Mike Deigan /
The Cursor /
An Analysis of Wonder
posted 2020-10-19

What is it to won­der? I’m go­ing to de­fend an analy­sis that I think might be right: to won­der is to want to know. This should be of in­ter­est es­pe­cial­ly to epis­te­mol­o­gists, giv­en won­der­ing’s cen­tral role in in­quiry.There seems to be some­thing of an in­quis­i­tive turn go­ing on in epis­te­mol­o­gy, con­tin­u­ing a rich tra­di­tion that goes back to Pla­to and Aris­to­tle. See, e.g., re­cent pa­pers by Jane Fried­man ((2013) "Ques­tion-di­rect­ed at­ti­tudes”, (2017) "Why Sus­pend Judg­ing?, (2019) "In­quiry and Be­lief”, (forth­com­ing) "The Epis­temic and the Zetet­ic”), and Christoph Kelp ((2014) "Two for the Knowl­edge Goal of In­quiry" (2019) "In­quiry and the Trans­mis­sion of Knowl­edge”, (2020) "The­o­ry of in­quiry”).

Why An­a­lyze?

Be­fore get­ting into the analy­sis, a quick method­olog­i­cal pre­am­ble. I agree with those who think epis­te­mol­o­gists shouldn’t get too hung up on find­ing analy­ses of the var­i­ous states and re­la­tions they are in­ter­est­ed in: knowl­edge, be­lief, ev­i­dence, etc.E.g., Williamson (2000) Knowl­edge and Its Lim­its and Pas­nau (2017) Af­ter Cer­tain­ty. What we real­ly want are good de­scrip­tive and nor­ma­tive the­o­ries of these things, or what­ev­er things in the ball­park turn out to be most in­ter­est­ing or im­por­tant. And we shouldn’t pre­sume that the good the­o­ries will make sig­nif­i­cant use of re­duc­tive analy­ses. For­tu­nate­ly, most epis­te­mol­o­gy go­ing on these days doesn’t pre­sume this, and doesn’t fo­cus too much on find­ing analy­ses.

But this isn’t to say that if there is some true re­duc­tive analy­sis in­volv­ing im­por­tant epis­temic no­tions that it isn’t worth know­ing. Analy­ses make for nice the­o­ret­i­cal con­nec­tive tis­sue, let­ting you port over in­de­pen­dent­ly mo­ti­vat­ed views about the an­a­lyz­ing no­tions to the an­a­lyzed one and vice ver­sa. They also at least give you the op­tion of claim­ing a kind of par­si­mo­ny: De­fined Thing, you might say, just is Defin­ing Thing, so we no longer need to treat these as sep­a­rate on­to­log­i­cal com­mit­ments. Or if you pre­fer lush to arid land­scapes, you have the op­tion of say­ing that we have giv­en a real de­f­i­n­i­tion, re­veal­ing some­thing’s essence. Analy­ses aren’t the end-all, be-all, nor are they nec­es­sary pre­lim­i­nar­ies to the­o­riz­ing, but they’re good if you can get it’s ’em.

The Analy­sis

Okay, so here it is:

WTK The­o­ry of Won­der
S won­ders Q iff S wants to know Q.

For in­stance, I want to know who will win the elec­tion, so it fol­lows that I won­der who will win the elec­tion. I won­der how many de­grees of warm­ing there will be this cen­tu­ry, so I must want to know how many de­grees of warm­ing there will be this cen­tu­ry. It’s not the case that I want to know ex­act­ly how many blades of grass there are in the near­by park, so that’s not some­thing I won­der. And I don’t won­der whether it’s rain­ing in Tope­ka, so it must be that it’s not the case that I want to know whether it’s rain­ing in Tope­ka.

