From self to craving (three characteristics series)

Bud­dhists talk a lot about the self, and also about suffer­ing. They claim that if you come to in­ves­ti­gate what the self is re­ally made of, then this will lead to a re­duc­tion in suffer­ing. Why would that be?

This post seeks to an­swer that ques­tion. First, let’s re­cap a few things that we have been talk­ing about be­fore.

The con­nec­tion be­tween self and craving

In “a non-mys­ti­cal ex­pla­na­tion of ‘no-self’”, I talked about the way in which there are two kinds of goals. First, we can ma­nipu­late some­thing that does not re­quire a rep­re­sen­ta­tion of our­selves. For ex­am­ple, we can figure out how to get a truck on a column of blocks.

In that case, we can figure out a se­quence of ac­tions that takes the truck from its ini­tial state to its tar­get state. We don’t nec­es­sar­ily need to think about our­selves as we are figur­ing this out—the ac­tual se­quence could just as well be car­ried out by some­one else.

I men­tioned that these kinds of tasks seem to al­low flow states, in which the sense of self be­comes tem­porar­ily sus­pended as un­nec­es­sary, and which are typ­i­cally ex­pe­rienced as highly en­joy­able and free from dis­com­fort.

Alter­na­tively, we can think of a goal which in­trin­si­cally re­quires self-refer­ence. For ex­am­ple, I might be feel­ing sad, and think that I want to feel happy in­stead. In this case, both the ini­tial state and the tar­get state are defined in terms of what I feel, so in or­der to mea­sure my progress, I need to track a refer­ence to my­self.

In that post, I re­marked that chang­ing one’s ex­pe­rience of one’s self may change how emo­tions are ex­pe­rienced. This does not nec­es­sar­ily re­quire high lev­els of en­light­en­ment: it is a com­mon mind­ful­ness prac­tice to re­frame your emo­tions as some­thing that is ex­ter­nal to you, in which case nega­tive emo­tions might cease to feel aver­sive.

I have also pre­vi­ously dis­cussed ther­apy tech­niques that al­low you to cre­ate some dis­tance be­tween your­self and your feel­ings, mak­ing them less aver­sive. For ex­am­ple, one may pay at­ten­tion to where in their body they feel an emo­tion, keep their at­ten­tion on those phys­i­cal feel­ings, and then al­low a vi­sual image of that emo­tion to arise. This may then cause the emo­tion to be ex­pe­rienced as “not me”, in a way which makes it eas­ier to deal with.

The next post in my se­ries was on crav­ing and suffer­ing, where I dis­t­in­guished two differ­ent kinds of mo­ti­va­tion: crav­ing and non-crav­ing. I claimed that crav­ing is in­trin­si­cally as­so­ci­ated with the de­sire to ei­ther ex­pe­rience pos­i­tive valence, or to avoid ex­pe­rienc­ing nega­tive valence. Non-crav­ing-based mo­ti­va­tion, on the other hand, can care about any­thing; not just valence.

In par­tic­u­lar, I claimed that dis­com­fort /​ suffer­ing (un­satis­fac­tori­ness, to use a generic term) is cre­ated by crav­ing: crav­ing at­tempts to re­sist re­al­ity, and in so do­ing gen­er­ates an er­ror sig­nal which is sub­jec­tively ex­pe­rienced as un­satis­fac­tori­ness. I also sug­gested that non-crav­ing-based mo­ti­va­tion does not re­sist re­al­ity in the same way, so does not cre­ate un­satis­fac­tori­ness.

Put­ting some of these pieces to­gether:

  • Crav­ing tries to en­sure that “the self” ex­pe­riences pos­i­tive feel­ings and avoids nega­tive feel­ings.

  • If there are con­sciously ex­pe­rienced feel­ings which are not in­ter­preted as be­ing ex­pe­rienced by the self, it does not trig­ger crav­ing.

  • There is a two-way con­nec­tion be­tween the sense of self and crav­ing.

