[Statistics & Neuroscience] Error Interval(s) and Cognitive Bias

Zilla360

What is the Error Interval on the number of different kinds of Cognitive Bias that a person can contain simultaneously? I guess I might also be thinking of the Confidence interval?

I suspect the answers to this question might get very philosophical, rather quickly... :tell_me_more:

TychoCelchuuu

Confidence intervals attach to specific applications of a statistical measurement technique. So for instance if you randomly sample people from a group and ask them how much they like potatoes, we can calculate a confidence interval for the number you end up with to describe how confident we should be that the group's number is close to the number you end up with. But there is no way to calculate a confidence interval for "the number of different kinds of Cognitive Bias that a person can contain simultaneously." That is not a specific application of a statistical measurement technique. If you come up with a statistical measurement technique to calculate how many cognitive biases a person can "contain" simultaneously, and if you apply that technique to a person, then we can calculate a confidence interval. Until then the question is nonsensical.

I do not think an "error interval" is even a relevant concept here.

In any case, the idea of people "containing" cognitive biases is not very illuminating. A cognitive bias is a process of reasoning, and it's not clear what it would mean to "contain" a process. Processes are things which are carried out. Perhaps we would want to speak of someone being disposed to, or liable to, engage in certain processes, or of someone currently engaging in certain processes, or of someone having engaged in certain processes over a period of time. But the "containment" idea does not strike me as a fruitful direction of inquiry.

Zilla360

You're right. There is no sample size large enough for good data. Thanks for your answer.

Zilla360

Clarification: By 'containing' a process I was talking about artificial neural networks and quantum computing. Each program, executed in the context of a modern POSIX-compliant operating system is labelled as a 'process'.

For each process, thermal bias exists.

Burnage

I'm not trying to be a goose here but I have a PhD in cognitive psychology and I'm genuinely struggling to parse what you're asking.

TychoCelchuuu

Zilla360 wrote: »

You're right. There is no sample size large enough for good data. Thanks for your answer.

This is not what I said. The question is not about getting "good" data. The question is about what it means to calculate a confidence interval. A confidence interval is calculated for some sort of statistic conclusion that we draw. Whether the data we draw the conclusion about is "good" or not is entirely irrelevant.

As @Burnage has pointed out, it is not clear you are thinking about these things in precise enough terms for what you have in your head to actually mean anything. Among the various cognitive biases that people are typically afflicted with, one of the most persistent is that we can convince ourselves that one of our cherished ideas makes sense long before it in fact makes sense, because we come up with our cherished ideas in the privacy of our own head, informed by our own understanding of things, without necessarily checking whether our understanding of things matches up with how things actually are.

I suspect in this case you have many, many deep misconceptions about many, many topics, and on the basis of these misconceptions you've arrived at whatever idea you are trying to ask about here.

Unfortunately I think in fact there isn't an idea here: there are so many misconceptions built into your approach here that, once we go through and clear up all the issues, this will destroy the foundations for what you're getting at here and you won't have anything like the question you currently have.

Zilla360

*Sighs*. Very well, I shall pursue this elsewhere. Thank you for your help, in any case. If a mod wants to lock this thread, please do.

Zilla360

Burnage wrote: »

I'm not trying to be a goose here but I have a PhD in cognitive psychology and I'm genuinely struggling to parse what you're asking.

https://en.wikipedia.org/wiki/Complexity_class

Zilla360

I think it's (the question I was asking) is another one of those P versus NP problems. Bah.

Mods can lock this thread now, if they want to.

Burnage

It's still not exactly clear to me what it is that you're actually asking. There has been a considerable amount of research done within the scientific literature on to what extent cognition might be limited by computational tractability (and I'm thinking here predominantly of Iris van Rooij's work) but it doesn't really make obvious sense to me to consider that in terms of the number of cognitive biases that a person may contain (?).