Predict, Then Press, Sometimes

Why I build these

Most of what my students need to understand in biology happens where no one can see it. A signal molecule lands on a cell, and a few seconds later the cell is doing something new. The textbook gives them a crowded diagram with eleven arrows and asks them to hold the whole sequence in their heads.

So I build tools that take the process apart one step at a time. Each one lives on this site, under my name, and each one gets used live in my classroom before I decide it is finished. I watch where students stall, where they skip ahead and where they get it wrong with confidence. Then I rebuild.

I build them because the tools I have do not fit how I teach. A Gizmo can take a long time to load, sends students down multiple paths, and does not sit directly inside Canvas or let me pause it and explain a concept to the room. I am reaching for something that is both static and dynamic, so I can use it in a way that is both teachable and interactive.

PhET simulations are my favorite. I worked with Carl Wieman on them and directed the National Academies study on games and simulations, Learning Science Through Computer Games and Simulations (2011). PhET tools are very teachable, but they do not cover the content I need. HHMI’s BioInteractive tools are wonderful interactives, but they depend on a lesson build that is not mine, and not built for the Virginia Biology SOL or for teaching undergraduate cell structure and function, a subject I wrote a book on, Cell Structure and Function: Mastering the Big Ideas (2017).

I build for students first, then for teachers, then for parents. I do not really rank them, but students come first.

This fall I reached the point where the tools are good enough that the next improvement has to come from the research on how people learn from pictures that move. The research confirmed some of my instincts, caught a few habits I had not noticed, and gave me a short list of changes for every build from here on.

How one gets made

Every animation starts as a still figure, and the still decides almost everything that follows.

1. I redraw the figure. Molecules are drawn at the level of their parts, never their atoms: a hexagon for a sugar, a circle for an amino acid or a phosphate, a capsule for a fatty-acid tail. Each class of molecule keeps its colour all year, so a student who learned the lipid colour in September still reads it in March.
2. I write a build brief with Claude. The brief names each step, the sentence that goes with it, the question students answer before they see it, and the rules the drawing has to keep.
3. Claude builds it. A coding session of Claude turns the still into a step-through page on this site, using the same geometry as the still, so the picture students see in the animation is the picture on their worksheet.
4. I test it on an iPad and a computer. My students work on iPads, and I teach from a projector. Both have to work before anything goes live.
5. It goes into Canvas and onto video. The original stays here under my copyright. Canvas gets a copy with a link back, and I record myself talking through it so students can watch at home and come back to it later.

My work with Claude is co-work in the full sense. I need to bounce ideas off someone, or something, and Claude is very smart. I think out loud with it, and it challenges me. I am training it, with memory that carries across tasks and projects, to push back on my ideas, and that pushing back is part of how these tools get made. The decisions about what students need to see, in what order, and what they get wrong stay mine, and they come from my classroom.

What the research says about learning from animations

An animation is not automatically better than a good drawing. Across 26 studies, animations beat static pictures by a modest margin, and the margin was larger when the motion showed the content itself rather than decorating it (Höffler and Leutner, 2007). A later and larger review found a smaller advantage, still on the side of animation (Berney and Bétrancourt, 2016). How an animation is built matters more than the fact that it moves, and eight findings shaped my next builds.

• Watching is not enough. In a study of physics demonstrations, students who only watched did barely better on explanations than students who saw nothing. Students who first wrote down a prediction gained about four times as much on explanations, for about two extra minutes of class (Crouch, Fagen, Callan and Mazur, 2004).
• A prediction works through surprise. Being confidently wrong makes the right answer stick, so the prediction has to be a real commitment, not a passing thought (Brod, 2021).
• Motion disappears. A picture stays on the page; an animation shows something and then it is gone, and the learner has to hold it in memory. Researchers call this the transient information effect (Ayres and Paas, 2007). Letting students control the pace, one segment at a time, lowers that load (Spanjers, van Gog and van Merriënboer, 2010).
• Pointing helps. Highlighting what matters at each moment, and quieting everything else, improves both recall and the ability to use the idea later (Richter, Scheiter and Eitel, 2016; Schneider, Beege, Nebel and Rey, 2018).
• Doing beats watching, and making beats doing. Clicking through a sequence is active. Explaining it in your own words or drawing it from memory is constructive, and explaining it to a classmate is interactive. Students learn more at each step up (Chi and Wylie, 2014).
• Molecules do not aim. Students already believe that molecules head toward their targets on purpose, and that belief grows more confident through college (Gauthier and colleagues, 2019). Animations that send molecules in straight lines to their destinations feed it. Molecules arrive by random collisions (Jantzen, McGill and Jenkinson, 2024).
• Name the wrong idea out loud. Physics students who watched a video that stated a common misconception and then took it apart learned more than students who watched a clear, correct explanation alone. The clear version felt easier and taught less (Muller, Sharma and Reimann, 2008).
• Short videos get watched. Students watch about six minutes of a video before attention falls off, and an instructor’s face at the right moments holds them better than slides alone (Guo, Kim and Rubin, 2014).

