coming across the border to kill you"


\(H_1:\) "Immigrants are coming across the border to kill you" ~ Donald Trump \(H_2:\) "One immigrant committed murder" \(H_3:\) "Two immigrants committed murder" \(H_4:\) "Immigrants are less prone to murder than native born citizens"


  1. \((H_1, H_2, H_3, H_4 \dots H_k) \rightarrow\) Murder committed by an "illegal immigrant"

  2. There has been a murder by an "illegal immigrant" see here

  3. Therefore, \(H_1, H_2, H_3, or\ H_4\) could be valid


\(H1:\) "An invisible and undetectable Flying Spaghetti Monster created the universe "after drinking heavily". According to these beliefs, the Monster's intoxication was the cause for a flawed Earth. Furthermore, according to Pastafarianism, all evidence for evolution was planted by the Flying Spaghetti Monster in an effort to test the faith of Pastafarians."


  1. \(H_1 \rightarrow ?\)

  2. \(?\)

  3. \(?\)

What observable implication of this could be proven wrong?


\(H_1:\) "According to Pastafarian beliefs, pirates are absolute divine beings and the original Pastafarians. Global warming, earthquakes, hurricanes, and other natural disasters are a direct effect of the shrinking numbers of pirates since the 1800s."


  1. \(H_1 \rightarrow\) Increase in Pirates would decrease Global Warming

  2. \(not\) Increase in Pirates produce decrease in Global Warming

  3. \(not \ H_1\)

In principle, this is could be falsified. (What kind of claims is \(H_1\)?)


\(H_1:\) "Human beings are rational actors who maximize their utility."

  1. \(H_1 \rightarrow\) People make decisions that maximize their expected monetary earnings.

  2. In games, people choose to take monetary losses to punish people who violate expected social norms (\(not \ O_1\))

  3. \(not \ H_1\)?

  4. Frequently, scholars say: utility functions also place value on adherence to social norms (preferences over process, not outcome)

  • Rational choice theories may not be falsifiable