05/12/2014 04:41 pm ET | Updated Jul 12, 2014

Being Second-Guessed by a Machine: Leveraging the Lie Detector

Matt Donahue is a senior at Wesleyan University double majoring in Neuroscience and Psychology who will soon begin a career in market research upon graduation at Kantar Retail.

We deceive ourselves and others on a daily basis. It's just how we cope. We lie about the little things in life, and given the right situation or opportunity, we lie about important things too. It's just too easy. And we can't help it anyway; we're addicted.

It is disillusioning, however, when we are lied to by someone we trust, especially the scientist. Life may be full of lies, but we never think that the laboratory will be. Why? Because: it's sterile, controlled and contained. It's a place of academic ardor and esteem -- a place of respect. Ah, but the experimenter has a social self as well; he's a person, and it just so happens he is one that likes to tell lies. Fortunately for us though, in most cases they are justifiable. Deception serves a purpose in the psychological laboratory. Synonymous to a surgeon's scalpel, the social psychologist uses deception to excise truth out of subjects.

The mid-20th century was a time of anxiety in the psychological laboratory. Experimenters found subjects cooperative, obedient and conditioned, yet also inattentive, unengaged and deviant; in short, experimenters were confused. They found subjects and findings contradictory. They worried that subjects were "faking it." The validity of findings in psychological experiments were questioned, and far worse, doubted (Sigall, Aronson, & Van Hoose, 1970). Furthermore, with the introduction of "Unconscious Experimenter Bias" by Rosenthal, experimenters not only doubted subjects and findings, but also themselves (Rosenthal, 1956). They identified the problem: the social design of the experimenter-subject relation. They named it many things, usually adjectives followed by the words bias or effect, but removing the problem proved much more difficult. There was no simple fix. Experimenters identified problem subjects, methodological missteps and statistical errors. Yet, discarding deviant data, disregarding problem subjects and re-designing laboratory procedure alone didn't seem to be enough.

Enter, center stage: Jones and Sigall. They had an idea, a new methodology -- a contraption that would solve the problem: the bogus pipeline. It was the social psychologists' dream come true: a machine that could prove subjects were acting deceptive in the laboratory. It was a modified lie detection device that did not detect lies at all. The bogus pipeline comprised a set of procedures that sought to deceive or lead subjects to believe that experimenters have developed a "new," "powerful" and "infallible" lie detecting machine -- a machine that could read into a subject's inner thoughts. This belief in the machine was then used to leverage subjects to be more truthful regarding issues such as attraction, anxiety or race. With the rise of technology, following the advent of the computer, experimenters didn't believe subjects would be given much reason to doubt the power of their electromyographic (EMG) device. As Jones and Sigall assert: "The paradigm is based on the simple premise that no one wants to be second-guessed by a machine" (1971). The goal being to convince the subject that it is plausible such scientific advances in lie detection have been made.

The BPL wasn't an easy sell to experimenters though. Standard paper-pencil questionnaires that promised anonymity were in wide use; they were the status quo: inexpensive, easy and effective. The bogus pipeline was elaborate, complicated and effortful; it required deception, a machine, a Confederate and effective experimenter delivery. It was developed in the wake of Milgram and his torture device. Deception had a bad rapport. The BPL was novel and unsubstantiated. Yet, in the 1970s, when S-E anxieties were peaking, it seemed like a possible solution to social desirability effects. The BPL had some completive edges too. EMG technology was blossoming in the 1970s and lie detection devices were common tools of criminologists. Subjects came to the laboratory with rudimentary understanding of lie detection and the science behind the electromyograph; as Jones and Sigall assure us, the "task is greatly facilitated by public knowledge and stereotypes about the 'lie detector' and its use in criminal investigations" (Jones & Sigall, 1971). The BPL was an easy sell for subjects with simple reinforcement of prior knowledge and confirmation of experimenter statements with a rigged trial.

What ensued after the advent of the bogus pipeline is best explained as lively dialogue in the psychological literature. Since the methodology was new and its scope of use not understood, it was the target of immense criticism, praise and questioning in the decade following its discovery. Criticisms of the BPL fell into the following categories -- the BPL is flawed in the sense that it: induces subject compliance, is not empirically supported, is methodologically defective and is merely unethical.

