Food Selectivity in Autism: What Interventions Actually Work

The actual sequence Dr. Holly Gover uses for food selectivity in autism goes like this: run an ISCA (an interview-style cousin of the functional analysis that pins down what reinforces food refusal), set up a small choice board with steps from "look at it" up to "chew and swallow," score each try as red, yellow, or green and pay out at three reinforcement levels, and budget three to four months of short sessions. That is the whole spine.

This is not a "tips for picky eaters" page. This is the page you read when a parent calls and says their seven-year-old has eaten chicken nuggets, French fries, and applesauce for three years and the family can't go anywhere. That is the kid Gover built her process for.

What food selectivity actually looks like#

Food selectivity is not the same as total food refusal. A food-refusing kid has a G-tube or is at risk of one. A food-selective kid eats. They just eat a tiny pinned-down list, and that list is making the family's life small. Gover's own definition in the talk is the cleanest one out there:

The kids who are eating all the chips in the world, chicken nuggets and French fries and maybe applesauce, but aren't eating any vegetables or meat, and they can't take their kids anywhere. From the talk — Dr. Holly Gover

That second half matters. "Can't take their kids anywhere" is the real referral driver. The family packs a cooler for every outing, skips birthday parties, stops going to grandma's. The eating is one problem. The shrinking world around it is the bigger one.

A few markers that say "this is selectivity, not refusal":

• The kid swallows food, chews food, and has been cleared by their pediatrician.
• They eat solid table-textured foods, just a short list of them.
• The list is mostly beige and crunchy or smooth and bland.
• No feeding tube, no choking history, no swallow-study red flags.

If any of that is missing, this is not the right page and probably not the right intervention. More on that in step one.

Step 1: Rule out medical issues first#

Before any shaping happens, the kid gets medically cleared. If there is any chance of an oral-motor problem, swallow issue, or GERD, a BCBA (a board-certified behavior analyst who designs the plan) should not be the lead. An SLP or OT watches the kid eat a snack. A swallow study gets ordered if needed. Only after the medical team says "this is behavioral" does the plan start.

A specific red flag: gagging that does not fade. A little on a new texture in the first few sessions is normal. Still gagging three to five sessions in is not. Gover shared a case where a clinician kept running the procedure with persistent gagging, finally sent the kid for a second swallow study, and the kid did have a swallowing problem. If gagging will not go away, stop and refer.

This step is not paperwork. It is the line that keeps the kid safe.

Step 2: The ISCA, in plain words#

ISCA stands for interview-informed synthesized contingency analysis. It is the assessment piece. In plain words, the team interviews the caregiver about what the kid actually gets when they refuse food and what they actually want instead. The team then runs short test trials to confirm. The output is a list of the reinforcers the kid is working for at mealtime, mashed together.

Gover's point is that with feeding, the ISCA is the easy version:

An ISCA with a feeding issue is one of the easiest you can run because your EO is already identified and your reinforcers are pretty much identified. What you do through talking to the caregiver and everything is add all the nuance. From the talk — Dr. Holly Gover

EO is "establishing operation," the thing that makes the reinforcers matter in that moment. With feeding, it is staring you in the face. The EO is "we are at the table and a non-preferred food is in front of you." The reinforcers are "escape from this bite" plus "the preferred food I actually want" plus whatever the parent says happens next at home: the iPad, the popcorn, leaving the table. The interview adds the nuance. Which exact foods are in play. Which preferred items count. What the family wants mealtimes to look like.

DR, short for differential reinforcement (paying out different amounts for different responses), is the engine that the ISCA feeds. The ISCA tells you what to pay out. The DR system tells you when.

Step 3: Choosing foods the family actually eats#

Picking the target foods is not a clinician choice. It is a family choice. Gover lets parents drive the bus. The team works on the foods the family actually eats, in the textures the family actually makes them, at the times the family actually serves them.

A few rules from her case files:

• If you never make spinach for dinner, you don't work on spinach.
• If the family is South Asian and eats with their hands, the protocol uses hands.
• If the family is in Nicaragua and gallo pinto (a rice-and-beans staple) is in every meal, gallo pinto goes on the list, both for the kid who only eats it and the kid who refuses it.
• Three or four target foods at once, not one. A preference order will show up inside the targets, and the team can drop the food the kid truly hates without guessing whether the protocol broke.

That last one matters. With one target food, a stuck step could mean the kid hates it or the procedure is off. With three or four, the team can see that two are climbing and one is flat. The flat one gets dropped without losing momentum on the others.

Step 4: Shaping with red, yellow, green levels#

This is the part that does the work. The kid sees a small visual choice board. Down the board is a step list, from easiest to hardest: look at the food, touch it, lift it, put it to the lips, put it on the tongue, bite it, chew it once, chew it five times, swallow it. The kid picks what they want to do with the bite. Every pick is scored a color.