    • On one hand, ex­pe­rienc­ing a strong sense of self trig­gers crav­ing, as feel­ings are in­ter­preted as hap­pen­ing to the self.

    • From the other di­rec­tion, once crav­ing is trig­gered, it sends into con­scious­ness the goal of ei­ther avoid­ing or ex­pe­rienc­ing par­tic­u­lar feel­ings. As this goal is one that re­quires mak­ing refer­ence to the self, send­ing it into con­scious­ness in­stan­ti­ates a sense of self.

Let’s look at this a bit more.

Crav­ing as a sec­ond layer of motivation

A ba­sic model is that the brain has sub­sys­tems which op­ti­mize for differ­ent kinds of goals, and then pro­duce pos­i­tive or nega­tive valence in pro­por­tion to how well those goals are be­ing achieved.

For ex­am­ple, ap­praisal the­o­ries of emo­tion hold that emo­tional re­sponses (with their un­der­ly­ing pos­i­tive or nega­tive valence) are the re­sult of sub­con­scious eval­u­a­tions about the sig­nifi­cance of a situ­a­tion, rel­a­tive to the per­son’s goals. An eval­u­a­tion say­ing that you have lost some­thing im­por­tant to you, for ex­am­ple, may trig­ger the emo­tion of sad­ness with its as­so­ci­ated nega­tive valence.

Or con­sider a situ­a­tion where you are suc­cess­fully car­ry­ing out some phys­i­cal ac­tivity; play­ing a fast-paced sport or video game, for ex­am­ple. This is likely to be as­so­ci­ated with pos­i­tive valence, which emerges from hav­ing suc­cess at the task. On the other hand, if you were failing to keep up and couldn’t get into a good flow, you would likely ex­pe­rience nega­tive valence.

Valence looks like a sig­nal about whether some goals/​val­ues are be­ing suc­cess­fully at­tained. A sub­sys­tem may have a goal X which it pur­sues in­de­pen­dently, and de­pend­ing on how well it goes, valence is pro­duced as a re­sult. In this model, be­cause valence tends to sig­nal states that are good/​bad for the achieve­ment of an or­ganism’s goals, crav­ing acts as an ad­di­tional layer that “grabs onto” states that seem to be par­tic­u­larly good/​bad, and tries to di­rect the or­ganism more strongly to­wards those.

There’s a sub­tlety here, in that crav­ing is dis­tinct from nega­tive valence, but crav­ing can cer­tainly pro­duce nega­tive valence. For ex­am­ple:

  • You are feel­ing stressed out, so you go for a long walk. This helps take your mind off the stress, and you come back re­laxed.

  • The next time you are stressed and feel like you need a break, you re­mem­ber that the walk helped you be­fore. It’s im­por­tant for you to get some more work done to­day, so you take an­other walk in the hopes of calming down again. As you do, you keep think­ing “okay, am I about to calm down and for­get my wor­ries now”—and this ques­tion keeps oc­cu­py­ing you through­out your walk, so that when you get back home, you haven’t ac­tu­ally re­laxed at all.

I think this has to do with crav­ing in­fluenc­ing the goals and in­puts that are fed into var­i­ous sub­sys­tems. If crav­ing gen­er­ates the goal of “I should re­lax”, then that goal is taken up by some plan­ning sub­sys­tem; and suc­cess or failure at that goal, may by it­self pro­duce pos­i­tive or nega­tive valence just like any other goal does. This also means that crav­ing may gen­er­ate emo­tions that gen­er­ate ad­di­tional crav­ing: first you have a crav­ing to re­lax on a walk, but then go­ing on the walk pro­duces frus­tra­tion, cre­at­ing crav­ing to be rid of the frus­tra­tion...

