Bogus Pipeline, Bona Fide Pipeline

Re­lated to: Never Leave Your Room

Per­haps you are a psy­chol­o­gist, and you wish to do a study on racism. Maybe you want to know whether racists drink more coffee than non-racists. Sounds easy. Find a group of peo­ple and ask them how racist they are, then ask them how much coffee they drink.

Prob­lem: ev­ery­one in your study says they’re com­pletely non-racist and some of their best friends are black and all races are equally part of this vast mul­ti­col­ored tapestry we call hu­man­ity. Maybe some of them are stretch­ing the truth here a bit. Un­til you figure out which ones, you’re never go­ing to find out any­thing in­ter­est­ing about coffee.

So you build a fore­bod­ing look­ing ma­chine out of gleam­ing steel, cov­ered with wires and blink­ing lights. You sit your sub­jects down in front of the ma­chine, con­nect them to its elec­trodes, and say as con­vinc­ingly as pos­si­ble that it is a lie de­tec­tor and they must speak the truth. Your sub­jects look doubt­ful. Didn’t they hear on TV that lie de­tec­tors don’t re­ally work? They’ll stick to their ve­he­ment as­ser­tions of tol­er­ance un­til you get a more im­pres­sive-look­ing ma­chine, thank you.

You get smarter. Be­fore your ex­per­i­ment, you make the sub­jects fill in a sur­vey, which you se­cretly copy while they’re not look­ing. Then you bring them in front of the gleam­ing metal lie de­tec­tor, and dare them to try to thwart it. Every time they give an an­swer differ­ent from the one on the sur­vey, you frown and tell them that the ma­chine has de­tected their fabri­ca­tion. When the sub­ject is suit­ably im­pressed, you start ask­ing them about racism.

The sub­jects start grudg­ingly ad­mit­ting they have some racist at­ti­tudes. You have in­vented the Bo­gus Pipeline.

The Bo­gus Pipeline is quite pow­er­ful. Since its in­ven­tion in the 70s, sev­eral differ­ent stud­ies demon­strate that its vic­tims will give sig­nifi­cantly less self-en­hanc­ing an­swers to a wide va­ri­ety of ques­tions than will sub­jects not con­nected to the ma­chin­ery. In cases where facts can be checked, Pipeline sub­jects’ an­swers tend to be more fac­tu­ally cor­rect than nor­mal sub­jects’.

In one of the more in­ter­est­ing Bo­gus Pipeline ex­per­i­ments, Millham and Kel­logg wanted to know how much of a per­son’s av­er­age self-en­hance­ment is due to self-de­cep­tion bi­ases, and how much is due to sim­ple ly­ing. They asked peo­ple some ques­tions about them­selves un­der nor­mal and Pipeline con­di­tions, us­ing the Mar­lowe-Crowne scale. This scale re­ally de­serves a post of its own, but the short ver­sion is that it asks you some loaded ques­tions, and if you take them as an op­por­tu­nity to say nice things about your­self, you get marked down as a self-en­hancer. There was a cor­re­la­tion of .68 be­tween Mar­lowe-Crowne scores in nor­mal and Pipeline con­di­tions. If we ac­cept that no one de­liber­ately lies un­der the Pipeline, that means we now know how much self-en­hance­ment is, on av­er­age, self-de­cep­tion rather than de­liber­ate false­hood (ten­dency to­wards de­liber­ate false­hoods cor­re­lated .37 with Mar­lowe-Crowne.1)

In­ter­est­ing stuff. But you still don’t know whether racists drink more coffee! Your Bo­gus Pipeline only elimi­nates part of the self-en­hance­ment in your sub­jects’ an­swers. If you want to solve the coffee ques­tion once and for all, you can’t count on a fake mind-read­ing de­vice. You need a real mind-read­ing de­vice. And in the mid 90s, psy­chol­ogy fi­nally de­vel­oped one.

The Bona Fide Pipeline is far less im­pres­sive-look­ing than the Bo­gus Pipeline. Though the Bo­gus Pipeline tries as hard as it can to scream “mind-read­ing de­vice”, the Bona Fide Pipeline has a vested in­ter­est in pre­vent­ing its vic­tims from re­al­iz­ing their minds are be­ing read. It is a sim­ple com­puter ter­mi­nal.

The Pipeline uses a com­pli­cated pro­cess to dis­guise it­self as an or­di­nary study on dis­trac­tion or face recog­ni­tion or some­such, but the ac­tive in­gre­di­ent is this: the sub­jects play a game where they must hit one key (per­haps “A”) if the screen dis­plays a good word (for ex­am­ple “won­der­ful”), and a differ­ent key (per­haps “L”) if the screen dis­plays a bad word (for ex­am­ple “ugly”).

But be­fore it gives you the word, it shows you a pic­ture of a white per­son or a black per­son. Re­mem­ber prim­ing? That pic­ture of a black per­son is go­ing to prime your brain’s con­cept of “black per­son” and any con­cepts you as­so­ci­ate with “black per­son”. If you have racist at­ti­tudes, “bad” is one con­cept you as­so­ci­ate with “black per­son”. You’re go­ing to have a very easy time rec­og­niz­ing “ugly” as a bad word, be­cause your “bad” con­cept is already ac­ti­vated. But you’re go­ing to have a harder time rec­og­niz­ing “won­der­ful” as a good con­cept, be­cause your brain is already skewed in the op­po­site di­rec­tion. It’s not im­pos­si­ble, it’s just go­ing to take a few hun­dred more mil­lisec­onds. Each of which the Bona Fide Pipeline is record­ing and pro­cess­ing. At the end, it spits out a score tel­ling you that you took an av­er­age of three hun­dred mil­lisec­onds longer to rec­og­nize good words when primed with black peo­ple’s pic­tures than white peo­ple’s pic­tures.

Does this ac­tu­ally work? The origi­nal study (Fazio et al, 1995) tested both whites and blacks, and found the whites were more likely to be prej­u­diced against blacks than the blacks were, which makes sense. In the same study, a black ex­per­i­menter con­versed with the sub­jects for a while, and rated the qual­ity of the in­ter­ac­tion by a typ­i­cally rigor­ous rubric. This fuzzy un­scien­tific mea­sure of racist be­hav­ior cor­re­lated well with the Pipeline’s data for the in­di­vi­d­u­als in­volved. A study by Jack­son (1997) find that peo­ple who score high on prej­u­dice by Pipeline mea­sures on av­er­age give lower scores to an es­say writ­ten by a stu­dent known to be black.

The Bona Fide Pipeline has lately been su­per­seded by its younger, sex­ier, Har­vard-ed­u­cated cousin, the IAT. More on that, the as­so­ci­ated con­tro­versy, and the rele­vance to ra­tio­nal­ity to­mor­row.


1: I doubt that de­cep­tions can be sep­a­rated cleanly into self-de­cep­tion and de­liber­ate false­hood like this. More likely there are many differ­ent shades of grey, and the Bo­gus Pipeline cap­tures some but not all of them.