Re­sults so plau­si­ble as to seem triv­ial. Per­haps this is why no philoso­phers have de­fend­ed this analy­sis.As far as I know. Please tell me if I’m miss­ing some lit­er­a­ture on this. Note also that some lin­guists—Kar­tun­nen (1977), Guer­zoni and Sharvit (2007)—have sug­gest­ed a se­man­tics for won­der along these lines. Ue­ga­ki (2016) ar­gues that on this analy­sis, plus some as­sump­tions about want’s pre­sup­po­si­tions and their pro­jec­tion, one can ex­plain why we can’t say that some­one won­ders that p. This is a cool re­sult, but see Theil­er, Roelof­sen, & Aloni (2019, §A.2) for rea­sons to think this won’t gen­er­al­ize to oth­er clause-em­bed­ding pred­i­cates that don’t take de­clar­a­tives. They also raise a prob­lem to do with pre­sup­po­si­tion pro­jec­tion which I might ad­dress at some lat­er date. But in fact the analy­sis is in­ad­e­quate as it stands. We need to get clear­er on what kind of want­i­ng is in­volved, which will help us avoid some ap­par­ent prob­lems for the analy­sis. We also need to deal with a prob­lem about metacog­ni­tion.


Here’s a class of po­ten­tial coun­terex­am­ples: some­times we find our­selves won­der­ing some­thing when we don’t real­ly want to know. Sup­pose you’re won­der­ing how many years you have left to live. But you also think that if you could some­how know the an­swer to this, it would cast such a pall over the rest of your life that any ben­e­fits to know­ing would be out­weighed. If some­one of­fered you a glimpse into the crys­tal ball, you would turn it down, be­cause you don’t want to know. Nev­er­the­less, you could still be won­der­ing how many years you have left to live. So it looks like we have won­der­ing with­out want­i­ng to know. An­oth­er analy­sis bites the dust, it seems.

Not so fast! Dis­tin­guo: all-things-con­sid­ered want­i­ng and some-things-con­sid­ered want­i­ng.

Sup­pose you in­vite me to play ten­nis on Tues­day at 2:00. I would real­ly en­joy this, need the ex­er­cise, and would be hap­py to have an ex­cuse to catch up, but I’m sched­uled to teach at that time. As Wayne Davis points out,Davis (1984) "The Two Sens­es of De­sire”: p. 239. I can tru­ly de­scribe my men­tal state in a pair of con­trast­ing ways:

  • “I want to play, but I have to teach.”
  • “I don’t want to play, since I have teach.”

The for­mer re­ports that I some-things-con­sid­ered want to play, where­as the lat­ter re­ports that I all-things-con­sid­ered want not to play.

This dis­tinc­tion al­lows us to avoid the coun­terex­am­ple. Won­der­ing, we should say, is want­i­ng to know some-things-con­sid­ered. When you won­der when you will die, you want to know some-things-con­sid­ered, even if all-things-con­sid­ered you’d rather not know. That seems right to me.


Here’s an­oth­er con­cern about the WTK The­o­ry.Fried­man (2013, p. 154) ob­jects along these lines. Thanks also to Zoltán Sz­abó for rais­ing it in con­ver­sa­tion. The verb to won­der seems to ex­press an ac­tiv­i­ty, where­as to want to know ex­press­es a state. They have dif­fer­ent as­pec­tu­al class­es.For a clas­sic di­vi­sion of verbs (or real­ly, verb phras­es) by as­pec­tu­al class, see Zeno Vendler (1967) Lin­guis­tics in Phi­los­o­phy: Ch. 4.

One mark dis­tin­guish­ing these class­es of verbs is that sta­tives don’t take well to the pro­gres­sive, where­as non-sta­tives do. Pic­ture a scene: your friend is sit­ting at their desk, pulling the Thinker pose. Con­trast the fol­low­ing as de­scrip­tions of what’s go­ing on.

  • “They’re think­ing that ex­pres­sivism might be true.”
  • “They’re be­liev­ing that ex­pres­sivism might be true.”

The lat­ter sounds ter­ri­ble! This is be­cause be­lieve is a sta­tive verb, in con­trast with think, which is non-sta­tive. Be­lief is a state you can be in, where­as think­ing is some kind of ac­tiv­i­ty—some­thing that hap­pens.

Now let’s try this test for won­der and want.

