Can the prosecution call a cop as an expert witness to testify in a drug-trafficking case that most drug couriers — not necessarily the one on trial — know they are carrying drugs? The Supreme Court has said yes, despite a clear federal rule that says an expert can’t testify as to the defendant’s state of mind. The decision is wrong, because it invites the jury to conflate abstract statistical probabilities with the specific circumstances of the individual case.

Delilah Diaz, a U.S. citizen, was stopped in the car she was driving across the border into the U.S. from Mexico. The customs officer tried to roll down the rear window and discovered, hidden inside the rear door panel, 54 pounds of methamphetamine worth roughly $370,000. These were the facts of the case, and they were undisputed.

What was disputed in Diaz v. United States was whether Diaz knew the meth was there. (She claimed she didn’t.) To show Diaz knowingly smuggled the drugs, the government introduced an expert witness, a Homeland Security investigation agent. The agent testified that Mexican cartels “generally do not entrust large quantities of drugs to people who are unaware they are transporting them” because they don’t want to take that kind of a risk with lots of money on the line.

Federal rule of evidence 704(b) says that “in a criminal case, an expert witness must not state an opinion about whether the defendant did or did not have a mental state or condition that constitutes an element of the crime charged or of a defense.” The lower federal courts held that the agent’s testimony didn’t violate the rule because he wasn’t testifying about whether Diaz herself knew the drugs were there, but only about the general practice of cartels. The Supreme Court, in a 6-3 decision written by Justice Clarence Thomas, agreed.

The rule itself has a quirky history: Congress enacted it after a jury acquitted John Hinckley of the attempted murder of Ronald Reagan on grounds of insanity. Apparently, the worry was that juries would take experts too seriously when they said a defendant lacked the requisite mental capacity to commit a crime.

Interestingly, Justice Ketanji Brown Jackson, a former federal trial judge, joined Thomas’s opinion. She wrote a separate concurrence to explain that it is important for defendants to be able to introduce expert testimony showing that a person with, say, a given mental condition is generally unable to form the mental state needed to commit a given crime. This approach suggests sensitivity to the anti-defendant impulses of the post-Hinckley rule.

Justice Neil Gorsuch dissented, joined by Justices Elena Kagan and Sonia Sotomayor. He framed his dissent as an exercise in textualism, the theory of statutory interpretation that says the words must be read to say what they mean. According to Gorsuch’s reading of the rule, the agent’s testimony was precisely “about” the defendant’s state of mind.

Readers of this column know I don’t have much patience with textualism. The majority opinion also had a plausible textual reading of the statute, namely that the testimony was about most drug couriers, not the defendant.

The better basis for the dissent is that allowing expert testimony about what most couriers know encourages the jury to infer on a probabilistic, statistical basis that the defendant is guilty — because after all, most couriers are. That sort of statistical generality is not an appropriate basis for proof beyond a reasonable doubt in a criminal case.

The whole point of requiring proof beyond a reasonable doubt is to demand the government establish that this particular defendant has committed the particular acts charged with the particular state of mind alleged.

Gorsuch’s dissent is at its best when he explains why the majority’s opinion is flawed: “Prosecutors can now put an expert on the stand — someone who apparently has the convenient ability to read minds — and let him hold forth on what ‘most’ people like the defendant think when they commit a legally proscribed act. Then, the government need do no more than urge the jury to find that the defendant is like ‘most’ people and convict.” Notably, Gorsuch was willing to discuss the consequences of the majority’s rule — a welcome deviation from a textualism that is supposed to care only about words, not purposes or effects.

It’s not like the majority’s rule is necessary to get convictions. In this case, there were other, specific, ways to try and prove Diaz’s guilt. For example, she claimed to be driving the car of a boyfriend whom she had met only a couple of times and whose phone number she did not have. Juries are well-suited to making determinations of credibility on assertions like these.

In life, it’s often appropriate and even necessary to rely on statistical generalizations, especially under conditions of uncertainty, where we don’t have all the facts. Civil cases fall into this category: juries are often asked to determine what is more likely than not to have happened.

Criminal cases are different. The law carves them out as a special domain of specific knowledge. We need that protection. Otherwise, juries might observe that most people arrested for a crime are guilty of something and conclude that they should convict any defendant before them.

Noah Feldman is a Bloomberg Opinion columnist. A professor of law at Harvard University, he is author, most recently, of “To Be a Jew Today: A New Guide to God, Israel, and the Jewish People."