Basic Mindsets

post by ExCeph · 2020-06-06T00:44:58.188Z · LW · GW · 5 comments

Contents

  Mindsets
    “Relevant” Hypotheses
  Primary mindsets
    Operation mindset
    Synthesis mindset
    Analysis mindset
    Organization mindset
  Secondary mindsets
    Empathy mindset
    Semantics mindset
    Tactics mindset
    Strategy mindset
  Recap
  Conclusion
None
5 comments

(Co-authored with Sailor Vulcan)

Long ago, in the previous article [LW · GW], we discussed the processes of guessing and checking, how they explore the unknown and the known, and how they can run in a distinct or subliminal mode. In this article, we’ll see how these processes form different mindsets that handle different types of tasks.

Mindsets

Mindsets are feedback loops of guessing and checking running in various modes. Their function is to create different types of maps that can be used for different purposes.

When using any mindset, the mind searches through hypothesis space in rough order of relevance to information from the territory (guessing) and redistributes its confidence in each hypothesis based on how well it compares against information from the territory and against other hypotheses that it holds in high confidence (checking).

When either of these processes takes place in the subliminal mode, it runs implicitly. It leaves no record of the iteration or comparison: only the results. When either process takes place in the distinct mode, it leaves an explicit memory of what information was accessed by the process in order to produce the results it did. Each of these modes has advantages and disadvantages.

Both the processes of guessing and checking are necessary for forming models of reality, but the modes in which these processes run influence what aspects of the model are updated and applied.

It should be noted that these mindsets do not have solid borders between them. The mindsets described below are analogous to primary colors, or a basis for a multidimensional vector space. They are the simplest way to define the range of skills people can access and to identify the strengths and weaknesses of a person’s repertoire.

As such, any limitations ascribed to a mindset are for the purpose of defining the purest and most basic building block form of that mindset. Any person’s thought process can blur the distinction between adjacent mindsets, or combine any number of mindsets together. Many of these compound processes I’ve already classified as hybrid mindsets, but for the sake of brevity we’ll leave those for another article.

“Relevant” Hypotheses

Before we get into the mindsets, it should first be noted that throughout this article we will use the word “relevant” several times in the context of describing hypotheses. What does it mean for a hypothesis to be relevant?

The process of checking (ideally) uses certain criteria when it considers a hypothesis for inclusion in our map of a particular territory. Based on these criteria, checking will accept or reject the hypothesis to a varying degree.

One of these criteria should be accuracy: hypotheses contradicted by evidence will make for a map that doesn’t match the territory. Checking should reject these hypotheses.

Simplicity should be another criterion, arguably tied to accuracy by Bayesian reasoning (or Occam’s Razor, if you prefer): assuming they are already consistent with the available evidence, hypotheses should not be more complex than the evidence supports. Otherwise, they are likely to make inaccurate predictions about the territory. All else equal, checking should consider a more complex hypothesis less likely than a less complex one.

A third criterion is relevance to our values: a hypothesis should not be included in our map of a particular territory we care about if it doesn’t make predictions about that territory, even if the hypothesis makes accurate predictions about a different territory. The more information a hypothesis gives us about the territory we’re focusing on, the more relevant it is for the map of that territory.

(There may be more criteria people should use, but these three are sufficient for the purposes of this article.)

To illustrate what we mean by a relevant hypothesis, we can use the example of inventing incandescent lamp (light bulb) technology. If we value having artificial light, then hypotheses that make predictions about what will or won’t serve as an incandescent filament are relevant to that value. We test and reject the ones that are inaccurate (“paper will work as an incandescent filament”) and replace them with hypotheses that are consistent with the evidence (“paper will not work as an incandescent filament”).

Accurate hypotheses about what will not work are still relevant, as each failure incrementally increases our knowledge. However, the accurate hypotheses about what will work give us the most information about the territory (“tungsten will work as an incandescent filament”) and so are usually the most relevant to our values, depending on the situation.

Accurate hypotheses about territories unrelated to the value of having artificial light, such as the hypothesis that penicillin has antibiotic effects, will be rejected as irrelevant by a checking process that is directed towards mapping the territory of artificial light technology.

This understanding of criteria that the checking process can use is important for describing the relative strengths and weaknesses of the mindsets.

Primary mindsets

The four primary mindsets are the simplest pairings of guessing and checking in the distinct and subliminal modes. They define the most basic functions that a mind can perform.

Operation mindset

Operation mindset uses the subliminal mode to store all its information about the territory in a cache.

