If you read my blog regularly you know that I am constantly singing the praise of consistent use of positive reinforcement for all of those correct and appropriate behaviors that we see from our kids. Remember, we want to see more and more of those behaviors so adding that positive reinforcement is key to keep those behaviors coming. But what do we do when our student gets an incorrect answer? How do we respond then? Providing consistently and thought-out error correction to incorrect responses is almost just as important as giving loads of positive reinforcement for the correct responses.
Remove Reinforcement
So this should seem obvious but sometimes it’s not. When a child responds incorrectly, remove reinforcement. That means no praise no “great trying.” Yes they did do a great job trying but that’s confusing.
I’ve watched many amazing teachers provide such sweet and praise filled “no you didn’t get that right” that I wasn’t even sure if the student got the answer right or wrong. I am by no means advocating being mean or harsh. But there is a big difference between a nasty, “That’s wrong. You’re horrible” (said in low, evil dictator voice) and “You’re amazing but that is also maybe not totally correct” (said in princess sunshine voice). Let’s shoot for somewhere in the middle. My goal is indifferent. Keep in mind many of our students struggle with complex receptive language, so keep it simple. I like a straightforward, “no” and move on. It’s not being mean. It’s teaching them a new skill and we want them to demonstrate the new skill correctly.
Block or Interrupt Response (if possible)
This might not always be possible depending on the type of skill you are teaching but try to catch them before they emit the incorrect response. If you are are working on teaching a student to match colored shapes, as he starts putting the blue circle on top of the green triangle provide a partial physical prompt and maneuver that hand right over to the correct spot. Again, not always feasible but the goal is to not give him the opportunity to make a mistake (remember errorless learning?).
Help Them Get it Right
Here is where that prompting stuff comes back. I know, you thought you were done with prompting forever. But prompting is a teaching tool remember? And your student had an incorrect answer so we know he needs some help. After your student had an incorrect answer, you can say a neutral “no” or “that’s not right” if you want. Then represent the same teaching cue (or contrive the natural cue again). So if you asked your student to find blue and he pointed to green ask him again to point to blue. Then, immediately provide a prompt so he gets the answer correct. So immediately use a verbal, gestural, or physical prompt to ensure he gets the answer correct this time. Then provide that reinforcer right away.
Distract and Assess
So he got the answer correct on the second time with your help (prompt). What do you do know? Move on forever and forget it happened? Nope. Time to double check. I like to throw in a distractor trial here. Something you know your student can do that is unrelated to the current task. If you are working on naming personal info, have them clap their hands or point to a body part. Something quick and easy. The point is to distract only for a quick minute.
After you’ve done a distractor trial, time to represent that original trial again – that same exact original question that he got wrong. You want to see if your prompt was successful and helpful. So provide the originally question again. If he gets it wrong, start back over at the top with the prompting.
This whole shebang of questions and answers still counts as ONE TRIAL. And it all counts as incorrect. So even though you are providing a prompt and the student is getting it correct and then error correction and re-providing the prompt – it’s all part of that original question. You are just doing some additional teaching.
Save the Big Time Reinforcement for the Correct Responses
Remember step 1? Removing Reinforcement can be tricky when we add in all these other steps. You can provide some low magnitude reinforcement like praise in my examples for the promoting trial, distractor trial, and assessment trial but keep it minimal. Your goal should always be to save the big, major, fun, amazing reinforcement for the correct responses. We don’t want to teach our students that you give a wrong answer, the teacher helps you, and you still get a gummy bear. He won’t buy the cow if he gets the milk for free. Make sure that reinforcer is only coming for those great, appropriate, correct responses that we want to see more and more of!
- Using TAH Curriculum for Homeschooling from a Homeschooling Parent - September 10, 2022
- Using The Autism Helper Curriculum for Homeschool - August 8, 2022
- Literacy Subject Overview in The Autism Helper Curriculum - August 2, 2022
Sasha – thanks for a great review on error correction. Just what I needed!
Thanks for reading! 🙂
Wow this came at the perfect time! I have been struggling to find a way to correct errors for a student and never thought to go back and reassess before the reinforcement.