Psychologists didn't know whether to buy in to the paradigm or retain the status quo. Ostrum (1973) directly addresses this concern in a piece entitled "The Bogus Pipeline: A New Ignis Fatuus?" published in 1973. Comparing the BPL to "swamp gas generated by rotting organic matter," Ostrom expressed concerns that the methodology was yet another failed attempt at finding the perfect attitude measurement tool. He concluded after comparing two initial studies conducted by Sigall and Page on subject attitudes regarding race with prior research conducted by Karlins, Coffman & Walters (1969) that the pipeline presents more problems than solutions, is not superior than rating scales on the bases of "relative sensitivity, functional comparability, ethical character, and practicality." He was convinced that the bogus pipeline was not and would not be a replacement for the status quo. Ostrom was not alone in his beliefs either; most studies conducted in the decade following the advent of the BPL were more intrigued with the validity, sensitivity and appropriate use of the methodology than using it in lieu of a paper-pencil rating test. Jones and Sigall did not intend for the BPL to be a "superior" methodology, not right of the bat at least; as they stated in a reply to Ostrom's work: "the 'bogus pipeline' is defended, not as an all-purpose substitute for conventional rating scales, but as a procedure useful in certain settings to inhibit social desirability influences or to explore the affective components of attitudes." This didn't change the landscape or scope of conversations surrounding the BPL though; experimenters were fascinated with its potential use and fixated on its superiority. After all, if the BPL methodology is to be used over the conventional scale, it must be better, right?

After dissecting piece after piece of literature, it wasn't hard to tell that there was tension over this methodology; however, the most curious product of the literature in the first decade following the advent of the bogus pipeline was the idea that the use of the machine creates a new type of social desirability. In the seminal piece introducing the BPL paradigm, Jones and Sigall identified five key problems with rating scale measures: the generosity effect (subjects' tendency to be generous on rating scales, especially when evaluating others), evaluation apprehension (subjects' tendency to produce socially desirable responses, as a "mature and rational person would"), experimenter demand (subjects' tendency to try to please the experimenter by confirming his hypothesis), thoughtlessness (subjects' tendency to be unengaged by rating scales), and errors of the 'psychologic' (subjects tendencies to correlate dimensions, balance rating scores, etc.) (Jones & Sigall, 1971). They believed that their paradigm would begin to solve these problems, and for the most part other psychologists agreed that the BPL did improve the accuracy of subject responses.

The problem that they couldn't seem to solve, though, was that of response constraint and accuracy constraint. The use of the machine may remove social cues, but the very premise of the methodology is based in subjects' belief in the machine, which as Brigham and his colleagues point out, is very problematic. How can we tell that subjects aren't just telling the machine (and in turn, the experimenter) what they think it wants to hear? Perhaps the BPL is just conditioning subjects to re-conceive themselves as more racist, sexist, ableist or critical. After all, that is what we truly know them to be -- deviant and misguided, right? Brigham argues that the machine creates a new social context, a new paradigm and in fact "provides few observable advantages over more conventional measurement techniques in the average measurement situation, even when socially sensitive attitude issues are involved" (Brigham, Bloom, Gunn, & Torok, 1974).

This brings us back to the essential question at hand: who is deceiving whom? Is the subject deceiving the experimenter? Or perhaps the experimenter is fooling himself that the BPL could even work? Who is being second-guessed by a machine? We know that the experimenter who uses the BPL is lying to their subjects, we just hope that they aren't lying to themselves, and more importantly, us. The decline in the use of the bogus pipeline in social psychology can be accounted to any number of possibilities. Perhaps it was an ineffective laboratory methodology; perhaps the pipeline was truly bogus. It is also possible that experimenters viewed the methodology as impractical, ungrounded or unethical (Roese & Jamieson, 1993). Only further research will tell.

--Matthew Donahue

The HuffPost College Thesis Project gives students a chance to share with a wide audience the fruit of their hard academic work. The project is launching with about a dozen partner schools, which comprise students from public and private, two- and four-year colleges. To read all posts in the series, visit here.