Default to Truth: Why We Trust Strangers

In 1984, two communication researchers at Michigan State — Steven McCornack and his mentor Malcolm "Mac" Parks — were running an experiment on how dating couples detect lies in each other. Couples were asked to identify which of their partner's statements were honest and which were fabricated. The data came back with a strong tilt the researchers hadn't expected. People were good at calling true statements true, and bad at calling lies lies. Even when subjects were warned in advance that some statements were false, they leaned toward believing. McCornack and Parks called this the truth bias and wrote it into their model of deception detection.

The number stayed the same when later researchers ran the experiment on hundreds of populations. The 1981 meta-analysis by Miron Zuckerman, Bella DePaulo, and Robert Rosenthal documented the same "truthfulness bias" across studies. Pull together the next four decades of work and human accuracy in detecting deception sits at roughly 54 percent — barely better than the 50 percent you'd get by flipping a coin. It doesn't matter much whether the judge is a college student or a customs officer or a federal agent. Trained interrogators sometimes do worse, because they swing the other direction into a lie bias.

Timothy R. Levine, now Chair of Communication Studies at the University of Oklahoma, formalized the explanation in 2014 with a paper in the Journal of Language and Social Psychology called "Truth-Default Theory." Levine's argument was that the truth bias isn't a cognitive defect. It's the design. Most communication is honest — Bella DePaulo's 1996 diary studies, published in the Journal of Personality and Social Psychology, found that the average person tells one or two lies a day, the vast majority of them small ones to spare feelings or smooth interactions. If your default were suspicion, you would burn enormous mental effort double-checking everything you heard, and you would be wrong almost every time, because almost everything you hear is true. Default-to-truth is faster, and across a lifetime it produces more accurate beliefs than constant skepticism would.

The mechanism Levine proposed is specific. People don't actively decide a statement is true. They simply fail to consider the possibility that it's false. The truth-default holds until something — a contradiction, a reputation, a third party, a piece of physical evidence — kicks the listener into what Levine calls a "trigger event." Only then does suspicion enter, and only then does the listener actually evaluate. Without a trigger, the message gets accepted without inspection.

This explains the veracity effect Levine, Park, and McCornack documented in a 1999 Communication Monographs paper. Run a deception-detection experiment with a 50-50 split of true and false statements and accuracy on truths runs around 65 percent while accuracy on lies runs near 35 percent. Run it with 80 percent truths and accuracy soars on the truths and collapses on the lies. The judges aren't learning anything about the speakers — they're just defaulting toward truth, and that default happens to match reality more often when reality contains more truths.

Truth-default theory got its general-audience hearing in 2019 with Malcolm Gladwell's book "Talking to Strangers," which built three of its longest case studies around Levine's work. Bernie Madoff ran a $65 billion Ponzi scheme for at least sixteen years while financial analyst Harry Markopolos wrote five separate detailed memos to the SEC starting in 2000 — none of which triggered a real investigation, because the people receiving them defaulted to trusting that a NASDAQ chairman wasn't running a fraud. Penn State assistant football coach Jerry Sandusky abused boys for at least fifteen years while parents, head coach Joe Paterno, and university administrators received warning signs they did not, in Levine's terms, treat as triggers. Defense Intelligence Agency analyst Ana Montes spied for Cuban intelligence from 1985 to her arrest in September 2001 — sixteen years inside the agency, polygraphed in 1994 and passed, all while feeding Havana the names of US intelligence assets.

What unites the three cases isn't that the deceivers were unusually good. It's that they presented honest demeanor — what Levine, in his 2011 work, called sender demeanor — and operated inside institutions where the cost of false suspicion was high. Mistrust your boss and you risk your career; mistrust your sister-in-law's investment manager and you insult your sister-in-law. The truth-default holds because suspicion has costs that the listener has to pay first.

Not everyone accepts Levine's account. The Adaptive Lie Detector model, proposed by Chris Street in a 2015 paper in the Journal of Applied Research in Memory and Cognition, argues people don't default to truth — they default to whatever the available evidence supports, which usually points toward truth because individual cues to lying are weak and lies are statistically uncommon. ALIED predicts the opposite of TDT in environments where deception is the norm: prison interviewers and customs agents both develop a documented lie bias, judging more statements as false than true. The 2009 study by Bond and colleagues on prison populations confirmed this directly.

The empirical work keeps pulling in the same direction. Lying takes longer than telling the truth — it consumes more brain activity, more cognitive load, more time to formulate. The truth comes out first. Listeners, downstream, take it without checking. Bella DePaulo's 2003 Psychological Bulletin meta-analysis of 158 studies on cues to deception found that of more than 100 supposed tells — eye contact, fidgeting, vocal pitch — none reliably distinguishes liars from truth-tellers. Pinocchio's nose does not exist.