Totally needed this post!! I definitely was getting stuck after the error correction part and counting it as one whole trial!!! Always reading my mind over here! 🙂
Hey,
I wasn’t sure if I should post this here or email, but here it goes. I’m a BCBA and I love that you are explaining the 3-term contingency here, however I’m not sure it all comes through completely correct. The A-antecendent is the teachers question “Where is your nose?” and the B-behavior is the student response of pointing to ears. The C-consequence is the teachers corrective feedback “no”. That is the end of trial 1. The beginning of trial 2 starts when the teacher represents the question A-antecentdent “Where is your nose?” + the prompt (physical, model, gesture, picture, whatever it may be) then B-behavior is the students response, correctly pointing to his nose. THEN the C-consequence must be delivered… reinforcement. This is the key to the second trial to increase the likelihood that an unprompted trial in the future will result in a correct response. I think the difference here with what I’m saying is that you would mark the first trial as incorrect and the second one as prompted… AND you would always reinforce the prompted trial and THEN follow it up with an unprompted trial that you REALLY HIGHLY reinforce when that trial elicits a correct response. 🙂 Reinforcement is what increases the likelihood that a behavior will increase in the future, not prompts. Prompts are just helping the student get to that correct response so we have something to reinforce.
Thanks for reading! Glad this was helpful!
Haha! I always read your mind! 🙂
Hi Christine, Love hearing from fellow BCBAs! Thanks for sharing that information. I have gone over the concepts of reinforcement and prompting in depth earlier this month in this blog series. During the prompted trials, you would of course still provide reinforcement like you said. And as I mentioned, I would typically provide a lower magnitude reinforcer if possible. As far as the difference on taking data (whether to count this as one trial or multiple trials), this is the way I have always run DTT programming. Several other BCBAs I have worked with also follow this same procedure. I think that for teachers especially keeping data as simple and streamlined as possible (ie. only incorrect or correct responses) for certain programs helps maintain consistency and clarity. Of course, some programs may need more detailed data collection procedures. Thanks for commenting!
This is very clear instruction on error correction specially I need more help in teaching special children.
This was very helpful. I have been counting all steps of error correction as separate trials. Thanks for clarifying…
Hello Sasha,
First year para Elizabeth, these examples of different trials for different students are great. Sometimes a student does need prompting and/or wants to master the skill and can so by helping others with encouragement. There is one thing I learned this week a student in our class decided to use some slang language and I inserted a different word with what she was saying. We all then ignored it for the rest of the day when the slang word came out and by the end of the week she slowly used the different word I had given her and smiled as she knew her response was a praise well deserved she did her on her own and succeeded. Thanks, Beth
Great example. Thanks for sharing, Beth!
Do you have a flow chart for easy visual for staff?
I have some in this guide: https://www.teacherspayteachers.com/Product/Paraprofessional-Training-Manual-PART-2-5186999
Hi Sasha, I have a question regarding end of the error correction procedures. If we are implementing the error correction during token board, if the client gives the correct independent response end of the error correction procedure, should we give the client token for this response, or should we give another High- P and then give the token.
You could go either way! Just save the high-powered reinforcer (maybe token plus praise or multiple tokens) for the correct response.
Hi Sasha, I noticed a previous comment (from Christine on Jan 28 of 2017) that mentioned recording the initial trial and any subsequent error correction trial(s) as separate trials, rather than counting it all as one trial (as you replied on Feb 6 of 2017 that this is how you collect DTT data). Up to this point, I have always done it the way you mentioned, but have just now learned of the way Christine mentioned. Are there different names for these different ways of data collection? Are there reasons one way of DTT data collection might be preferred over another?
I think it totally depends on the teacher and student. I like to in general keep it simple and taking data on the subsequent trials adds extra steps. But you may want to know how many subsequent trials error correction takes. I’d say pick what works best for you and stay consistent! 🙂
Hi Sasha! So today was my first day doing DTT sessions with a client and I got stuck every time he got one wrong. This really helped! However I still have a question. So say we provide the SD, client gets it wrong, we say “no” and ask again providing immediate prompting. after error correction, if they get it wrong again, do we again provide that most assistive prompt or are we going to want to move down to a less assistive prompt, aka most to least? & if they get it wrong again after that is error correction done again? Thanks so much!
Great question – I wouldn’t jump to a more assistive prompt right away. I would basically start over and give the correct answer again and see if you could evoke the correct answer through modeling. Take data and follow the client’s lead.