The processes of guessing and checking take place subliminally, building high-resolution maps of localized situations that most people call “intuition”. These maps shape effort: they guide a person’s movements to conform to the environment to get their desired results. The way to use operation mindset is to repeatedly practice interacting with environments that give consistent and immediate feedback.

One advantage of operation mindset is that the cache reduces the burden on your working memory, so you can use operation mindset for a long time before becoming mentally fatigued. With enough practice you can perform rote behaviors with operation while simultaneously using working memory for other purposes. Another advantage of operation is that a cached map can guide complex movements in real time and provide fast and accurate answers in complex situations, provided the situations are familiar.

However, operation also has several disadvantages compared to other mindsets.

Firstly, the cached map cannot be directly accessed. Its information is stored as implicit correlations between myriad factors rather than as explicit statements, which makes it difficult to share with others. It also means that to review and update your intuition for a changing situation requires further repetitive training. Without the ability to accept ad hoc edits, intuition alone is powerless in situations with changing parameters and delayed feedback.

Furthermore, the process of subliminal guessing does not range far in its exploration of hypothesis space, so non-local maxima of accuracy or relevance may go unnoticed. Operation may attempt to refine an ultimately futile method rather than approaching from a different angle.

Finally, the process of subliminal checking does not register nuance regarding which aspects of a hypothesis don’t match the situation, and so operation can reject hypotheses that could be redeemed with only minor revisions. Operation can only map what it can directly interact with, which means abstract reasoning and logical principles are outside its grasp.

Examples of use:

Synthesis mindset

Synthesis mindset uses the distinct mode to handle its guessing, and the subliminal mode to cache its check, so to speak.

The distinct mode of guessing roves through hypothesis space based on the information fed to it. It can branch out and build on the ideas (hypotheses) it finds because it has an explicit record of where it’s been. Subliminal checking runs some basic comparisons to make sure that the hypotheses are coherent enough to be expressible. In this way, synthesis mindset generates ideas that form maps of wide-ranging possibilities for our consideration.

The way to use synthesis mindset is to allow your mind to wander through free association while suspending judgment of the accuracy or relevance of the hypotheses you encounter. It also helps to deliberately remove assumptions about certain ideas, and to toy with mixing multiple ideas together. This process is often called “being creative” or “using your imagination.”

One advantage of synthesis mindset is that in its travels through hypothesis space, it reveals non-local maxima of accuracy or relevance, even if it cannot recognize them as such. The non-obvious hypotheses it comes up with may ultimately turn out to be more accurate or relevant than those you already have, and without using your imagination you would not have considered them.

The obvious disadvantage of synthesis mindset is that it has little power to screen the hypotheses it finds for accuracy or relevance. Synthesis cannot learn to smoothly navigate situations, nor to compare various options, nor to judge its hypotheses against the territory. For many serious problems, synthesis is necessary but insufficient to identify an effective solution.

Examples of use:

Analysis mindset

Analysis mindset uses the subliminal mode to cache its guessing, and the distinct mode to pick up the check.

The subliminal mode of guessing keys off a situation and offers up the closest hypotheses to hand. The distinct checking process uses working memory to inspect each idea (hypothesis) to make sure it matches the observed data in all ways and to explicitly account for its findings. For any discrepancies, it may reject a hypothesis outright or request the guessing process to produce a revision to patch the flaw. In this way, analysis mindset can evaluate ideas against large samples of data from the territory to produce explicit, nuanced maps of the generalizable principles governing the territory’s causal relationships.

One way to use analysis is to take note of discrepancies between ideas and reality. A good starting point is to describe each mismatch accurately and with nuance, acknowledging which aspects of an idea are compatible with evidence and which are not, and why. It helps to continue by honestly investigating anything that challenges your map, and seeking out more information to test your map against it. In the process, you can consider alternative ideas that could contribute to an explanation of the evidence you see. It’s also important to explore the logical implications of each idea: what else would you expect to see if the idea were true?

A major advantage of analysis mindset is that the maps it forms of principles and causality can be used to figure out how processes and systems work. These maps can also predict the outcomes of unprecedented situations, sometimes with explicit probability estimates. Furthermore, analysis mindset helps correct errors in your models of reality by rejecting hypotheses that contradict observation or that are gratuitously complex. Upon discovering new information, you can update your maps ad hoc.

However, analysis has some disadvantages compared to other mindsets.

Firstly, analysis does not have access to non-obvious hypotheses unless they have already been brought to its attention. If the hypothesis most closely matching the territory isn’t already within reach of analysis’s cached guessing, analysis will either rule out all its hypotheses and fail to map the territory, or it will match the closest hypothesis to hand, either with overcomplicated patches or with exposed flaws labeled as anomalies.

