Building a Stimulus Response Learning Module
Sep 04, 2017
We are in the midst of developing many projects for a complete curriculum for Physicians and Nurses in Modern TB Diagnosis, Treatment and Management.
There is one unit that is different from the rest, and we would like to be able to use a different learning model for that and wondered if anyone had ever programmed that model with any of the Articulate tools or could advise us how to do it.
Description of Learning Problem
There are almost 20 different drugs that are used to treat various kinds of TB today, Each has different side effects. Learners must know the side effects for each particular drug. There is no logic and no concepts; it is just rote learning.
In the late 1950s and early 1960s B F Skinner developed an interesting learning model for this kind of problem. He even invented a machine to present it with. It was a straight "Stimulus - Response Model". It had excellent results but only for this very limited learning situation. But that is what we face in this one instance.
The Skinnerian Method
The student was given a card of all the stimuli and the proper responses. Then student was presented with a mechanical machine that presented the stimuli at random (it was actually pseudo random) and the responses as multiple choices or true false questions.
The student would look up the correct responses on the chart and make a response on the machine. Student would be given immediate feedback. The machine would cycle through the set of stimuli N times (the most efficient Ns turned out to be 2 or 3. Very bright Harvard graduate students optimized at N being 1 or 2.
Then the chart was removed, and the cycle began again. However, this time if the student answered a stimulus correctly that item was dropped out of the cycle and the process repeated over and over until all stimuli were answered correctly. Then, the process began again and was repeated again and again until the student could go through the cycle with no errors. The range of times ranged from 1 to over 10 but 80 percent of the students took between 3 and 5 cycles to meet the criterion.
Many retests were done after different elapsed time and the memory held up pretty well.
What I would like to do
I would like to create a Quiz Bank or an equivalent of the stimuli and proper responses.
Then present it to the student with feedback but somehow keep track of which items the student got right and flag it so in the next cycle that item would not be displayed. When there are no more items left, the routine ends but can be restarted by the program (that part is easy).
Using Story line much of this can be done, but here is what I do not know how to do yet:
- Drawing the questions in a random order
- Flagging the items that are correct in each run of the Quiz draw and then using the flag to bypass that item thereafter.
- Using Quiz results to determine whether the student should recycle back through the process WITHOUT the Student seeing the results. In other words the process just goes on and on and the only feedback the student gets is on each item. I would like him/her to never even see the results slide.
Hopefully, someone has done this already and I could use that. If not, someone might have done one of the 3 above for this or other reasons.
If not, hopefully I can get some good suggestions of how I might go about programming any of these processes.
I will appreciate any help I could get.
6 Replies
Hi Lewis,
Wow, this is a very ambitious project! I've never created or seen anything quite like this, unfortunately so I won't be able to link you to a similar example or download. However, I will attempt to answer your questions below:
1. To draw questions randomly, simply check the "draw question randomly" checkbox in your question bank settings (see screenshot)
2. You would need to set up a combination of variables and triggers to get this to work. Here are some tutorials to help you get the hang of how variables work.
3. You can format your quiz results slide however you want. You can even take off the score so the learner never sees it. I recommend disguising it as a regular slide.
I hope that helps!
I think this is simple to build, Alison has given you the basis.
Add a variable for each question Q1-20 set all to false.
On first Question (Randomisation will not affect this) on base layer
Set a trigger to change state of radio button or checkbox for correct answer/s to selected if Q1 equal to True
Set trigger to submit interaction if Q1 equal to True
On Correct layer set triggers in this order
Jump to next slide if Q1 equal to True
Set Q1 to True
Continue this for all other questions.
Of course, now I get it. Thanks so much Phil.
Thank you much Allison
Hi Lewis,
It looks like others have already addressed the implementation details, so let me take a moment to point out that some important research has been done since the days of Skinner, and in light of that more recent research, there is likely a more efficient way to achieve the learning results you seek.
A key research paper in this area is Karpicke, J. D., & Roediger, H. L. (2008). The critical importance of retrieval for learning. Science, 319, 966-968. I highly recommend this paper to you because it almost exactly addresses the situation you are designing for!
