In 1942, the United States Army had a problem. Millions of young men had just been drafted, and nobody was sure they actually understood what they were fighting for. So the Army did what any institution with a budget and a sense of drama would do: they hired Frank Capra, the most celebrated director in Hollywood, to make a series of propaganda films. The result was Why We Fight — seven installments of sweeping orchestral scores, stirring narration, and carefully edited footage designed to convince soldiers that this war mattered. These were not amateur productions. They were cinema.
And they didn’t work.
Not in the way the Army hoped, anyway.
🧠 The Experiment the Army Didn’t Want
The psychologist tasked with measuring the films’ impact was Carl Hovland, a young Yale researcher recruited by the Army’s Research Branch. Hovland’s team did something unprecedented: they gave soldiers questionnaires before and after watching the films, measuring both factual knowledge and attitude change. The results were published in 1949 as Experiments on Mass Communication, and they contained a finding that should still make every advertiser sweat.
The films were excellent at teaching facts. Soldiers who watched Why We Fight knew significantly more about the war’s background, the Pearl Harbor attack, and the reasons the US entered the conflict. But when it came to the thing the Army actually cared about — motivation to fight, willingness to serve, belief in the cause — the numbers barely budged. Hovland measured attitude change on the order of single-digit percentage points. Sometimes zero.
The most expertly crafted propaganda in history had turned soldiers into better-informed soldiers. It had not turned them into more motivated soldiers.
🤔 Information Is Not Persuasion
This seems obvious in hindsight, but at the time it was a bombshell. The intuitive model of persuasion — the one most people still operate on — is that if you give someone enough good information, they’ll come around. Show them the facts. Make a logical case. They’ll agree. Hovland’s data said: no, they won’t.
After the war, Hovland returned to Yale and launched what became known as the Yale Communication and Attitude Change Program. Over the next decade, he and his colleagues ran a systematic research program that identified the key variables that actually drive persuasion. Not the message content alone, but the interaction between four factors: the source (who’s talking), the message (how it’s framed), the channel (what medium carries it), and the receiver (who’s listening). The SMCR model, he called it. It sounds obvious now only because it became the foundation of every advertising strategy course.
His specific findings were sharper still: high-credibility sources produce immediate attitude change that decays over time. Two-sided arguments work better on educated audiences. Moderate fear is more persuasive than extreme fear. State the conclusion explicitly — unless your audience distrusts you. These weren’t abstractions. They were experimental results, tested and replicated.
🔗 The Source You Trust vs. The Source You Forget
Here’s the practical truth for anyone who builds products, writes copy, or designs experiences: who says something determines how much it matters. The same feature description, the same piece of advice, the same recommendation — when it comes from a trusted source, it shifts behavior. When it comes from an unknown source, it gets ignored.
Every AI companion, every recommendatory interface, every personalized message relies on the same mechanism Hovland mapped out in the 1950s. The advice doesn’t work because it’s accurate. It works because the user has come to see the source as credible — and source credibility, once established, creates a persuasive multiplier on every subsequent interaction. Hovland would have recognized an LLM’s conversational framing immediately. It’s source credibility on a feedback loop.
And there’s a darker side, which Hovland also discovered. In his studies of low-credibility sources, he found something strange: over time, the persuasive effect of a low-credibility message increased. People forgot where they heard it, but remembered the content itself. He called it the sleeper effect. It means that a lie from an untrustworthy source, given enough time, can become as persuasive as a truth from a credible one — because the source tag fades faster than the message. This is the psychology behind why debunked rumors don’t die. The debunking is the source tag, and it decays. The rumor is the content, and it persists.
🎲 The Party Trick
On your next night out, try this: ask someone to name the most persuasive piece of advertising they’ve ever seen. They’ll probably describe an emotional image or a compelling story. Then ask them what product it was for. Half the time, they won’t remember. Because the ad that changes your attitude doesn’t need you to remember anything about it. The sleeper effect means you can forget the pitch entirely and still feel its pull.
Hovland’s great discovery, the one that made him uncomfortable, was this: persuasion doesn’t happen in the rational part of your brain. It happens in the part that decides whether to trust the person talking. Everything else — the facts, the logic, the arguments — is just the vehicle for that trust.