1. SD is presented to the learner (+ any other accompanying stimuli)
2. The learner errors on response or doesn’t respond within selected time frame (ex. 5 second latency)
3. SD is presented again and prompting is included from RBT (using Least to Most or Most to Least, depending on the client’s specific program description)
4. Transfer trial is performed (represent SD and fade some part of the prompt)
5. Remove stimuli (if applicable) and begin Expanded Trials (1-3 distractor trials)
   1. you can think of these as high probability trials (like what we use when we do behavior momentum. These should be 1-3 very easy tasks that is within the learner's repertoire)
   2. for example, if the learner has mastered body parts you may ask then to "touch nose", "touch head", "show me your ears", while simultaneously providing reinforcement (praise) after each correct response.
6. Represent SD (and any accompanying stimuli) and allow independence (i.e., do not prompt on this trial)
7. If learner is correct-reinforce and move on to another trial/program
8. If learner is incorrect-go through previous steps until correct or 3 sequences total occur.
   1. if you go through 3 ECP sequences and the learner is still incorrect, do not say anything or react, simply move on to another program.

Again, it is important to note that no matter how many ECP sequences you may run (1, 2, or 3) and no matter if they end up getting the answer right, the sequence(s) are only marked on CentralReach as one data point. This data point will always be a minus or incorrect, since the client errored originally.

Below is a breakdown of what the ECP sequence looks like. Note that this is only one sequence and, if continuing to error, ECP allows for a total of 3 sequences. Therefore, you would just repeat this sequence as many times as applicable to the particular situation.