My model about crav­ing and non-crav­ing seems some­what similar to the pro­posed dis­tinc­tion be­tween model-based and model-free goal sys­tems in the brain. The model-based sys­tem does com­plex rea­son­ing about what to do; it is ca­pa­ble of pretty so­phis­ti­cated analy­ses, but re­quires sub­stan­tial com­pu­ta­tional re­sources. To save on com­pu­ta­tion, the model-free sys­tem re­mem­bers which ac­tions have led to good or bad out­comes in the past, and tries to re­peat/​avoid them. Un­der this model, crav­ing would be as­so­ci­ated with some­thing like the model-free sys­tem, one which used “did an ac­tion pro­duce pos­i­tive or nega­tive valence” as a short­hand for whether ac­tions should be taken or avoided.

How­ever, it would be a mis­take to view these as two en­tirely dis­tinct sys­tems. Re­search sug­gests that even within the same task, both kinds of sys­tems are ac­tive, with the brain adap­tively learn­ing which one of the sys­tems to de­ploy dur­ing which parts of the ac­tivity. Crav­ing valence and more com­plex model-based rea­son­ing also seem to be in­ter­twined in var­i­ous ways, such as:

  • It is pos­si­ble to un­learn par­tic­u­lar crav­ings, as the brain up­dates to no­tice that this spe­cific crav­ing is not ac­tu­ally use­ful. The un­learn­ing pro­cess seems to in­volve some de­gree of model-based eval­u­a­tion, as the brain seems re­sis­tant to re­lease crav­ings if it pre­dicts that do­ing so would make it harder to achieve some par­tic­u­lar goal.

  • Model-based sub­sys­tems may act in ways that seem to make use of crav­ing. For ex­am­ple, in an ear­lier post re­view­ing the book Un­lock­ing the Emo­tional Brain, I dis­cussed the ex­am­ple of Richard. A sub­sys­tem in his brain pre­dicted that if he were to ex­press con­fi­dence, peo­ple would dis­like him, so it pro­jected nega­tive self-talk into his con­scious­ness to pre­vent him from be­ing con­fi­dent. Pre­sum­ably the self-talk had nega­tive valence and caused a crav­ing to avoid it, in a way which con­tributed to him not say­ing any­thing. I sus­pect that this is part of why mind­ful­ness prac­tice may re­lease buried trauma: if you get bet­ter at deal­ing with mild aver­sion, that aver­sion might pre­vi­ously have been used by sub­sys­tems to keep nega­tive mem­o­ries con­tained—and then your mind gets flooded with re­ally un­pleas­ant mem­o­ries, and much stronger un­satis­fac­tori­ness than you were nec­es­sar­ily pre­pared to deal with.

It is also worth mak­ing ex­plicit that pur­su­ing pos­i­tive valence and avoid­ing nega­tive valence are goals that can be pur­sued by non-crav­ing-based sub­sys­tems as well.

For ba­si­cally any goal, there can be crav­ing-based and non-crav­ing-based mo­ti­va­tions. You might pur­sue plea­sure be­cause of a crav­ing for plea­sure, but you may also pur­sue it be­cause you value it for its own sake, be­cause ex­pe­rienc­ing plea­sure makes your mind and body work bet­ter than if you were only ex­pe­rienc­ing un­hap­piness, be­cause it is use­ful for re­leas­ing crav­ing… or for any other rea­son.

From self to crav­ing and from crav­ing to self

I have mostly been talk­ing about crav­ing in terms of aver­sion (crav­ing to avoid a nega­tive ex­pe­rience). Let’s look at some ex­am­ples of crav­ing a pos­i­tive ex­pe­rience in­stead:

  • It is morn­ing and your alarm bell rings. You should get up, but it feels nice to be sleepy and re­main in bed. You want to hang onto those pleas­ant sen­sa­tions of sleep­iness for a lit­tle bit more.

  • You are spend­ing an evening to­gether with a loved one. This is the last oc­ca­sion that you will see each other in a long time. You feel re­ally good be­ing with them, but a small part of you is un­happy over the fact that this evening will even­tu­ally end.