Secondly, analysis cannot respond smoothly to situations requiring subtle movements or fine-tuned navigation in real time. The process of distinct checking takes too much time meticulously comparing each hypothesis to the territory, and by the time it finds the correct move the situation has already changed.

Lastly, analysis is ill-suited to come up with plans that fulfill multiple goals with the resources at hand. Even with its accurate maps of the principles governing a situation, the cached guessing process of analysis mindset does not range far enough to quickly identify and explore the different general approaches, or to consider non-obvious options. Moreover, the ability of analysis to identify discrepancies is insufficient for forming value judgments on what risks are subjectively worth taking. It is usually necessary to use explicit guessing in some capacity to explore the various options opened up by a decision in order to gauge its desirability relative to the alternatives.

Examples of use:

Organization mindset

Organization mindset uses the distinct mode to handle both guessing and checking, employing working memory to direct and track the activities of these processes.

The distinct processes of guessing and checking build explicit maps of systems and their most salient features. These maps help allocate efforts to enable you to achieve as many desired goals as possible.

To use organization mindset, you can review your various goals and the resources that are directly and indirectly available. You can then consider different options for using those resources to attempt to achieve as many goals as possible, and what the relative tradeoffs and constraints are for each option.

One advantage of organization mindset is that it suggests different plans and explicitly compares their marginal costs and benefits. Unlike with operation mindset, these suggestions and assessments can incorporate expectations about delayed or remote effects, and can be updated ad hoc when the situation changes. Organization can also anticipate how different efforts may affect each other and what can be done to make them compatible.

A disadvantage of organization mindset is that your working memory has tighter restrictions than your cache on how much information can be processed at once (and on how long it can be used before needing to rest, in biological brains). To cope with these limitations, organization used properly will ignore any details that aren’t obviously important.

Additionally, compared to synthesis mindset, organization mindset explores less of hypothesis space through guessing, so it may still fail to notice some non-local maxima of accuracy or relevance. In other words, it may not be imaginative enough to see less familiar possibilities.

Conversely, compared to analysis, organization mindset may miss some points of mismatch between its maps and reality, and can fail to apply enough distinct checking to catch flaws in its plans.

Examples of use:

Secondary mindsets

Mindsets can be combined in a variety of ways. When you combine two primary mindsets which aren't opposites, you get a secondary mindset. The four secondary mindsets are described below.

Empathy mindset

Empathy mindset uses the subliminal mode to handle both guessing and checking, and uses the distinct mode to handle guessing. It combines operation and synthesis to create maps of immediate possibilities and to navigate them intuitively. These maps help individualize interactions by exploring situations with hidden variables and handling their idiosyncrasies.

To practice empathy mindset, you must interact with an entity that changes internal states in response to interactions and to its environment. One way to use empathy is to borrow existing intuitions regarding similar situations in order to help model the current entity, and use imagination in order to tailor those intuitions to fit. Another approach to empathy is to imagine scenarios and then intuit how the entity you’re dealing with might respond to them. Yet another approach is to start with desired responses, and then imagine and intuit ways in which the entity might be led to respond in such a way. Cautiously introduce variations in how you interact with the entity to see how that affects what it does.

One advantage of empathy mindset is that it can more effectively interact with and influence subtle and temperamental entities than any other basic mindset, and achieve otherwise unlikely outcomes. Empathy can also maintain desired behavior from entities even in variable situations and environments. It combines the graceful navigation of operation with the expanded hypothesis space of synthesis to dynamically generate and try out creative options for what approaches will be most effective.

A disadvantage of empathy mindset is that the individualized and implicit maps it creates are often difficult to explain or generalize. In addition, not all entities have idiosyncrasies that allow a person to influence them through skillful interaction; some entities are highly consistent. Finally, it is possible for attempts at empathy to backfire, depending on how far outside your local map you have ventured in the exploration of possibilities.

Examples of use:

Semantics mindset

Semantics mindset uses the subliminal mode to handle guessing and checking, but also uses the distinct mode to handle checking. It combines operation and analysis to create streamlined maps of general principles that can be applied intuitively to simplify interactions.

One way to use semantics mindset is to learn a lexicon of labels and rules, then practice applying them in appropriate situations until they become intuitive. Another way to use semantics is to learn and practice applying basic ideas or techniques until they become intuitive, then work your way up to more complicated situations which require you to use them together as a functional vocabulary.