In a nutshell, what Karpicke and Roediger did was teach people vocabulary words in Swahili. Like your situation, this task amounted to pure, rote memorization. The researchers wanted to know what the best way to do this would be, so they set up 4 different groups of students and tried a different strategy with each group.
The general setup for all 4 groups was that learners were provided with the full set of vocabulary words in Swahili and English for a certain amount of time to study them. Then they were quizzed on the full set of words.
After the each quiz, they went back to study words again in another "study" cycle, and then were quizzed again in another "quiz cycle". They continued in this way for a set number of study/quiz cycles, while the researchers carefully tracked their results on each quiz.
Where things differed between the four groups was in how each subsequent study and test cycles after the first adjusted the list of remaining words for the learners to study and/or be quizzed on:
Group 1 never adjusted. At each subsequent study and test cycles they always studied the full set of words and were tested on the full set, regardless of how many words they may have already correctly identified in previous test cycles.
Group 2 dropped words from subsequent test cycles after they got those words correct on a test cycle. But they continued to study the full set of words during subsequent study cycles. This matches your design pretty closely because you plan to drop items after learners successfully "get" them on a quiz cycle.
Group 3 dropped words from subsequent study cycles after they got those words right on a test cycle. However, they continued to be tested on these words at each test cycle, even though they were no longer given those words and their matches to study during the study cycles.
Group 4 dropped words from both the study and test cycles after they got those words correct on a test cycle. It would be great if this method turned out to be the most effective, wouldn't it? Because this method requires the least effort from the learner over the span of all the study and test cycles. Unfortunately, this didn't prove to be the best strategy.
When the results were all tallied up and compared, the researchers found that a very significant improvement came from dropping words only from the study cycles but not from the test cycles! For example, Group 3 had about a 150% better retention for these words than Group 2, even though they each dropped a similar number of words at each cycle.
The key finding, therefore, is that when you repeatedly ask learners to retrieve answers (as, for example, on a quiz) it is actually much more effective than repeatedly telling them things.
To me, it suggests that a better strategy for your "Skinnerian" quiz is to not drop any questions from cycle to cycle. Instead, end the looping when the learner gets all questions correct on a the same iteration of the quiz. Your feedback can remind learners of the correct answers to questions they miss (this simulates the "study" cycle in Karpicke & Roediger's research), but continue to ask all questions on each cycle, mimicking Group 3, which had the better retention results.
Whew. Hope that wasn't too long-winded and technical. You can see the charts and more detail in the paper itself if you follow the link.
Cheers!
-Ray
This is working well. I am now trying to generalize it so I can use it for different content at different times.
I have one question about how the questions slides in a slide draw behave:
1. Assume I have a trigger on a question slide X that activates when the Timeline starts and under certain conditions it Jumps to the Next Slide. Then I put that slide in a Question Bank. Then I draw from the question bank so that the questions are randomly selected. Now how does question slide X with the trigger behave if the if conditions are met? I assume that when slide X was drawn, the Timeline starts and the slide X does not even appear. I further assume that “Next” in the case of a random draw does not men the slide that is immediately after Slide X in the Question Bank, but it means whatever question slide is randomly drawn “next”, Is that correct?
2. Can I put triggers on the Quiz Results Slide like any other slide. For example, can I add tigers when the timeline starts of a quiz results slide?
3. If so when does the Quiz results slide timeline actually start? After the test is complete? When the test starts ie the first question slide appears? Something else?
1. And then I have one general question. I would like to embed slides with notes to other team members developing the course, Would this method work: We design a Notes Slide. On that slide we add a trigger to goto the Next Slide when the timeline starts. In that way, the Notes Slide would always be visible in the Story View or Slide View, when published, the learner will never see it. Would that mess up anything? I was going to request a new feature, but when I do that I always try and figure out how I would program the feature. When I started to think about that, this idea stuck me and no new features are needed. We can use the heading of the slide to denote the members of the team who need pay attention, and responses can by put on the same slide if there is room or just add another of my “Note” slides. Will that mess up anything?
Lewis D. Eigen
Lewis D. Eigen
lewiseigen@gmail.com
(301) 983-0228
This discussion is closed. You can start a new discussion or contact Articulate Support.