Two Explorations

post by alkjash · 2020-12-16T21:27:42.790Z · LW · GW · 8 comments

This is a link post for https://radimentary.wordpress.com/2020/12/16/two-explorations/

Contents

  Part 1: Distance to Safety
  Part 2: Explore Versus Exploit
None
8 comments

Much has been written about the fundamental opposition between explore and exploit, chaos and order, yin and yang. In this post I make two observations about the psychology of this opposition.

In the first part, I challenge the metaphor of the comfort zone: a slowly-changing region in activity-space where everything inside is comfortable and everything outside induces anxiety. The point is that anxiety depends not only on the spookiness of the activity itself, but on one's proximity to safety. I am afraid because it is dark; I am terrified because the light switch is all the way down the hall. In particular, it is possible to reduce anxiety by bringing your comfort zone with you in the form of a safety behavior or a trusted companion. This serves as an alternative to Comfort Zone Expansion.

In the second part, I note that explore and exploit are often embodied in the human personality as two competing subagents [? · GW]. In almost everyone I've met, one of these subagents dominates the other. I tell four typical stories of this imbalance, and then suggest that something better is possible. This is perhaps the central example of integrating disagreeing subagents [? · GW].

Part 1: Distance to Safety

1. Mother and Child

(The following is my retelling of an old story, dating back probably to at least Jean Piaget. I first encountered a variant of this story in The Monkey Wars, where identical behavior was observed in primates.)

A mother brings her child, a girl of perhaps three or four years of age, to an empty playground at the park. Autumn has progressed to the stage where leaves fall in twos and threes. The girl peeks out of the folds of her mother's coat, staring now at the swing set, now at the neon tube slide. With a nudge, the mother pushes her daughter onto the mulch and gives her an encouraging nod. After glancing back to her mother several times, the girl slowly approaches the playground and begins to play.

Soon, she is clambering about, but periodically looks back at the bench where her mother is sitting. Each time their eyes meet, the girl waves, pig-tails dancing, and the mother waves back. The girl then returns to free play.

At some point, the mother briefly leaves her post to chat with an acquaintance. The girl, pepping herself up to go down the slide for the first time, looks back, finds her mother gone, and panics. She curls up into a ball at the top of the playground and fights back tears, trying to make herself as small as possible. The thought of going down the slide vanishes from her mind.

2. The Difficulty of Dark Souls

Dark Souls (III) is one of those difficult video games which spawns endless internet debates about whether every game needs an easy mode. ("I have enough challenge in my day job, I just want to relax when I play a game," says the one. "Git gud scrub," says the other.) Dark Souls is not in fact mechanically difficult, or at least not exceptionally so. Just off the top of my head, Celeste, XCOM 2, and FTL were all significantly more difficult mechanically for me than Dark Souls. And yet I do believe that Dark Souls was the hardest game I ever played.

Call me a coward, but the main difficulty of Dark Souls for me is the psychological difficulty of its dominant aesthetic: loneliness and nihilism. You, the protagonist, are born "Nameless, accursed Undead, unfit even to be cinders. And so it is... that ash seeketh embers." All the friendlies in the world have been sitting in the same positions for eons before you died, and will be sitting there long after you return to ash. The brief encounters with friendly NPCs in the wild are ephemeral: each person passes in and out of your life, and you are as likely to meet them again as to stumble upon their ashes. The entire rest of the world is Out to Get You: treasure chests that turn out to be mimics, halberdiers hide between crates, face-eaters hang from ceilings. Your job is to save a world that doesn't care to be saved.

I could never play Dark Souls for more than a couple hours at a time, and found myself constantly teleporting back to base. I told myself I came home to level up, to upgrade gear, and to purchase items, but the truth of the matter is: I kept coming home just to hear a friendly human voice again.