  • You are at work on a Fri­day af­ter­noon. Your mind wan­ders to the thought of no longer be­ing at work, and do­ing the things you had planned for the week­end. You would pre­fer to be done with work already, and find it hard to stay fo­cused as you cling to the thoughts of your free time.

  • You are sin­gle and hang­ing out with an at­trac­tive per­son. You know that they are not into you, but it would be so great if they were. You can’t stop think­ing about that pos­si­bil­ity, and this keeps dis­tract­ing you from the ac­tual con­ver­sa­tion.

  • You are in a con­ver­sa­tion with sev­eral other peo­ple. You think of a line that would be a great re­sponse to what some­one else just said. Be­fore you can say it, some­body says some­thing, and the con­ver­sa­tion moves on. You find your­self still think­ing of your line, and how nice it would have been to get to say it.

  • You had been plan­ning on go­ing to a fa­mous mu­seum while on your va­ca­tion, but the mu­seum turns out to be tem­porar­ily closed at the time. You keep think­ing about how much you had been look­ing for­ward to it.

  • You are hun­gry, and keep think­ing about how good a par­tic­u­lar food would taste, and how much bet­ter you would feel af­ter you had eaten.

Th­ese cir­cum­stances are all quite differ­ent, but on a cer­tain ab­stract level, they share the same kind of a mechanism. There is the thought of some­thing pleas­ant, which trig­gers crav­ing, and a de­sire to ei­ther get into that pleas­ant state or make sure you re­main in it.

Let’s say that there is this kind of a pro­cess:

Sense con­tact. You have some par­tic­u­lar ex­pe­rience, such as the sen­sa­tion of be­ing sleepy and in bed, or the thought of how it would feel if the at­trac­tive per­son in front of you were to like you. In third-per­son terms, this sen­sa­tion/​thought is sent to the global workspace of your con­scious­ness.

Valence. An emo­tional sys­tem clas­sifies this ex­pe­rience as be­ing pos­i­tive, and paints it with a cor­re­spond­ing pos­i­tive gloss as it is be­ing broad­cast into the workspace.

Crav­ing. A crav­ing sub­sys­tem no­tices the pos­i­tive valence, and in­ter­prets the valence as be­ing ex­pe­rienced by you. The sub­sys­tem reg­isters the imag­ined state the valence is as­so­ci­ated with, so it sets the in­ten­tion of get­ting the self to achieve that state, and the re­sult­ing pos­i­tive valence. For ex­am­ple, it may set the in­ten­tion of get­ting into a re­la­tion­ship with that at­trac­tive per­son.

Cling­ing and plan­ning. This in­tent clings in your mind and is fed to plan­ning sub­sys­tems. They make plans of how to get to the state which has been eval­u­ated as be­ing bet­ter. (My pre­vi­ous post was largely a de­scrip­tion of this stage.)

Birth. As these plans are broad­cast into con­scious­ness for eval­u­a­tion, they con­tain a rep­re­sen­ta­tion of your cur­rent state, cre­at­ing a stronger ex­pe­rience of hav­ing a dis­tinct self that wants a par­tic­u­lar thing. Sub­sys­tems may em­pha­size par­tic­u­lar as­pects of their model of you that seem rele­vant for achiev­ing the goal: for ex­am­ple, if the other per­son seems to be drawn to­wards mu­si­ci­ans, your skill at this may be­come high­lighted. Maybe you, be­ing a mu­si­cal kind of per­son, could play the gui­tar and make a good im­pres­sion on them… em­pha­siz­ing your “mu­si­cian na­ture” in the self-rep­re­sen­ta­tion that ap­pears in con­scious­ness.

Death. Even­tu­ally, you stop pur­su­ing the goal, at least for the time be­ing. Maybe you get what you wanted, it turns out to be un­vi­able right now, or you just get dis­tracted and for­get. This par­tic­u­lar goal-state and plan dis­ap­pear from con­scious­ness, and with it, the sense of self that was track­ing its com­ple­tion dis­ap­pears as well. Be­fore long—or maybe even si­mul­ta­neously—an­other self will be cre­ated, one which cares about some­thing com­pletely differ­ent: maybe you got hun­gry, and now the crav­ing only cares about get­ting some food, cre­at­ing a food-hun­gry self...

