When I see a question in an Aphantasic questionnaire it makes me laugh internally because the poor Phantasics probably have to visually see a real horse or video) before they can draw it. What comes to my mind when I an asked such a question is not an image of a particular horse, in fact I don’t see any images at all, but what I get is a sort of database of related instantly-recallable facts such as: a horse is an animal, an equine animal, it has 4 legs and a tail, it was used as a ‘power source’ in earlier time, but nowadays is mostly used for sport or recreation, etc etc, some particular horses, especially race horses had names, which might be remembered etc etc. From the data I could improvise a generic picture of a horse, but it would no be a specific horse. So asking me that question would be a waste of time and would no doubt be misinterpreted by a Phantasic researcher.
Other responders in here have described the creation of a drawing, bit by bit, function by function; that’s exactly what I would do, even in drawing something simple like a house.