Code Vein is one of many Dark Souls knockoffs, notorious mainly for the addition of anime waifus to the familiar formula. For all its abysmal enemy variety and boring level design, I loved playing Code Vein for one simple reason: I can bring a companion. Bringing a companion on my journey solved all my anxiety in-game. It didn't matter so much that my digital companion got stuck in corners and committed suicide in boss fights and repeated the same half-dozen canned lines. The feeling that someone has my back let me enjoy a Souls-like world for entire afternoons at a time, and I almost never felt the need to teleport back home.

Upshot

Human beings explore too little. One common solution to this problem is Comfort Zone Expansion, gently straying beyond the boundary of your comfort zone to notice things out there are not as scary as you thought. This can be a fine solution, but is inadequate for situations where you are required to immediately leave your comfort zone far behind for long periods of time.

Perhaps you travel internationally for the first time. Perhaps you attend a self-improvement workshop with cult-ish vibes in a remote location. Perhaps you try to prove the Riemann hypothesis, and are beset on all sides by the diabolical malice inherent in the primes.

If so, remember that exploration anxiety is a function of how far (you feel like) you are from safety.

The little girl comfortably clambers around the playground when her mother is nearby. When the mother disappears, the girl curls up in fear and is incapable of sliding down the exact same slide she went down a minute ago. The slide didn't change, the girl's distance to safety did. Playing Dark Souls, I teleport to base after finding each new bonfire (checkpoint). Playing Code Vein, a companion follows me around and the psychological need return home disappears. Every time a married person gives an acceptance speech, they thank their spouse for being "their rock." This is a rather unflattering term for "unwavering center of my comfort zone."

What are concrete applications of this principle?

  1. One purpose of collaboration is for each collaborator to serve as a mobile comfort zone for the others. This might explain why most successful startups are founded by a small group and not an individual or a larger group. The comfort zone effect hits rapidly diminishing returns past one trusted collaborator. In this lens, the purpose of open communication in collaboration is simply to feel psychologically close to the other person.
  2. The optimal solution to comfort zone expansion may be planting a small number of well-spaced "bases of operation." Instead of continuously expanding one connected chunk of activity-space, plant comfort flags on the points of an ε-net. Comfort zones, like lighthouses and highway truck stops, cover more space if you place them far apart. If anxiety is a major limiting factor for you, consider focusing your energy on a small number of extremely different activities so that the comfort zones that radiate out of each one together cover as much space as possible.

Part 2: Explore Versus Exploit

The previous part upgrades the usual model for comfort zone expansion, taking for granted the value of exploration.

This part turns to a new topic altogether: the internal conflict between the drive to explore and the drive to exploit.

Hereafter I assume a multi-agent model [? · GW] of the mind, and refer to two common subagents: the "explore" subagent which tends towards freedom, creativity, contrarianism, and chaos, and the "exploit" subagent which tends towards structure, discipline, lawfulness, and order. Whether to interpret these subagents as full-blown independent subpersonalities or merely conflicting desires within a single mind is entirely up to you, and should not affect the meaning of this post.

I begin with four archetypal stories to illustrate varying levels of imbalance between "explore" and "exploit." These are amalgams of real stories from my own life and others'.

1. The Unmoored

The Unmoored was stifled as a girl, surveilled at every quarter-hour by parents, teachers, tutors, and coaches. Every day from dawn to dusk was packed with activity that would gentle her mind and ennoble her condition. As her fingers marched from piano to textbook to tennis racket, her mind danced farther and farther away into Wonderland.

But even her dreams are turned against her. After she shows a passing aptitude for story-telling, her parents sign her up for creative writing classes and poetry jams. The beloved characters in her flights of fancy are clamped into straightjackets and paraded onto stage to be judged by panels of condescending curmudgeons.

When she finally escapes her shackles, there is no temperance to her wildness, no second-guessing, no backward glances. She drops out of pre-med, then out of college, then out of polite society altogether. The world is joy and light and one-way airfares.

Many years later, she snaps awake in the lap of a street urchin in the outskirts of Ulaan Bataar. She'd had a strange dream: she was back in front of the piano playing Mendelssohn, and she liked that feeling of rote and mindless obedience to the sheet music. She shakes off that absurd notion and takes another hit.