  • “They’re won­der­ing whether ex­pres­sivism can make sense of nor­ma­tive in­quiry.”
  • “They’re want­i­ng to know whether ex­pres­sivism can make sense of nor­ma­tive in­quiry.”

Again, the lat­ter sounds ter­ri­ble! It seems that want is a sta­tive verb and won­der is not. De­sire is a state you can be in, where­as won­der­ing is some kind of ac­tiv­i­ty—some­thing that hap­pens.

This looks bad for the WTK The­o­ry of Won­der. A suc­cess­ful analy­sis of won­der will have to iden­ti­fy won­der not with a state like de­sire, but with some­thing non-sta­tive, some kind of ac­tiv­i­ty or hap­pen­ing.

Here’s one al­ter­na­tive pro­pos­al: try­ing to find out. This gets many nor­mal cas­es right. When I’m won­der­ing what I should cook for din­ner, I am try­ing to find out what I should cook for din­ner by think­ing through what in­gre­di­ents I have, how long var­i­ous op­tions would take and how much ef­fort they would in­volve. And it’s not easy to think of cas­es where one is try­ing to find some­thing out but doesn’t won­der the rel­e­vant ques­tion.

That said, I don’t think it gets all the cas­es right. It has trou­ble, first of all, with the mere­ly some-things-con­sid­ered want­i­ng to know cas­es. You can be won­der­ing when you will die while pur­pose­ful­ly not try­ing to find out the an­swer to that. Sec­ond, it has trou­ble with cas­es of won­der­ing what you re­al­ize you’ll nev­er know. I some­times won­der what Mozart would have com­posed had he not died young, but I’ve nev­er tried to find out, since it’s so trans­par­ent­ly hope­less.

I think a bet­ter way to go here pre­serves the WTK the­o­ry, again by mak­ing a dis­tinc­tion be­tween types of want­i­ng. Dis­tin­guish, as Gold­man does,Gold­man at­trib­ut­es the dis­tinc­tion to Al­ston, but no doubt it goes back fur­ther, in some form or oth­er. be­tween oc­cur­rent and non-oc­cur­rent wants. Here’s how he in­tro­duces the dis­tinc­tion:

John is con­cen­trat­ing on fin­ish­ing the lawn by six o’clock. He is giv­ing all his at­ten­tion to mow­ing it as quick­ly as pos­si­ble, with the thought of get­ting it done by six. Dur­ing this pe­ri­od, whether it be a few sec­onds or a whole minute, John has an oc­cur­rent want to fin­ish mow­ing the lawn by six o’clock. The thought of fin­ish­ing the lawn by six oc­curs to hum, oc­cu­pies his at­ten­tion, fills his con­scious­ness. It is a dat­a­ble event or process. Dur­ing the same pe­ri­od it would also be cor­rect to say that John want­ed to be pres­i­dent of his com­pa­ny. But his want­i­ng to be pres­i­dent of the com­pa­ny is not an oc­cur­rent want dur­ing this time, for John has not been think­ing about the com­pa­ny or his fu­ture at the com­pa­ny. Nev­er­the­less, we could say that John wants to be pres­i­dent of his com­pa­ny be­cause he has a dis­po­si­tion to have oc­cur­rent wants to this ef­fect, and this dis­po­si­tion is present while John is mow­ing the lawn, though it is not be­ing man­i­fest­ed. I shall say that John has a stand­ing want to be pres­i­dent of his com­pa­ny.

Alvin Gold­man (1970, p. 86)
A The­o­ry of Hu­man Ac­tion

Pin­ning down what pre­cise­ly this dis­tinc­tion amounts to is no easy task,See Bartlett (2018) "Oc­cur­rent States" for one re­cent at­tempt. but I take it to be clear enough as is for my pur­pos­es. Note that this kind of want is not just a philoso­pher’s pos­tu­late. We do have words for at least some such want­i­ng events. In Eng­lish: long­ings, crav­ings, han­ker­ings, lust­ings, yearn­ings. It would be a bit over the top to say that John was yearn­ing to fin­ish mow­ing the lawn, but it is some­thing that could be said if he want­ed it bad­ly enough (and per­haps met some oth­er con­di­tions re­quired for yearn­ing).