Crav­ing and the self-model

There’s a straight­for­ward rea­son for why crav­ing should be tied into a con­cep­tion of the self: its pur­pose is to mo­ti­vate you to ac­tion. If your brain pre­dicts that some­one other than you would ex­pe­rience pos­i­tive valence, this does not trig­ger crav­ing in the same way. (Imag­ine get­ting the one thing that you most de­sire at the mo­ment. Then imag­ine some com­plete stranger get­ting a similar thing. Not quite the same, is it?)

Your crav­ing sub­sys­tems have been wired to make you ex­pe­rience pos­i­tive valence /​ avoid nega­tive valence. Each crav­ing sub­sys­tem con­tains an im­plicit schema along the lines of “I will strive to­wards a more pos­i­tive ex­pe­rience, and then I will have that more pos­i­tive ex­pe­rience in the end”. If a crav­ing sub­sys­tem pre­dicted that its ac­tions would pro­duce some­one else a more pos­i­tive ex­pe­rience, this would not fulfill the goal con­di­tion in that schema. (Of course, crav­ing may have the in­stru­men­tal goal of mak­ing some­one else happy, if it pre­dicts that leads to a pos­i­tive con­se­quence for you.)

At the same time, there is some­thing in­ter­est­ing go­ing on with the fact that mind­ful­ness and cog­ni­tive de­fu­sion prac­tices work: if you can men­tally trans­form a source of nega­tive valence into some­thing that feels ex­ter­nal to you, it may not bother you as much. In par­tic­u­lar, it feels like you don’t need to re­act to it: that is, there is less of a crav­ing to get it out of your con­scious­ness. The in­fer­ence of “if I get this emo­tion out of my mind, I will feel bet­ter” is never ap­plied, as the emo­tion is not ex­pe­rienced as be­ing “in my mind” in the first place.

In other words, the mere ex­pe­rience or pre­dic­tion of pos­i­tive/​nega­tive valence alone does not seem to be enough to trig­ger crav­ing. The valence also needs to be bound into a par­tic­u­lar ab­stract rep­re­sen­ta­tion of the self, and in­ter­preted as hap­pen­ing to “you”.

At this point, we need to dis­t­in­guish be­tween two differ­ent senses of “hap­pen­ing to you”:

  • Hap­pen­ing to the sys­tem defined by the phys­i­cal bound­aries of your body; for con­scious ex­pe­riences, ap­pear­ing in the global workspace lo­cated within that body. I will call this “hap­pen­ing to the sys­tem”.

  • Hap­pen­ing to “the self”, an ab­stract tag which is com­puted within the sys­tem, and which feels like the en­tity that in­ter­nal ex­pe­riences such as emo­tions hap­pen to; typ­i­cally only in­cludes part of the con­tent in the global workspace. I will call this “hap­pen­ing to the self-model”.

Crav­ing re­acts to valence that is ex­pe­rienced by the self-model. It may seem to also re­act to events that hap­pen to the sys­tem, such as get­ting to sleep for longer, but this always takes place through some­thing that bot­toms out at the self-model ex­pe­rienc­ing valence. E.g. the thought of sleep­ing longer cre­ates pos­i­tive valence, and the crav­ing re­acts to the pos­i­tive valence be­com­ing in­cor­po­rated in the self-model. (At the same time, the pre­dic­tions that crav­ing is based on are not nec­es­sar­ily ac­cu­rate or up to date, so there is also crav­ing for things that we do not ac­tu­ally en­joy.)