2. The Dreamer

Like the Unmoored, the Dreamer yearns to be free, to one day live fully unconstrained to pursue his creative vision. He has not quite decided whether he wants to write screenplays, musicals, or novels; his creative side finds the even the idea of making such a decision fettering. In the meantime, he works an programming job that makes good money.

Unlike the Unmoored, the Dreamer knows the instrumental value of discipline and constraint. Under the flickering lamplight, he writes short stories with a tomato timer by his side, following a classic book of writing prompts. The more exotic the prompt, the more alive he feels bending around it.

At times, the Dreamer worries that his day job is changing him. He cannot help but take joy in turning on his monitors in the morning, in passing code review on the first try, in following procedures to the letter to build something that comes alive before his very eyes.

When he notices himself feeling this way, the Dreamer clamps down on this joy and reminds himself, "I hate this boring, technical job. I'm only working it to get my art off the ground. One day, I will finally be free of this drudgery."

The tomato timer goes off and he goes back to writing. He never considers that part of his soul might not want to be free.

3. The Magpie

Unlike the Unmoored and the Dreamer, the Magpie is primarily motivated by the joy and comfort of the known. She delights in decorating and redecorating her cozy little apartment, in organizing her books and plants in tidy rows, and in folding origami animals that she can leave around all the rooms, each a permanent addition to her little family.

One day, she hopes to add a few extra bedrooms and a full bathtub to that apartment, and a partner and children to that family. She is the Dreamer's coworker, but unlike him she hopes to stay at this programming job for the rest of her career. She imagines inviting direct reports into her well-lit corner office to admire her cactus collection and ask for her advice on the operating system she helped architect.

The Magpie understands the instrumental value of exploration and creativity, but fears it. Every year, she takes a vacation to a scary new place to challenge herself, but more importantly to bring back souvenirs, pictures, and memories, to better decorate her place. At work, she pushes herself to learn new technologies and programming languages, but she yearns for the day she won't have to do this any more.

The Magpie believes that at the end of the day, one should only explore until one finds the best place to nest.

4. The Recluse

Like the Magpie, the Recluse is primarily motivated by comfort and familiarity. But his world was constantly in flux for too long, so he does everything in his power to hide away in his comfort zone and wall off the rest of the world.

His parents divorced before he entered middle school, and he ping-ponged between their two new families, never truly belonging to either. Traveling, learning new things, meeting strangers all terrify him, and yet he's forced to do more and more of all of these to survive.

When he finally finds a place to settle down, he never leaves it again. In every relationship, he becomes deeply codependent. He is a consistent member of a few local clubs; new faces join and leave, but he is one of the few who remain through the years. If he had his way, the clubs would meet in his living room and he'd never have to go outside.

Occasionally, he reconnects with an old friend who is a Dreamer or an Unmoored, and becomes deeply fascinated with their alien way of being. It is hard for him to understand how comfort and familiarity can be claustrophobic, but he's glad he has these friends. They bring him new knowledge and experiences in a safe and digested way, or if not, at least they send him postcards.

Upshot

In each of the above four stories, there is an imbalance between the "explore" subagent and the "exploit" subagent.

The Unmoored lives to "explore." The "exploit" subagent is greatly suppressed or externalized.

The Dreamer also lives to "explore", but understands the instrumental value of "exploit." He views "exploit" as a unsightly means to an end, and suppresses the needs of the personality (comfort, safety, order) associated with it.

The Magpie lives to "exploit," but understands the instrumental value of "explore." She views "explore" as a dangerous means to an end, and suppresses the needs of the personality (freedom, creativity, chaos) associated with it.

The Recluse also lives to "exploit." The "explore" subagent is greatly suppressed or externalized.

