The fact that you said my experience training dogs is not at all relevant to the topic of why computer science education is failing tells me you don’t have enough behavioral science background to evaluate that. There, is that better?

You can expand many of the concepts of clicker training to teaching computer science, such as back chaining (teaching how to do the end part of the process first, then working back from there). I just didn’t feel a need to enumerate every single example that was possible.

Reading isn’t really a unique way of learning — it’s just a variation on evaluating the environment to discover things we can’t directly see. If I smell something burning, I know there’s a fire nearby. That’s something I’ve learned by the normal mechanisms.

You could argue that when dogs respond to cues, they’re doing the same sort of thing — there’s a meaningless thing the human does that has been arbitrarily assigned a value that translates for the dog into “if you do ____, you will get what you want.” Or, for old-school trainers, “if you do ____ I won’t make you uncomfortable/put you in pain.”

But let’s assume for the sake of argument that this is not “enough” of a proof that dogs can understand concepts and you don’t buy into the idea that reading is along a continuum that dogs and other animals fall on.

Match to sample is very similar to reading. Several animals can read simple symbols, including modifier symbols (left vs right, up vs down). My own dogs understand left vs right, clockwise vs. counter-clockwise, though I haven’t done much modifier work with them. What I have done under concept work is teaching my dog to do the thing I just did, which you could argue is akin to learning from a lecture. I couldn’t find any videos showing the end state, but I did find this one of being able to retrieve toys he could not see by name. That’s fairly abstract.