#yale autism study

An examination of Yale’s defensive statement, released after public outcry over their study on autistic toddlers.

On December 6th, 2020, a study conducted by three researchers from Yale University’s Child Study Center was published in the Official Journal of the International Society for Autism Research. It’s called “Attend Less, Fear More: Elevated Distress to Social Threat in Toddlers With Autism Spectrum Disorder.” (https://europepmc.org/article/med/33283976) It is a peer-reviewed study, and it was approved by Yale’s Institutional Review Board.

The goal of the study was to measure autistic toddlers’ fear responses, in order to compare their emotional reactivity to that of neurotypical toddlers. The idea was that this could give some insight into how anxiety and depression develop in autistic people later in life. That’s a decent goal, but those insights could have been sought in much better ways, and without conflating autism with mood disorders (suggesting that autism itself requires treatment, as the authors of this study did).

As it was, the authors ended up conducting a study in which toddlers (42 autistic and 22 neurotypical) endured 10 trials of frightening stimuli to measure their fear responses. The methodology was based on the Lab-TAB - Locomotor Version, which is a standardized method of measuring the general temperament of young children.

In other studies (1,2) that use some variation of Lab-TABtechniques, the stimuli used to induce fear include a mechanical toy dog, and a toy robot. In those studies, none of the fear-inducing stimuli were introduced for more than three trials, and the durations of those trials were significantly shorter (15 seconds and ~10-30 seconds respectively, compared to 60 seconds in the Yale study). In the Yale study, not only did the toddlers endure 10 trials each, but the stimuli used were much more frightening.

From Yale’s study:

“TheStranger probe involved a female stranger wearing dark clothing, a hat, and sunglasses entering the room, approaching the child, and leaning toward the child for approximately 3 s (one trial). The Objects condition included Spider (large mechanical spider crawling toward the child, three trials) and Dinosaur (mechanical dinosaur with red light-up eyes approaching the child, three trials).Masks involved a female stranger dressed in dark clothes and wearing three grotesque masks in succession (e.g. vampire, Star Wars character) entering the room briefly and maintaining an approximate 1.5-m distance from the child (three trials).”

“Each probe lasted approximately 60 s with the effective exposure to threat time of approximately 30 s. Breaks were instituted between each probe, with a minimum of 30 s and an average of 75 s (SD = 36 s) needed to ensure that the child’s affect returned to neutral before proceeding to the next probe.”

So, ten trials. Assuming that “trial” means “repetition of the probe,” each one lasted around 60 seconds. Since 60 seconds is one minute, that’s ten minutes of exposure to threat; five minutes if we’re being conservative and going with their estimation of 30 seconds of effective exposure. There was an average of 75 seconds between each trial, however. So that’s 9x75, which adds 675 more seconds to the total time of the experiment. 675 seconds is 11.25 minutes. Add that to the previously calculated ten minutes, and that brings the total time of the experiment to around twenty-one minutes.

All of this is in contrast to what the authors wrote in their defensive statement (https://medicine.yale.edu/news-article/29344/):

Let’s pick apart each aspect of this paragraph.

1. “The events used to elicit emotional responses were very brief [and] had low intensity.”

Based on my calculations, the combined amount of time that the kids were exposed to the events was 5 to 10 minutes. That’s not brief compared to other studies measuring similar things. And surely a large mechanical spider, a dinosaur with red eyes, and a vampire are much more intense stimuli than a mechanical dog?

2. “[The events] were interspersed with playtime, and mirrored what the children might encounter in the real world. For example, a Halloween costume or a new mechanical toy.”

There is absolutely NO mention of playtime anywhere in this paper. Nothing. Not a word about it. The only thing that could be potentially be seen as playtime is the 30 to 75 seconds between trials. But really? There’s not even a mention of the kids being given toys between trials. All the study says is that they waited for the toddlers’ demeanor to become neutral again.

As for the “Halloween costume” and “mechanical toy” euphemisms here: the study literally says “grotesque masks.” Grotesque. And “toy” sure is an interesting way to say “large mechanical spider crawling toward the child.” We don’t have photographs of the masks or toys used in this study, but from the way they were described in the paper itself, the words “Halloween” and “toy” put a much too positive spin on things.

3. “The entire task reported on in the paper lasted approximately two minutes with several additional minutes for breaks and transitions.”

This is the part that baffles me. Anyone can look at the paper and see where they wrote that each probe lasted for 60 seconds, and that there were 10 trials. Even if there had only been one 60-second trial of each probe, that still would have been 4 minutes (for the 4 probes), not 2 minutes.

What they might be doing here is only counting “effective exposure to threat” (30 seconds), and then multiplying that by 4 for each probe. That would be 2 minutes. But if that’s true, they’ve still failed to explicitly state how long each trial of each probe was. Because there were 10 trials, not 4. And why would a trial of the Stranger probe last 60 seconds, while a trial of the other probes would only last 20 seconds? (60 seconds divided by 3 trials). The math works out, yes. But if that’s the case, this is an issue that should have been more clearly addressed in the paper itself. Clarification on what’s meant by the terms “probe” and “trial,” in addition to information on the duration of each trial, should have been established. And, “effective exposure to threat time for each probe” (a subjective measure to begin with) is not even close to what’s implied by the phrase “the entire task.”

Here’s the last bit I want to touch on:

There is no mention of physiological responses in this study. According to what they wrote in the paper, the authors observed external behavior, not internal bodily changes. The toddlers were not hooked up to any sort of technology that would have measured their heart rate, breathing, etc.

And perhaps the mildly distressed children had an easy time calming down. But what about the trials that had to be terminated and excluded from the results due to “the child’s negative affect” or “parental noncompliance (i.e. parent interfering with probe administration)”? If they literally had to end trials because the kids were so upset, or the parents intervened to comfort their children, then how is it possible to say that “none” of the children experienced extreme negative emotions?

Yale’s statement is full of holes, and creates more questions than it answers. The autistic community is calling for full transparency on the methods used, an explanation of the reasons behind those choices, and detailed answers to our questions about the ethical legitimacy of what happened. Inquiries about the study should be directed to the authors at these two email addresses: [email protected], and [email protected].

Thank you for reading.

~Eden