What’s im­por­tant is that there is a kind of want­i­ng that is a “dat­a­ble event or process”, a non-sta­tive kind of thing. We can nat­u­ral­ly use the pro­gres­sive here. John is oc­cur­rent­ly want­i­ng to fin­ish mow­ing the lawn by six, crav­ing choco­late cake, etc.

We can solve the non-sta­tive prob­lem for the WTK The­o­ry, then, by say­ing that the kind of want­i­ng it iden­ti­fies won­der­ing with is oc­cur­rent want­i­ng. To won­der is to oc­cur­rent­ly want to know. Mod­u­lo the ex­ag­ger­a­tion and oth­er con­no­ta­tions, we might say it is to crave or yearn for knowl­edge.

Be­sides get­ting the non-sta­tive-ness right, this also works for the mere­ly-some-things-con­sid­ered and the hope­less won­der­ing cas­es. Crav­ing some­thing but not want­i­ng it all-things-con­sid­ered is very fa­mil­iar, as is yearn­ing for what you know will not hap­pen.

Why knowl­edge?

I’ve spent all this time talk­ing about the kind of want­i­ng in­volved in won­der­ing, but what it is sup­posed to be a want for? Why think it’s knowl­edge? To be hon­est, this is main­ly just a hunch. There are var­i­ous al­ter­na­tives worth con­sid­er­ing.

Maybe all we want is be­lief, as Peirce claims.Peirce (1877) "Fix­a­tion of Be­lief”: §IV. Peirce is talk­ing about ‘doubt’, but it seems to me that he has won­der­ing in mind. Or maybe true be­lief, or jus­ti­fied true be­lief, or sta­ble cre­dence, or ra­tio­nal sta­ble cre­dence, or some oth­er non-knowl­edge-en­tail­ing state. But I don’t think so. Peirce ob­serves that one stops won­der­ing once one has set­tled one’s opin­ion into a be­lief. But one also stops yearn­ing for what one mis­tak­en­ly be­lieves one has at­tained. The de­sire is not sat­is­fied in such cas­es, but mere­ly dis­si­pates. This is what strikes me as go­ing on in cas­es where one stops won­der­ing but doesn’t know.

This is not yet a sub­stan­tive ar­gu­ment against these oth­er op­tions, but I sus­pect one can be giv­en (maybe in an­oth­er post, if I can think of an in­ter­est­ing one).

I’m less con­fi­dent that the ob­ject of a won­der­ing de­sire shouldn’t be tak­en to be some­thing some­thing that en­tails knowl­edge, but in­volves more. Per­haps won­der­ing Q is want­i­ng to learn Q or to find out Q. The usu­al way one goes from not know­ing to know­ing is through learn­ing or find­ing out, so it wouldn’t be sur­pris­ing if most cas­es of want­i­ng to know also in­volve want­i­ng to learn or find out. But it does seem pos­si­ble to come to know with­out learn­ing or find­ing out. Sci-fi ex­am­ples come to mind: one may come to know through some kind of di­rect up­load­ing with­out learn­ing or find­ing out. I will leave try­ing to fig­ure out if one’s won­der­ing is sat­is­fied or mere­ly dis­si­pat­ed in such cas­es to an­oth­er time. Per­haps in the end we should fa­vor the WTF (Want To Find out) over the WTK The­o­ry.

These are still cas­es of com­ing to know through com­ing to have a be­lief. An­oth­er way one might come to know with­out learn­ing is go­ing from hav­ing a be­lief that is not jus­ti­fied to one that is, or go­ing from Get­teried jus­ti­fied true be­lief to un-Get­teried jus­ti­fied true be­lief. It’s some­what un­clear whether one can real­ly want to un­der­go such a change, for Moore­an Para­dox-ish rea­sons, but if one can I sus­pect we wouldn’t want to call it won­der­ing. If so, I think the WTK the­o­ry can han­dle this with ap­peal to some form of con­trastivism. Some­thing like want­i­ng to know Q as op­posed to re­main­ing in sus­pend­ed judge­ment about Q.

