Conversation

Replying to

and

I’m slightly surprised not one person has asked: “so what are the prescriptive models in the literature?” I won’t say that using a cognitive bias is itself a bias. Understanding the descriptive models and the normative models are necessary for using the prescriptive ones.

Replying to

What are the most common prescriptive models? I'm imagining Confirmation Bias as descriptive and Thinking From First Principles as normative (situationally at least), but what gets you from A to B? (feel free to choose different A and B)

Replying to

and

It'll probably take a blog post to do justice, but here's a collection from Baron's Thinking and Deciding. (I'm probably breaking copyright law for uploading an entire chapter, so I'll take this down as some point) dropbox.com/s/32n320o6k0qg

Replying to

and

Note: requires understanding of the search inference framework, which I've summarised here: commoncog.com/blog/putting-m IMO, Baron's approach of arresting the 3 errors of search-inference is more useful than the goal of 'achieving perfect instrumental rationality'.

commoncog.com

Putting Mental Models to Practice Part 3: Better Trial and Error

Instrumental rationality is the sort of thinking that allows you to achieve your goals. We take a closer look at what decision science says is the 'best' way to pursue this purpose.

Replying to

and

Oh, the decision making/search inference process is the preceptive "model"? I was thinking of something more red-pillish, but I think I'm grasping it now. Would the Scientific Method fit that mold too?

Replying to

No, the search inference framework is a description of human thinking. The process of searching and selecting the optimal result using a utility function is normative; the fact that we fail in one of 3 ways is descriptive. Chapter 9 is Baron’s prescriptive model.

Replying to

A prescriptive model is more of a learning process than a doing process? I don't mean to make you keep going in circles, I think I need to make an eli5 post. Descriptive: the red pill Normative: living outside the matrix Prescriptive: giving/taking the red pill?

Replying to

and

Now I'm thinking that was still wrong, I've heard of satisficing before, would you agree with this description? google.com/amp/s/www.rese

Replying to

and

Back to

original question, maybe biases are useless as outputs ("that was X bias, you're wrong"), but useful as inputs ("because of these biases I should shade my decision/option like so")? And the prescription is ~do your best? cc

and

Replying to

Yes this wraps it up neatly.

Replying to

Eep, I realise that 'do your best' is your take on satisficing. That's not the right conclusion — satisficing is descriptive! Do reread the link+screenshot you've attached.

3:14 PM · Jul 27, 2019Twitter Web App

Replying to

The term satisficing is descriptive, but "do your best" is (a description of) a prescription: to use biases as inputs and improve your actual outputs vs. just calling them out or slavishly sitting in stasis with infinite bias loops. Prescriptive model ~= compromise of desc+norm?