At the same time, non-crav­ing-based mo­ti­va­tion may re­act to any­thing, in­clud­ing events that hap­pen to the sys­tem rather than the self-model. Even if the thought of get­ting to sleep for longer didn’t cause crav­ing, you could still stay in bed in or­der to get more rest. Be­cause this in­volves some de­gree of ab­stract rea­son­ing, crav­ing is prob­a­bly some­thing that hu­mans need to have in place be­fore non-crav­ing-based sys­tems have had a chance to ma­ture and ac­quire so­phis­ti­cated mod­els. A tod­dler is not ca­pa­ble of in­tel­lec­tu­ally figur­ing out the con­se­quences of most harm­ful things and how they should thus be avoided, but the tod­dler does still have visceral crav­ing to avoid pain.

Ear­lier in this se­ries, I men­tioned a metaphor of ex­pe­rienc­ing your­self as the sky, and all the emo­tions and thoughts as things that are on the sky but which do not af­fect the sky. This was in­tended as a metaphor for how it feels once your mind comes to iden­tify with your un­der­ly­ing field of con­scious­ness, as op­posed to the con­tents of that con­scious­ness. I noted that this raised a ques­tion which I promised to come back to: why would this kind of a shift in iden­ti­fi­ca­tion re­duce suffer­ing?

What I have out­lined above sug­gests an an­swer: be­cause crav­ing sub­sys­tems are ac­ti­vated by a pre­dic­tion of pos­i­tive or nega­tive valence be­ing ex­pe­rienced by the self-model. If the sys­tem’s self-model changes so that ex­pe­riences are no longer in­ter­preted as hap­pen­ing to the self-model, crav­ing will not trig­ger, nor will it pro­duce a feel­ing of un­satis­fac­tori­ness. At the same time, non-crav­ing-based mo­ti­va­tion can still rea­son about the con­se­quences of those events and re­spond ap­pro­pri­ately.

In these ar­ti­cles, I have used the term “no-self”, be­cause that seems to be the es­tab­lished trans­la­tion for anatta; but I could also have used the ar­guably bet­ter trans­la­tion of“not self”.

In other words, the brain has some­thing of a built-in model defin­ing what kinds of crite­ria a piece of con­scious ex­pe­rience needs to fulfill in or­der to be in­cor­po­rated into the self-model. With suffi­ciently close study, one may come to no­tice the way in which ac­tu­ally no con­scious con­tent fulfills this crite­ria: there is noth­ing in con­scious­ness that is “self”… and if there is noth­ing that is in­cor­po­rated into the self-model, then there is noth­ing that could trig­ger crav­ing.

At the same time, some “no-self” ex­pe­riences also seem to also be in­ter­preted as “ev­ery­thing is self” rather than “noth­ing is self”. For ex­am­ple, my ear­lier post on no-self quoted an ex­pe­rience de­scribed as “this is all me [...] my iden­tity is liter­ally ev­ery­thing that I could see through my eyes”. Crav­ing also seems re­duced in these situ­a­tions.

With emo­tions, the raw ex­pe­rience of the emo­tion is dis­tinct from the pro­cess of nam­ing the emo­tion; the same emo­tion may have differ­ent names in differ­ent lan­guages. Like­wise, the sub­sys­tem which cre­ates the ab­stract bound­ary that crav­ing re­sponds to, and the sub­sys­tem which pro­duces the ver­bal la­bel of that bound­ary, may be dis­tinct. Thus, if the ex­pe­rience of a bound­ary di­vid­ing self and non-self dis­ap­pears, then this might on a ver­bal level be in­ter­preted as ei­ther “all is self” or “noth­ing is self”, de­pend­ing on which frame­work one is us­ing and which fea­tures of the ex­pe­rience one is pay­ing at­ten­tion to.

This is the fourth post of the “a non-mys­ti­cal ex­pla­na­tion of in­sight med­i­ta­tion and the three char­ac­ter­is­tics of ex­is­tence” se­ries. The next post in the se­ries is “On the con­struc­tion of the self”.

Some of the stick figures and mu­si­cal notes in this post were bor­rowed from xkcd.com.