In all four cases, one subagent or the other dominates the personality, and holds the other subagent and its needs in contempt. Internal alignment of the two subagents can only occur if the whole person recognizes not only the instrumental value of each subagent, but respects their needs as ends in themselves. Here is my loose prescription for alignment, which might be attempted with an exercise like Internal Double Crux [LW · GW]:

  1. If you are near the extremes (the Unmoored or the Recluse), learn to recognize at least the instrumental value of the suppressed subagent. If you lean heavily towards exploring, recognize that more systematic exploiting can often make you better at exploring in the long run. Similarly, if you lean heavily towards exploiting, recognize that more systematic exploring can often make you better at exploiting in the long run. Hopefully, you will level up into the Dreamer or the Magpie.
  2. If you are near the middle (the Dreamer or the Magpie), learn to respect the needs of the weaker subagent as ends in themselves. If you lean towards exploring, realize that it's genuinely ok for you to enjoy checking boxes, following rules, and tidying things up. If you lean towards exploiting, realize that it's genuinely ok for you to enjoy trying crazy things, breaking rules, and making a mess.

8 comments

Comments sorted by top scores.

comment by Kaj_Sotala · 2020-12-17T19:17:50.835Z · LW(p) · GW(p)

Besides looking at a person's overall balance of explore/exploit, it's also interesting to look how that balance plays out in the case of specific issues.

An observation that e.g. IFS therapists have made is that often when a person seems to be struggling with two subagents that are pushing in opposite directions, those two subagents are ultimately optimizing for the same goal, but with opposite strategies. (It feels more natural to describe their general class of strategies as something like "safety / passivity" vs. "adventure / activity", but those do have a resemblance to explore / exploit as you've described it here.)

For example, someone may have a workaholic subagent that pushes them to do nothing but be productive, and a slacker subagent that never wants to do any work at all. Both may be reactions to an insecurity over whether the person has been productive enough - with one of the subagents trying to eliminate that insecurity by working all the time, and the other trying to eliminate it by deciding that actually work doesn't matter at all and it's fine never to work.

A list of some such typical opposite-paired agents that I wrote down during some lecture of my IFS training:

  • Obligated <-> Ambitious, risk-taking
  • Speak up <-> Silence
  • Stay <-> Leave
  • Striver/busy <-> Avoidant
  • Engaged/emotional <-> Numbing
  • Emotion <-> Reason
  • Selfish/egocentric <-> Caretaker
  • Overindulgent <-> In control
  • Inner critic <-> Inner defender
comment by romeostevensit · 2020-12-17T04:28:19.095Z · LW(p) · GW(p)

One benefit of nomading that I think makes it endorseable for far larger numbers of people is that it encourages building the skill of taking your comfort with you, and rebuilding it when it is damaged in different contexts. I am sad I didn't do it sooner.

comment by DirectedEvolution (AllAmericanBreakfast) · 2020-12-17T04:12:26.778Z · LW(p) · GW(p)

I've been exploring with memorizing my textbooks this year. Part of the value, I've found, is the sense of comfort and familiarity that comes with every theorem I commit to memory. With the bulk of the material memorized, filling in the gaps isn't a scary venture into the unknown, but stitching together a new panel on a quilt that's already grown sizeable.

comment by adamShimi · 2020-12-17T12:52:22.946Z · LW(p) · GW(p)

Thanks for this insightful (and well-written) post! 

This is perhaps the central example of integrating disagreeing subagents [? · GW].

Thanks for pointing to that post, it's exactly what I needed to read to write my own post on integrating together my different identities.

At some point, the mother briefly leaves her post to chat with an acquaintance. The girl, pepping herself up to go down the slide for the first time, looks back, finds her mother gone, and panics. She curls up into a ball at the top of the playground and fights back tears, trying to make herself as small as possible. The thought of going down the slide vanishes from her mind.

At that point, I thought that I also know kids who don't even look around before trying stuff. You actually discuss this later in the post, but I would have enjoyed a pointer to that so that this example feels less like you remove half the cases.

The optimal solution to comfort zone expansion may be planting a small number of well-spaced "bases of operation." Instead of continuously expanding one connected chunk of activity-space, plant comfort flags on the points of an ε-net. Comfort zones, like lighthouses and highway truck stops, cover more space if you place them far apart. If anxiety is a major limiting factor for you, consider focusing your energy on a small number of extremely different activities so that the comfort zones that radiate out of each one together cover as much space as possible.