Green was the full reinforcement contingency. The yellow was just attention. So just staying at the table with us, but you could chat about whatever you want to chat about. And then the red was no access to any positive reinforcers. From the talk — Dr. Holly Gover

A green response is the hardest step at that level (say, "chew it five times"). It pays out the full reinforcer mix: toys, iPad, a preferred snack, free play with mom. A yellow response is a middle step. It pays out attention only. The clinician stays warm at the table and chats. A red response is the lowest step (often "look at it"). The trial just moves on. No fun, no toys, but also no force, no spoon held to the lips, no blocking the kid from leaving.

There is one more piece: the opt-out space. The room has a beanbag, books, blocks. The kid can leave the table any time and go there. No reinforcers in the opt-out zone, but no pressure either. The opt-out is what makes the rest of the system safe. It is the reason crying drops out. It is also a data point. A kid using opt-out a lot is a kid telling the team the step is too big. Shrink the step.

Sessions are short. Three or four blocks of trials, around thirty trials a day, three days a week.

Why one-size-fits-all interventions fail#

There is a real temptation to take this protocol and run it on every kid who eats poorly. Gover is careful to say no. The early cases were a narrow group on purpose:

The cases that we took early on were really specific in that way, which I think has been important in how the research has developed and how we talk about what procedures are more appropriate for certain clients than others. From the talk — Dr. Holly Gover

The kids who fit the protocol could talk, could follow instructions, ate a range of table-textured foods, and had no medical issue driving the refusal. For a kid with a G-tube, the answer is a hospital feeding team. For a two-and-a-half-year-old who cannot yet imitate, the answer is a year of imitation and play skills first, then come back to feeding. Gover sent that exact kid home with that exact plan. He was food selective, but it was not an emergency, so the family worked on prerequisites and circled back later.

The matching cuts the other way too. A kid like Luke, in Gover's third case, finished in a fraction of the time because his selectivity was really a control issue. The second he had the opt-out and the choice board, he tried the food, decided burritos were delicious, and was done. A kid like Allie needed micro-steps the team had to invent on the fly. She chewed one tiny noodle twenty times so it would just disappear in her mouth, because swallowing was a wall she could not cross any other way.

The protocol is the spine. The kid in front of you sets the steps.

What to expect over 3 to 4 months#

Three to four months is the honest timeline. That is what Liam took. That is what Allie took. Roughly three one-hour sessions a week, around thirty trials a session, six target foods, with the team allowed to drop a food if the kid clearly hates it after a fair shot.

What progress looks like inside that window:

• Weeks one to two: assessment, preference checks, ISCA, first easy steps. Kids are usually fine because they are choosing to look at, touch, or sniff foods, which is a low ask.
• Weeks three to six: most of the shaping ladder. This is where stuck steps show up. The team task-analyzes the stuck step into smaller ones and stays there until it moves.
• Weeks seven to ten: chewing and swallowing for most of the target foods. Some kids cruise. Some kids need creative micro-steps.
• Weeks eleven to sixteen: parent involvement at the table, generalization into a family meal, light home protocols, decision on which foods to keep and which to drop.

What is not in the window: a fully fixed eater for the rest of childhood. Generalization to home is still an open question in the research. Some kids carry the new foods home well. Some do it only with the clinician. Light parent training and a small home protocol help, but the honest answer is that the field is still building the long-term piece.

The win is more modest and more real: by month four, the family serves the kid the same dinner as everyone else, and they can go to grandma's again.

Frequently asked questions#

How long does food selectivity treatment take?

In Gover's cases, three to four months of roughly three one-hour sessions per week. That is the honest average for a food-selective kid (not total refusal) running the shaping plus differential-reinforcement protocol with an opt-out. Faster kids finish in weeks. Slower kids need more micro-steps and may run longer.

Should we work on one food at a time or multiple?

Multiple, usually three or four at once. Gover is direct on this. With one food, a stuck step is a guessing game. Is it a bad procedure or a true non-preference? With several foods, the team can see two climbing and one flat, drop the flat one, and keep moving. A single-food approach also wastes the kid's time because clear non-preferences only show up against other targets.

Can food selectivity go away on its own as kids get older?

For some kids, yes. Plenty of picky eaters outgrow it. For the food-selective kids in Gover's referral pattern, the ones whose food list is so narrow the family can't go anywhere, waiting is usually not the win. The family loses years of normal mealtimes, and the kid's preferred-food list often hardens with age. If the medical work is clear and the family is asking for help, four months of structured work tends to beat four years of waiting.

If the kid is medically cleared, eats some table foods, and the family is losing parts of life to mealtimes, the protocol fits. If the kid is on a G-tube, has a swallow issue, or cannot yet imitate, refer out, build prerequisites, come back.

Want to watch Dr. Holly Gover walk through her case files, share the graphs, and explain the messy moments that pushed her to add the opt-out room? The full CEU is about an hour and a quarter and earns one credit.