What predict, then press looks like in my room

Frankly, it is something I am still learning to do. Read, talk, write is not a simple process. I worked out a version of it with college students through student-authored modules, but with younger students it is hard to get them talking about claim, evidence and reasoning in a way that is constructive. That only happens when the reading is very good, often done as homework, and then we work with a tool and predict from good prior understanding.

What these tools show is so new to ninth graders that they do not yet have a concept of the cell. In many ways this is their first experience of these ideas, and you cannot predict much about something you have never met. So the first meeting is a video. Predict, then press comes on the second pass: when we revisit a process, when a student needs remediation, or as self-study. In the room it is usually a single, pointed question that I ask, and then students turn and talk about it.

What the first animation already does well

The first piece built this way is a G-protein-coupled receptor, the kind of receptor that answers adrenaline and many hormones. A signal molecule lands outside the cell and never enters, and six steps later calcium has switched on enzymes inside. Measured against the research, it already gets several things right.

• A prediction before every step. Each one is written as a real choice: where does the signal molecule stop, outside, inside the membrane or in the cytosol?
• Students set the pace. Nothing moves until they press Next, and Back replays any step.
• The record stays on the screen. Each finished step leaves a numbered marker and a one-line caption, so the sequence builds instead of vanishing.
• It shows what a still cannot. After the last step, the cell keeps making messengers while the one signal molecule sits unchanged outside. That is amplification, and it only shows up over time.
• One picture, many uses. The still, the animation, the projected Teaching view and my narrated video all share the same drawing, so students carry one image from the screen to the worksheet to the test.

What it does not do yet, and what changes

The research also showed me where the first version falls short. Each of these changes goes into the shared engine, so every animation after this one inherits it.

Two content fixes ride along. The last step’s box read “The cell does the job,” which names no job. At the college level it will name protein kinase C, with “the cell responds” beside it. And the switch between the high school and college versions will show real differences instead of the same picture twice.

The “so what” I have not solved

That last box exposed a problem I do not have an answer for yet. In ninth grade, the end of this pathway is not protein kinase C; it is simply that the cell responds. By eleventh grade, protein kinase C gets named, and the honest student reaction is: so what?

At the college level the same thing happens all the time. Students learn the information without the context around it, because the context keeps going and keeps extending, and what a lower-level course was preparing them for is often not clear to them until much later. When I find a spot like this one, there is a hum that is clear to me, and I think: I need to help that kid with that. What do I really need to show there? I do not know yet. I need to talk to students, and hear their own questions, before I rebuild this part.

Going back over the older tools

The receptor animation, built in October 2026, is the first model built and then checked against this research. I keep a log of where each improvement starts, so it is clear which tools on this site follow the newer pattern and which came before it.

The five animations that follow it are built on the improved engine from the start. The earlier classroom tools, the Membrane Patch, the Water Patch and the Energy Patch among them, stay up and stay in use. Next year I plan to go back through them one at a time and bring each up to the newer pattern. Their web addresses will not change, so links from Canvas and from past students keep working.

Using these in class

If you teach and want to use these tools, the research and my own classroom point to a simple routine.

1. Make the first meeting a video. When an idea is brand new, students need to see it once before they can predict anything about it.
2. Save the predictions for the second pass. On a revisit, in remediation or in self-study, have students commit before every step: out loud, on a whiteboard or by tapping a choice.
3. Ask one pointed question in the room. Project the Teaching view, stop at a single step, and have students turn and talk.
4. End with something they make. One sentence explaining the whole process, or the blank figure labelled from memory.
5. Talk about how molecules really move. Ask whether the messenger knew where it was going. It did not; it got there by bumping into things.

Not every teacher wants this, and I understand why. Many of us grew used to printed, cut-and-paste activities, the kind sold on Teachers Pay Teachers, and they have their place. But our students now work on iPads, and how we teach has to evolve with the technology that is part of their learning. I am still in the middle of that conversation.

Sources: Ayres & Paas (2007), Applied Cognitive Psychology 21, 695–700 · Berney & Bétrancourt (2016), Computers & Education 101, 150–167, open copy · Brod (2021), Psychonomic Bulletin & Review 28, 1839–1847, free text · Chi & Wylie (2014), Educational Psychologist 49, 219–243, pdf · Crouch, Fagen, Callan & Mazur (2004), American Journal of Physics 72, 835–838 · Gauthier and colleagues (2019), CBE—Life Sciences Education, doi · Guo, Kim & Rubin (2014), Proceedings of Learning at Scale, pdf · Höffler & Leutner (2007), Learning and Instruction 17, 722–738 · Jantzen, McGill & Jenkinson (2024), Frontiers in Bioinformatics, free text · Muller, Sharma & Reimann (2008), Science Education 92, 278–296 · National Research Council (2011), Learning Science Through Computer Games and Simulations, free text · Richter, Scheiter & Eitel (2016), Educational Research Review 17, 19–36 · Schneider, Beege, Nebel & Rey (2018), Educational Research Review 23, 1–24 · Spanjers, van Gog & van Merriënboer (2010), Educational Psychology Review 22, 411–423, doi.