3 ms·

This is a bit off topic but I have a question about causality. Is causality really a fuzzy concept? For example I assume if action A causes event B I assume th

This is a bit off topic but I have a question about causality. Is causality really a fuzzy concept? For example I assume if action A causes event B I assume the connection is 100%. Can we really say that action A has a 30% chance of causing event B? I ask because the more I think about it, when someone says action A causes a 30% chance of event B occurring what he is technically saying is action A is one causative factor that must occur and that we're missing information about other causative factors. In the case of this article. A causative link is established between EBV and MS to a fuzzy probable degree. This seems to me that technically what's actually occurring is that a fuzzy causal link simply means that there are other causative factors we don't yet know about, and likely this is a specific type of immune system that reacts to EBV in a certain way. Would my assessment be accurate? All causal connections are either 100% and any fuzziness just means we're missing information about other joint causal events that must occur to trigger the outcome. Does anyone who's a statistician know?

echelon 5y ago
> we're missing information about other causative factors. Welcome to biology. Perhaps EBV is particularly immunogenic, but it has to reactivate in the presence of CD8+ T-cells or CD20+ B-cells along with some other self-signal or co-infection. Perhaps there's a chain of immune signaling that must happen upstream in other immune cell populations. Perhaps other agents can trigger autoimmune behavior in these cells, and EBV isn't strictly required. Maybe other viruses in the herpesvirus family (HSV, zoster, etc.) There could be more than one set of causes, and they might not even look similar. In the crazy absurd limit, maybe nothing at all. Maybe autoimmunity sporadically arises with no causal agent. Random radiation hits the cell at the wrong time. All or none of these could be the case. Diseases can have a multitude of causes, sometimes with complex interactions and interdependencies, sometimes not. With cancer it's a progression of increasingly worse state changes, and that could be the case here too.
peter303 5y ago
There was a similar debate in the 1880s. Dr. Robert Koch claimed some diseases were caused by bacteria and developed a procedure to prove it. The germ theory of disease was controversial until then. Its suspected several diseases without known causes may be caused by a bacterium, virus or protein. But lack the evidence to satisfy the Koch postulates. These diseases include some cancers and many human nervous system diseases.
csee 5y ago
This is more in the domain of philosophy than statistics. I wrote something in response, but then saw that the Wiki article on causality was far better than what I wrote, so have a look at that. https://en.wikipedia.org/wiki/Causality https://en.wikipedia.org/wiki/Causality https://en.wikipedia.org/wiki/Necessity_and_sufficiency https://en.wikipedia.org/wiki/Necessity_and_sufficiency
gwern 5y ago
Is it helpful to ask about 'percentages' here? Is oxygen 0% or 100% of the cause of you posting here? What about gravity? Whenever you have a causal question, often taking a Pearlean perspective and asking about interventions would be more useful. For example, there is an EBV vaccine being worked on right now. If you believe OP that EBV infection is a necessary but not sufficient condition for 90% of MS cases, then it would be reasonable to say something like "if we intervened by making everyone got the EBV vaccine, then there would be ~90% fewer MS cases than in the counterfactual universe where everything else was the same but no one got the EBV vaccine." That sounds much more interesting and meaningful to me than talking about how "EBV is necessary but not sufficient".
sva_ 5y ago
> All causal connections are either 100% and any fuzziness just means we're missing information about other joint causal events that must occur to trigger the outcome. I think one theory is, that the immune system has a way of approximating the surface of a protein, and in this way recognizes what to build antibodies against. This process could be a source of fuzziness as the "protein signatures" of a virus may vary from person to person. There is also the question how those signatures are stored and retrieved in the immune system, and why they -presumably- lead to the immune system misrecognizing myelin sheaths as a threat (in some people, some of the time.)
dataflow 5y ago
> Is causality really a fuzzy concept? For example I assume if action A causes event B I assume the connection is 100%. Can we really say that action A has a 30% chance of causing event B? The world is probabilistic at the quantum level so I don't see how it could be otherwise. > I ask because the more I think about it, when someone says action A causes a 30% chance of event B occurring what he is technically saying is action A is one causative factor that must occur and that we're missing information about other causative factors. To my knowledge this has been physically disproven: https://en.wikipedia.org/wiki/Hidden-variable_theory#Bell's_theorem https://en.wikipedia.org/wiki/Hidden-variable_theory#Bell's_... (Note I'm just responding to your philosophical question, not the biological one. In particular I'm not suggesting anything about the practical relevance of quantum effects to the biology discussion here, or lack thereof.)
- beaconstudios 5y ago
  this doesn't occur above the quantum level though.
beaconstudios 5y ago
simple, 1:1 causality (if I flick a light switch, the light turns off) is not fuzzy at all. Complex causality is more complex (obviously) and most things in real life have complex causality.
- deltaonefour 5y ago
  In science nothing can be proven. That means in order to to prove that light switch turns the light off you must flick the switch and observe the light turning off infinite times. If in one of these infinite observations flicking the switch off turns the light on your hypothesis is disproven. Hence disproof exists but proof is the domain of maths and logics. From this perspective 1:1 causality cannot be verified to exist.