One advantage of semantics mindset is that when calibrated to a set of situations meeting certain requirements, it is faster and easier than analysis mindset. Semantics uses assumptions in the form of cached checking to reduce the amount of working memory that it uses on checking. This enables large, complex problems to be simplified into modules. Thus equipped, semantics can quickly infer information that is not immediately evident.

Another advantage to semantics is that it is more generalizable than operation, enabling you to intuitively solve different problems within the same paradigm by adjusting the input parameters. Semantics is capable of conveying nuanced information and updating it ad hoc by applying the distinct checking of analysis.

A major disadvantage of semantics mindset is that like operation, it starts producing the wrong answers when the assumptions that it is based on are invalid. Additionally, semantics requires explicit rules and labels in order to form useful maps of situations. In a situation where no consistent rules can be inferred explicitly, semantics mindset cannot function effectively any more than analysis can.

Examples of use:

Tactics mindset

Tactics mindset uses the subliminal mode to handle checking and the distinct mode to handle both guessing and checking. It combines organization and synthesis to map in more detail the possibilities afforded to you based on your available resources. It uses its maps to identify ways in which you can redirect paths.

One way to use tactics mindset is to start by taking inventory of what goals you have and the options your resources afford you. You can then start questioning your assumptions about what you can’t do with those resources, or about what you can call a resource in the first place.

Another way to use tactics is to start by brainstorming clever applications of resources you’d like to try out or feats you would like to accomplish. You can then think about how you can apply your existing resources to implement those ideas.

Tactics combines the pragmatic weighing of options that organization provides with the wider range of hypotheses from synthesis. The major advantage of tactics mindset is that it allows you to come up with plans to accomplish feats that are not obviously possible or available. If a feat appears impossible, tactics is your best bet to find the way to make it happen. If your range of options seems narrow, tactics can expand your awareness of your options.

A major disadvantage of tactics mindset is that the possibilities it opens up may have side effects or points of weakness that the mindset won’t notice. If a person gets too clever and neglects strategy, the solution provided by tactics may cause more trouble than the original problem. Another disadvantage of tactics is that its grounding in your available resources may limit how much of hypothesis space it explores, preventing it from effectively handling abstract ideas as well as synthesis alone might. Finally, compared to organization, the approaches that tactics comes up with may be less efficient. That’s not a problem when the solution wouldn’t even be found if not for tactics, but an inexperienced user of tactics may often overlook a sensible and practical solution in favor of a clever one.

Examples of use:

Strategy mindset

Strategy mindset uses the subliminal mode to handle guessing, and uses the distinct mode to handle both guessing and checking. It combines organization and analysis to map the effects of different options in more detail, the requirements for them to work, the relative costs and benefits of those implications, and further options that would be relevant to dealing with them. It uses these maps to identify ways in which you may want to fortify paths.

One way to use strategy mindset is to start by considering various goals and options, and then asking yourself what the logical consequences of those options would be and what the assumptions are that they rely on. Another way is to start by considering the different matches and mismatches of various hypotheses, and then considering the marginal benefit of increased accuracy, discarding questions that are not likely to be worth the effort invested into them.

A major advantage of strategy mindset is that it allows you to see and address weaknesses in plans, by spending resources to set things up so that even if an assumption fails and a contingency occurs, the plan can still work. The foresight strategy affords allows you to prevent unwanted outcomes and thereby make the desired outcomes more likely. Strategy mindset also increases the efficiency of investigating discrepancies in knowledge by prioritizing the effort based on your own goals.

A disadvantage of strategy is that it is less geared towards getting accurate pictures of generalized principles than regular analysis is. Another disadvantage is that strategy may rely too heavily on its foresight compared to organization, causing analysis-paralysis when the best option may be to choose something good enough and reevaluate it later. Furthermore, while strategy mindset is useful for closing unwanted possibilities, it is less able to open possibilities of success, leading a person to conclude that all options have unacceptable side effects without realizing there are ways to overcome those problems.

Examples of use:

Recap

Now we’ve looked at the eight primary and secondary mindsets, the different maps they build and the processes they use to do so, and their relative strengths and weaknesses.

Conclusion

In the introduction to this sequence [LW · GW], I expressed my frustration that people consistently failed to apply what I considered “common sense” to solve critical problems. I described how I had set out to figure out what they were missing, and announced the results of my efforts: a toolbox of fundamental concepts and paradigms, with which people can learn the skills they need to effect positive change. The basic mindsets were the first eight tools in that toolbox.