How do you think this apply to intellectual pursuits? I have in mind research advising: in my experience, some people that I think could be great researchers are terrified of exploring some part of knowledge where there is no answer yet. And even we established researchers can easily be afraid of learning a new subject or a new technique that would help them tremendously. Maybe the comfort flags should be links with stuff that the graduate student/researcher knows well? Anecdotally, people seem more open to learning about what you want to say if you link it to their own field.

If you are near the middle (the Dreamer or the Magpie), learn to respect the needs of the weaker subagent as ends in themselves. If you lean towards exploring, realize that it's genuinely ok for you to enjoy checking boxes, following rules, and tidying things up. If you lean towards exploiting, realize that it's genuinely ok for you to enjoy trying crazy things, breaking rules, and making a mess.

I would say I'm more of a Dreamer myself, at least naturally. And indeed, most of my internal progress came in the form of accepting a part of routine, of rituals and time scheduling, so that I can explore what I care about and create. Damn, my last sentence shows that I'm not there yet in accepting my exploiting subagent. But I'm getting better.

Replies from: alkjash
comment by alkjash · 2020-12-17T15:59:25.940Z · LW(p) · GW(p)

How do you think this apply to intellectual pursuits? I have in mind research advising: in my experience, some people that I think could be great researchers are terrified of exploring some part of knowledge where there is no answer yet. And even we established researchers can easily be afraid of learning a new subject or a new technique that would help them tremendously. Maybe the comfort flags should be links with stuff that the graduate student/researcher knows well? Anecdotally, people seem more open to learning about what you want to say if you link it to their own field.

I don't pretend to be an established researcher, but here is what I had in mind. Most researchers at one point or other spend some amount of time white-knuckle learning things that are outside their comfort zones, but usually these things are just barely outside. My suggestion would be that all other things equal, some of that time should be spent learning things really far out instead.

Also I think learning in pairs is a very helpful tool. The active ingredient is to have someone you trust enough to freely share your ignorance and ask basic questions, and the easiest way to get this trust is to find someone who is also obviously ignorant of the same thing.

Replies from: adamShimi
comment by adamShimi · 2020-12-17T17:18:45.261Z · LW(p) · GW(p)

My suggestion would be that all other things equal, some of that time should be spent learning things really far out instead.

I agree, but I was more asking about how you think your insight about the "distance to safety" can help with that.

Also I think learning in pairs is a very helpful tool. The active ingredient is to have someone you trust enough to freely share your ignorance and ask basic questions, and the easiest way to get this trust is to find someone who is also obviously ignorant of the same thing.

Interesting. My own approach is usually to collaborate/ask someone who knows the subject you want to learn. But that does require being okay with asking stupid questions.

Replies from: alkjash
comment by alkjash · 2020-12-17T17:31:19.860Z · LW(p) · GW(p)

I agree, but I was more asking about how you think your insight about the "distance to safety" can help with that.

Well, after a bounded number of initially difficult "far-out explorations" that cover the research landscape efficiently, the hope is that almost everything is reasonably close to safety henceforth.

Interesting. My own approach is usually to collaborate/ask someone who knows the subject you want to learn. But that does require being okay with asking stupid questions.

Yes, I think your approach is ideal for the efficiency of learning if anxiety was not a factor. Unfortunately the people who know the subjects I want to learn best are people I care about impressing and/or people so well-versed in the subject that they have difficulty bridging the inferential abyss between us. At least for me it is hard to treat them as a "psychologically nearby" companion who has my back.

Even after getting much better at asking stupid questions, it feels like the maximum I feel okay with asking in a meeting with someone who knows a subject already is ~3, and not ~40, which is the number I want to ask.

comment by Ben Pace (Benito) · 2020-12-16T22:44:36.953Z · LW(p) · GW(p)

I love your writing. That is all.