Is there any limitation to what vertebrate you can clicker-train besides ability to hear?
It’s not even limited to vertebrates! Any animal that can learn from it’s surroundings can theoretically be trained using operant conditioning, as long as you can find a) a reinforcer that’s relevant to it, b) something it can perceive to use as a bridge, c) and enough long term memory for you do be able to do anything with it. People tend to use the term ‘clicker-training’ to refer to the physical act of actually using a clicker, but it’s really just applied operant conditioning and that doesn’t require a specific type of bridge. We already alternate bridges all the time based on the situation we’re using them in - marine mammal trainers use whistles because they’re hands-free and can be heard at a greater distance than a clicker or a verbal cue. Deaf dogs are a great example of animals that can’t perceive a traditional cue at all still being able to be trained - they can’t hear, so instead people use things like the flash of a small light or a vibrating collar to tell the dog when it’s successful.
But back to non-vertebrates! Octopus are probably the most well known example, but we’re not limited to super smart aquatic eldritch organisms! If you google ‘insect training’, for instance, you get some incredible examples of non-vertebrate training, such as this entire Wikipedia page about “olfactory detection bees.” Turns out we’ve been using honeybees as a model organism for studying classical conditioning for fifty years!
There’s also this paper where they used classical and operant conditioning to test short- and long-term memory in chambered nautilus. It was pretty simple: light flashes, they got food. So in that case, if the light is always accompanied by food, it could function as a bridge. Now, they found out that nautilus have a short-term memory span of about an hour and a long-term memory span of 6-24 hours, so while you can theoretically train a nautilus this way, you’d be hard pressed to get much done given the limited number of reinforcers you’d have (they’d be full) and the time span you’d have to work in.
And apparently someone has been training nematode worms, and I’ve just gotta throw out this quote from the abstract because I adore it so much: “Until 1990, no one investigated the possibility that C. elegans might show behavioral plasticity and be able to learn from experience. This has changed dramatically over the last 14 years! Now, instead of asking “what can a worm learn?” it might be better to ask “what cannot a worm learn?””
I went to go look up the a big news article recently about scientists training a spider to jump long distances on cue, assuming it would be appropriate for this response - but it turned out they did not actually train the spider using operant conditioning! As per the paper: “We did not use prey as a bait item as this would mean one jump per week based on the observed diet of the spider. Instead, the spider was manually transported between the take-off and landing platforms until it became familiar with the challenge. No form of stimulation (e.g. air blowing) was used to induce a jump. The spider did not fail any of the jumps; it was either a jump or no-jump situation. Hence, once a jump was performed it was deemed a successful trial.” By the way,the spider’s name was Kim.
(I wrote a whole bunch on how training basically is hacking into how animals learn for maximum efficiency, you can check that out here if you’re into technical stuff).