First I’ll try to define the meaning of the term “reality” hopefully better than my previous [above] attempt, then I’ll try to deal with how that fits with what people mean when they refer to this or that as “real”:
A model, a hypothesis, etc. has predictions. We can imagine whatever models we want. We can compare models to experience. Sometimes they match, sometimes not.
If our models and experience always matched 1 to 1 (no matter what we made our models to be), there would never be surprises. Since there are surprises, we posit something as the cause of this. Something capable of supplying experience besides our models and expectations. We call that “reality”. Reality is defined as the thing that has that capability.
Yes, this concept that “there’s a thing called reality that can do such and such” is a human concept. But that concept does not (can not) itself (nor in combination with other models) produce all the experienced differences between a successful prediction and a failed prediction. Or else we’d always have a 1 to 1 match between models and experience.
Now, in everyday language, instead of talking about all of reality, we are usually talking about part of it. So a “real duck” is the part of reality that causes the experiences which will be surprising or familiar depending on our model. A “real decoy duck” causes experiences which will be surprising or familiar depending on our model. If your model is “real decoy duck” when there’s a biological duck, you’ll be surprised. If your model is “real duck” when there’s a “completely” unknown alien object there, you might be surprised. And so on.
I’m pretty sure this is what people mean. I hope I’ve tied enough of the loose ends.