For now, then, I’m in­clined to ac­cept some­thing along the lines of

WTK The­o­ry of Won­der (elab­o­rat­ed)
S won­ders Q iff S oc­cur­rent­ly, some-things-con­sid­ered wants to know Q rather than re­main in sus­pend­ed judge­ment about Q.

Metacog­ni­tion With­out Cog­ni­tive Con­cepts

One more po­ten­tial coun­terex­am­ple. If Sam­son the pup­py lacks the con­cept of knowl­edge (plau­si­ble), he can’t have at­ti­tudes to­wards con­tents about knowl­edge. So he can’t have de­sires about knowl­edge. So he can’t, e.g., want to know stuff. But he does seem to won­der. He won­ders what made that sound, which is why he’s go­ing to the oth­er room. But ap­par­ent­ly he can’t want to know what made that sound. So won­der­ing can’t be want­i­ng to know.

More gen­er­al­ly it seems that we’re com­fort­able de­scrib­ing var­i­ous non-hu­man an­i­mals as won­der­ing, but we wouldn’t be so com­fort­able as­crib­ing to them a con­cept of knowl­edge. Such crea­tures lack the ca­pac­i­ty for metacog­ni­tion, and so must lack the ca­pac­i­ty for metacog­ni­tive de­sires. So won­der­ing can’t be a metacog­ni­tive de­sire like want­i­ng to know.Fried­man (2013, §3) makes this ar­gu­ment, as does Car­ruthers (2018) "Ba­sic Ques­tions”. Whit­comb (2010) "Cu­rios­i­ty was Framed" makes the same point about cu­rios­i­ty, build­ing on Kvan­vig (2003) The Val­ue of Knowl­edge and the Pur­suit of Un­der­stand­ing: pp. 143ff.

At­tribut­ing to non-hu­man an­i­mals men­tal states with par­tic­u­lar con­tents is fraught with philo­soph­i­cal dif­fi­cul­ty, so I’m not con­fi­dent what the best view is here. How­ev­er, I’m un­per­suad­ed that the analy­sis is re­fut­ed.

One rea­son I’m un­per­suad­ed stems from how I my­self feel about the cas­es. In most con­texts, I’m to­tal­ly hap­py to say that Sam­son does want to know what made that sound. “Why’s Sam­son try­ing to go there?” “He wants to know what made that sound.” This just seems true to me.

Does that mean I think he pos­sess­es the con­cept knowl­edge? Well, if all that’s meant by that is that he can, e.g., want to know what made that sound, then sure. But if some­thing more ro­bust is meant, then maybe not.

In oth­er con­texts—like ones where philo­soph­i­cal is­sues about an­i­mals’ men­tal states are raised—I ad­mit I am hes­i­tate to say he real­ly wants to know what made that sound. But in those same con­texts I’m sim­i­lar­ly hes­i­tant to say he won­ders what made that sound. Re­gard­less of con­text, an­swer­ing the ques­tion about why Sam­son is go­ing to the oth­er room with “He wants to know what made that sound” seems about equal­ly ob­jec­tion­able as “He won­ders what made that sound”. This is as the WTK The­o­ry would pre­dict.

The oth­er rea­son I’m not per­suad­ed is more the­o­ret­i­cal. I think there are good rea­sons to think that at least some kinds of metacog­ni­tion are less cog­ni­tive­ly de­mand­ing than the skep­tics make it out to be, and that Sam­son and oth­er non-hu­man won­der­ers do rep­re­sent their own knowl­edge states in some of their goal rep­re­sen­ta­tions. But that gets into is­sues about psy­chose­man­tics best left for their own post.

In the mean­time, email me if you’ve got oth­er coun­terex­am­ples to the WTK The­o­ry of Won­der.

Send comments to mike.deigan@rutgers.edu.