Here is the objection this post is going to draw, so I deal with it directly instead of dodging it: raising questions about the reliability of memory sounds like it’s providing cover for abusive parents. It sounds like another way of telling a child, or an adult survivor, that what they remember probably isn’t real. That reaction is fair, and any framework that ends up functioning that way in practice is a bad framework.
But that is not what the underlying research supports, and it is not the argument this post is making. The argument is narrower and, I’d suggest, more useful to everyone involved, including children who have genuinely been harmed: courts should test the process that produced an account, rather than either accepting a label at face value or dismissing it at face value. Skepticism about unexamined process is not skepticism about the child. Getting that distinction right is the difference between protecting children and failing them twice.
The Betty K. Study—and What Happened When Someone Tried to Replicate It
A frequently cited 1978 study by psychologists Mark Snyder and Seymour Uranowitz had 212 college students read a detailed biography of a fictional woman, Betty K.—her childhood, relationships, education, social life. A week later, some participants were told Betty was now living a lesbian lifestyle; others were told heterosexual; a control group got no label. When memory for the original biography was tested, participants’ answers skewed toward whatever label they’d been given afterward. Snyder and Uranowitz described it as systematic retrospective distortion: a belief formed after the fact shaping the apparent recollection of information received before it.
That is the version of this study you’ll see cited everywhere, usually as if it settled the question. It didn’t. In 1981, Leslie Clark and Stanley Woll attempted both an extension and a direct replication of the design, across two experiments with 90 participants each. They found no effect of the stereotype label—including in the experiment that replicated the original procedure most closely. Clark, L., & Woll, S. B. (1981). Stereotype Biases: A Reconstruction of the Determinants of the Effect. Journal of Personality and Social Psychology, 41(6), 1064–1072.
That failed replication matters more than the original finding does, for purposes of this post. It doesn’t mean labels never color memory—the broader literature on reconstructive memory (Bartlett; Loftus) is well established, and Snyder and Uranowitz’s own finding may hold under some conditions. What it means is that the specific, vivid, one-study version of “labels rewrite memory” that circulates in pop-psychology form is not settled science. It’s a contested, partially disconfirmed finding being used with far more confidence than it’s earned. Treat it as a caution rather than a rule, in either direction—including the direction of assuming a label is always correct because it feels sincerely held, and the direction of assuming a label is always suspect because a study once suggested people can be primed.
The Label Is Not the Evidence
In child custody disputes, labels routinely arrive before the underlying facts have been tested. A parent gets called narcissistic, emotionally abusive, toxic, controlling, unsafe, alienating, traumatizing. Sometimes the label is accurate. Sometimes it’s exaggerated. Sometimes it’s wrong. Once adopted, though, a label becomes an organizing principle—ordinary discipline reinterpreted as domination, an unpleasant disagreement reinterpreted as abuse, a parent’s awkwardness reinterpreted as manipulation. The same distortion runs the other direction: a child’s genuine fear can get relabeled as “coaching” the moment it’s inconvenient for a court’s caseload or a parent’s preferred outcome.
Utah law (at least as written) is not agnostic about how this gets resolved. Under Utah Code § 81-9-204, the court determines whether a custody or parent-time arrangement is in the child’s best interest by a preponderance of the evidence—not by which side’s vocabulary is more current. Subsection (3)(a) requires the court to consider evidence of domestic violence or abuse in accordance with § 81-9-104, which means abuse has to be established as evidence, not simply asserted as a characterization. Where a child’s preference is part of the picture, § 81-9-206 governs how much weight that preference gets, and ties it to the child’s maturity—not to which adult introduced the vocabulary the child is now using. And § 81-9-202‘s advisory guidelines favor arrangements that preserve the child’s relationship with both parents absent a reason not to, which means a label that functions to sever that relationship carries a real cost the court has to weigh, not a free pass because it sounds clinical.
Sincerity Is Not Accuracy
The hardest reality in family litigation is that a witness can be completely sincere and still be mistaken. A child can honestly report what the child now believes happened. A parent can honestly believe a child’s account proves abuse. A therapist can honestly believe a child’s statements fit a clinical theory the therapist was trained to recognize. None of that answers the evidentiary question.
The questions that do: What exactly happened? When was it first reported, and to whom? How did the description change over time, and in what direction? What questions were asked, and by whom? What terminology did the child use originally—and did the child supply the label, or did an adult supply it first? Is there anything outside the account itself—texts, school records, medical records, photographs, recordings, a neutral witness—that corroborates or contradicts it? Can the child distinguish what was personally observed from what was inferred from what someone else later explained?
These questions don’t attack the child. They’re what stands between a child’s account and a court basing a permanent custody decision on an account nobody has actually tested.
That’s the same gap I’ve written about elsewhere in this blog—the unrecorded custody evaluator interview, the PGAL summary nobody can cross-examine, the Utah Rules of Evidence Rule 702 gatekeeping that gets skipped because the witness has a license. This is the same structural problem wearing different clothes: information reaches the court pre-filtered through someone else’s interpretation, and the interpretation gets treated as if it were the underlying fact.
Therapy Should Treat Harm, Not Build a Litigation Narrative
Therapy helps children dealing with genuine abuse, grief, conflict, anxiety, family breakdown. It is not a forensic investigation, and a therapist is not automatically positioned to determine historical truth. A therapist properly helps a child name feelings and build coping skills. The problem is when therapy becomes a search for confirming evidence—was your father controlling? did your mother make you feel unsafe? do you think that was emotional abuse? Questions framed that way can teach a child how an adult expects the story to come out. That doesn’t make the resulting account false. It does mean the process that produced it needs to be examined before that account is used to restrict or end a parent-child relationship—which is exactly what §§ 81-9-204 and 81-9-206 require the court to do anyway, if the statute is actually followed rather than deferred past.
Test the Account, Don’t Predetermine the Answer
I could not find a Utah appellate decision squarely addressing forensic interview reliability or the weight of a child’s preference against the process behind it—and that gap is itself worth sitting with rather than papering over. It’s consistent with what this series has been arguing since the first post on custody evaluator accountability: Utah courts have detailed statutory guidance on what to weigh and comparatively little appellate guardrail on how the underlying account has to be tested before it’s weighed at all. That absence doesn’t cut in favor of either side of a custody dispute. It cuts in favor of building the record now, in the trial court, because there’s less appellate law to lean on than the confidence with which labels get thrown around would suggest.
Reconstructive memory research should not become a weapon for dismissing abuse allegations. The seriousness of an abuse allegation should not excuse a court from examining how the allegation developed. Both failure modes are live risks, and neither is solved by picking a side in advance.
Start with the child’s earliest words. Separate the event from the interpretation. Compare the account across time against contemporaneous records. Examine the interviewing process, not just its conclusion. Identify what corroborates and what contradicts. Make experts explain the limits of their own methods before their conclusions get treated as fact. Hold off on the global label until the underlying conduct is actually established.
A child may be telling the truth. A child may be sincerely mistaken. A child may be repeating an interpretation an adult supplied. More than one of those can be true at once. Permanent decisions about custody and parent-time should rest on facts that were tested—not on whichever label made an uncertain past sound like the most finished story.
Utah Family Law, LC | divorceutah.com | 801-466-9277