Contaminated memories and criminal injustice.

In 1968, Elizabeth Loftus discovered what she wanted to do with her life. She wanted to experiment on people.

Loftus was 24. For several years, she had studied psychology  and helped her professors with their research. She had been a cog in the academic system. But now, in her third year of graduate school at Stanford, she was finally getting a chance to design, run, and analyze her own experiments. She knew the thrill of training a rat to press a lever. But this was different. Now she was working with human beings.

Her topic was semantic memory. She was trying to find out how people's brains stored and retrieved words. She couldn't see inside their heads, but she could administer inputs and measure outputs, as she had done with her rat. The inputs were questions, and the output was response time. Sometimes she asked her subjects to name a "yellow fruit." Sometimes she asked them to name a "fruit that is yellow." On average, they answered the latter question a quarter of a second faster than the former. From this, she drew an inference: The brain organized such information by the noun, not the adjective.

Loftus loved the whole thing: conceiving the experiment, trying it out on people, measuring their performance, drawing conclusions about the mind. But not everyone was impressed. Shortly after earning her Ph.D., she had lunch in New York with her cousin, a lawyer. When her cousin asked about her work, Loftus proudly told her about the yellow-fruit study. Her cousin dryly asked how much it had cost the government.

The conversation stung Loftus. She was running her own memory experiments, but they were just about words. Why did it matter how people recalled yellow fruit? Wasn't there something more worthwhile to study?

What did she really care about? As an experimental psychologist, she decided to answer the question by studying her own behavior. What did she talk about when the topic was hers to choose? What did she like to bring up at parties? The answer was crime. She loved books, movies, TV shows, and news stories about it. Maybe she could become a crime expert. She could use the science of memory to help the justice system.

The first step was to find a project somebody would pay for. The Department of Transportation was offering money to study car accidents. Accidents weren't crimes, but they involved eyewitnesses, so she started there. She showed people films of collisions and quizzed them about what they had seen. Sometimes she asked how fast the cars had been going when they "hit" or "contacted" each other. Sometimes she asked how fast they had been going when they "smashed" into each other. The "smashed" question produced estimates 7 miles per hour faster  than the "hit" question and 9 miles per hour faster than the "contacted" question. The questions were skewing the answers.

In another experiment, she showed her subjects a multi-car collision and asked some of them, "Did you see a broken headlight?" She asked others a leading version of the question: "Did you see the broken headlight?" Of the six questions posed in this dual format, three referred to things that weren't in the film. Compared with subjects who heard the question with an "a," those who heard it with a "the" were twice as likely to say they had seen a bogus item. (To experience one of Loftus' traffic experiments, click on the adjacent slide show, "Follow That Car.")

Still, that was just lab work. Loftus wanted to get involved in a real court case. In 1973, after moving to the University of Washington, she called up the Seattle public defender's office and volunteered to help as a memory expert. In exchange, she got to watch the case unfold. It was a murder trial that hinged on conflicting memories over how much time had elapsed for premeditation. It ended in acquittal.

Loftus packaged her memory expertise with her accident studies in an article for Psychology Today. She challenged the reliability of eyewitness testimony, mentioned her work in the Seattle murder case, and noted that the defendant had been acquitted. It was practically an advertisement. Attorneys read the article and picked up their phones. Her career in legal consulting was launched.

She was exactly what defense lawyers needed. The chief threat to their clients was incriminating witness testimony. Loftus could shake the jury's faith in such recollections without attacking the witness personally. Memory errors were natural. The witness, like the defendant, was innocent. Even police, who caused misidentifications by contaminating witnesses' memories with mug shots and lineups, often didn't realize what they were doing.

Over the next 35 years, Loftus testified as a memory expert in more than 250 hearings and trials. She worked dozens of famous cases: Ted Bundy, O.J. Simpson, Rodney King, Oliver North, Martha Stewart, Lewis Libby, Michael Jackson, the Menendez brothers, the Oklahoma City bombing, and many more.

Why did a woman who had endured assault in her own life defend accused predators? Part of it was the structure of criminal law: Her work created doubt, and doubt was an ally of the defense. Part of it was her empathy for the accused: She had always been suspicious of criminal allegations and lenient toward small-time offenders. And where empathy failed, scientific rigor took over. Memory's fallibility was a fact. By testifying to that fact, she believed she was serving justice.