5 comments

Comments sorted by top scores.

comment by Pattern · 2020-06-06T19:08:39.593Z · LW(p) · GW(p)
The obvious disadvantage of synthesis mindset is that it has little power to screen the hypotheses it finds for accuracy or relevance. Synthesis cannot learn to smoothly navigate situations, nor to compare various options, nor to judge its hypotheses against the territory. For many serious problems, synthesis is necessary but insufficient to identify an effective solution.

An explicit set of assumptions can be fiddled with more directly. (A triangle may, by definition, have non-zero area, or it may not.)

Conversely, compared to analysis, organization mindset may miss some points of mismatch between its maps and reality, and can fail to apply enough distinct checking to catch flaws in its plans.

The goal of 'utility' can be replaced with information (expected or otherwise).

goats.

goals.


This post as a whole was great, and a lot of information. I think the time it takes to process might be a limiting factor on comments.

Replies from: ExCeph
comment by ExCeph · 2020-06-06T23:04:03.851Z · LW(p) · GW(p)

Thanks for the input!

1. You mean we can fiddle with the explicit assumptions we use with synthesis mindset? That can help, but to get the full benefit of synthesis I find it's often better to let go of explicit assumptions, and then apply other mindsets with those explicit assumptions to the results yielded by synthesis.

Otherwise our explicit assumptions may cause synthesis to miss hypotheses that ultimately point us in a helpful direction, even though the hypothesis itself violates the explicit assumptions. Sometimes the issue is that we make too many assumptions and need to remove some of them, and practicing synthesis is a good way to do that. Does that address your point?

2. I'm not sure what you mean by replacing the goal of 'utility' with information. Can you please elaborate on that?

3. Fixed, thanks. Not sure how the goats got in there, but I'll check the latch on the gate.

4. That's encouraging. I'll stand by for more feedback. Glad you liked it!

Replies from: Pattern
comment by Pattern · 2020-06-07T18:50:33.065Z · LW(p) · GW(p)

1.

You mean we can fiddle with the explicit assumptions we use with synthesis mindset?

I haven't fully digested your framework yet. (Connecting 'synthesis', and these other mindsets, to experience.) I mean that, if you have explicit assumptions then:

  • They're easier to examine
  • they can be messed with (as a way of generating hypothesis/ideas) It's 'random' but constrained enough that it has better odds of hitting something useful, or figuring out why something is wrong improves your understanding.

The way to use synthesis mindset is to allow your mind to wander through free association while suspending judgment of the accuracy or relevance of the hypotheses you encounter. It also helps to deliberately remove assumptions about certain ideas, and to toy with mixing multiple ideas together.

(Emphasis added.)

I might just be describing your 'synthesis' concept. (Or something similar, with a more systematic focus.*)


Math examples:

  • If multiplication is repeated addition, then what's repeated multiplication? Repeated powers?
  • What kind of space doesn't obey the triangle inequality?
  • If the sum of the interior angles of a shape are always the same, then squash the shape flat to find the sum.

The obvious disadvantage of synthesis mindset is that it has little power to screen the hypotheses it finds for accuracy or relevance. Synthesis cannot learn to smoothly navigate situations, nor to compare various options, nor to judge its hypotheses against the territory. For many serious problems, synthesis is necessary but insufficient to identify an effective solution.

*These don't seem like problems in math. (Except with untranslated or high level hypothesis. And smooth navigation takes time to build up.)


2.

replacing the goal of 'utility' with information

I read this:

"Conversely, compared to analysis, organization mindset may miss some points of mismatch between its maps and reality, and can fail to apply enough distinct checking to catch flaws in its plans."

and thought if you focus on gaining information instead of some other goal, that downside might go away. Information is a funny resource, but it can be accumulated over time. And before trying to do XYZ where X, Y, and Z are simple and therefore XYZ is simple, sometimes there's the option of first doing them individually (which should be easy because they are simple in theory).

Replies from: ExCeph
comment by ExCeph · 2020-06-08T04:06:17.498Z · LW(p) · GW(p)

1. Ah, now I see. Yes, removing assumptions is one good way to direct one's use of synthesis mindset. It helps with exploring the possibilities.

2. Organization can gather information efficiently, but integrating it all and catching contradictions is normally a job more suited for analysis. It's still possible to combine the two. That can end up forming strategy or something similar, or it could be viewed as using the mindsets separately to support each other.

Does that make sense?

Replies from: Pattern
comment by Pattern · 2020-06-11T14:43:27.543Z · LW(p) · GW(p)

Yes.