echelon 5y ago

> we're missing information about other causative factors. Welcome to biology. Perhaps EBV is particularly immunogenic, but it has to reactivate in the presence of CD8+ T-cells or CD20+ B-cells along with some other self-signal or co-infection. Perhaps there's a chain of immune signaling that must happen upstream in other immune cell populations. Perhaps other agents can trigger autoimmune behavior in these cells, and EBV isn't strictly required. Maybe other viruses in the herpesvirus family (HSV, zoster, etc.) There could be more than one set of causes, and they might not even look similar. In the crazy absurd limit, maybe nothing at all. Maybe autoimmunity sporadically arises with no causal agent. Random radiation hits the cell at the wrong time. All or none of these could be the case. Diseases can have a multitude of causes, sometimes with complex interactions and interdependencies, sometimes not. With cancer it's a progression of increasingly worse state changes, and that could be the case here too.

peter303 5y ago

There was a similar debate in the 1880s. Dr. Robert Koch claimed some diseases were caused by bacteria and developed a procedure to prove it. The germ theory of disease was controversial until then. Its suspected several diseases without known causes may be caused by a bacterium, virus or protein. But lack the evidence to satisfy the Koch postulates. These diseases include some cancers and many human nervous system diseases.

csee 5y ago

This is more in the domain of philosophy than statistics. I wrote something in response, but then saw that the Wiki article on causality was far better than what I wrote, so have a look at that. https://en.wikipedia.org/wiki/Causality https://en.wikipedia.org/wiki/Causality https://en.wikipedia.org/wiki/Necessity_and_sufficiency https://en.wikipedia.org/wiki/Necessity_and_sufficiency

gwern 5y ago

Is it helpful to ask about 'percentages' here? Is oxygen 0% or 100% of the cause of you posting here? What about gravity? Whenever you have a causal question, often taking a Pearlean perspective and asking about interventions would be more useful. For example, there is an EBV vaccine being worked on right now. If you believe OP that EBV infection is a necessary but not sufficient condition for 90% of MS cases, then it would be reasonable to say something like "if we intervened by making everyone got the EBV vaccine, then there would be ~90% fewer MS cases than in the counterfactual universe where everything else was the same but no one got the EBV vaccine." That sounds much more interesting and meaningful to me than talking about how "EBV is necessary but not sufficient".

sva_ 5y ago

> All causal connections are either 100% and any fuzziness just means we're missing information about other joint causal events that must occur to trigger the outcome. I think one theory is, that the immune system has a way of approximating the surface of a protein, and in this way recognizes what to build antibodies against. This process could be a source of fuzziness as the "protein signatures" of a virus may vary from person to person. There is also the question how those signatures are stored and retrieved in the immune system, and why they -presumably- lead to the immune system misrecognizing myelin sheaths as a threat (in some people, some of the time.)

dataflow 5y ago

> Is causality really a fuzzy concept? For example I assume if action A causes event B I assume the connection is 100%. Can we really say that action A has a 30% chance of causing event B? The world is probabilistic at the quantum level so I don't see how it could be otherwise. > I ask because the more I think about it, when someone says action A causes a 30% chance of event B occurring what he is technically saying is action A is one causative factor that must occur and that we're missing information about other causative factors. To my knowledge this has been physically disproven: https://en.wikipedia.org/wiki/Hidden-variable_theory#Bell's_theorem https://en.wikipedia.org/wiki/Hidden-variable_theory#Bell's_... (Note I'm just responding to your philosophical question, not the biological one. In particular I'm not suggesting anything about the practical relevance of quantum effects to the biology discussion here, or lack thereof.)

beaconstudios 5y ago
this doesn't occur above the quantum level though.

beaconstudios 5y ago

simple, 1:1 causality (if I flick a light switch, the light turns off) is not fuzzy at all. Complex causality is more complex (obviously) and most things in real life have complex causality.

deltaonefour 5y ago
In science nothing can be proven. That means in order to to prove that light switch turns the light off you must flick the switch and observe the light turning off infinite times. If in one of these infinite observations flicking the switch off turns the light on your hypothesis is disproven. Hence disproof exists but proof is the domain of maths and logics. From this perspective 1:1 causality cannot be verified to exist.