Verbal Overshadowing and The Art of Rationality

To be­gin, here are some Fun Psy­chol­ogy Facts:

Peo­ple who were asked to de­scribe a face af­ter see­ing it are worse at rec­og­niz­ing the same face later.

Peo­ple who are asked to de­scribe a wine af­ter drink­ing it are worse at rec­og­niz­ing the same wine later.

Peo­ple who are asked to give rea­sons for their prefer­ences among a col­lec­tion of jel­lies are worse at iden­ti­fy­ing their own prefer­ences among those jel­lies.

This effect, known as Ver­bal Over­shad­ow­ing, oc­curs pri­mar­ily when a prin­ci­pally non-ver­bal pro­cess is dis­rupted by a task which in­volves ver­bal­iza­tion. The above gen­er­al­iza­tions (and Ver­bal Over­shad­ow­ing effects more gen­er­ally), do not oc­cur among what we can term “Ver­bal Ex­perts”: in­di­vi­d­u­als who are as good at ver­bal­iz­ing the rele­vant pro­cess as they are at do­ing it im­plic­itly or au­to­mat­i­cally. This seems like it will be very im­por­tant to keep in mind when cul­ti­vat­ing our own Ra­tion­al­ity.

Here’s an over­sim­plified pic­ture of what this means: We’ve got an im­plicit fa­cial recog­ni­tion pro­cess, IFRP, which is pretty good. We’ve also got a gen­er­al­ized ex­plicit ver­bal think­ing pro­cess, GEVTP, which is good for lots of things, but isn’t es­pe­cially good at rec­og­niz­ing faces. Nor­mally, IFRP is in charge of fa­cial recog­ni­tion, but there are some things we can do, like, try­ing to put a face into words, that wakes up GEVTP, which then mus­cles IFRP out of the way, and all of a sud­den, we are a lot worse at rec­og­niz­ing faces.

The good news is that GEVTP can be trained. To take the wine case, peo­ple who put in the time and effort can be­come ver­bal ex­perts about wine. This isn’t to say they au­to­mat­i­cally have bet­ter judg­ments about wine. Rather, it means that their GEVTP is on par with their im­plicit wine recog­ni­tion, be­cause it has been trained to do the same qual­ity job as the the im­plicit pro­cess.

As a crude metaphor, imag­ine the differ­ence be­tween the nat­u­ral pro­cess by which you go about walk­ing, ver­sus hav­ing to keep track of each and ev­ery in­struc­tion that needs to be sent to differ­ent joints and mus­cles if you had to con­sciously is­sue each one.

Now, ob­vi­ously the spe­cific stud­ies men­tioned are im­por­tant for wine tast­ing, eye-wit­ness iden­ti­fi­ca­tion, or de­ter­min­ing one’s own jelly prefer­ences, but the phe­nomenon of Ver­bal Over­shad­ow­ing has a much larger, more sys­tem­atic im­por­tance for th Art of Ra­tion­al­ity.

Let’s bridge to the broader point with a quote from David Hume, a man whose in­sights were of­ten far ahead of their time: “I shall add [...] that, as this op­er­a­tion of the mind, by which we in­fer like effects from like causes, and vice versa, is so es­sen­tial to the sub­sis­tence of all hu­man crea­tures, it is not prob­a­ble, that it could be trusted to the fal­la­cious de­duc­tions of our rea­son, which is slow in its op­er­a­tions; ap­pears not, in any de­gree, dur­ing the first years of in­fancy; and at best is, in ev­ery age and pe­riod of hu­man life, ex­tremely li­able to er­ror and mis­take. It is more con­formable to the or­di­nary wis­dom of na­ture to se­cure so nec­es­sary an act of the mind, by some in­stinct or me­chan­i­cal ten­dency, which may be in­fal­lible in its op­er­a­tions, may dis­cover it­self at the first ap­pear­ance of life and thought, and may be in­de­pen­dent of all the laboured de­duc­tions of the un­der­stand­ing. As na­ture has taught us the use of our limbs, with­out giv­ing us the knowl­edge of the mus­cles and nerves, by which they are ac­tu­ated; so has she im­planted in us an in­stinct, which car­ries for­ward the thought in a cor­re­spon­dent course to that which she has es­tab­lished among ex­ter­nal ob­jects; though we are ig­no­rant of those pow­ers and forces, on which this reg­u­lar course and suc­ces­sion of ob­jects to­tally de­pends.”

In short, Hume is say­ing, in the field of in­fer­ence and rea­son­ing, our Im­plicit Rea­son­ing Pro­cess of­ten out­paces our GEVTP. I’m not sug­gest­ing that our im­plicit rea­son­ing is perfect (it is, af­ter all, fraught with its own bi­ases), but, sup­pos­ing that Ver­bal Over­shad­ow­ing is a gen­eral phe­nomenon, it would ap­pear that, with re­spect to our rea­son­ing and in­fer­ences more gen­er­ally, our situ­a­tion is one in which try­ing to talk about what we are do­ing is li­able to mess us up.

The ob­vi­ous sug­ges­tion, then, is that we be­come ver­bal ex­perts on the sub­ject, so that our think­ing about ra­tio­nal­ity doesn’t mess up our think­ing ra­tio­nally.

“Aha,” I hear you all say, “then your ad­vice is un­nec­es­sary, for what is it that we Ra­tion­al­ists are already do­ing, if not train­ing our­selves to think ex­plic­itly about ra­tio­nal­ity?” And that would be a good re­ply, but for one cru­cial fact: we are not train­ing our­selves cor­rectly to be­come ver­bal ex­perts.

One does not be­come a ver­bal ex­pert about wine by tast­ing only strange vin­tages or the wine of ab­nor­mal grapes. One does not be­come a ver­bal ex­pert about fa­cial recog­ni­tion by prac­tic­ing only on the stun­ningly gor­geous or the hideously de­formed. And like­wise, one does not be­come a ver­bal ex­pert on Ra­tional think­ing by fo­cus­ing on the edge cases (i.e. The Epistemic Pri­soner’s dilem­mas, The Get­tier Cases, The High Stakes sce­nar­ios, etc.). Ver­bal Ex­perts get trained, pri­mar­ily, on the paradigms.

In fact, the stud­ies on In­sight Puz­zles in par­tic­u­lar (i.e. ver­bal over­shad­ow­ing with re­spect to ex­plain­ing the ac­tual pro­cess by which one achieved the solu­tion to a prob­lem), sug­gest that those of us who en­gage in ver­bal­iza­tion tasks re­lat­ing to our rea­son­ing and in­fer­ences (say, those of us ded­i­cat­ing a lot of time and en­ergy to writ­ing posts or com­ments about it), had bet­ter figure out how to train our Gen­er­al­ized Ex­plicit Ver­bal Think­ing Pro­cess not to drop the ball when it comes to think­ing about rea­son­ing.

I am not a psy­chol­o­gist, but I do know that our cur­rent plan (of, for ex­am­ple, think­ing about the brain­teaser cases), is definitely not the way to de­velop ac­tual ex­per­tise.